The current educational landscape is undergoing subtle changes, and smart classrooms urgently require upgrades and innovations. This study takes School A, which is undergoing educational reforms, as an example. First, using a weekly time unit, we collected a series of data on students’ consumption behavior (including consumption frequency, consumption time, and consumption amount), online behavior (including online frequency and online time), and daily routines (including start time of activities and preparation time for rest) over two semesters, thereby constructing a multi-source dataset of students’ daily behaviors. The data is preprocessed and corresponding features are extracted. Then, a multi-feature combination-based online course performance prediction model is proposed. Finally, a student academic performance management system based on multi-source information fusion technology is constructed. Through experiments, the accuracy rate of the student performance prediction model based on feature combinations can reach over 90%. The student academic performance management system constructed based on this model is accepted by most students and can significantly improve student performance compared to traditional models. Additionally, applying multi-source information fusion technology to teaching, learning, and management processes can promote the joint development of schools and students.
In recent years, China’s high-speed train sets have developed rapidly, with numerous trains operating across the nation’s “Eight Vertical and Eight Horizontal” high-speed rail network, posing significant challenges for train operation, maintenance, and inspection. This paper establishes a defect detection plan based on subway bogie inspection standards, conducting inspection and analysis on components such as the axle box front cover and anti-roll torsion bar. Following the general bogie defect detection process, the inspection plan designed by the operation depot is implemented. A defect detection model is constructed using a convolutional neural network (CNN) algorithm. Through a three-stage detection process, defect detection in the bogie region is accomplished. Performance metrics are used to quantify the model’s performance on the test dataset. On the test set v2, the predicted values for the three metrics—MAE, MAPE, and RMSE at a 5-step length—are 1.1125, 3.0421, and 1.9866, respectively, outperforming other models. Simultaneously, the model maintains tracking sensitivity above 90% during train emergency braking scenarios, demonstrating its high prediction accuracy.
As the Belt and Road Initiative gains momentum, the real estate industry urgently requires internationally oriented, specialized English professionals in its globalization process. Optimizing university English talent cultivation models is therefore paramount. Based on ESP teaching principles, this article constructs an ESP curriculum model for the real estate sector. After analyzing challenges in cultivating English talent for the industry, it introduces an output-oriented approach to establish a university English talent development model utilizing the Yu Classroom platform. To validate this teaching model’s effectiveness, a pedagogical experiment was conducted with two real estate major classes at a university. Results revealed that after the experiment, the mean vocabulary diversity in essays of the experimental group increased from 14.211 to 22.357. An independent samples t-test showed t=3.971 (P=0.018<0.05), with vocabulary density also significantly improving (P<0.05). Based on this teaching model, students’ reading and writing scores, as well as oral expression abilities, showed significant improvement (P<0.05). The average satisfaction score for the teaching model reached 4.26 points. Therefore, in the international development of the real estate industry, giving full attention to ESP course design and output-oriented approaches can significantly enhance the quality of English talent cultivation for the real estate sector.
Against the backdrop of the digital economy, this study examines rural areas across five northwestern provinces of China. It constructs an evaluation framework for rural financial services based on four dimensions—financial service penetration, usage, coverage, and innovation capacity—using 2024 data. Principal component analysis is employed for measurement. Empirical findings reveal that regions with higher rural economic development generally exhibit stronger overall economic growth. Moreover, rural economic growth stimulates the development of rural financial institutions. Financial institutions in economically advanced regions gain greater capital and motivation to optimize and upgrade financial services, thereby enhancing their own economic development. Concurrently, robust institutional capital, comprehensive and extensive financial service infrastructure, superior financial products, and efficient deposit-to-loan conversion rates collectively elevate the region’s financial service capacity, quality, and efficiency, ultimately driving economic advancement. Furthermore, theoretical analysis indicates that financial service innovation significantly promotes rural economic development, with the extent of this promotion varying depending on the specific rural industries empowered by such innovation.
The digital economy not only injects new momentum into regional economic growth but also serves as a crucial catalyst for advancing high-quality regional economic development. This paper identifies the direct effects of the digital economy on regional economic coordination and its indirect effects in promoting technological innovation to support highquality regional economic development. Based on this framework, data from multiple provinces in China spanning 2015–2023 were selected. Comprehensive empirical tests were conducted using bidirectional fixed effects models, mediation effect models, and geographic detector models. The results indicate that the digital economy significantly promotes high-quality regional economic development, with pronounced regional heterogeneity—particularly benefiting central, western, and southern regions. Technological innovation serves as a key mediating pathway for digital economy-driven development, and fostering interregional innovation cooperation further advances coordinated high-quality regional economic development. This study provides an innovative theoretical foundation for cross-regional digital economy collaboration.
To address rescue challenges caused by material shortages, traffic disruptions, and complex terrain in disaster zones, this study proposes a tethered turbojet UAV rescue system based on an improved active disturbance rejection control (ADRC) method. The UAV employs a tethered power supply to sustain engine operation, enabling precise nozzle deflection and thrust control for enhanced endurance, stability, and payload capacity. A nonlinear dynamic model is established, and an enhanced ADRC is designed by optimising the extended state observer (ESO) with a klnfal function, with its convergence proven via Lyapunov stability theory. Simulations show that under strong disturbances, the improved ADRC reduces attitude control mean square error by 75.81% compared to PID and conventional ADRC, significantly improving robustness for precise material delivery.
