Volume 46, Issue 4

Minority areas are generally backward areas with lagging economic development level, low level of industrial structure and obvious dual structure characteristics, but these areas often contain rich natural resources and unique national culture, so it is of great significance to explore the path of upgrading industrial structure in minority areas. The article firstly introduces the input-output analysis method and constructs a multi-objective evolution model of industrial structure adjustment in ethnic minority regions from the perspectives of low carbon, employment and economy, and solves the model by using the multi-objective genetic algorithm with improved genetic operation (I-NSGA-II). Using the method proposed in this paper to carry out comprehensive and detailed quantitative calculations and effect analysis on the industrial structure of ethnic minority areas, the influence coefficient of computers, audio-visual equipment, communication equipment, and cultural and office supplies machinery is larger as an industry with higher scientific and technological content, and each increase in one unit of production has a larger pulling effect on all other sectors, and it is considered that low-carbon development, economic growth and promotion of employment are equally important. Minority areas can upgrade their industrial structure in the aspects of strengthening basic agriculture, upgrading traditional industries and accelerating the development of service industry.

In recent years, virtual power plants have rapidly emerged as a flexible and efficient form of intelligent energy management. This paper focuses on the optimal scheduling of virtual power plant resources, after explaining the virtual power plant resources and their characteristics, it proposes a new energy consumption model of virtual power plant with hybrid energy storage system, establishes a joint optimal scheduling model of thermoelectricity and electricity based on the hybrid storage system, and selects the improved particle swarm algorithm for the optimization and solving. Collecting the relevant spatio-temporal data of the system power generation for example analysis, the improved particle swarm algorithm has better convergence, and the optimal allocation of energy storage is realized when the electric energy storage and thermal energy storage are 9MWh and 35MWh, respectively, and then the annual profit of the virtual power plant can be increased by 605,610,000 yuan. In addition, after the aggregated scheduling of the model, the electric load demand response and the electric energy storage work together to maintain the balance of the electric output of the virtual power plant. The proposed optimization control strategy can realize the complementary advantages of distributed resources, improve the flexible regulation ability of the virtual power plant, alleviate the pressure of power supply preservation, and ensure the safe and smooth operation of the power grid.

With the continuous development of Internet e-commerce, reasonable inventory arrangement for different warehouses or retailers and dynamic pricing of goods have gradually become key factors affecting the profitability of each company. The study proposes a reinforcement learning-based approach for dynamic pricing and inventory control in e-commerce retailing. The problem is modeled and converted into a Markov decision process by incorporating e-commerce retailing characteristics, and a joint inventory control and dynamic pricing algorithm for e-commerce retailing is designed based on the Deep Deterministic Policy Gradient (DDPG) method. The results of numerical experiments show that the joint inventory control and dynamic pricing strategies based on deep reinforcement learning have the best performance in terms of gains, with gains of 0.197 and 0.035, respectively. The numerical experiments validate the performance effectiveness of the algorithms proposed in this paper, and the DDPG algorithm significantly outperforms the traditional methods. This research can improve enterprise revenue and effectively promote the landing of reinforcement learning in the field of revenue management, which has practical application value.

In the era of artificial intelligence, more and more enterprises cooperate with universities to form a collaborative innovation alliance for the integration of industry and education. In this paper, using the evolutionary game theory, we establish the payment matrix of the collaborative innovation game led by the government, led by universities, and participated by enterprises, and carry out the analysis of replicated dynamic equations and stability of equilibrium points to construct the evolutionary game model of the integration of industry and education with the participation of multiple subjects. Through numerical example analysis, the role of different factors on the evolutionary equilibrium and the behavioral strategies of each game subject is explored. The results show that the evolution of industry-industry integration increases with the increase of the willingness of enterprises and universities to participate, and when the willingness to participate is greater than 0.5, the evolution of industryindustry integration tends to be 1. In addition, the enthusiasm of the university and enterprises to participate in the co-construction increases with the increase of the cooperation benefits created by the input resources, and the scientific and reasonable distribution coefficients of the cooperation benefits can help to increase the willingness to cooperate of the main subjects. In the R&D stage and the production and commercialization stage, when the penalty is greater than 6.25~6.5 and 4~6 respectively, it is easy to lead to the deep integration of the university and enterprise sides. In order to ensure the construction of the integration of industry and education, it is recommended to establish a complete information channel, design a good incentive-punishment mechanism, as well as improve the policies and regulations of the integration of industry and education, so as to promote the optimization and development of the organizational form of the integration of industry and education.

In this paper, we utilize the features of YOLOv3 combined with multi-size prediction of feature pyramid network, which fuses the feature information of multiple feature maps of different sizes through up-sampling from top to bottom to improve the resolution of feature maps of different sizes. The converged target classification loss function and target regression loss function are used to train the YOLOv3 algorithm in combination with the homemade defect mapping dataset to meet the design requirements of the surface defect detection algorithm. Finetuning is applied as a pre-training model to optimize the loss function of the improved YOLOv3 algorithm. Analyze the experimental performance of the improved YOLOv3 algorithm with different Image Size parameters, different numbers of Anchors, and different sizes of defect areas. Compare the index performance of the improved YOLOv3 algorithm proposed in this paper with the YOLO series algorithms. The detection precision and recall of the improved YOLOv3 algorithm for different defective regions are 0.9324 and 0.8589, respectively, and the improved algorithm meets the requirements of defect detection algorithm design.

This paper proposes a MIDI automatic composition framework that integrates multi-track clustering algorithm and WaveNet model. The main melody is extracted by multi-track clustering algorithm, and the iterative prediction mechanism of pitch sequence is constructed based on WaveNet model. The model designed in this paper is used to generate Yunnan Ethnic Minority Music and explore its specific application effects. Deconstruct the mapping relationship between pitch and physical parameters, and quantify the short-time energy and spectral characteristics. The skyline algorithm is selected as a control to test the improvement effect of multi-track clustering algorithm in training efficiency and accuracy. Combined with user ratings and melody line visualization, the performance level of music generation of this paper’s model is analyzed. The results show that the music generated by this paper’s model improves about 24%~58% on five subjective evaluation indexes respectively compared to skyline model, and about 12%~22% on five subjective evaluation indexes respectively compared to KD3 model. This study provides a solution with both technical suitability and cultural fidelity for the digital inheritance of ethnic music, which is of great significance for the inheritance and development of Yunnan Ethnic Minority Music.

In this paper, distributed power supply and electric vehicle charging load are studied in depth, and the distributed power supply output model and electric vehicle charging load prediction model are constructed, which are integrated into the power system balance calculation model. In the framework of traditional power balance calculation, flexible calculation method is introduced for improvement, a new power system model is constructed, and a new power system power balance method based on flexible calculation is proposed. The application of the flexible calculation method in power systems is understood through the flexible balance calculation and example study of typical regional power grids.The variation interval of the load flexibility parameter in region A is [20350, 25000] MW, and the minimum value on the left side of the flexibility inequality (24524.6) is larger than the maximum value on the right side (23875.6), which realizes the flexible balance.The annual load of the power system of province Z is dominated by the summer and winter as the The annual load of the power system in Province Z is peaked in summer and winter, and troughs in spring and fall. The daily load basically shows the pattern of “double peaks and double valleys”. The UHV DC corridor in Province Z delivers power according to the 100/75 curve, with 100% capacity in January, July, August and December, and 65% capacity in other months. The UHV AC channel delivers power on a 100/70/55 curve throughout the year. Photovoltaic output is large in spring and small in winter, and the maximum output in a day is around 13:00. Wind power output is large in winter and small in summer, with large output at night and small output during the day.

Grid planning plays an important role in the long-term development of electric power enterprises, is an important part of the national economy and social development, and plays a fundamental role in supporting the development of the whole country. In this paper, the day-ahead-intraday scheduling model is constructed with the objective of integrated energy system operation cost and the coordinated optimization of integrated flexible loads, cogeneration units and wind power from the perspective of multi-timescale scheduling. The genetic algorithm is improved based on the concept of adaptive, and the improved adaptive genetic algorithm is used to solve the system. The optimization model is verified to be effective for both load adjustment and small load fluctuation through examples, while the improved adaptive genetic algorithm can be effectively applied to the integrated energy system optimization and operation problem, and the proposed multi-timescale optimization and operation scheme has the advantages of reducing the operation cost and improving the consumption of renewable energy.

Digital technology is accelerating the deep transformation of agriculture, and improving production efficiency has become the key to the high-quality development of agriculture. Crop growth prediction, as an important link in precision agriculture, can effectively guide agricultural production decisions and improve yield and quality through the integration of intelligent algorithms. In this paper, we constructed a BNM-PNN model combining Bayesian network and process neural network, and established a crop growth and development prediction model by collecting and analyzing the hyperspectral data, SPAD data and leaf area index of Italian shoot-tolerant lettuce under different nitrogen fertilizer application conditions. The study adopts the improved neural network initialization method and learning algorithm to solve the output locking problem of the Sigmoid-type excitation function and improve the convergence speed of the model. The results show that the BNM-PNN model has superior performance in crop growth prediction, with the coefficient of determination and regression estimation errors reaching 0.927 and 1.436 g/pl, respectively, with an average relative error of 3.24%. In the model performance evaluation, the accuracy of the BNM-PNN model reaches 96.15%, and the DICE score and sensitivity are 92.15% and 91.78%, which are significantly better than the traditional deep learning models such as GoogleNet and ResNet-50. The study shows that the fusion of Bayesian network and process neural network can effectively extract the key features of crop growth and provide technical support for accurate management of agricultural production.

Global urbanization process leads to ecological habitat fragmentation and biodiversity reduction. This study investigates the mechanism of multi-functional ecological corridor construction in enhancing urban ecological resilience, taking the ecological restoration project of Chengdu Ring Road Ecological Zone as an example. Using the theory of landscape ecology, the study constructed an ecological resilience assessment model from the three dimensions of ecological source identification, resistance surface construction and corridor identification, combined with the “patch-corridor-substrate” model, to analyze the role of ecological corridors in enhancing the ecological resilience of the city. The results show that the critical area for restoration of Chengdu Ring Road Ecological Zone is 277.77km², of which Jinjiang District accounts for 85.20%; the total length of the critical area for corridor restoration is 416.0717km, and the restoration needs of Jinjiang, Wuhou, and Jinniu Districts account for 81.53% of the total length of the critical area for corridor restoration; The Ecological Priority Development (ELP) scenario, with the shortest length of ecological corridors (9.2498km) and the highest network transmissibility, showed the best ecological resilience. The study confirms that multifunctional ecological corridors significantly enhance urban ecological resilience by improving habitat connectivity, enhancing network structural stability, and improving damage resistance. The ecological network with ecological priority development shows higher connectivity robustness and vulnerability robustness in the face of both random and deliberate attacks. The results of the study provide a scientific basis for the construction of urban ecological security patterns and the enhancement of resilience.

The high-speed railroad on the Qinghai-Tibetan Plateau is located in the highest altitude region in the world, and the permafrost roadbed has been subjected to freeze-thaw cycles for a long time, which leads to disasters such as freezing and thawing of the roadbed, and threatens the safety of railroad operation. In this study, a risk prediction model of freezing and thawing disaster on the roadbed of the Qinghai-Tibet Plateau High-speed Railway is constructed to provide technical support for ensuring the safety of railroad operation in high-altitude areas. The study firstly clarifies the definition and classification of freeze-thaw disaster, which is divided into five types based on the freeze-thaw mechanism: cold weathering-gravity freeze-thaw disaster, freeze-thaw creep-gravity freeze-thaw disaster, freeze-thaw frost cracking disaster, thawing and sinking freeze-thaw disaster, and freezing and sliding disaster. By combining wavelet decomposition, ARIMA and BP neural network methods, a high-precision prediction model was established to assess the temperature change of roadbed and the risk of freeze-thaw hazard. The study collected 685 sets of data for model training and testing, and the results showed that the correlation coefficients R² of the training and testing data reached 0.98 and 0.97, respectively, and the percentage of error less than 1% accounted for 90% of the total data. The results applied to the Totuohe and Nagqu station areas on the Tibetan Plateau show that the predicted value of the roadbed surface thaw index in 2020 in the Totuohe area is 1715°C·d, the freezing index is -2024°C·d, and the thaw ratio is 0.8473, which is an increase of more than 7% compared with that in 2000; the predicted value of the thaw ratio in the Nagqu area reaches 1.7768, up more than 10% compared with 2000, indicating that the frozen soil of the roadbed in this region is developing toward thawing, and corresponding protective measures need to be taken to cope with the potential risk of freeze-thaw disaster.

With the advancement of digital transformation, online courses have a profound impact on higher education. This study explores the impact of online course construction in higher education on the transformation of students’ learning styles, aims to analyze the current situation of online course construction and its value, proposes a course recommendation model based on knowledge graph, and explores its role in the transformation of students’ learning styles. The study combed the development history and construction content of online courses through the literature analysis method, constructed a knowledge graph graph convolutional network recommendation model (KGCN-CNSH) integrating common neighbors and structural holes, and conducted learning situation analysis based on 149,561 valid questionnaire data from 32 colleges and universities in 15 provinces. The results show that the KGCN-CNSH model improves the two metrics of Recall and NDCG by 0.83% and 0.39%, respectively, compared to the previous best-performing SGL algorithm on the Last-FM dataset; 75% of college students participated in at least one online course and took an average of 4.107; and the recommendation of online courses is the main factor influencing students’ gains in online learning one of them (p<0.001). The study conclusions show that online courses provide students with greater autonomy of choice while breaking through time and space constraints; the course recommendation model based on knowledge graph can effectively improve the level of learning personalization; and students' maintenance of concentration and persistence in the process of selfselecting courses is a key factor influencing the effectiveness of online learning.

With the development of educational technology, there are problems such as feedback lag, subjective evaluation, and training fragmentation in traditional piano art instruction. This paper explores the influence of realtime feedback generation algorithm on the improvement of playing skills in piano art instruction. The study designs and implements a deep learning-based piano hand fingering recognition system, which uses the YOLOv3 target detection algorithm to identify playing errors, combines with the HRNet network for gesture estimation, and compares and scores the player’s playing with the standard data through the DTW algorithm. Evaluation on the FHAD dataset shows that the average joint error of this system is 16.18 mm, which is better than the comparison methods NTIS of 16.88 mm, Crazyhand of 22.03 mm and BT of 27.24 mm.When the error threshold is above 26 mm, the estimation effect of this system exceeds that of all comparison methods, which indicates that it has an excellent performance in the piano hand fingering feature extraction and fusion with excellent performance. Experiments on two piano art classes (100 students in total) in the College of Music and Arts showed that the average performance of the students in the experimental class that utilized the system for assisted teaching increased by 9.00 points (from 65.45 to 74.45), while the control class that did not utilize the system only increased by 0.79 points, which is a significant difference (P=0.041<0.05). In addition, teaching based on the piano hand fingering recognition system significantly increased students' interest in learning, with the experimental class increasing its interest score from 38.54 to 48.49. The study confirms that the deep learning-based piano hand shape fingering recognition system can effectively improve students' piano playing skills, stimulate learning interest, and provide a new teaching aid for piano art instruction.

