Transformer Familial Defect Detection Method Based on SVM Improved by Apriori Algorithm—Based on Analysis of Dissolved Gases in Oil

Transformer familial defects are different types, different specifications, different series and even different types of transformers produced by the same manufacturer that have the same type of defect. In this paper, the dissolved gas in transformer oil is analyzed by using the improved SVM (support vector machine) algorithm based on apriori (frequent item set), and a transformer familial defect detection method is proposed. Apriori method was used to analyze the correlation of dissolved gas in oil, and the variation of dissolved gas content was normalized to obtain the confidence degree and correlation degree of various gases under different transformer fault types, so as to judge the weight of transformer faults. Then, according to the gas type and content, the classifier is established by using SVM method. Considering the influence weight obtained by apriori method, the accuracy of transformer fault diagnosis can be 83.78%∼91.58%. Finally, the classifier is used to diagnose the transformer familial defects.