Fault diagnosis method for energy storage mechanism of high voltage circuit breaker based on CNN characteristic matrix constructed by sound-vibration signal

Aiming at the problem that some traditional high voltage circuit breaker fault diagnosis methods were over-dependent on subjective experience, the accuracy was not very high and the generalization ability was poor, a fault diagnosis method for energy storage mechanism of high voltage circuit breaker, which based on Convolutional Neural Network (CNN) characteristic matrix constructed by sound-vibration signal, was proposed. In this paper, firstly, the morphological filtering was used for background noise cancellation of sound signal, and the time scale alignment method based on kurtosis and envelope similarity were proposed to ensure the synchronism of the sound-vibration signal. Secondly, the Pearson correlation coefficient was used to construct two-dimensional image characteristic matrix for the expanded sound-vibration signal. Finally, the characteristic matrix was trained by utilizing CNN. Local Response Normalization (LRN) and core function decorrelation were utilized to improve the structure of CNN model, which reduced the bad impact of large data fluctuation of energy storage process on the diagnostic accuracy of circuit breaker energy storage mechanism. Compared with the traditional method, the proposed method has obvious advantages, whose total accurate rate up to 98.2 % and generalization performance is excellent.