Power industry software, as a core tool for modern power equipment control and management, is facing increasingly severe cybersecurity threats. Distributed ledger technology provides new ideas for power software security detection due to its decentralization, transparency and tamper-proof characteristics. This paper discusses the application of distributed ledger technology in the security detection of software development in the electric power industry, and proposes a trusted traceability and quality access control reinforcement method based on distributed ledger. The research designs the traceability data model and smart contract system to realize the trusted collection, storage and verification of security data; at the same time, it proposes the sensitive data aggregation method based on homomorphic encryption and the tamper-proof technology of RSA asymmetric encryption, and constructs the data communication structure of Overlay structure, which guarantees the complete transmission of electric power software security detection data and traceability tracking. The experimental results show that compared with SHA256 algorithm and DyRH model, the average value of the error localization time of this method is reduced to 9.23ms, which is 8.6ms and 4.1ms less than the control group, respectively; the accuracy rate of the error localization reaches 98.33%, which is improved by 4.77% and 1.79%; and in the test of the anti-attack performance, the average number of tampered data is only 189, which is respectively reduced by 184 and 156. The study proves that distributed ledger technology can effectively enhance data credibility, strengthen traceability, and enhance the strength of system quality access control in software development security detection in the power industry, which provides a new technical path and solution for the information security of the power system.

With the growth of power demand, software development in the power industry is becoming increasingly important, but the network security problem is severe. Based on the basic principles of quantum mechanics and quantum key distribution protocol, this paper proposes a quantum key adaptive immune distribution strategy based on SDN, which is comprehensively tested and simulated to verify in the laboratory and current network environment. The experimental results show that under the condition of 15dB line loss, the key formation rate of QKD device reaches 107.014kbps, which is much higher than the standard requirement of 1kbps; in the test of overhead fiber, the cumulative loss of the quantum optical channel after adding the dispersion compensation module is only 12.02dB, which is 23.3% lower than that of the synchronous optical channel of 15.67dB. Simulation analysis shows that under the condition of service intensity of 600k and k=4, the key resource utilization of SDN-based quantum key adaptive immune distribution strategy reaches 0.536, which is about 9% higher than the classical key distribution strategy. The study shows that quantum encryption technology can effectively overcome the limitations of traditional encryption methods, provide unconditional security for software development in the power industry, with strong anti-interference ability, flexible key management and other characteristics, which can significantly improve the network security protection ability of the power system, and provide a new way of thinking for the construction of network security barriers for the security detection tools of software development in the power industry.

With the accelerated digital transformation of power systems, traditional protection measures can no longer cope with complex attack methods. In this paper, for the problem of data privacy protection and security defense in software development in the power industry, a privacy protection and collaborative defense system based on artificial intelligence federated learning framework is proposed. The study adopts differential privacy technology to protect client data privacy, designs a differential privacy federation learning method based on knowledge distillation, and constructs a collaborative DNS defense system based on blockchain technology. The experimental results show that the proposed method achieves 86.7% and 95.9% security detection accuracy on MNIST and Fashion Mnist datasets, respectively, which is 4.6 and 4.5 percentage points higher compared to the FedMatch method; in terms of the accuracy rate of different types of samples, the accuracy rate of the attack event, natural event, and no-event type reaches 82%, 81%, and 95%, an improvement of 5, 8 and 5 percentage points over the FedMatch method, respectively; and significantly outperforms the traditional differential privacy mechanism in terms of model convergence speed. The study provides a new idea for data security and efficient circulation in software development in the power industry, which can effectively deal with cyber security threats and guarantee the stable operation of the power system.

Open source software has become an important part of enterprise information systems due to its low cost, openness and transparency, and ease of customization. However, the open source software supply chain faces complex security risks, including management challenges brought about by multi-developer collaboration and difficulties in controlling third-party dependencies, which may lead to data leakage, system paralysis, and business interruption, and bring huge losses to enterprises. This paper proposes an artificial intelligence-driven open source software supply chain security risk identification and protection technology system. The study adopts AHP-entropy combination assignment model to assign weights to supply chain security evaluation indexes, and constructs a security risk identification model based on PSO-SVM, and finally designs a supply chain security protection system based on trusted computing. The results show that the weight of open source code component management is 0.478, which is the most important first-level evaluation index, followed by open source code quality management with a weight of 0.422; among the second-level indexes, open source code submission frequency, self-developed code size and percentage and component vulnerability severity have the highest weights. The PSO algorithm obtains the optimal parameters after 136 iterations, and the constructed risk assessment model has a test set of The assessment accuracy rate reaches 90%, only one sample is misclassified, and the squared correlation coefficient of the regression analysis is 0.96432. The conclusion of the study shows that the combined empowerment method reduces the influence of subjective or objective bias of single empowerment, the PSO-SVM model can accurately identify supply chain security risks, and the end-to-end protection system based on trustworthy computing can realize the trustworthy monitoring of the whole process of business communication, which provides enterprises with a comprehensive and accurate open source software supply chain security management solution.

The progress of technology promotes the energy storage power station to play an increasingly important role in the energy system. In this paper, a multi-objective cooperative control strategy based on improved differential evolutionary algorithm is proposed for the optimization of power conversion system (PCS) equipment of lithium iron phosphate battery energy storage power station. An operation model considering the dynamic energy efficiency characteristics of the battery is constructed, and the optimal operation strategy is converted into an objective function solving problem. The differential evolutionary algorithm is introduced and combined with adaptive parameter adjustment and hybrid mutation strategy to optimize the power allocation and charging/discharging scheduling of the PCS equipment. It is shown that the improved differential evolutionary algorithm can effectively regulate the PCS equipment under the influence of two power steps: 0.9 MW output and 15 MW of discharge and 15 MW of charge, and the algorithm can still stably output effective optimization strategies under the two extreme conditions of SOC close to the extremes of 0.75 and 0.85. Comparison of the four objective evaluation indexes shows that the improved differential evolutionary algorithm has better performance than the other algorithms.

With the deepening of electricity market reform, the diversity of users’ electricity demand and the complexity of electricity retail packages have put forward higher requirements for package design and recommendation technology. In this paper, we propose a three-stage collaborative approach that integrates the modeling of users’ electricity consumption behavior, the global optimization of differential evolution (DE) algorithm and the intelligent recommendation of attention factor decomposition machine (AFM), aiming at achieving the dynamic design and accurate recommendation of electricity retail packages. An extensible package product family GBOM is constructed based on the quintuple information expression model and modular design rules, and the differential evolution (DE) algorithm is used to efficiently search for the optimal bidding strategy in the highdimensional solution space, combined with the constraint rules to adaptively deal with the complex coupling relationship between modules. An intelligent recommendation algorithm based on AFM is further proposed to enhance the model’s sensitivity to the key features of users’ electricity behavior by introducing an attention mechanism to dynamically assign feature cross weights. The experimental results show that the differentiated pricing strategy has a significant effect in peak and valley time regulation, and the peak time price is 45.4% higher than the fixed price (e.g., the peak time price of user 1 is 453.28 yuan/MWh), and the valley time price is reduced by 47.3% (the valley time price of user 1 is 164.23 yuan/MWh).The AFM recommendation algorithm combines the user’s load characteristics with the monthly electricity consumption grading, and the recommendation accuracy rate is reaches 94.09% (when the number of historical purchases Te=7), which is a significant improvement over traditional methods (e.g., 89.41% accuracy rate of the mean weight method). Through comparative analysis, the DE-AFM method performs optimally in terms of balancing error and practicality (RMSE=0.0144, accuracy rate 94.09%), which verifies its stability and accuracy in complex scenarios.

Language modeling provides a resource carrier for students’ English learning. In this paper, ZO-VRAGDA algorithm is designed to reduce the complexity of multi-task solving for English language models. By calculating the complexity of the model processing task, the intelligent body is guided to decompose the task into multiple subtasks. The efficiency and accuracy of the model in completing the task are optimized by invoking appropriate tools and reminding the error-prone points. The English language model is introduced in the language classroom to recommend personalized learning resources for students and improve teaching quality. The study shows that with different numbers of neurons and iterations, the training time of the model based on computational complexity analysis in this paper is 5.54s-7.05s, 698.53s-1213.94s and 115s and 2722s in the 2 datasets, respectively, which is better than the comparison model. In different complex task processing, the confusion degree is reduced to 41 with only 99.22s, 104.21s, 97.91s. The similarity degree is improved to 27 with only 113.53s, 60.77s, 93.31s.

The reasonable configuration of plant community structure in landscape garden is the core factor to maintain ecological balance and landscape sustainable development. Starting from the spatial structure of plant communities, this paper proposes an evaluation index system for the spatial structure of plant communities consisting of four different dimensions, namely, horizontal structure, vertical structure, tree species composition structure and seasonal structure. The AHP-PCA entropy weight combination model was introduced as a method to assign weights to the indicator system and a test method for the indicator structure. A total of 120 landscape garden sample plots in three districts and counties of K city were selected as the research objects, and the questionnaire on the structure of landscape garden plant communities was designed by combining the natural conditions of K city with the evaluation index system of the spatial structure of plant communities. Based on the questionnaire data and AHP-PCA entropy weight combination modeling method, the evaluation index system of landscape garden plant community structure was determined. Using this index system to evaluate the plant community structure of 120 landscape garden sample plots, the highest evaluation values are between evaluation levels 1-3, indicating that the overall configuration of plant community structure in landscape gardens in K city is more reasonable.

Under the background of rapid development of digital economy, cross-border e-commerce has become an important engine of global trade, and the efficiency of its logistics network directly affects customer satisfaction and enterprise cost control. Aiming at the multi-objective optimization of cross-border e-commerce logistics network, this study innovatively introduces the quantum annealing algorithm, which breaks through the local optimal limitation of the traditional algorithm through the quantum tunneling effect and the superposition state characteristic, and constructs a multi-objective optimization model focusing on the logistics cost, time value and customer satisfaction. The study takes the overseas warehouse mode maritime logistics network as the object, and quantifies the multiple cost items such as storage, transportation, tax and fee through the mechanism of task decomposition and resource synergy. In the simulation example, the network robustness is verified by destruction resistance simulation. Under random attack, the network connectivity drops to 49.29% (node attack) and 56.55% (edge attack) when the node deletion ratio reaches 50%. In the deliberate attack, the improved maximum node mediator attack maintains a connectivity rate of 43.10% at a deletion ratio of 20%, which is better than the 20.79% of the traditional strategy and the 11.04% of the node-degree attack, indicating that the optimized network possesses a strong anti-interference capability. The quantum annealing algorithm has an optimal solution ratio of 79.23% in large-scale problems, which is significantly higher than the 56.19% of the genetic algorithm and the 50.18% of the heuristic algorithm, and the average number of solutions is 27.3, which is also ahead of similar algorithms. Although the CPU running time is slightly higher than that of the genetic algorithm, its global search ability and the quality of the solution have a significant advantage. This study provides support for the intelligent optimization of cross-border e-commerce logistics network and helps enterprises to reduce cost and increase efficiency and sustainable development.

Digital protection and display of ancient buildings is an important technical means of cultural heritage transmission, and “Gongshu Hall” as a typical wooden ancient building, its wall materials are complex and diverse, and the precise identification and protection of surface damage is in urgent need. This study proposes a set of intelligent media fusion technology framework based on image recognition and computer vision algorithms, and realizes high-precision material recognition and damage detection by optimizing the network structure, loss function and algorithm fusion strategy. Firstly, the EfficientNet v2 network is improved, and the collaborative attention mechanism (CA) is introduced to replace the original SE module to enhance the spatial location perception of the feature map. To solve the problem of insufficient bounding box regression accuracy of YOLOv7 in crack detection, the sample gradient contribution is balanced by the normalization factor and monotonic focusing coefficient, which improves the model convergence speed and location accuracy. The two-level detection-segmentation joint algorithm is further constructed by combining the pixel-level segmentation capability of UNet3+ network. The model achieves a test accuracy of 93.87% on a dataset containing eight types of materials, with the highest recognition rate of metal (97.04%), followed by blue brick (95.13%) and stone (95.27%), but rammed earth (89.72%) and glazed glass (89.43%) are misclassified due to the complexity of surface features. Experiments show that the algorithm has excellent comprehensive performance in the detection of “spalling”, “phthalate” and “crack”, with an average F1 value of 97.21%, of which the F1 value of crack detection is the highest (97.64%), and the spalling accuracy (99.47%) and phthalate recall rate (95.96%) are outstanding.

The concrete structural design effect of shear wall plays a supporting role in the safety of high-rise houses. In this paper, the shear wall assembly model of high-rise housing building is constructed, and the structural arrangement of the model is adjusted to optimize the design of shear wall force through trial calculation analysis. The time course analysis method is introduced to review the design results of the shear wall structure. Seismic tests are conducted on the house structure by selecting seismic waves. Evaluate the response performance of shear walls under earthquakes by means of seismic pattern decomposition reaction spectrum method and static elasticplastic analysis, and backward optimize the design of concrete structure. The changes in structural performance of the reinforcement measures proposed in the model can be accurately calculated using the 2 methods, seismic pattern decomposition response spectrum method and static elasto-plastic analysis method. The optimization of structural reinforcement resulted in a 42.5% reduction in maximum horizontal displacement, a 42.0% reduction in maximum inter-story displacement, a 25.8% reduction in maximum deflection angle, a 13.1% increase in maximum nodal shear force, and a 29.5% increase in maximum bending moment.

Fusion of multimodal images for the detection of lung cancer lesions can synthesize different modal image features and break through the limitation of single modality. In this paper, 500 lung cancer patients in a hospital were selected as research objects, and ROI segmentation and image filtering were successively performed on their lesion image group data to complete the preprocessing of the image group data. Then Pyradiomics technique is used to extract image features with clear contour details from tumor regions in magnetic resonance (MRI) type images based on five filters. Subsequently, a hierarchical multimodal feature and classifier fusion framework is proposed based on the MCF algorithm. The filtered image features are input into the framework, and each modal feature is selected individually, and the model training and information fusion are carried out in a hierarchical manner to build a prediction model based on multimodal features and classifiers. Compared with similar modeling algorithms, the correlation of most of the features extracted from CT and PET images by the prediction model in this paper reaches 0.2 or above, which shows excellent performance of multimodal image feature screening for lung cancer.

The increasing improvement and maturity of 5G network technology provides a new developable direction for real-time processing of agricultural production data. For the monitoring of sensor data in the agricultural production process, this paper proposes a data collection network (GCN) system consisting of a perception layer, a network layer, and an application layer as a method for collecting research data. After obtaining the research data, a GCN with a forest-like network structure is used to construct the GCN components for data exchange between the data collection system and the gateway as well as the optimization of large-scale data transmission between networks. The ARMA time series data model with high accuracy and short training time was selected to perform short-term time series data prediction in agricultural production scenarios by describing the dynamic characteristics of time series data. AMRA modeling is performed on the experimental data, and the calibration accuracy of the model is 88.81% on average for 10 runs, and it can be as high as 89.16% and as low as 88.27%.

The development of digital intelligence technology is a powerful driving force for breaking down professional barriers and cross-fertilization among multiple disciplines. This paper takes the talent cultivation of art and design majors as the research objective, and explores the optimization and improvement of students’ learning methods as well as instructional design respectively. In the recommendation of learning resources, a feature matrix of art education information technology assessment resources is established to realize the cluster analysis of assessment art resource themes. And the four common problems of learners in personalized e-learning are constructed as multi-objective optimization problems to build a personalized e-learning resource recommendation problem model. Combined with the multi-objective particle swarm optimization algorithm (NEMOPSO), the recommendation model of art and design professional learning resources is constructed. With the assistance of the proposed recommendation model and the professional course design method, the performance of students in the experimental group reaches 4.00 and above in 10 creative force indicators, which verifies the high feasibility and reliability of the designed talent cultivation program.

