Insulation State Identification of Vacuum Circuit Breaker Based on Surface Discharge Sound Analysis by Gammatone Filter

Abstract

Insulation devices such as insulators inside the vacuum circuit breaker will generate surface discharge in the event of a serious insulation problem. In this paper, the creeping sound recognition algorithm based on the principle of human cochlear basement membrane and Hilbert transform are proposed to judge the insulation state of the circuit breaker. The surface discharge audio signal passes through the 36dimensionalgammatone filter bank, so the 36 sets of time domain components are obtained. The energy of each component is calculated by Hilbert transform; The Fisher criterion is used to reduce the dimension of 36-dimensional data. After selecting the parameters with large discriminant degree to reconstitute the new feature vector, and the surface discharge sound signal is recognized by one-class support vector machine classifier. The single-class support vector machine maps the energy feature vector of the Surface discharge to the high-dimensional linearly separable feature space through the kernel function, and recognizes the creeping discharge sound through the high-dimensional feature space. The experimental results show that the proposed creeping sound recognition algorithm based on gammatone filter bank and Hilbert transform can effectively identify the insulation state of vacuum circuit breaker.