Intelligent fault diagnosis of high voltage circuit breaker coils based on multiphysics simulation

High voltage circuit breakers (CBs) play an important role in power grid stability and fast clearing faults on this grid. Therefore, real-time monitoring of these critical components is necessary to prevent possible failures and uundesirable interruptions on power grid. In this paper, the possibility of monitoring of HVCBs as well as fault diagnosis and classification are investigated using artificial intelligence. Accordingly, failure prevention is possible, therefore maintenance costs will be reduced. Here, monitoring of CBs has been implemented through condition-based maintenance strategy (CBM) using close/trip coil current signal (CC) of a real 72.5 kV CB, which helps to detect and predict faults in different parts of CB such as supply voltage, coil winding, latch and auxiliary contacts. Through simulation of the coil in COMSOL Multiphysics software and linking it with MATLAB software, a wide range of faults is obtained for training of machine algorithms. These algorithms are being used to detect faults, such as: logistic regression, support vector machine (SVM), decision tree and nearest neighbor (KNN). Results indicate that among these algorithms, SVM algorithm does not perform well because of overlapping, and the most accuracy is related to KNN algorithm, so this algorithm is selected for the fault diagnosis system.