Deep auto-encoder network for mechanical fault diagnosis of high-voltage circuit breaker operating mechanism

Abstract To improve the accuracy of mechanical fault diagnosis operating mechanism and fully exploit characteristic information in vibration signal high-voltage circuit breaker, a method breaker based on deep self-encoding network is proposed. First, switch extracted, wavelet packet conversion performed, each frequency band divided into equal times. The energy time–frequency subplane then calculated, distribution used as switch. Finally, failure diagnostic model self-coding established. Pretraining tuning 126 kV are to simulate different types faults validate method. Experimental results show that this can acquire sample data perform diagnosis, rate reaches 97.5%. pierce signal.