Electronic communication signals are susceptible to noise and interference in complex electromagnetic environments, resulting in a decrease in feature extraction accuracy. This paper proposes a cyclic frequency feature extraction algorithm that integrates wavelet noise reduction and LCL-FRESH filter. The signal quality is optimized by the adaptive wavelet threshold noise reduction algorithm, and the wavelet coefficients of the characteristic waveforms are screened by combining the fluctuation statistics method. The separation of time-frequency overlapped signals is realized by using LCL-FRESH filter, and the feature extraction ability is enhanced by combining the cyclic cumulative volume reconstruction technique. The Matlab test platform is selected for simulation experiments, and the curve obtained by extreme value threshold noise reduction has more burrs, and the overall is obviously not as smooth as the curve after wavelet threshold noise reduction. The SNR advantage of the LCLFRESH algorithm for the same reconstruction error is about 2dB. The accuracy of the anti-jamming cyclic frequency extraction algorithms for electronic communication signals are all above 92%, and the average completeness of the extraction results reaches 94.2%.

This paper preprocesses missing and anomalous power load data to reduce the error of power load forecasting. Loess smoothing-based time series decomposition algorithm (STL) is introduced to initially decompose the data into trend component, period component and residual component according to the inner and outer loop process. Aiming at the noise limitation of the residual component, the method of reconstruction is carried out by improving the STL-ICEEMDAN quadratic decomposition. Combined with sample entropy and maximum information number, it is quadratically divided into high-complexity component and low-complexity component to improve the prediction accuracy. The results show that the trend component autocorrelation coefficients are 0.786 and 0.729, the correlation is high, there is periodicity, and the decomposition method is effective. The root mean square error, average absolute error, average absolute percentage error, and relative absolute error of power load prediction of this paper’s model are 91.66kW, 77.91kW, 0.88%, and 0.98%, respectively, and the prediction error is smaller than the comparison model.

Aiming at the problems of insufficient control accuracy and low energy efficiency caused by load fluctuations in mechanical hydraulic systems under complex working conditions, this paper uses fuzzy control algorithms as a research tool to explore the optimization path of load adaptation in hydraulic systems. After completing the modeling of load-sensitive valves, the principle of implied parallelism is borrowed to evaluate the adaptability capability of individual components in the mechanical hydraulic system. At the same time, genetic operators are defined to realize the mathematical theory of genetic algorithms in mechanical hydraulic systems as well as optimization. A new load control algorithm for mechanical hydraulic system is proposed by establishing a fuzzy PID control algorithm and integrating the genetic algorithm. The proposed algorithm is used to design a fuzzy adaptive PID controller for the hydraulic cylinder of the loader rocker arm, and simulation experiments of the loader load hydraulic system are carried out. After the flow and power parameters are stabilized, the flow rate of the multiway valve port is stable at 73.5539L/min and the power saving is stable at 0.8691kW, which shows the high effectiveness of the proposed algorithm in the optimization of mechanical hydraulic system load.

Based on the dynamic planning model and multi-player MZS game theory, this study constructs an adaptive optimization framework oriented to the stability of social community, and proposes a computable conflict reconciliation strategy by quantitatively analyzing the dynamic evolution law of social conflicts. First, the cooperativecompetitive relationship among social members is mapped as a state-constrained tracking control problem in the game system, the control barrier function (CBF) is used to define the safe set boundary, and the Nash equilibrium strategy is solved by the dynamic planning recursive equation to realize the mathematical modeling of the conflict evolution. In order to verify the effectiveness of the model, simulation experiments were carried out in combination with the enterprise competition scenario, and the iterative dynamic programming (IDP) algorithm was used to solve the multi-stage optimal control problem, and the results showed that the optimal control performance index J=1.3470, and the error between the control strategy and the analytical solution was controlled in the order of 10−3, indicating that the model can accurately approximate the optimal solution and ensure the evolution of enterprise behavior within the compliance boundary. Further analyzing the initial value sensitivity of the system under the chaotic state, it is found that the initial value difference of only 0.01 (x0=3.50 vs. x0=3.51) can trigger the trajectory dispersion, arguing for the risk of social disorder when the security set constraints fail. In addition, delay strategy experiments show that a single firm’s delay can raise its short-term profit by 4%, but the competitor’s profit decreases by 3%-5%, which needs to optimize resource allocation through dynamic planning to balance the interests of multiple parties. The study proves that by integrating game rule design, real-time risk warning and diversified mediation mechanisms, conflicts can be effectively reconciled and the dynamic stability of the social community can be maintained.

Under the development trend of age-friendly society, the living needs of the elderly have gradually become a hot spot of concern. Among them, indoor space, as the main activity place for the elderly, has a crucial impact on the quality of life of the elderly. In this paper, the objective function is determined according to the task of optimal selection of indoor layout candidates, and the problem of solving model parameters is transformed into a structured regression problem to solve the parameter problem of the model, and the modeling method of indoor space layout estimation under the optimal selection of layout candidates is proposed. Meanwhile, for the conflicting needs between family members and the elderly in the ageing indoor space retrofit and the associated paths of different functional spaces, the basic idea of dynamic programming is used to solve the optimal paths. By combining the dynamic planning method and the layout estimation modeling method, a retrofit design method for aging-adapted indoor spaces is formed. Using this method to design a retrofit program for the aging space, the proportion of “very satisfactory” rating reached 41.98% in the satisfaction evaluation of the actual layout features, which can effectively adapt to the needs of the elderly.

While hand in hand with the Internet of Things (IoT) to provide people with daily convenience, technologies such as big data are also assisting illegal information to invade and interfere with IoT. In this paper, we take the construction of multi-polar intrusion detection system as the ultimate research goal, design the global search strategy of intrusion, the difference degree detection model and the association rule analysis model to analyze the intrusion information in an all-round way. With this data preparation, support vector machine (SVM) classifier is established and fused with similar cluster-based sample size approximation algorithm to build FSVM model as the detection method of abnormal data. The Gray Wolf Optimization (GWO) algorithm is used for the optimization of support vector machine algorithm parameter selection to form the GWO-SVM algorithm, which in turn proposes an IoT intrusion detection system based on the GWO-SVM algorithm. The designed intrusion detection system shows excellent suitability for IoT security multi-pole intrusion detection with accuracy up to 100.00% and F1 value up to 99.01% in the performance evaluation.

Under the background of globalization and multicultural coexistence, the construction of cultural community has become an important path to promote the inheritance and innovation of national culture. This paper takes ethnic music cultural exchange as the research object, and proposes a shortest path model based on the Improved Approximate Neighbor Propagation (IAP) algorithm to optimize the efficiency and structure of cultural communication. The study quantifies the interaction characteristics of nodes and the propagation law by constructing a complex network of ethnic music and cultural communication, and the IAP algorithm solves the limitations of noise interference and modularity resolution to realize the optimization of cultural communication paths. Experiments show that the IAP algorithm has better influence diffusion ability under the DWA model, and the final calculation of influence diffusion in the three datasets is higher than that of the Forward algorithm by 3.4%, 2.6%, and 2.7%, respectively, and the actual number of activated nodes is improved by 3.4%, 3.4%, and 2.5%, respectively. The improved IAP algorithm is used to optimize the network influence problem, and the average shortest path length is only 1508.77 when G=2000, which provides theoretical support and practical solutions for the universal communication and cross-regional mutual appreciation of ethnic music culture.

This study focuses on the potential impact of new media language on the semantic evolution of Chinese, constructs the Chinese Lexical Semantic Core Knowledge Base (CLSKB_Core), which integrates the resources of multi-source dictionaries, and puts forward a semantic similarity computation model based on conceptual graphs. The semantic evolution of Chinese in the new media context is explored through the analysis of ephemeral frequency and the validation of the co-conceptual map retrieval algorithm. The experiments show that the algorithm recognizes sentence types with simple structure such as definite middle, preposition, number, gerund, etc. with better efficiency, and the average correctness rate reaches 0.899. In eight types of problems, the average Fβ value reaches 67.025%. For concept maps with imbalanced attribute descriptions, the algorithm in this paper achieves an average compatibility of 0.886, which is 0.6% higher than the original algorithm. The semantic migration trajectory of the typical case “Tuhao” reveals the penetration mechanism of Internet buzzwords into the semantic field of traditional vocabulary, and reflects the feasibility of adopting natural language processing algorithms to study the semantic change of Chinese.

This paper is based on the position of the elderly group in Anhui Province, and combines the performance of the ease of use of furniture products in the interior design of senior living buildings to propose a set of furniture ease of use evaluation system consisting of 4 dimensions of practicality, economy, aesthetics, and sociability, as well as 12 secondary indexes and 36 tertiary indexes. Subsequently, based on the performance data of furniture products in 10 senior living places in Anhui Province, the entropy weight-TOPSIS method was used to calculate the entropy weights and weights of each index, and the furniture was ranked according to its ease of use performance data. Five commonly used furniture products in the 10 senior living places were selected as experimental objects, and the correlation analysis algorithm was used to calculate the confidence level of the rules between the furniture products and the 36 third-level indicators. Among them, the indicators of “ergonomic furniture scale” and “purchase price” have higher confidence level with multiple furniture products, which indicates that the most concerned factors in the purchase decision of elderly furniture products are the comfort of the elderly and the price of the products.

Heilongjiang Province is an important agricultural production base in China, and its soil organic carbon stock occupies an important position in the country. In recent years, the spatial distribution of soil organic carbon has gradually received wide attention with the environmental changes and the transformation of agricultural production methods. By studying the spatial distribution characteristics of soil organic carbon in erosion gully area, it can provide scientific basis for regional soil protection and carbon sink management. This paper discusses the spatial distribution characteristics of soil organic carbon in erosion gully area and its simulation method based on the geostatistical method in Heilongjiang Province. The study collected soil samples from different points in the erosion gully area, and used geostatistical methods to spatially analyze and simulate the soil organic carbon data. The study firstly determined the best semi-variance function model by calculating the spatial correlation of the sample points, and then predicted the spatial distribution of soil organic carbon by using the Kriging interpolation method. The results show that the spatial distribution of soil organic carbon in the erosion gully area has obvious regional differences and is jointly influenced by topography, vegetation cover and land use type. Through the simulated spatial distribution maps, areas with higher organic carbon storage and plots with higher carbon sink capacity can be clearly seen. This study provides a scientific basis for further soil protection measures, carbon sink management and sustainable agricultural development.

This paper constructs a simulation model of indoor sound field based on ODEON software, and proposes an iterative updating strategy for the distribution of sound-absorbing materials in combination with the topology optimization algorithm. Combined with 2D boundary element model and finite element analysis, the validity of sound field simulation and optimization effect of sound-absorbing materials are verified. Through the 2D boundary element model and the sound barrier optimization case, it is verified that the reverberation time of each room is adjusted under the effect of room coupling, and the difference between the reverberation time of the master bedroom and the living room under the same frequency is not more than 0.05 s. The optimized material has a surface density of 0.39 kg/m², which has the largest sound absorption coefficient, and it has a very good acoustic absorption effect for the mid-frequency band as well. In order to achieve the same acoustic effect, glass fiber acoustic cotton needs to be 80mm thick. The results highlight the importance of optimizing the design of the ratio of sound-absorbing materials.

As the core equipment of high-voltage direct current transmission system, the operation reliability of normal direct current converter valve directly affects the stability of power system. In this paper, a composite processing strategy integrating tensor decomposition theory and deep learning is proposed for the fault diagnosis problem under the missing data scenario of normally straight converter valve. A multidimensional data interpolation model is constructed based on Tucker decomposition, and the efficient recovery of high-dimensional missing data is achieved through the co-optimization of core tensor and factor matrix. Design 1DCNN-BiLSTM hybrid network with attention mechanism to enhance the time-frequency characterization of fault features. As verified by the analysis of simulated and measured vibration data on the PSSE platform, the average relative error of the Tucker decomposition and its rate of change are both minimized in the comparison models, and the average MRE is 2.29 in the random missing data scenario. The MRE is reduced by 38.75% compared to the suboptimal model in the PMU fault scenario with 25% high missing rate. The method in this paper can successfully isolate the fault features of severe faults. Moreover, there are rich fault feature modulation bands in addition to the fault feature frequency.

With the continuous development of high-voltage direct current (HVDC) transmission technology, the Conventional DC Converter Valve is becoming more and more important to maintain system stability. This paper proposes an integrated monitoring method based on acoustic signal processing and multimodal timing analysis. The DC maximum current time series feature model is constructed to extract the maximum current eigenvolume and other characteristics. The acoustic signal of the converter valve is collected by combining the homogeneous electrode acoustic detection device and digitized. The TimesNet time series analysis model is introduced to convert the two-dimensional depth characteristics of the time series to realize the accurate positioning of the commutator valve phase change abnormality and vibration noise. Research shows that when the power load is 400MW, the frequency of the converter valve sound signal is concentrated within 3kHz, and the frequency increases with the increase of power. During the monitoring process, the output current of the converter valve fluctuates abnormally in 2-3s, while the voltage fluctuates abnormally in that time. The voltage ripple adjusted according to the monitoring results re-maintained the coherent state.

In this paper, the BERT model containing a large number of encoders is used to complete the preprocessing operation of literary text data and generate the corresponding word vectors. The frequency-intended document frequency (TF-IDF) algorithm and the sentiment computation model SnowNLP are introduced to count the word frequency and calculate the sentiment polarity. Build a metaphorical sentiment polarity computation modeling framework. Embedding culturally relevant attributes in word vectors, modeling contextual semantics by combining long and short-term memory networks, and using the attention mechanism to train to get the attention weight matrix of the word items to accurately classify the emotional polarity of the word items. Taking the selected comparison works as an example, the average word length of classical Chinese literature is 1.08-1.20, and the proportion of real words accounts for 74.15%. The percentage of negative sentiment for the 6 keywords was 20%, 15%, 15%, 21%, 15%, and 10%.Five of the six represented pieces are dominated by negative emotions. The average word length in Western Romantic literature is 2.26-2.46, with 78.27% real words. The 6 keywords positive emotions accounted for 15%, 5%, 20%, 25%, 20%, 15%. 5 of the 6 represented works were dominated by positive emotions.

Traditional human resource allocation methods often rely on empirical judgments and lack the support of scientific quantitative analysis, making it difficult to find a balance between cost control and benefit maximization. Aiming at the cost-benefit balance problem in hospital human resource allocation, this study proposes a multiobjective optimization model based on the improved NSGA-II algorithm. By introducing the cosine similarity to adjust the congestion distance ranking, the multi-objective decision-making model with the objectives of minimizing human resource costs and maximizing project benefits is constructed, and the multi-dimensional chromosome coding and adaptive parameter selection strategy are used to solve the problem. An empirical study is conducted in three departments of respiratory medicine, neurology and orthopedics in a hospital, and the results show that the improved algorithm reduces the human resource cost by 6.81% to 23.90%, improves the project benefit by 10.95% to 41.07%, and converges to the optimal distance of about 20 at the 150th iteration cycle. Compared with the traditional NSGA-II algorithm, the improved algorithm shows significant advantages in both Pareto frontier quality and convergence performance, and provides an effective quantitative analysis method for the dynamic optimization of hospital human resources.

