A new fault diagnosis method for HV circuit breakers based on wavelet packet-neural network

This paper introduces a new method based on wavelet packet analysis and neural network theory to solve the mechanical faults of high voltage circuit breakers. Firstly, the vibration signal is decomposed after removing noise at three levels, and the integral energy of each junction is calculated. Secondly, 8-dimension characteristic vectors are constituted with the ratios of 8 sub-band energy to total energy. Finally, the characteristic vectors are classified by BP neural network. The experiments of HVCB without loading verify that this method is simple and accurate to detect the faults of circuit breaker, and it also can check out the mechanical fault rapidly.