中文摘要: In view of the lack of effective fault early warning and prediction methods for substation equipment, this paper proposes Pearson correlation analysis, one-way ANOVA, statistical comparison and other methods to analyze the correlation of various substation equipment patrol data, and then establishes a logistic regression fault prediction model to realize the advanced early warning of equipment status. The classification depth learning neural network model based on depth learning can accurately judge the possible fault types of early warning equipment. Finally, the substation intelligent operation and maintenance assistant decision-making big data platform is built to visualize the early warning model of substation equipment failure; the risk matrix analysis method is used to manage and control the equipment at different levels, to assist in optimizing the operation and maintenance strategy, so that the important equipment can get the key operation and maintenance, and realize the differential operation and maintenance of substation equipment. A large number of practical tests show that the accuracy of this method reaches more than 90%, which is about 20 times higher than the previous manual experience. The equipment failure rate is reduced by 75% and the equipment availability coefficient is 99.988%. This method greatly improves the equipment health level and the equipment operation and maintenance efficiency.