Abstract:In this paper, wavelet packet (WP) band energy analysis method is firstly used to get feature band of field circuit breaker (CB) vibration signal with heavy noises disturbing which is done with pretreatment of de-noising by WP method. The determined feature band signal is then analyzed by the non-parametric, ARMA model, and AR model bi-spectrum analysis respectively. With the field data collected from different period, the results show that the AR model bi-spectrum analysis is better among three methods mentioned above for CB state detection, which can be used to be a means to diagnose possible CB fault. © 2008 IEEE.

Application on circuit breaker vibration signal procession based on wavelet packet band energy analysis and bi-spectrum estimation