Transformer Fault Diagnosis Based on Intelligent Algorithm and Partial Discharge

Abstract

Abstract As an important part of power transmission and transformation equipment and power systems, transformers’ operating conditions will have a direct impact on the stability and reliability of the power system. In view of the problems of high diagnosis cost and low accuracy of diagnosis results in existing fault diagnosis technology, this paper uses the pulse current method to detect the pulse signal generated by the discharge model based on the partial discharge experimental platform data, analyzes and studies the discharge characteristics of different discharge defects, and draws For the partial discharge phase distribution (PRPD) spectrum of each defect type, the statistical characteristic parameters characterizing the characteristics of the spectrum were extracted. The partial discharge statistical characteristics were used as classification feature quantities and input into three network models: RBF, GRNN, and PSO-GRNN. The results show that PSO-GRNN has a good recognition effect and is suitable for partial discharge type identification. It has important practical significance and application value for the all-round health management of power transformers.