Traditional power system faces problems such as data dispersion, low monitoring efficiency, and insufficient prediction accuracy. The development of 3D visualization technology and artificial intelligence provides new ideas for grid operation status monitoring and fault prediction, and the realization of intuitive display and intelligent analysis of grid information has become a demand for the development of the industry. In this paper, a grid panoramic display platform is constructed based on the information integration method of SOA architecture, and a grid operation state monitoring and prediction model is designed by using convolutional neural network (CNN), which is combined with three-dimensional visualization technology to realize smart grid monitoring. In the data processing stage, the data quality is ensured by pre-processing steps such as denoising, normalization, normalization, etc. The CNN model contains an input layer, two convolutional layers, an activation layer, a pooling layer, a fully-connected layer, and an output layer, which realizes real-time monitoring and prediction of power parameters. The results show that during the fault time period (17:35-18:00), the average prediction absolute error of active power of the CNN method reaches 0.917, and the relative error absolute value reaches 151.13%, which is significantly higher than that of the time series method. The platform performance test shows that when the number of concurrency is 100, the dataset throughput rate reaches 315.5 bit/s, and the response time is 355.7 s. The system successfully recognizes the distribution pattern of high and low-frequency events in the practical application of X city and J city. The conclusion shows that the system realizes the effective integration and display of grid information, improves the accuracy of fault prediction, and provides reliable technical support for the intelligent management of power grid.
