Fault State Evaluation Method of Transformer Using Genetic Algorithm and Fuzzy C-Means Algorithm

Abstract

To improve the accuracy of oil-immersed transformer fault diagnosis, a new transformer fault condition assessment method is proposed based on Dissolved Gas Analysis (DGA), combined with Fuzzy C-Means (FCM) algorithm and Genetic algorithm (GA). The method uses the dissolved gas data of the transformer as the sample data, and performs the weighted processing on the sample data, and performs the FCM clustering processing through the weighted data to obtain the fault classification model, wherein the weight is obtained through a Genetic algorithm. The experimental results show that the transformer fault classification method proposed in this article can distinguish between different transformer faults and have higher fault judgment rate and unweighted FCM clustering Class method to improve the accuracy of fault classification, and to join the other models in the existing fault to discover new fault types. The proposed method incorporates unsupervised machine learning methods into transformer fault classification. Unsupervised learning allows transformer fault assessment models to discover new faults autonomously under existing fault conditions, providing a new idea for fault condition assessment of transformers.