Mechanical Faults Diagnosis of HVCB Based on Auditory Features and Sparse Representation Classification

The high-voltage circuit breaker(HVCB) plays an important part in power system. Detecting the mechanical faults of HVCB is of great significance to the operation of power grid. In this paper, the closing sound signals of HVCB are taken as the target signals of signal processing. In the feature extraction stage, the Gammatone Frequency Cepstrum Coefficients(GFCC) of target signals are extracted according to the response characteristics of cochlea basement membrane. Then the linear discriminant analysis(LDA) is applied to enhance the feature representation ability of GFCC. In the feature recognition stage, the sparse representation classification algorithm is used to identify the improved GFCC due to the dimension of GFCC is too high. Finally, a 110kV HVCB is taken as the experimental object to implement field experiments. The jamming and refusing to operate faults of HVCB are set artificially. And the experimental results show that the recognition method proposed in this paper can complete the mechanical faults diagnosis of HVCB well.