Research on Fault Diagnosis of Switchgear Contacts Based on BP Neural Network

The switchgear is the key equipment in the power system. The overheating of the switchgear contacts is the main cause of the failure of the switchgear. Since the switchgear is relatively closed so that the internal contacts cannot be directly measured, which need to push out the contact status through a portion that is easy to measure. Therefore, this paper uses the collected temperature of the infrared window to push out the contacts temperature based on the BP neural network to determine whether the switchgear contacts are faulty. Firstly, the ANSYS Workbench finite element software is used to simulate the switchgear and obtain the contacts temperature and infrared window measurement temperature as the BP neural network test data. Then, the BP neural network model is trained to determine the model parameters. Finally, the trained BP model is used to test and determine the accuracy of the fault diagnosis of the switchgear contacts based on the BP model. The experimental results show that the maximum error between the expected data and the predicted data based on the BP neural network model does not exceed 1. 5 °C, which is within the normal range. Therefore, based on this model, the operation and maintenance staff can acquire the contacts state of the switchgear by measuring the infrared window temperature. It is of great significance to the safe operation of the switchgear.