Research on Prediction and Identification Method of Latent Fault Characteristic Gas in High Voltage Switchgear Based on Neural Network

With the development of uHV projects in China, the construction scale of power grid is gradually increasing. High voltage switchgear has been widely used in substations and power industry, which has gradually become a key part of the power system. In order to effectively identify the latent fault in the switchgear, a latent fault test model based on machine learning is designed. Get training samples to establish a BP neural network with machine learning ability, and the network is optimized through genetic algorithm to establish the mapping relationship between the characteristic gas and the latent fault in the switchgear. The test results show that the optimized network can accurately identify the potential faults in the switchgear, which plays a very important role in improving the safe operation level of the high-voltage switchgear.