Fine-grained image classification (FGIC) is increasingly important in computer vision, driven by its extensive use across various domains. However, existing methods often struggle to achieve both discriminative feature representation and semantic consistency, primarily due to subtle inter-class differences, complex object structures, and background clutter. To tackle these issues, this paper proposes a innovative framework named CARE-Net (Cross-level Adaptive Recalibration and Enhancement Network), which enhances feature learning through three synergistic mechanisms: multi-scale fusion, crosslayer guidance, and explicit feature reconstruction. Specifically, CARE-Net extracts multi-scale features from different semantic levels and employs a guidance enhancement module to recalibrate shallow features using high-level semantic cues. A lightweight attention-based module is then introduced to adaptively fuse features across scales, reinforcing responses in key discriminative regions. Finally, an auxiliary reconstruction branch is incorporated to enforce structural consistency across semantic layers under supervision. Experimental results on the CUB-200-2011 and Stanford Dogs datasets show that CARENet achieves Top-1 classification accuracies of 76.3% and 75.8%, respectively, outperforming several mainstream baselines. Ablation experiments provide further evidence for the effectiveness and complementary nature of each module. These results demonstrate that CARE-Net provides an efficient and interpretable solution for FGIC in complex visual environments.
In the digital age, higher education is undergoing significant transformations, with textbook digitization becoming a crucial focus. The “Outline of the Education Power Construction Plan (2024-2035)” underscores the urgency and importance of digital and intelligent transformation in textbooks. By analyzing the results of the National Outstanding Textbook Selection, this study explores the disciplinary distribution of digital textbooks in Chinas higher education system. The research reveals notable heterogeneity in award-winning digital textbooks across disciplines: medical and natural sciences demonstrate higher award rates, while humanities and agricultural sciences lag behind. Additionally, there are distinct preferences in award-winning formats—natural sciences favor “paper + online courses”, whereas humanities and social sciences prefer “paper textbooks with digital teaching resources”. The analysis suggests that disciplinary characteristics, developmental needs, and technological advancements are key factors influencing award distribution. Therefore, the development of digital textbooks in higher education requires differentiated strategies: tailoring initiatives to disciplinary features, establishing discipline-oriented standards, enhancing resource support, improving evaluation mechanisms, elevating textbook quality, promoting interdisciplinary collaboration, and stimulating collaborative innovation.
Climate change exacerbates global wildfire risk, particularly in the wildland–urban interface (WUI), which constitutes a zone of heightened wildfire susceptibility. Understanding the spatiotemporal evolution patterns and fire vulnerability characteristics of the WUI is therefore crucial for regional ecological security and sustainable development. This study focuses on the impacts of wildfires on WUI ecosystems and human health. To assess these impacts, we developed a comprehensive fire vulnerability index. This index provides a framework for systematically evaluating the spatiotemporal evolution of China’s WUI and its associated fire vulnerability patterns under various climate change scenarios. Climate change will significantly influence the future spatiotemporal evolution of China’s WUI. Under the SSP1-2.6 and SSP2-4.5 scenarios, the national WUI area exhibited a significant decreasing trend, with rates of 8030 km² per decade (p < 0.01) and 7060 km² per decade (p < 0.01), respectively. By the end of the 21st century, China’s WUI is projected to shift eastward. Concurrently, the boundaries between forests and urban areas are expected to become increasingly diffuse, and anthropogenic influences on surrounding forest areas will intensify. The WUI fire vulnerability is highest in southern China, particularly in Southwest (0.27–0.31), whereas northern China presents relatively lower values. Under the high-emission scenario SSP5-8.5, the national WUI fire vulnerability showed a significant increasing trend (0.00016 per year, p < 0.01), with particularly pronounced increases in Southwest and Northwest China. Both the spatial distribution of WUI areas and their fire vulnerability exhibit significant regional variations and scenario dependence under climate change. High-emission scenarios increase China’s WUI fire vulnerability, thereby exacerbating the risk of fire-related losses across regions. This study elucidates the complex impacts of climate change on China’s WUI and its fire vulnerability, providing a crucial reference for developing targeted fire prevention strategies across different regions.
The accuracy of underwater garbage detection and identification plays a very important role in improving the garbage cleaning work carried out by underwater robots. Based on this, this paper proposes an improved underwater trash detection network model based on YOLOv5s. In order to improve the recognition performance of underwater garbage images, this paper also proposes a weighted fusion-based underwater image enhancement algorithm, which fuses the CLAHE algorithm and Retinex algorithm on the basis of weighted logarithmic transformation and adaptive Gamma correction, so as to improve the quality of underwater garbage images. For underwater garbage detection, GhostNet is introduced to improve the backbone network of YOLOv5s to enhance the feature extraction capability, and combined with the ECA attention mechanism and CARAFE up-sampling mechanism to further realize the model lightweighting and enrich the features and semantic features. The results show that the YOLOv5s-G-E-C model improves the detection average accuracy from 60.92% to 86.19% and the model computation reduces the model computation from 18.42 GLOPs to 15.78 GLOPs compared to the YOLOv5s model.It is feasible to apply the improved YOLOv5s model to underwater garbage detection with better detection performance.