Panoramic Assessment Method of Substation Equipment Health Status Based on Multisource Monitoring and Deep Convolution Neural Network under Edge Computing Architecture

Abstract

In view of the low efficiency of the traditional manual evaluation method of substation equipment status under the background of complex environment, a panoramic evaluation method of substation equipment health status based on multisource monitoring and deep convolution neural network under edge computing architecture is proposed. Firstly, a panoramic sensing system for substation equipment is built based on edge computing, and an edge computing server is deployed in the substation to process the massive data obtained from multisource monitoring nearby. Then, the improved YOLOv4 network is used to detect the equipment state in the substation, in which the Squeeze-and-Excitation attention module and deep separable convolution are used to optimize the YOLOv4 network. Finally, based on the status image of substation equipment, the health status of equipment is evaluated on the panoramic platform of substation combined with the characteristics of multisource data, and four states are divided according to the evaluation criteria. Based on the selected dataset, the experimental analysis of the proposed method is carried out. The results show that the index values of accuracy, recall, and mean precision are 91.53%, 93.07%, and 92.28%, respectively. The overall performance is better than other methods and has certain application value.