Mechanical Fault Diagnosis Method of High Voltage Circuit Breaker Operating Mechanism Based on Deep Auto-encoder Network

In order to improve the accuracy of mechanical fault diagnosis of operating mechanism, the characteristic information in the vibration signal of high voltage circuit breaker is fully excavated, and a mechanical fault diagnosis method for operating mechanism of high voltage circuit breaker based on deep auto-encoder network is proposed. Firstly, the vibration signal of the circuit breaker operating mechanism is extracted to perform wavelet packet transformation. Then, the time frequency sub-planes of vibration signal are obtained by equal time segmentation of vibration signal in each frequency band, and their energy is calculated. The time-frequency energy distribution is taken as the characteristic quantity of fault diagnosis. At last, the fault diagnosis model of circuit breaker based on deep auto-encoder network is built with two steps of pre-training and fine-tuning, and simulation experiments are performed in a 126 kV high voltage circuit breaker under different fault types to verify the effectiveness of this method. The experimental results show that this method can get the fault sample data for fault diagnosis, and the fault diagnosis accuracy rate is 97.5%. The deep self-coding network can fully excavate the deep features of the vibration signal in circuit breakers, and can be used to diagnose the mechanical faults more accurately and effectively than the traditional shallow networks. © 2020, High Voltage Engineering Editorial Department of CEPRI. All right reserved.