With the arrival of the digital era, the digitization of history education resources has become an important means to improve the quality of classroom teaching. Through experimental design and data analysis, this paper explores the application effect of digital history education resources in secondary school history classroom teaching. The study adopted a learning recommendation model based on knowledge graph and conducted a semester-long teaching experiment in Shanghai A Middle School. The experimental results showed that the experimental class using digital resources demonstrated significant advantages in learning motivation, learning achievement and classroom participation. Specifically, the final grades of the experimental class increased by 5 points on average, and the grades were more centrally distributed with lower standard deviation, showing better learning effects and higher learning stability. This paper also analyzes the accuracy of different recommendation algorithms by comparison, proving that the proposed algorithm is superior to other algorithms in terms of recommendation accuracy, with lower MAE and RMSE values, respectively, verifying the feasibility and effectiveness of the application of the model in history education. The study shows that digital history education resources can not only enhance the interactivity of classroom teaching, but also enhance students’ interest in learning and improve their mastery of history knowledge.

Medical image detection plays a crucial role in disease diagnosis; however, traditional manual interpretation is often limited by subjectivity and low efficiency. Interpretable artificial intelligence (AI) techniques, grounded in image processing algorithms, demonstrate strong potential for enhancing objectivity and reliability in this domain. In this study, we propose an enhanced YOLOv12 algorithm that integrates both attention mechanisms and a residual feedback structure. Combined with a selective contextual transformer-enhanced SCTNet segmentation network, these components form a unified and intelligent medical image processing system. Methodologically, the AGs-ECSA hybrid attention module enhances feature extraction by incorporating Efficient Channel Attention (ECA) and spatial attention mechanisms. A loopback residual structure is introduced to preserve original feature information, while bounding box regression is improved using the Complete Intersection over Union (CIoU) loss function. Additionally, the Selective Contextual Transformer (SCT) module captures both local and global semantic dependencies.Experimental results on the DeepLesion dataset demonstrate that the proposed method achieves an average sensitivity of 88.97%, surpassing the strong baseline MULAN by 0.46% and consistently achieving higher detection accuracy at lower false positive rates. On the GLAS segmentation dataset, SCTNet achieves an mF1 score of 0.8285 and an mIoU of 0.7246, representing improvements of 4.98% and 8.64%, respectively, over existing mainstream methods. The system was further validated on cerebral hemorrhage CT scans, accurately localizing lesions and estimating their size. These findings demonstrate the effectiveness of interpretable AI in medical image detection and offer a reliable framework to support clinical diagnosis.

The standardization and personalization needs of sports training have driven the application of virtual reality technology in the field of sports. This study designs and implements an athletic training simulation system based on virtual reality technology, which uses 3D human motion simulation simulation technology for training simulation, creates an action database through motion capture, applies offset mapping technology to correct the design of training actions, checks the reasonableness of the actions by using the Newtonian Eulerian motion model, and arranges the standardized technical action sequences through the method of motion splicing. The system integrates the core functional modules of VR perception interaction, motion capture acquisition, training environment establishment and action reproduction. In the study, 100 athletes were experimentally selected to verify the training effect, and the results showed that the error of the system’s action data acquisition was controlled within 0°15′, the mastery of the training content reached 96%, the training cost was maintained at 10,000 yuan, and the error of the system’s response speed was no more than 5%. The system in this paper can effectively improve the training effect of athletes, significantly reduce training costs, and provide an efficient and economical technical solution for sports training.

The security of the rolling stock network is directly related to the stability and security of the railroad transportation system. With the continuous development of the rolling stock network, the network security problem is becoming more and more prominent. Based on this, this paper proposes a security situational awareness model based on multi-source data fusion. First, based on the network topology, host information and alarm information, the fusion framework of multi-source heterogeneous data is established. Second, the network security posture is evaluated and predicted by applying algorithms such as Bayesian network and Kalman filter. The experimental results show that the model can effectively improve the accuracy of security posture assessment when dealing with multi-source data. By comparing the detection results of different methods, the model proposed in this paper shows high accuracy and low false detection rate in a variety of network attack scenarios, especially in the types of attacks such as privacy data stealing and network bandwidth consumption, the recognition effect is most significant. The experimental data show that the proposed method has an error of less than 0.03 during the evaluation process and has good real-time performance and stability. Therefore, the security situational awareness method based on this model can provide more accurate security protection support for the rolling stock network.

Accurate load forecasting can not only help microgrids improve the utilization efficiency of energy resources, but also ensure the stability and reliability of power supply. In this paper, a deep learning-based load forecasting model for microgrids is proposed, and its application effect in different microgrids is verified through experiments. First, deep learning algorithms such as LSTM, BiGRU and CNN are used to construct a hybrid prediction model, and TVFEMD technique is introduced to signal decompose the load data to reduce the influence of noise. Through comparative experiments, the results show that on microgrid 1, the proposed model has a higher prediction accuracy with a minimum MAPE value of 2.0485% compared with the traditional methods, while on other microgrids, the model still maintains a more stable performance. In microgrid 3, the prediction results are more reliable in general, although there is a large error. Based on the experimental results of the model, this paper also discusses the interpretability of the model and its potential application in real microgrid scheduling. Ultimately, the proposed deep learning model can effectively improve the accuracy of microgrid load prediction with strong adaptability and stability.

In order to enhance the public’s awareness of network security, innovative educational means are particularly important. Combining digital art and animation elements as the carrier of network security propaganda, and utilizing advanced technologies such as style migration and self-attention mechanism, a network security propaganda platform with fun and interactivity can be designed so as to improve the dissemination effect of the platform and user participation. This paper proposes a design method of network security publicity platform based on digital art animation elements. The animation elements are extracted by style migration technology, and the generator and discriminator structures are optimized by combining the self-attention mechanism to achieve highquality animation image generation. The evaluation metrics used include PSNR and SSIM, and the experimental results show that compared with the traditional GAN model, the improved AnimeGANv2-Self-Attention model significantly improves the image stylization effect, with a 1.39% increase in PSNR and a 2.99% increase in SSIM value. In addition, the cybersecurity publicity platform designed in this paper combines animation elements and cybersecurity education, which has a high degree of user attraction and participation. By analyzing the actual user data, the difference between the experimental group and the control group in terms of publicity effect is significant, indicating that the platform is more effective than traditional publicity means in enhancing cybersecurity awareness.

Currently, aesthetic education in higher education in China faces the dilemma of dispersed resources and poor results. The allocation of aesthetic education resources is not balanced, the degree of social participation is not high, and there is a lack of effective articulation between correctional education in prisons and aesthetic education in colleges and universities. In this paper, we constructed the “prison – university – community” trinity synergistic aesthetic education resource integration mechanism, and explored the optimization of social aesthetic education resource allocation path. Using experimental method, interview method, questionnaire survey method and other research methods, with 856 sophomore students in key undergraduate colleges in province A as the research object, 60 students were randomly selected into the experimental group and the control group of 30 people each, using one-way analysis of variance method for data processing. The study constructed a synergistic pathway led by universities, corrected by prisons, and expanded by communities, which was practically verified through the three dimensions of aesthetic education literacy, aesthetic education skills, and basic theoretical knowledge of aesthetic education. The results showed that the mean value of aesthetic literacy in the experimental group reached 3.891, significantly higher than that of 2.60 in the control group, with a P-value of 0.004; the mean value of aesthetic skills in the experimental group was 3.981, and that of the control group was 2.694; and the P-value of the test of basic theoretical knowledge of aesthetic education in the experimental group was 0.001, which showed a significant difference. The trinity synergistic path is effective in enhancing the comprehensive quality of students’ aesthetic education, and provides an effective practical model and theoretical support for the integration of social aesthetic education resources.

At present, ecological environmental protection and rational use of natural resources have become an important strategy for national development, and the establishment of a scientific system for calculating the damage compensation of natural resource assets is of great significance for promoting the construction of ecological civilization. In this paper, we constructed a natural resource asset damage compensation value assessment model based on IWOA-BP neural network, and established an accounting index system including land resources, biological resources and water and air regulation functions. The study adopts the improved whale optimization algorithm to optimize the weights and thresholds of the BP neural network, improves the uniformity of the initial population distribution through Sine chaotic mapping, and constructs a comprehensive accounting method for the value of land assets, the value of biological assets and the value of water and air regulation assets. The validity of the model was verified with Yulin City as the empirical research object. The results show that the IWOA-BP model is stabilized after 77 iterations, and has a faster convergence speed and higher prediction accuracy than the traditional BP, GWO-BP, and WOA-BP models. The average relative error of the model is controlled within 9.7452%, the average absolute error is 0.3013, and the root mean square error is 0.2241, and the assessment accuracy is significantly better than other algorithms. The total value of natural resource assets in Yulin City increased from 577,452.16 billion yuan to 577,934.63 billion yuan, and the value of land resources accounted for 64.94%. The model can effectively solve the problems of low accuracy and slow convergence of traditional methods in the calculation of natural resource asset damages, and provides a scientific and feasible calculation method for natural resource asset damages.

The traditional way of watching sports events can no longer meet the modern audience’s demand for interactivity and immersion. The environment interaction perception algorithm in virtual reality provides new ideas to enhance the immersion of live sports events. Using Leap Motion and other sensing devices combined with virtual reality technology, the gestures of the audience can be recognized in real time, thus improving the interaction between the user and the live broadcast environment and enhancing the audience’s sense of presence and participation. This study explores the application of virtual reality technology based on environment interaction perception algorithms in live broadcasting of sports events, using Leap Motion sensors to capture the user’s hand movements and interacting with the virtual reality environment through gesture recognition algorithms. In the experiment, 18 participants watched live sports events using different media and completed the corresponding interactive tasks. The experimental results show that after applying the interactive perception algorithm, the viewer’s operation efficiency is improved by about 30%, and the interaction process is smoother. Through regression analysis, it was found that perceptual algorithm response, algorithm awareness and immersion experience were significantly positively correlated, while perceptual algorithm indifference and intrusiveness were significantly negatively correlated. The study shows that the live sports event broadcasting system based on this algorithm can significantly enhance the immersion of the audience and reduce the operation burden, which provides a useful reference for the development of virtual reality live broadcasting technology.

Sports injuries have become an important factor affecting athletes’ competitive performance and career, and traditional prevention methods mainly rely on empirical judgment, which lacks scientificity and precision. Constructing an efficient sports injury prediction model and formulating corresponding rehabilitation strategies are of great significance to improve athletes’ health management. In this study, we constructed a sports injury prediction model using Improved Whale Optimization Algorithm Optimized Support Vector Machine (IWOA-SVM), and analyzed it based on 1000 athletes’ records in Kaggle dataset. The traditional whale optimization algorithm was improved by Circle chaos strategy initialization, inertia weight adjustment and Cauchy variation strategy, and the prediction model was built by combining with support vector machine. Correlation analysis showed that training intensity was significantly correlated with injury likelihood (p=0.007). The results of model performance evaluation showed that the IWOA-SVM model had an accuracy of 93.92%, a precision of 92.79%, a recall of 93.52%, and an AUC value of 95.45%, which were better than the traditional machine learning methods in all indicators. The feature importance analysis showed that height, weight and training intensity were the key predictors, and the influence weights were more than 0.24. Based on the prediction results, personalized rehabilitation treatment strategies including strains, abrasions, joint sprains and contusions were developed. The prediction model provided a scientific basis for the prevention of sports injuries, and the rehabilitation strategies provided a systematic guide for the athletes’ post-injury recovery.

As a discipline that combines artistry and education, the teacher competencies of music education majors have a significant impact on students’ educational opportunities. This paper examines the impact of teacher competence on students’ educational opportunities in music education majors and investigates the mediating role of learning inputs in this context. The study used Structural Equation Modeling (SEM) to analyze the data, based on a questionnaire survey of teachers and students majoring in music education in a university, and 411 valid questionnaires were collected. It was found that teacher competence had a significant positive effect on students’ educational opportunities (path coefficient = 0.163, p<0.001) and that teacher competence also had a more significant effect on students' learning input (path coefficient = 0.184, p<0.01). Further analysis showed that learning input played a partial mediating role between teacher competence and educational opportunities and the mediating effect was 58.8%. In particular, all dimensions of learning inputs (behavioral inputs, cognitive inputs, affective inputs, and social interaction inputs) significantly affected educational opportunities, and the mediating effect of affective inputs was the least. The results suggest that improving teacher competence and student learning input can effectively improve the educational opportunities of music education majors, thus providing some practical basis for the improvement of educational quality.

Hepatocellular carcinoma (HCC) is a common and fatal malignant tumor worldwide, which is difficult to diagnose early and has a low survival rate. In this paper, we investigated the temporal expression pattern of AKR1B10 gene in patients with hepatocellular carcinoma (HCC) and constructed a GL-TGRN model using LSTM neural network combined with gene regulatory network. By analyzing multiple sets of student data (including mRNA expression data, miRNA data, and DNA methylation data), we conducted an in-depth exploration of the temporal expression of the AKR1B10 gene and its relationship with hepatocellular carcinoma development. The results showed that the GL-TGRN model performed excellently in inferring AKR1B10 gene expression, and the AUROC and AUPR values were increased by 26.23% and 35.69%, which were significantly higher than the comparison methods (e.g., GC-SIN and JUMP). In addition, through differential expression analysis, we screened 786 differential genes and 14 miRNAs related to hepatocellular carcinoma, and these molecules are closely related to hepatocarcinogenesis. Ablation experiments demonstrated that the fusion of multi-omics features in the GL-TGRN model significantly improved the accuracy of gene regulatory inference. This paper provides new data support for early diagnosis and personalized treatment of hepatocellular carcinoma.

Under the cloud computing environment, traditional security mechanisms are difficult to effectively protect the confidentiality and integrity of sensitive data. Aiming at the security protection of sensitive data in cloud storage environment, this paper proposes an optimization algorithm for sensitive data discovery and integrity checking based on Intel SGX trusted execution environment. Methodologically, the SGX-based trusted execution environment framework is constructed, the integrity verification scheme combining the multi-branch path tree (MBT) data structure and bilinear pair algorithm is designed, and the third-party verification organization and blockchain network are introduced to realize data integrity verification. The data verification in challenge-answer mode is realized through the smart contract mechanism, and the file version verification is optimized by combining the version sequence number parameter. The results show that when the number of files reaches 1900, the total execution time of this paper’s scheme is reduced by 51.29% and 32.76% compared with the B-PDP and SA-PDP schemes, respectively. Tests based on the MNIST dataset show that the time overheads of the storage and validation phases are 0.728ms and 0.291ms, respectively. The overall performance of the Trusted Execution Environment network reaches 95.48% of the original Fabric, with only a 6.42% increase in latency and a 5.03% decrease in throughput. The conclusion proves that the scheme can significantly improve the efficiency of sensitive data verification under the premise of guaranteeing security, and provides an effective guarantee for data security in cloud storage environment.

With the continuous development of technology, personalized learning has gradually become an important trend in the field of education. This study proposes a Bayesian network-based personalized learning path planning method for art education. First, the personalized needs of learners are analyzed, and a Bayesian network model is used to determine learners’ mastery of art knowledge points and recommend adaptive learning resources. Then, the study uses the Felder-Silverman learning style model and Bayesian classifier, combined with the learner’s online behavior, to calculate the similarity between the learner and similar learners, so as to recommend relevant learning resources for the target learner. The experimental results show that the personalized recommendation system significantly improves the students’ art scores, and the scores of the students in the experimental group increase from 6.94 to 8.93 in the pre-test, which is an improvement of 1.99 points. Through the personalized recommendation of learning resources, learners’ learning efficiency is significantly improved and their learning attitude is more positive. The study shows that the personalized learning path recommendation system based on Bayesian network can effectively improve the learning effect of students and promote the mastery of their art knowledge and ability.

