Intelligent diagnosis algorithm of power equipment based on acoustic signal processing

Abstract

The operational state determination of power equipment is a key prerequisite to realize maintenance. On studying the relationship between power equipment state and its acoustic wave mutation character, a new diagnosis scheme of power equipment fault has been put forward. After the running acoustic signal acquired, MFCC coefficient has been selected the acoustic signal various band energy feature, and dynamic time warping (DTW) is utilized to determine equipment type. Then local energy band based wavelet packet decomposition is used in fault feature extraction. According to these feature parameters values and expert experience scoring, the knowledge based of fault database was established to diagnosis power equipment state and its fault level. Lastly, By 200 group transformer measured acoustic signal analysis experiments have been completed, and the results show the series acoustic treatment of methods is effective, and the diagnosis scheme of equipment failures have great practical value.