Diagnosis of Transformer Faults Using Feature Extraction and Automatic Division Based Fuzzy Clustering

Abstract

Power transformer which has an important influence on the entire power transmission system is related to the safety and stability of the entire power system. However, transformer may fail if it works for a long time. In order to improve the accuracy of fault diagnosis and guarantee system security, an optimal fuzzy clustering method was put forward to diagnose faults. The relationships between characteristic gas in power transformer oil and different fault types were analyzed, and fuzzy clustering was optimized to improve the algorithm convergence speed. It was found that the accuracy of fault diagnosis based on analysis using optimal fuzzy clustering algorithm was 92.23%, suggesting an increase of 23.3% compared to the traditional gas feature based fault diagnosis. The method can accurately diagnose faults of power transformers and ensure the stability of electric system.