Research on Transformer Fault Diagnosis Based on SMOTE and Random Forest

Abstract

The rapid development of artificial intelligence provides a new method with higher accuracy for transformer fault diagnosis, but the existing fault diagnosis models are not conducive to handling unbalanced data sets. In order to improve the accuracy of transformer fault diagnosis, a diagnosis method combining SMOTE and random forest is proposed. The SMOTE algorithm is used to expand the minority fault samples of transformer oil chromatography fault data set to balance the data quantity of each fault type. Then, the random forest classifier is used to identify the faults of the data that have not been expanded and the data that have been expanded by SMOTE respectively. The diagnosis results show that the accuracy of fault diagnosis can be significantly improved by using SMOTE to expand the unbalanced transformer oil chromatography fault data set before fault diagnosis. In addition, the results of several other fault diagnosis models are added to verify the above conclusion. At the same time, it is concluded that the random forest classifier is the model with the highest diagnostic accuracy among several fault diagnosis models, so it is an ideal choice for transformer fault diagnosis.