In today’s rapid development of digital economy, e-commerce platforms are facing the challenges of diversified user demands and fierce competition. Traditional marketing methods have problems such as high cost, poor effect, low conversion rate, and it is difficult to meet the demand for personalized services. For the problems of poor accuracy of user group division and poor marketing effect of e-commerce platform, this paper proposes an improved K-means clustering algorithm integrating genetic algorithm. The method optimizes cluster center selection by genetic algorithm, which solves the local optimal problem caused by the randomness of the initial point selection of traditional K-means algorithm. Based on 4000 user consumption data, the study constructed a user portrait model containing 8 dimensions, such as gender, age, average monthly total consumption, etc., and utilized the improved algorithm to divide the user groups. The experimental results show that the improved algorithm achieves an average accuracy of 90.16% on the Iris dataset and 80.94% on the Wine dataset, which is 13.41% and 14.37% higher than the traditional method, respectively. The user groups were successfully divided into four clusters, with the highvalue user group spending an average of 3,500 yuan, accounting for 11.08% of the total. The study formulated differentiated marketing strategies for different user groups, providing an effective solution for e-commerce platforms to achieve precision marketing.

Low-light image processing is an important task in computer vision, which is widely used in many fields. Due to the high image noise and blurred details in low illumination environment, it is difficult for conventional image enhancement techniques to effectively recover the image quality. In this paper, a method for automatic detail enhancement and visual effect improvement of low illumination images based on hierarchical image processing algorithm is proposed. The method combines Retinex theory and convolutional neural network, and realizes the effective enhancement of low illumination images by constructing the decomposition and enhancement module of reflection component and light component. In the experiments, LOL and LOLv2 datasets were used for subjective and objective evaluation, respectively. The results show that on the LOL dataset, the PSNR and SSIM of this method reach 25.42 and 0.911, respectively, and on the LOLv2 dataset, the PSNR is 23.09 and the SSIM is 0.932, which are better than other existing algorithms. In addition, the subjective evaluation also shows that the proposed method has obvious advantages in terms of noise removal, line clarity, color naturalness and image realism. The study shows that the method can effectively improve the detail performance and visual effect of low illumination images.

The rapid development of information technology promotes the digital transformation of the financial industry, and the deep integration of big data technology and the financial market forms a big data financial model. Traditional financial risk assessment methods have limitations such as insufficient accuracy and slow convergence speed when dealing with massive multidimensional data. Big data algorithms show great potential in the fields of transaction fraud identification, credit risk assessment, customer marketing and stock market prediction, etc. However, the existing assessment models still face challenges such as local extreme value traps and poor generalization ability, and the construction of efficient and accurate financial risk assessment models has become a current research focus. METHODS: A regional financial risk assessment model based on Improved Cuckoo Optimization BPNN Neural Network (ICS-BPNN) is constructed, and a risk assessment index system is established by selecting five first-level indexes and 26 second-level indexes for the local macroeconomy, the government sector, the financial sector, the real estate sector and the real economy. The principal component analysis method of downscaling and entropy value method are used to determine the weights, and the traditional cuckoo algorithm is improved by dynamic step size and abandonment probability to optimize the weights and thresholds of BP neural network. Results: the ICS-BPNN algorithm reaches the global optimal solution of 0.046 after 20 iterations, while the traditional BP algorithm needs 42 iterations to find the optimal solution of 0.105. The absolute errors of the ICS-BPNN algorithm are all under 3.85, the risk prediction accuracies are all over 0.90, and the average value of R² of the fit of the test set is 0.848. The average value of the financial risk prediction of the eastern part of the country is 0.848 for the years 2026 and 2027, respectively. The predicted values of financial risk of the region in 2026 and 2027 are 0.482 and 0.527 respectively, which are both in risky status. Conclusion: The improved cuckoo algorithm effectively improves the convergence speed and prediction accuracy of BP neural network, and the ICS-BPNN model shows excellent performance in regional financial risk assessment, which provides reliable technical support for financial risk management.

Traditional learning assessment methods rely on static tests, which cannot reflect students’ learning status in real time. In this paper, we propose an online learning assessment system based on Bayesian network, which can dynamically assess the learning effect of students and update their knowledge status in real time. First, the system collects data from the online learning platform through Python crawler technology, including learners’ test scores, homework scores and learning behavior data. Then, a Bayesian network model is used to model the learning process of the students, assess their knowledge mastery, and calculate the probability of answering questions correctly by combining with IRT theory. Through experimental validation, this system performs well in terms of assessment accuracy and prediction accuracy, with an assessment accuracy of 92.65% and a prediction accuracy of 90.84%. In addition, the system is able to track learners’ behavioral characteristics in real time and improve the effect of personalized teaching by analyzing learners’ learning patterns. The experimental results show that the online learning assessment model based on Bayesian network can effectively improve the accuracy of learning assessment and provide an efficient learning assessment tool for online education platforms.

Traditional quality management methods have problems such as insufficient prediction accuracy and slow response speed when facing complex production environments, which are difficult to meet the demand for refined management of modern production. In this study, the quality stability prediction model of cigarette production process based on Meta-DQN is constructed, which solves the deficiencies of traditional methods in small sample learning and environmental adaptability. 500 samples of working condition data such as temperature, humidity, airflow speed, etc. were collected through the production line sensors in cigarette factory H. The MQTT communication protocol was used for data transmission, and MinMaxScaler normalization was applied to ensure data consistency. The Meta-DQN prediction model is constructed by combining the Meta Reinforcement Learning MAML algorithm with the deep Q network, and the fast adaptation to different production tasks is achieved through the two-layer loop optimization mechanism. The experimental results show that the R² coefficients of determination of the training and test sets reach 0.992 and 0.957, respectively, and the model prediction accuracy is significantly improved. In the key parameters prediction validation, the average deviation of the predicted values of six process parameters from the real values is only 0.212, which is much lower than the standard setting of 0.565. Comparison experiments show that the Meta-DQN model can quickly converge in the initial operation stage of the equipment, effectively reducing the scrap rate, which is significantly better than that of the pure DQN algorithm. The method provides an efficient and intelligent solution for cigarette production quality management and has important engineering application value.

Multi-terminal grid system contains wind power, photovoltaic power generation, energy storage equipment and other distributed power sources, and its operating characteristics are complex and variable, so the traditional optimization method is difficult to effectively deal with the scheduling decision-making problem under multi-objective constraints. In this paper, an adaptive particle swarm optimization algorithm based on information interaction mechanism is proposed to address the problems of low information interaction efficiency and insufficient computational accuracy in multi-terminal grid scheduling optimization. The method constructs a multi-objective optimal scheduling model with the objectives of lowest operation cost, maximum environmental benefit and highest system security, adopts a chaotic initialization strategy to enhance the diversity of particle swarms, designs an information interaction mechanism to enhance the global search capability of the algorithm, and establishes an adaptive updating strategy to avoid local optimum. The performance of the algorithm is verified by the ZDT test function, and the results show that the SP value of the improved algorithm on the ZDT3 function is only 0.0096, which is 75% of that of the traditional PSO algorithm; the GD value is 1.9224e-04, which is better than that of the traditional algorithm, which is 2.0905e-04. In the actual multi-terminal grid scheduling experiments, the maximum load of the user is 2200kW, and the daily load rate is 44.5%. The total load fluctuation range of the grid is optimized from 400-900 MW to 560-760 MW after scheduling with the proposed method.This study provides effective technical support for the optimal scheduling of multi-terminal grids, and significantly improves the stability and economy of system operation.

With the rapid development of artificial intelligence technology, enterprises are facing a new brand competition environment. The enhancement of brand core competitiveness no longer relies solely on traditional marketing and brand management strategies, and digital transformation and technological innovation have become key factors. This paper utilizes the fsQCA method to study the path of enterprise brand core competitiveness enhancement in the context of artificial intelligence. Through a case study of 20 listed manufacturing enterprises, factors such as technological innovation, enterprise scale, marketing investment, corporate social responsibility and digitalization level are selected as antecedent variables, and how these factors affect the enhancement of brand core competitiveness through different combinations is explored. The results show that technological innovation, digitalization level and enterprise scale have a significant impact on brand core competitiveness, and four different grouping paths (S1, S2, S3 and S4) can explain more than 90% of the brand competitiveness enhancement. Multiple regression analysis further verified the positive influence of path 1 and path 2 on brand core competitiveness, especially the key role of technological innovation in brand value enhancement. In addition, the implementation of corporate social responsibility also plays an indispensable role in the long-term development of the brand. The conclusion of this paper points out that enterprises should pay attention to technological innovation and digital transformation, and build an all-round brand competitiveness enhancement strategy by combining enterprise scale and the implementation of social responsibility.

With the rise of digital finance, improving financial services through advanced technologies such as the Internet and big data has gradually become a key factor in optimizing the financing efficiency of small and mediumsized enterprises (SMEs). This paper analyzes the impact of digital finance development on the financing efficiency of Chinese SMEs through a two-stage network DEA model. The study selected the financial data of 200 SMEs during the period of 2020-2024, and used input indicators such as labor costs, fixed assets, and financial expenses, intermediate indicators such as equity financing amount and debt financing amount, and output indicators such as operating income and investment income. The results show that the financing efficiency of SMEs is generally low during the period of 2020-2024, among which the financing efficiency is the highest in 2022, and the pure technical efficiency reaches 0.823, which indicates that the management and technical level of the enterprise has improved in that year. Further regression analysis shows that the digital finance index (DFI) is significantly positively correlated with enterprise financing efficiency, indicating that the development of digital finance can significantly improve the financing efficiency of SMEs. Specifically, the breadth of coverage, depth of use and degree of digitization of digital finance all have a positive effect on the improvement of financing efficiency, especially the depth of use has the greatest impact. It is concluded that digital finance significantly contributes to the improvement of SMEs’ financing efficiency by reducing financing costs and improving the efficiency of capital utilization.

Under the background of the current energy structure transformation, the proportion of distributed photovoltaic power generation in the power system is increasing, but its intermittency and uncertainty bring serious challenges to the stable operation of the power grid. In this paper, for the traditional virtual power plant in distributed PV monitoring and regulation of the lack of real-time and data processing delay problems, constructed a virtual power plant based on edge computing technology distributed PV real-time monitoring and regulation system. The study adopts centralized and decentralized control mode, taking edge computing nodes as the low-level control and the virtual power plant energy management center as the high-level control. Five main influencing factors, namely, solar irradiance, ambient temperature, air humidity, wind speed and barometric pressure, are screened out by the gray correlation method, and an improved LSTM-TCN prediction model is constructed for the ultra-short-term output prediction of distributed photovoltaic. Based on the experimental data validation at five sites in Australia, the LSTMTCN model has an MAE of 0.0388 and an RMSE of 0.0759 in typical summer scenarios, which improves the accuracy by 2.59% and 6.21%, respectively, compared with the traditional LSTM model. In the IEEE 33-node distribution network example, the total system load is 12MW and the total installed capacity of PV is 9.0MW, the proposed method realizes the ideal characteristics of virtual power plant with strong internal coupling and weak external coupling. The results show that the virtual power plant architecture based on edge computing can significantly improve the prediction accuracy of distributed PV and the system regulation effect, which provides an effective technical solution for building a smart grid.

Accurate observation of the anatomical structure of the root system of aquatic plants is of great significance in understanding their adaptive mechanisms. In this paper, an image processing algorithm based on improved Retinex theory is proposed to address the problem of poor image quality and blurred details of underwater reed root system. The method combines the light-directed variable attention module (LDAT), the spatial-frequency domain feature fusion module (SDFF), and the medium transport module (MTM), and realizes the effective enhancement of the underwater image by introducing a perturbation term to simulate the underwater imaging environment and using saturation to distinguish between the artificial illumination and the natural light source. The experimental results show that compared with the traditional dark channel a priori algorithm and MSR algorithm, the algorithm in this paper improves the information entropy significantly, in which the information entropy of image 3 is improved from 6.762 to 7.827, and the standard deviation is increased from 18.978 to 58.973, and the average running time is only 6.696 s, which is significantly better than that of the dark channel a priori algorithm, which is 71.49 s. The anatomical structure of the root system is analyzed for the reeds under the different water depth conditions. By analyzing the anatomical structure of the root system under different water depth conditions, it was found that the proportion of root cortex aerated tissues was significantly positively correlated with the water depth, and the proportion of aerated tissues was significantly larger under deep flooding than non-flooding conditions. The conclusion shows that the algorithm can effectively improve the quality of underwater images and provide technical support for the accurate analysis of plant root structure.

Traditional time series analysis methods often face problems such as missing data and noise interference, which affects the predictive ability and accuracy of the model. Generative Adversarial Networks (GANs) have made significant progress in the fields of image generation and data restoration due to their excellent generative capabilities. In this study, a feature extraction and reconstruction method based on generative adversarial network (GAN) for time series data is proposed. By improving the structure of generative adversarial network and introducing the synergistic loss function of global optimization objective and single-channel optimization objective, combined with the Transformer architecture, the CTTS-GAN model is proposed. The model can better retain the features of the original data and improve the diversity of the data when generating time series data. The experimental results show that the data generated by CTTS-GAN exhibits a lower error (0.42) under the maximum mean difference (MMD) metric, showing a distribution closer to the real data. In addition, CTTS-GAN obtained better classification results than the traditional GAN model when using the generated data for the classification task, with a TSTR score of 0.83 on the Support Vector Machine classifier, which is significantly higher than other methods. This indicates that CTTS-GAN has a strong potential for application in generating high quality time series data.

Currently, the global economy is in a critical period of digital transformation, and governments have been pushing forward the reform of the fiscal and taxation system to adapt to the new development pattern. Based on the panel data of 30 provinces in China from 2014 to 2024, this paper empirically investigates the impact mechanism of digital fiscal and tax innovations on optimizing the regional tax system and promoting the high-quality development of the economy by applying the Linked Panel Equation Model (LPEM), the mediation effect test, and spatial econometric methods. The study found that: digital fiscal innovation is significantly and positively related to economic high-quality development, with a correlation coefficient of 0.205; the regression coefficient of governmental fiscal expenditure preference on economic high-quality development reaches 10.633, and the regression coefficient of tax effort is 0.114, both of which are significant at the 1 percent level; The Sobel test and Bootstrap method confirm that the regional tax system plays a mediating role in the process of digitalized fiscal innovation affecting the high-quality development of the economy; the spatial autocorrelation test shows that the Moran’s I index of high-quality development of the economy in 2024 is 0.066, which indicates that there is a significant spatial positive correlation. This study provides empirical support for deepening the reform of fiscal and taxation system and promoting regional coordinated development.

