Intelligence Expert System of Transformer Running State Diagnosis Based on Acoustic Signal Analyzing

Abstract

There are many kinds of external phenomena with power equipments running state changing, and one of the important characters is acoustic signal mutation when power transformer would get out of order. But it is difficult to judge the acoustic change degree and describe the failure character scientifically only by human subjective sensation. The acoustic monitoring can be realized with non-contact measurement, the committed step is the acoustic signal processing, and intelligence fault diagnosis scheme is presented in this thesis. After acoustic signal collected in substation, the layer threshold de-noising algorithm is utilized, the signal characteristics based on wavelet package is extracted to be the gist of knowledge acquisition and consequence, and the fault diagnosis expert system is designed by acoustic quantative analysis. Through acoustic signal analyzing experiments, the effects of the de-noising algorithm and signal characteristics extraction method have been verified. The acoustic wave diagnosis expert system can solve many problems because of the high voltage and powerful electromagnetic field of the power system, and it would be a good reference to many power equipment fault identify.