Eliminating White Noises from Acoustic Discharge Signals Using Adaptive Filter Techniques

Abstract

Acoustic partial discharge (PD) measurement is one of the accepted methods in detecting PD in power transformers. Comparing with the electrical method, the acoustic method has the ability of locating the source of PD and is immune to the electrical noise that affect the interpretation of the electrical measurement results. However, if the source of discharge is deep inside the transformer, the acoustic signal could be significantly reduced and buried inside the white noise generated by the detection system. In this paper the adaptive filter technique was used to eliminate the white noise from the signal captured by the acoustic sensor. MATLAB was used as the platform to develop the adaptive filter algorithm. The performance of the adaptive filter was assessed using the signals and noises generated by the MATLAB functions.