Bitcoin, as the most representative cryptocurrency, has an extremely volatile price, and traditional financial theories are difficult to fully explain its market behavior. The high volatility and complexity of Bitcoin’s price pose a great challenge to investment decisions. This study proposes a bitcoin exchange rate prediction method based on a combined ARIMA-LSTM model, which improves the prediction accuracy by combining traditional time series analysis with deep learning techniques. Methodologically, an LSTM neural network is first constructed to capture the nonlinear characteristics of the bitcoin price, then an ARIMA model is built to analyze the linear trend, and finally the prediction results of the two models are optimally combined by using the CRITIC weight assignment method. The experiment uses the bitcoin closing price data from September 1, 2021 to December 31, 2024 for validation. The results show that the combined ARIMA-LSTM model significantly outperforms the single model in terms of forecasting performance, with a mean absolute error (MAE) of 0.0002, a root mean square error (RMSE) of 0.0003, and a mean absolute percentage error (MAPE) of 0.0006, which are 0.0071, 0.004, and 0.0051 lower than that of the ARIMA model, respectively. Empirical analysis shows that the combined model can more accurately capture the changing law of bitcoin price by integrating the advantages of linear and nonlinear prediction methods, which provides effective technical support for digital currency investment.

Traditional music assessment methods rely on manual scoring, which is not only time-consuming but also highly subjective. And modern technology, especially the fusion of computer vision and audio processing, provides a new solution. In this paper, an intelligent performance evaluation system based on the fusion of computer vision and audio data is proposed. The system utilizes a deep learning model for music performance evaluation by combining improved visual feature extraction and audio acoustic feature extraction. In the audio feature extraction part, an improved Gammatone filter and FFT algorithm are used to optimize the audio feature extraction process; in the visual feature extraction part, lip features are extracted using a convolutional neural network (CNN), and sequential processing is carried out by an LSTM network. In order to improve the accuracy of the evaluation, the system also introduces a bimodal feature fusion technique, which further enhances the performance of the model by weighted fusion of audio and visual features. The experimental results show that the model in this paper performs well on the OAVQAD dataset, the training loss has reached convergence after 21 rounds, and the anti-jamming ability in the noisy environment is significantly higher than that of the other comparison models. The character error rate (CER) of this paper’s model is 0.32% in a high-intensity noise environment, which is much lower than that of the traditional model. The model’s pitch features and chord recognition are more excellent, and it can accurately capture the detailed features in music, providing reliable technical support for intelligent performance evaluation.

As the backbone of future basic education, the independent learning ability of English teacher trainees directly affects the quality of teaching and student development. However, the existing cultivation mode lacks personalized guidance, and it is difficult to accurately identify the learning state, resulting in poor cultivation effect. Focusing on the cultivation needs of independent learning ability of English teacher training students, this study constructs a deep knowledge tracking algorithm based on students’ state (DKT-ST) and an open learner model (ALS-OLM) for independent learning services. The model adopts a two-layer memory network structure, introduces the forgetting factor and attention mechanism, stores knowledge concept information through a static matrix, and updates students’ knowledge mastery through a dynamic matrix to realize accurate tracking of learning status. Experiments on three public datasets show that the AUC value of the DKT-ST model on the ASSISTments2009 dataset reaches 0.817, which is a 5.42% improvement compared with the DKT-DSC model, and on the KDDCup2010 dataset is a 2.21% improvement. The results of the teaching experiment show that the average improvement of students’ independent learning ability in the experimental class is 25.61 points, with an increase of 37.20%, which is significantly better than that of the control class, which is 14.89%. The analysis of students’ ability attributes shows that the key ability dimension is improved by 0.46. The study proves that the personalized learning model based on artificial intelligence algorithms can effectively improve the independent learning ability of English teacher training students, and provides technical paths and practical references for the reform of teacher education.

As a core component of terrestrial ecosystems, forests play a crucial role in global ecological balance, climate regulation and biodiversity conservation. In this paper, the spatial distribution characteristics and dynamic changes of forest resources in Jiangxi Province were studied through multi-source remote sensing data. The methodology utilizes spatial distribution pattern analysis, Gini coefficient, kernel density estimation method, and buffer analysis method, and combines with the forest productivity remote sensing model (VPM) to analyze the distribution characteristics of forest resources in the province in depth. The results showed that the forest resources in Jiangxi Province showed obvious geographical differences, and the distribution types in different areas ranged from aggregated to stochastic. According to the analysis, the north and west of Gan are randomly distributed, while the central area of Gan is of aggregated type distribution. The dynamic monitoring results showed that the overall forest area in Jiangxi Province showed an increasing trend between 2014 and 2024, but the rate of forest loss was still high, especially in areas with low rainfall. Specifically, arid regions with an average annual rainfall of less than 400 millimeters have a forest loss rate of 43.81%, whereas wet regions have a lower loss rate of 25.43%. These findings provide important data support for rational conservation and utilization of forest resources.

Non-exchangeable extremely large subgroups have important application value in the study of the nature of groups, which can reveal the intrinsic structure and behavioral patterns of groups. In this paper, a method of calculating the critical index of non-exchangeable extremal subgroups of a finite group is proposed. By introducing the Sylow subgroup of the group and the structural characteristics of the noncommutative extremal subgroup, several theorems and lemmas are proposed to derive the lower bound of the critical exponent. The validity of the method is verified by numerical simulation analysis, and the results show that the number of non-commutative extremal subgroups is significantly affected by the solvability of the group and the informality of the Sylow subgroup in different types of group structures. In the simulation, taking the karate club network as an example, the expectation value of 0.0682 is obtained through 1200 times of random graph simulation calculations, which indicates that the proposed method has high accuracy in practical applications. In addition, for the dolphin group network and the political book classification network, the simulation results also show that the calculation method can effectively reflect the intrinsic relationship and structure of the group. The study shows that the calculation method of the critical exponent of noncommutative extremely large subgroups has potential application value in finite group theory, and can provide an important basis for the classification and structure analysis of groups.

Test anxiety is an important factor affecting the learning effect and psychological health of higher vocational students. This study constructed a test anxiety intervention model based on the combination of cognitivebehavioral theory and neurofeedback technology. Using a 2×2 mixed experimental design, 50 higher vocational students were selected as the research subjects, divided into 25 each of the experimental group and the control group, and a five-week intervention experiment was implemented. The cognitive behavioral changes and EEG activity characteristics of the students before and after the intervention were analyzed through EEG signal acquisition and psychological scale measurement. The study used a portable three-lead EEG acquisition device to collect five types of EEG signals, Delta wave, Theta wave, Alpha wave, Beta wave and Gamma wave, and extracted linear and nonlinear features for difference analysis. The results showed that the experimental group showed significant improvement in cognitive test anxiety, and the group and time-point interactions reached a significant level (p=0.041); subject 1 scores improved significantly, with a significant interaction (p=0.021); obsessivecompulsive symptoms were effectively alleviated, with a significant interaction (p=0.011); and the Beta wave of subjects in the high-test anxiety group was significantly higher than that of subjects in the low-test anxiety group (p= 0.038). The study shows that cognitive behavioral therapy combined with neurofeedback technology can effectively reduce the test anxiety level of higher vocational students and improve their academic performance, which provides a scientific basis and practical guidance for test anxiety intervention.

Healthy development is not only about children’s physical health, but also involves mental health and social adaptability. In the preschool stage, the influence of multiple factors such as family, school and community has a profound impact on children’s health. In this paper, we used PLS-SEM combined with Bayesian network modeling to explore the mechanism of influencing factors on children’s health development in preschool education. The study first constructed a health influence mechanism model, taking into account the influence of multiple dimensions such as community environment, family support and school environment. The data of 600 preschool children were analyzed through a questionnaire survey and structural equation modeling was used to analyze the data. The results showed that four factors, namely, personal characteristics, community environment, family support and school environment, had a significant effect on children’s health, in which the direct effect of school environment was the largest, with a path coefficient of 0.305, and the direct effect of family support was the second largest, with a path coefficient of 0.221.The inference and diagnostic analysis of Bayesian networks revealed that changes in the school environment had a more significant effect on children’s health, especially when the children’s mental health and social adjustment are more negatively affected when the school environment is poor. The study suggests that children’s health development in preschool education is a complex system interwoven with multiple factors, in which the school environment plays a key role and requires special attention.

Currently preschool education faces the challenges of mismatch between traditional teaching mode and modern technology development, and structural contradiction between children’s cognitive development needs and educational resource allocation. The wave of digitization has swept through the global education field, providing a new opportunity to solve the problem of preschool education quality improvement. In this study, the questionnaire survey method and experimental comparison method were used to analyze the effect of digital technology application on 268 preschool children. By constructing a digital technology system containing four dimensions of AR literacy, audio storytelling, doodling programming and number and shape playground, combined with three cognitive development indicators of operational IQ, linguistic IQ and total IQ, Pearson’s correlation analysis and hierarchical regression analysis were used for data processing. The results showed that there was a significant positive correlation between digital technology and children’s cognitive ability, and the goodness of fit of the hierarchical regression model R² reached 0.9673, and the significant P-value of each digital technology indicator was 0.000. Analysis of the five-year longitudinal data showed that the average student score decreased from 87.75 to 64.84 under the traditional mode of instruction, and the standard deviation increased from 6.79 to 17.48, reflecting a significant difference in the effectiveness of the instruction. The study shows that the introduction of digital technology can effectively improve the cognitive development of preschool children, providing empirical support for the construction of a modern preschool education system.

The critical period of children’s early psychological development determines their future cognitive abilities, social skills and emotion regulation levels. In this study, 1430 preschool children from eight kindergartens in Jiangsu Province were selected by whole cluster random sampling method, and data were collected by using the Classroom Environment Scale and the Overall Well-Being Scale, and the mechanism of preschool environment on children’s psychological development was analyzed by combining multiple linear regression algorithms and big data techniques such as redundancy analysis (RDA). It was found that the intimacy of the preschool environment was significantly and positively correlated with children’s social initiative, verbal and nonverbal interaction skills (P<0.01), and the standardized regression coefficient of the emotional expression dimension was 0.124 (P=0.01) and the standardized regression coefficient of the recreational dimension was 0.128 (P=0.023), which were both positively predictive of children's verbal and nonverbal interaction skills. RDA ordination analysis showed that the cumulative percentage of explanation of the first four axes for the relationship between children's psychological development and the preschool environment amounted to 99.88%, with the correlation coefficients of self-concept, family relationships, school hardware, and maskedness with the first axis being -0.4127, -0.366, -0.3272, and -0.3085, respectively. The study showed that several dimensions of the preschool environment had a has a significant effect, in which emotional expression and entertaining environment are important factors to promote the development of children's language interaction skills, and the study provides a scientific basis for optimizing the design of preschool education environment.

In the current higher education environment, the traditional “one-size-fits-all” mode of ideological and political education is difficult to meet the diversified development needs of students. Students in higher education have significant individual differences and diversified ideological concepts, and traditional education lacks pertinence and effectiveness. This study builds a personalized teaching strategy system for ideological and political education in colleges and universities based on big data, and realizes accurate student profiling and classified policy through student information modeling, similarity comparison and K-means clustering analysis. The study collects 5,712 data on the performance of 408 students in the class of 2020 in a college of a provincial university over the past four years, and establishes an index system covering seven dimensions: academics, work, ideology and politics, economy, development, employment, and psychology. The 64 students were divided into five clusters by K-means cluster analysis, and independent samples t-test was used to verify the teaching effect. The results show that: the p-value of students in the experimental class in the six dimensions of healthy life, ecological civilization, patriotism, scientific spirit, social responsibility, civic literacy is 0.000, and the level of ideological and political awareness is significantly higher than that before the practice; the difference between the average score of the experimental class and the control class in terms of interest in learning Civics and Politics course is 8.92 points, and the difference is statistically significant; the profile coefficient converges most closely to 1 when the K-value is 5, and the clustering effect is the best. The study shows that the personalized teaching strategy based on accurate portrait can effectively enhance students’ ideological and political awareness and learning interest, and provides a new practice path for the Civic and Political Education in colleges and universities.

With the development of information technology and big data analysis technology, data mining has been widely used in various disciplines. In this paper, the literary themes and their trend changes in Chinese language literature are explored through big data analysis techniques, combined with the LDA theme model. The study used CiteSpace and other tools to analyze the relevant literature from 2000 to 2023, identify the major literary themes in the field of Chinese language literature, and discuss their trends in depth. It is found that the literary themes of Chinese language literature mainly focus on love, war, social reality, etc., and these themes show different research hotspots and trends in different time periods. For example, the number of literature publications reached 100 in 2020, which is 25 times more than the number of publications in 2000, showing the rapid development of this field in recent years. In addition, keyword analysis shows that “war”, “loyalty” and “society” are the main research concerns. Further, through the co-occurrence analysis of the literature, the changing patterns of literary themes are identified, providing new perspectives and directions for the study of Chinese language literary works. This paper provides strong evidence for academic research on Chinese language literature through data mining methods, revealing the major trends and hot topics in its development.

The process of global economic integration is deepening, and the international trade network presents a complex and diversified development situation. This paper constructs a value chain optimization model of “Belt and Road” foreign trade industry chain based on multi-objective planning algorithm, adopts complex network analysis method to construct trade network, and applies improved multi-objective Gray Wolf optimization algorithm to solve the industrial upgrading path. By analyzing the data of the United Nations Merchandise Trade Database from 2004 to 2024, it is found that the number of nodes in the trade network of “Belt and Road” has increased from 26 to 49, with a growth rate of 88.5%, and the number of network relations has increased from 28 to 107, which shows the trend of trade diversification. The study establishes a three-dimensional objective function system covering economic production benefits, social life benefits and ecological resource benefits, and designs the corresponding constraints. The empirical analysis shows that under the balanced transformation and upgrading path, the optimized value of industrial added value reaches 22456.4 in 2024, the energy consumption decreases to 16515.2, and the pollution emission index decreases from 21104.9 in 2009 to 18556.The study confirms that the multi-objective optimization algorithm is able to effectively achieve the coordinated development of “three-life space” and provide scientific decision support for the upgrading of the foreign trade industry of the “One Belt, One Road”. The study confirms that the multi-objective optimization algorithm can effectively realize the coordinated development of the “three living spaces” and provide scientific decision support for the upgrading of foreign trade industry in the “Belt and Road”.

Mechatronics systems are widely used in modern industries, and with the advancement of technology, the complexity and functional requirements of the systems are increasing. In this study, a vibration suppressionbased adaptive controller (VS-MRAC) is proposed for enhancing the vibration control performance of mechatronic systems. First, an adaptive decomposition method is used to reduce the noise of the vibration signal, and the controller design is optimized by combining the principle of minimum information quantity. The effectiveness of the proposed method in trajectory tracking and vibration suppression is verified through simulation analysis. In the comparison experiments, the VS-MRAC controller is able to significantly reduce the platform amplitude after 7 seconds and reach the steady state within 25 seconds, which provides superior control performance compared to the traditional PD controller. Specifically, the system vibration amplitude is significantly reduced with the proposed controller, and it is more robust to the system model uncertainty. Simulation results show that the VS-MRAC controller is able to reduce the burden on the controller while maintaining high accuracy, and improve the efficiency and safety of the mechatronic system. The method has strong applicability and popularization value, especially in the practical application of high-precision vibration control.

Under the background of Industry 4.0, intelligent manufacturing technology has become the core driving force to enhance the competitiveness of manufacturing industry. Aiming at the problems of low control accuracy and low resource utilization of traditional manufacturing systems in metal material processing, this paper designs an intelligent manufacturing system based on mechatronics technology. The research adopts digital intelligent control algorithm combined with PID control structure and state space equation to establish the system mathematical model, realizes the organic integration of enterprise layer, management layer, operation layer, control layer and field layer through the design of layered architecture, and constructs active real-time production plan management system. The performance test results show that the average memory occupancy rate of the system is 15.205% when processing 2,000 products, which is 12.702% and 18.054% lower than that of the system based on digital twin and cloud computing, respectively; the real-time data acquisition in the functional test is 2.5 seconds, and the accuracy rate of fault diagnosis reaches 98.8%. The study shows that the intelligent manufacturing system based on mechatronics technology is better than the traditional system in terms of resource occupation, processing efficiency and stability, and provides an effective solution for intelligent processing of metal materials.

