Study on the Identification of the Signal Component of Circuit Breaker Based on Blind Source Separation

Abstract

Acoustic analysis is an effective method for fault diagnosis of contacted high voltage circuit breakers. Acoustic signals are often mixed with different perturbations in the circuit breaker’s practical operation within the complex environment. The low frequency disturbance can be fully filtered by filtering equipment. Circuit breaker's error action or running state’s misjudgment may be caused by high intensity and the disturbance noise such as thunder, car horns. In this paper, a new blind source separation method is proposed to identify the signal component of acoustic signal. Firstly, the K-means algorithm is used to estimate the number of blind sources. Secondly, the IMF component is obtained by improving the EEMD decomposition signal, and then the signal is reconstructed to form a new multi-dimensional signal. Finally, the Fast ICA algorithm is used to realize the blind source separation of the signals.