Power Transformer Fault Detection and Classification Using SVM Classifier

Abstract

This paper mainly focuses on the internal fault detection and classification process of power transformer using the machine learning techniques. As the power transformers are the vital and prime equipment in power system, an internal fault occurs in a transformer has to be isolated as quickly as possible to protect the power system. Keeping in mind the cost of power transformer and ageing factor, machine learning algorithm has been developed with the Dissolved Gas Analysis (DGA) method diagnosing the incipient faults using the SVM classifier. Seven fault including one normal classification have been taken in this research work with the features extracted from the gases evolved from the insulating oil of transformer. Testing of the classifier has been done using the available data’s collected. The performance of the classifier yields comparatively better results of accuracy. 

Keywords: Dissolved Gas Analysis (DGA), SVM, Power Transformer.