The traditional distribution network structure is solidified and lacks flexibility, which is difficult to adapt to the new situation of large-scale access of distributed energy and rapid growth of electric load. This study proposes a topology optimization strategy for flexible interconnected low-voltage distribution network based on BTB-VSC, and constructs a mathematical model of distribution network with embedded DC and a VSC converter control system. In the methodology, a two-level voltage source converter is used to realize flexible interconnection between feeders, AC circuit equations and DC circuit equations are established, voltage deviation control and voltage sag control strategies are designed, and BTB-VSC optimization algorithm is developed for topology identification. The feasibility of the algorithm is verified by a 12-node network model and simulated in a large distribution network with 120 users connected to 4 stations. The results show that in a Gaussian noise environment with a signal-to-noise ratio of 20 dB, the BTB-VSC optimization algorithm reduces the number of time sampling points required to achieve 100% accuracy in topology identification from 235 to 200, which provides higher identification accuracy and faster convergence speed than the PCA algorithm. The strategy effectively solves the distribution feeder load imbalance problem, improves the terminal voltage quality, enhances the power supply reliability of the distribution network, and provides technical support for the friendly grid-connection of distributed power sources and the intelligent development of the distribution network.

Chinese national music culture has a long history and contains rich regional characteristics and artistic connotations. Regional cultural characteristics have important value in ethnic piano music creation, but the traditional creation method is less efficient. In this paper, a deep learning-based automatic generation system for ethnic piano music is constructed, which consists of four parts: a score feature extraction layer, a data enhancement module, a Word2Vec-CBOW feature fusion layer and a BiLSTM-CRF fingering generation layer. Specifically, pitch information and velocity information are obtained through score feature extraction, data enhancement is performed by utilizing the symmetry characteristics of the left and right hands, fusion feature vectors are trained using the Word2Vec-CBOW model, and fingering sequence generation is realized by combining the BiLSTM-CRF network. Experimental results show that the system achieves 94.88% accuracy and 92.18% F-value on the MAPS dataset when the task loss weight is 30. The pentatonic scale rate test shows that the pentatonic scale rate of the generated Chinese style piano music in the major key of Celadon reaches 100%, and that of Chrysanthemum Terrace reaches 91.563%. The subjective evaluation experiment in which 40 evaluators evaluated 20 pieces of generated music showed that this paper’s method is superior to the traditional method in terms of coherence and emotional expression. The study shows that the system can effectively integrate regional cultural characteristics and provides a new technical path for ethnic piano music creation.

Chinese national music culture has a long history, and pentatonic modulation, as an important part of traditional music, plays a unique role in modern piano music composition. In this study, a national tuning recognition system was constructed based on the pitch level distribution matrix (PCDM) and temporal adaptive neural network (TANN), and Chinese piano music was spectrally analyzed using the constant Q transform (CQT). The study establishes a Chinese ethnic modal classification system covering 360 combinations by experimentally analyzing eight MIDI samples of Chinese folk songs, including four pentatonic modal and four heptatonic modal works. The experimental results show that the accuracy of the proposed algorithm in recognizing Chinese folk music modes reaches 100%, in which the pentatonic modal works such as “Jasmine Flower” are accurately recognized as G levied pentatonic modes, and the septatonic modal work “It’s a Man of Our Knowledge”, with a 15.69% partiality, is successfully recognized as Yanle G levied septatonic modes. The study proves that the modal detection method based on PCDM features has significant advantages in the analysis of Chinese folk music, and provides an effective technical support for the digitization research of traditional music.

Traditional credit risk assessment mainly relies on expert experience and simple statistical models, which are difficult to effectively deal with complex nonlinear relationships in massive multidimensional data. The unbalanced dataset problem makes the model insufficiently capable of recognizing a few types of defaulting customers, resulting in limited risk control effects. Aiming at the risk prediction problem in financial credit asset management, this study constructs a credit risk prediction model based on extreme gradient boosting (XGBoost). The customer data of a financial institution for the observation period of 2023 and the performance period of 2024 are used, with a total of 124,980 samples, of which 13,000 are positive samples, accounting for 10.40%, and containing 51 variable features. The study adjusts the model parameters by hyperopt Bayesian optimization method, setting learning-rate to 0.06, gamma to 0.04, n-estimator to 4, and subsample to 0.8. An improved unbalanced data processing strategy is used in conjunction with XGBoost algorithm, which optimizes the second-order gradient information and regularization term using the objective function. The experimental results show that the AUC value of this paper’s method reaches 0.8463, which is significantly better than the 0.8324 of the traditional SMOTE oversampling method and the 0.8401 of the simple undersampling method.In the comparison with other machine learning algorithms, the AUC value of XGBoost combined with EasyEnsemble sampling is 0.8472, which outperforms the 0.8159 of the decision tree and the 0.8004 of the support vector machine’s 0.8004.The study verifies the effectiveness of the extreme gradient boosting algorithm in dealing with unbalanced financial data, and provides reliable technical support for credit risk management of financial institutions.

With the continuous changes in the educational environment, the traditional teacher development path model is facing the challenges of multidimensionality and dynamism. The application of artificial intelligence technology, especially machine learning and optimization algorithms, provides new perspectives and methods for teacher development in education. Based on artificial intelligence algorithms, this study proposes a model for optimizing teachers’ career development path. First, a mathematical model of the teacher career development path is constructed by combining the basic information of teachers, work performance and external environmental factors. Genetic algorithm and particle swarm optimization algorithm are used to optimize the path to maximize teachers’ career potential and the utilization efficiency of school resources. In order to further improve the prediction accuracy, ARIMA and LSTM models were combined to model the linear and nonlinear parts of the time series data, respectively. The ARIMA model was used to obtain a smooth sequence through difference processing and to make preliminary predictions on the teacher career development data. Subsequently, the residual data from the ARIMA model was input into the LSTM model to capture the nonlinear trend and achieve a more accurate prediction of teacher career paths. The experimental results show that the ARIMA-LSTM hybrid model has a mean square error (MSE) of 1.0795, a root mean square error (RMSE) of 0.9456, and a mean absolute error (MAE) of 0.4516, which is significantly better than the traditional ARIMA and LSTM models. The optimization model provides new methods and ideas for the scientific planning of teachers’ career development path.

Decompression sickness is a common disorder in special occupational groups such as divers and highland operators, and patients often experience acute onset of illness accompanied by severe psychological trauma and cognitive dysfunction. Methods: Eighty-six patients with decompression sickness were selected and treated with hyperbaric oxygen therapy combined with conventional rehabilitation in a 6-week treatment cycle. The Champion Health Belief Model Scale (CHBMS), Health Promotion Lifestyle Scale (HPLP-II) and Zung’s Anxiety Self-Assessment Scale were utilized to assess the patients’ psychological health status. A quantitative calculation model was established to analyze the changes in psychological indicators before and after treatment, and multifactorial logistic regression was used to analyze the independent factors affecting the efficacy. Results: After hyperbaric oxygen treatment, the patients’ total CHBMS score improved from 46.57 to 80.89, and the total HPLP-II score increased from 76.87 to 115.20. 74 out of 86 patients had excellent efficacy, accounting for 86.05%. Multifactorial analysis showed that age, disease duration, anxiety level, sleep quality, cognitive function and psychological stress were independent influences with ORs ranging from 3.74 to 8.59. Conclusions: Hyperbaric oxygen therapy based on a quantitative computational model significantly improves mental health status and health behavior performance in patients with decompression sickness. Factors such as age over 50 years, disease duration over 7 days, and severe anxiety are the main risk factors affecting psychological recovery. The model provides a scientific basis for individualized psychological recovery programs and helps to improve the precision and effectiveness of treatment.

The current global digital economy is developing rapidly, and all industries are experiencing profound digitalization changes. As a unique national industry in China, the traditional Chinese medicine industry faces many challenges in the modernization process, such as low production efficiency, difficulty in quality control, and fierce market competition. This study uses factor analysis to construct a profitability evaluation system for the traditional Chinese medicine industry, and analyzes the current situation and development potential of the industry’s profitability by establishing a digital twin technology-driven profitability improvement strategy model. The study selected 10 key financial indicators, used KMO and Bartlett’s test to determine the applicability of the sample, and extracted three principal factors for analysis. The results showed that the KMO statistic was 0.613, the Bartlett value was 668.651, the cumulative variance contribution rate amounted to 75.903%, and the first common factor variance contribution rate was 35.99%. The analysis found that the profitability factor of traditional Chinese medicine industry experienced significant fluctuations during 2018-2023, decreasing from the highest value of 32.8 to the lowest value of 5.8.The study shows that digital twin technology can effectively improve the profitability of traditional Chinese medicine industry through the three dimensions of Internet of everything, data integration and cloud platform construction, which provides a scientific theoretical guidance and practical path for the digital transformation of the industry.

As an emerging power system form, the power output characteristics of optical storage-charging integrated microgrids are characterized by significant intermittency and stochasticity, which bring new challenges to the safe and stable operation of distribution grids. Aiming at the problem of evaluating the impact of power fluctuation of optical storage-charging microgrids on the reliability of distribution grids, this paper constructs a full-time simulation and analysis model based on time series data and a hybrid CNN-GRU neural network prediction model. By analyzing different fault scenarios, a microgrid reliability assessment index system is established, and the method of extracting spatial features by convolutional neural network and capturing timing features by gated recurrent unit is applied to realize accurate prediction of PV power. The case analysis based on the IEEE RBTS test system shows that the SAIFI index of the distribution system decreases from 1.6354 to 1.4795 after microgrid access, and the power supply availability rate increases from 0.99936 to 0.99947, which significantly improves the system reliability. The prediction accuracies of the CNN-GRU model are better than those of the traditional methods in all four seasons, and the NMAI indexes of the CNN-GRU model are better than those of the traditional methods, and the NMAI indexes are better than the traditional methods in all four seasons. Attention model with a maximum reduction of 59.2% in NMAE metrics and 45.4% in NRMSE metrics. The results verify the effectiveness of the proposed model in microgrid output fluctuation assessment and prediction, and provide theoretical support for distribution network planning and operation.

Choral works, as an important form of musical expression, have a complex compositional process and require a deep foundation in music theory. The rapid development of artificial intelligence technology provides new possibilities for music creation. In order to solve the problems of low efficiency and limited creative ideas of traditional choral works, this study constructs a choral works creation model based on the combination of deep learning and knowledge graph. Methodologically, the Transformer and BART models are adopted as the core architecture, and the deep mining and representation learning of semantic information of choral works is realized through keyword extended graph generation, TransE knowledge representation encoder and two-layer graph attention network. Specifically, the keyword expansion graph is constructed by utilizing the knowledge map of ancient poems, the keyword semantic representation is enhanced by the multi-head graph attention mechanism, and the dual cross attention mechanism is incorporated in the BART decoder to improve the quality of text generation. The experimental results show that the model achieves 75.4%, 76.3%, and 78.4% in terms of accuracy, recall, and F1 value, respectively, which significantly outperforms baseline models such as SVM, BiLSTM, and TextCNN. The model achieves convergence at about 460 training sessions, and the convergence speed is significantly faster than the comparison model. The application practice shows that the average score of students’ choral works in the experimental group reaches 86.24, which is 5.58 points higher than that of the control group. The study shows that the model can effectively support the creation of multi-topic choral works, which provides a new technical path for the intelligent development in the field of music creation.

Modern architectural space is increasingly demanding for acoustic quality, and a good acoustic environment not only relates to the comfort of users, but also directly affects the effect of spatial function. In this paper, to address the issue of the influence of sound-absorbing material configuration on the sound field uniformity in the indoor acoustic environment, using the quadratic residual diffuser (QRD) material, through the establishment of the reverberation laboratory model, the design of seven different material arrangement combinations of the program, the use of VA-Lab6 acoustic test system to measure and analyze the sound pressure level, reverberation time and other key acoustic parameters. The results of the study showed that in the horizontal alternating arrangement scheme of the side walls, combination mode 3 achieved a total score of 38 in the loudspeaker source condition and 34 in the normal male voice condition, both of which were the highest scores; When the rear wall was arranged vertically and alternately, combination mode 4 scored 37 points in the loudspeaker condition, the best performance; The modal density analysis shows that the modal density is 37 in the 500Hz band and reaches 75 in the 1000Hz band, which meets the sound field diffusion requirements. The quantitative assessment of sound field uniformity reveals that the reasonable configuration of sound-absorbing materials can significantly improve the uniformity of indoor sound field distribution, which provides a scientific basis and practical method for the artistic design of indoor acoustic environment.

Driven by the green building concept, assembly building technology has gradually become an important direction for the transformation and upgrading of the construction industry. At the same time, the development of construction waste recycling technology provides a new way to solve the problem of construction waste disposal, and the application of recycled materials in the assembly building system can not only reduce the construction cost, but also reduce the environmental load and realize the goal of sustainable development. In this paper, a synergistic efficiency system of low-carbon assembled toilet system is constructed by combining the standardized modular design of integrated smart toilets with the application of recycled concrete. The study used ABAQUS finite element software to establish a numerical simulation calculation model for the seismic performance of the energy-consuming combination wall, analyzed the performance of recycled concrete using the concrete damage plasticity model, and realized standardized production through the principle of modular design. The test results show that the ultimate load of the roof cover reaches 109.72kN, the initial stiffness is 4.5kN/mm, and the energy-consuming capacity of the KJ-6 node is 12.2 times higher than that of the empty steel frame. The technical and economic analysis shows that the total cost of Project A with recycled concrete composite thermal insulation blocks is 1055.14 yuan/m², which is 48.6% lower than that of traditional concrete Project B of 2054.72 yuan/m², and the thermal insulation cost only accounts for 1.36% of the total project budget. The study shows that the modular design of the integrated smart toilet combined with the application of recycled concrete significantly reduces the construction cost and improves the efficiency of resource utilization under the premise of ensuring the structural safety performance, which provides technical support and economic feasibility verification for the development of low-carbon assembled buildings.

With the continuous development of the movie industry, the problem of restoring old movie films has become an important issue. Especially for damaged old movies, restoring their picture quality through modern technological means has become an important way to enhance the audience experience. This study proposes a method for restoration of old movie films based on recurrent multiscale convolutional mixer, which aims to effectively improve the image quality of old movie films using artificial intelligence algorithms. Methodologically the study uses a recurrent neural network (RTN) combined with a convolutional mixer network to cope with the common damage problems in old movie films. In the experiments, a self-constructed dataset of old movie films is used and restoration tests are performed on the 1996 movie Titanic and the 1980 movie Love of the Sea. The experimental results show that the method has a PSNR value of 23.22 and an SSIM value of 0.47 in the restored images, demonstrating a significant image quality improvement compared to traditional restoration methods. The temporal brightness curve of the restored movie changes smoothly, the flicker phenomenon is effectively suppressed, and the restoration effect is more natural. In addition, the performance of this method is also better than the comparison method in terms of hyper-segmentation and plaque removal. The study shows that the restoration method based on cyclic multiscale convolutional mixer shows strong potential for application in restoration of old movie films, especially in improving the clarity and naturalness of the picture.

