Abstract:

Aiming at the common insulation defects during the operation of electrical equipment,this article has produced several typical partial discharge models of defects,including tip discharge,creeping discharge,air gap discharge and floating discharge.By comparing the time-domain waveforms of the discharge acoustic emission signals of different defects recorded by the test system and the characteristics of the correlation graph of the parameters of the discharge acoustic emission signal,the positive and negative of the pulse amplitude,phase,shape and power frequency of the four types of discharge acoustic emission signals are obtained.There is a big difference in half-cycle symmetry;in response to this difference,18 acoustic emission waveform feature parameters are extracted,a multi-class support vector machine classifier is designed.The recognition test results show that for the recognition of four typical types of partial discharges of defects,the rate reaches 92%,so the partial discharge acoustic emission waveform characteristic parameters can be used for discharge type identification.

Key words: discharge type, acoustic emission, correlation analysis atlas, support vector machine