Application of Support Vector Machine Classifier for Transformer Winding Faults Diagnosis

Diagnosing faults based on artificial intelligence is one of the modern and effective methods, so it has become necessary to apply it to power transformers. This paper proposes a new methodology for detecting the type, location, and severity of power transformer faults, based on the analysis of frequency responses (FRA) and support vector machines (SVM). The method was tested on two faults that were simulated on a transformer winding model, where databases were formed to train the support vector machine (SVM) by collecting (FRA) signals in defective and healthy conditions and analyzing them by statistical indicators. The obtained results confirm the effectiveness of the proposed method in determining the type, location, and extent of faults with high accuracy, and its ability to contribute to the development of the application of machine learning in power transformers.