The globalization process promotes the rapid development of international Chinese language education, and the cross-cultural communicative competence of teaching personnel has become a key factor affecting the quality of teaching. This paper constructs a natural language processing-based assessment model of international Chinese teaching talents’ intercultural communicative competence. Methodologically, Python crawler technology is used to collect online comment data from Chinese university MOOC platform, and text mining is applied with LDA topic modeling to construct an evaluation system containing three primary indicators of language skills, cultural understanding, and communicative strategies and 10 secondary indicators. Then, the weights of the indicators are determined by the AHP-CRITIC game theory combination assignment method, and the evaluation model is established by combining the theory of mixtures elemental topable. The results show that listening comprehension has the highest weight of 0.145, cultural sensitivity has the highest weight of 0.163, and the dimension of communicative strategy has the highest importance among the three first-level indicators. Taking A international school as an example for empirical analysis, the overall correlation is -0.2751, and the assessment grade is “good”. The assessment model constructed in this paper takes subjective and objective factors into consideration, realizes the quantitative assessment of intercultural communicative competence, and provides a scientific basis for the cultivation of international Chinese teaching talents.

The development and update iteration of technology provide new possibilities for the integration of embedded systems and digital media. In this paper, on the overall design of the convergence hardware, we propose a hardware framework for embedded multimedia player composed of three main parts: hardware system, embedded real-time operating system and application program. And the basic programming model of computing based on service body-execution flow is used to accelerate the data transfer and cooperative work between processors and improve the overall performance of the embedded system. The streaming data transmission scheme of basic queue is introduced to construct the queue-based streaming data transmission framework by real-time parsing and caching multi-channel data to accomplish the multi-channel streaming transmission task. As a result, the overall design of the embedded system and digital media convergence framework model is completed. The design method is applied to construct the hardware dynamic power consumption model of embedded system and digital media, and the temperature fluctuates in a very small interval at about 42°C and the voltage is maintained at about 1.42V in the actual operation of the model. It shows that the proposed embedded system and digital media integration framework modeling method has high feasibility.

Taking digital technology as the core driving force, this paper systematically explores the innovative application of MIDI technology and computer music tools in classroom teaching, and proposes an intelligent retrieval framework based on semantic features and data mining and a resource sharing incentive model, aiming to optimize the efficiency of teaching resource allocation. The study realizes accurate retrieval of educational resources by constructing a music resource data extraction model and a semantic feature distribution structure model, combined with weight division and ontology library matching. The average retrieval time of the semantic feature extractionbased retrieval method for 10 high-frequency keywords is only 30.5ms, which is 75.9% and 55.9% shorter than that of the traditional method (126.4ms) and meta-search engine (69.2ms), respectively. The average optimization of the retrieval path length is 22.6%, which verifies its high efficiency in the massive data environment. Further testing of the system load performance by JMeter tool reveals that in the scenario of 4000 concurrent users, the throughput (TPS) of resource uploading and downloading reaches 268 items/s and 403 items/s, respectively, but the performance decreases after the number of users exceeds 4000, which suggests that it is necessary to optimize the allocation of server resources. Resource sharing tests for different file sizes show that IPFS latency increases significantly with file size, with upload/download latency of 0.24s/0.42s for 1MB files and 16.30s/27.64s for 150MB files, indicating that IPFS+blockchain architecture is more suitable for small and medium-sized file sharing. In addition, blockchain transaction latency is higher for resource uploading, e.g., 100MB file takes 11.58s, while download latency is mainly dominated by IPFS transmission, verifying the complementary advantages of the technical architectures.

In this paper, we propose a knowledge base Q&A system based on retrieval-enhanced generation (RAG) technology, which significantly improves the semantic comprehension ability and generation quality of the system by integrating bi-directional gated recurrent units (BiGRUs), neural state machines (NSMs), and hybrid RAG indexing optimization methods. To address the gradient decay problem of traditional recurrent neural networks (RNNs), BiGRU extracts deep features of the text through a bidirectional information transfer mechanism, while NSM simulates human causal thinking through probabilistic scene-graph reasoning to enhance the interpretability of the model. The hybrid RAG strategy further combines local knowledge base construction, context fusion and cue sample introduction to achieve dynamic knowledge enhancement. For indexing optimization, the HNSW-PQ composite indexing technique is adopted to significantly reduce the latency and storage overhead of highdimensional vector retrieval. In the complex knowledge base quiz tasks (CWQ and WebQSP), the model F1 scores reach 70.34% and 83.31%, Hits@1 were 73.18% and 85.33% respectively, which are fully ahead of the traditional methods and pre-trained models. The experimental results show that through the semantic reasoning capability of BiGRU-NSM, the dynamic knowledge enhancement of hybrid RAG and the efficient retrieval of HNSW-PQ indexing, the system achieves breakthroughs in multi-hop reasoning, complex semantic parsing and generative accuracy, and provides an efficient, interpretable and adaptable solution for knowledge base quiz tasks.

Based on the theory of persuasion, this paper analyzes the application of persuasive design on products. It constructs a product emotion modeling database and optimizes the construction of product design model by combining orthogonal design method. The improved SO-PMI algorithm is proposed, and the SO-PMI algorithm is extended to enhance the emotion recognition ability in Chinese context. Adopt stratified sampling and domain equalization strategy for systematic collection, and construct a hybrid emotion corpus covering 2 domains. For the electronics dataset, the model in this paper gains improvement in precision, recall, and F1 value metrics, which are 0.6693, 0.7011, and 0.6848, respectively.For the home furnishing kitchen and bathroom dataset, the model in this paper also achieves the best performance. The proposed model demonstrates significant advantages on all three sentiment categories of the electronics dataset. The accuracy of 99% is achieved in the Home Kitchen and Bath dataset, which is 12.6% better than the original model.

The development of online education has continuously promoted the level of intelligent analysis of foreign language teaching videos. This paper proposes a content comprehension analysis method for foreign language teaching videos that integrates spatio-temporal residual attention network (STRAN) and improved text rank algorithm (TextRank). The spatio-temporal features in the video are extracted by three-dimensional residual network (3DResNet), and the attention mechanism is combined to optimize the recognition accuracy of students’ classroom expressions. Improved TextRank algorithm is utilized to filter the keywords of teaching focus and integrate multisource data and storage technology to construct a knowledge graph of foreign language teaching. The results show that the recognition rate of all 6 kinds of student expressions of STRAN is more than 0.930. The 3 keyword extraction performance indexes of Improved TextRank are all greater than 90.00% in different numbers of keyword extraction. Comprehensively applying the method of this paper to assist teaching, the students’ concentration and excitement expression scores are more than 90 points, and the scores of confused, distracted, nervous, and sleepy expressions are around 70-85 points.

In the transformation of government functions to “service-oriented” construction, administrative efficiency is always the focus of the digital government project. This paper takes the collection, mining and integration of event data as the path to improve administrative efficiency under the digital government platform. The government event collection channels are sorted out and an event distribution model is formed to build a digital platform for government event distribution. Subsequently, the key technology framework for multi-source heterogeneous data fusion of sensitive data is established by cleaning inferior data and standardizing storage of integrated data. For the multi-source heterogeneous data provided by the technical framework, a heterogeneous database log parsing algorithm is proposed to meet the demand for change log data capture from multiple heterogeneous databases for the daily operation of government affairs. After completing the data preparation, a multi-source heterogeneous data mining model is constructed to carry out multi-source heterogeneous data mining based on fuzzy C-mean clustering to realize the deep mining of multi-source heterogeneous few class data sets. Compared with similar model algorithms, the data screening time of the multi-source heterogeneous data mining model is always under 20s, which assists in improving the administrative efficiency of the digital government platform with superior data processing speed.

Since the implementation of the policy of college enrollment expansion, the number of college students has increased dramatically, with significant individual differences and mental health problems becoming more and more prominent. At the same time, society’s demand for talent quality has been rising, and academic performance has become an important indicator for measuring the effectiveness of education. This paper analyzes the interaction mechanism between the two systems by constructing an evaluation model of the coupling and coordination of college students’ mental health and academic performance. The study used hierarchical analysis to determine the weights of evaluation indexes, and analyzed the data from 2014 to 2023 using the coupled coordination degree model. The mental health evaluation subsystem contains 4 primary indicators and 12 secondary indicators of social factors, self-factors, school factors, and family factors, with weights of 0.2667, 0.2488, 0.2724, and 0.2121, respectively. The academic performance evaluation subsystem covers 3 primary indicators and 13 secondary indicators of self-factors, family factors, and school factors, with weights of 0.3513, 0.2974, 0.3513.The results of the study showed that the degree of orderliness of the two systems fluctuated between 0.2018 and 0.6417 during the ten-year period, the degree of coupling ranged from 0.2530 to 0.5951, and the degree of coupling harmonization varied within the interval of 0.2563 to 0.5956.The lowest degree of harmonization was found in 2015 and 2018, with the values of 0.3918 and 0.2563, respectively. Both were in a state of heavy dissonance. Overall, the coupling coordination degree of college students’ mental health and academic performance is basically in the range between severe dysfunction and mild dysfunction, and the level of coordinated development of the two systems needs to be improved.

Root rot of tea tree is one of the major diseases affecting tea production and poses a serious threat to tea yield and quality. In order to investigate the screening of biocontrol microorganisms in tea plantation soil and their inhibitory effects on tea root rot, this study was conducted to isolate and characterize biocontrol microorganisms from tea plantation soil, and to screen Z-1 strain and evaluate its inhibitory effects on tea root rot. The study examined the inhibitory ability of Z-1 strain against the pathogenic fungi of tea tree root rot using plate standoff assay and found that its inhibition rate was as high as 42.9%. Further field trials showed that the Z-1 strain had a positive effect on tea tree growth, with root length, plant height and fresh weight showing significant improvement. Compared with traditional chemical control methods, Z-1 strain showed comparable control effects and lower toxicity. The study suggests that Z-1 strain has strong potential for disease resistance and can provide new ideas and methods for green control of root rot of tea tree.

With the increasing global energy demand and environmental problems, energy storage technology is gradually playing an important role in modern power systems. In this study, an optimal scheduling strategy based on genetic algorithm for grid-side energy storage system to participate in regional power trading is proposed. The strategy optimizes the charging and discharging decisions of the energy storage system by genetic algorithm to maximize the revenue. First, an optimization model of the energy storage system is constructed, including power trading, charging and discharging constraints of the energy storage device and the prediction of market electricity price. Then, the multi-objective optimization problem is solved by genetic algorithm to optimize the charging and discharging decisions, and applied to the actual scheduling of grid-side energy storage system. The experimental results show that the energy storage system is able to realize high returns under specific electricity price and load demand conditions. In the specific arithmetic analysis, the improved NSGA-II algorithm is used to optimize the scheduling of the IEEE 33-node power system. The optimized grid-side energy storage system achieves better returns in several time periods, including renewable energy consumption and peak-valley arbitrage in the 14:00-22:00 time period. Under different tariff fluctuation conditions, the energy storage system is able to efficiently dispatch resources, reduce operating costs and enhance market competitiveness. Through further optimization, the scheduling performance of the system is significantly improved and has strong application prospects.

As an emerging energy aggregation management technology, virtual power plant can integrate multiple distributed resources such as wind power, photovoltaic and energy storage. Aiming at the multi-objective optimization and uncertainty handling problems in the participation of virtual power plants in the auxiliary services of regional power grids, the study proposes a virtual power plant scheduling method based on multi-objective dynamic planning. The method constructs a virtual power plant model containing micro gas turbine, energy storage unit, wind power generation and demand response, establishes a multi-objective optimization function with the objective of minimizing the operating cost, and solves it with an improved particle swarm algorithm. Through the simulation verification of the IEEE33 node distribution system, the results show that, compared with the traditional peaking method, the proposed method is able to reduce the load variance from 2251kW to 2071kW, the operating cost from 41845.3 yuan to 39,574.63 yuan, and the network loss from 3041kW to 2711kW. In the analysis of the different confidence levels, the system operating benefit reaches 16753 when the confidence level is 0.87 When the confidence level is 0.87, the system operation benefit reaches 16753.68 RMB, while the benefit is 16148.47 RMB when the confidence level is 0.97, which verifies the negative correlation between the confidence level and the operation benefit. The research results show that the multi-objective dynamic planning method can effectively improve the economy and operational stability of virtual power plants, which provides theoretical support for virtual power plants to participate in the electricity market.

With the continuous development of groundwater remediation technologies for contaminated sites, selecting appropriate remediation technologies and evaluating their effectiveness are the keys to achieving water resource protection and sustainable utilization. In this paper, the groundwater remediation process was simulated by the finite difference method, and the effect of remediation by permeable reaction wall (PRB) coupled with pumping-injection hydraulic control technology was evaluated. The study used the steady flow model and solute transport model to simulate the groundwater flow field and contaminant migration process in groundwater. Before and after the remediation, the pollutant concentrations of Fe and Mn changed significantly, with the maximum concentration of Fe decreasing from 0.351 mg/L to 0.011 mg/L before remediation, and the maximum concentration of Mn decreasing from 0.239 mg/L to 0.007 mg/L, which complied with the groundwater standards, respectively. By modeling contaminant removal under eight scenarios, the results showed that the coupled remediation technology increased the removal rate by 3.96 to 4.99 times compared to the single PRB remediation technology. The removal efficiency reached up to 80.25% in all scenarios. Therefore, the coupled remediation technology has significant effect in groundwater remediation of contaminated sites and provides effective technical support for practical application.

Soil pollution control is a key area in environmental protection, and with the acceleration of urbanization, the demand for remediation of polluted land is increasing. This paper investigates multi-objective optimization algorithms in soil pollution control and proposes a hybrid multi-objective genetic algorithm combined with local search for remediation path design. The method includes a global search using NSGAII algorithm combined with HCS local search strategy to improve the quality of search efficiency and remediation. The study constructed a multi-objective optimization function for economic and environmental benefits based on different land types and restoration technologies, and optimized the decision-making by simulating the restoration costs and land benefits under different scenarios. The results show that during the restoration process, the unit land benefit increases and the restoration cost decreases gradually with the increase of volume ratio. The unit restoration cost when maximizing the land benefit is 0.05 million yuan/m², while the unit land benefit can reach 18,200 yuan/m² when minimizing the restoration cost. The optimized scheme significantly improved the economic benefits and effectively reduced the environmental impacts, providing a scientific basis for the remediation and redevelopment of contaminated soil.

Music composition under traditional culture not only involves complex artistic expression, but also needs to be combined with modern computational methods to improve the efficiency and quality of composition. This study proposes a music composition model based on the Improved Multitrack Sequence Generation Adversarial Network (RFGAN), which aims to improve the quality and coherence of the generated music. The model optimizes the music generation process by introducing a loop-structured generator and timing model, combined with a discriminative feedback mechanism inside and outside the tracks. Comparative experiments were conducted to evaluate the model’s note prediction using Top1, Top2 and Top3 accuracies, and the results showed that RFGAN achieved the highest 88.79% Top1 accuracy in note prediction. To further validate the effectiveness of the model, the study also used a twelve-mean rhythm comparison, and the results showed that the generated note distributions were similar to the real music data, indicating that the model was able to capture the regularity of the music. In addition, the music generated by the model also outperforms the traditional GAN and BiGRU models in terms of harmony, rhythm, and overall effect, verifying its advantages in music composition.