Research on insulation fault gas detection method of power equipment by integrating GWO-SVM

Abstract When overheating degradation occurs inside a transformer, the release of characteristic gases becomes a key fault warning signal. However, the accuracy of existing classification algorithms in transformer fault diagnosis still has room for improvement. The goal of this study is to enhance the accuracy and reliability of transformer fault characteristic gas identification, and a method based on a Gray Wolf Optimization Support Vector Machine classification model (GWO-SVM) is proposed. The performance of the three models (GWO-SVM, standard SVM, and BP neural network) is compared using various performance metrics, and their performance on the same dataset is evaluated. Experimental results show that the GWO-SVM model significantly outperforms other models in transformer fault characteristic gas identification, achieving higher classification accuracy and better stability.