A NOVEL APPROACH TO DISCRIMINATE BETWEEN INRUSH AND INTERTURN FAULT CURRENTS OF TRANSFORMER

Ganesh Bonde, Sudhir Paraskar, Saurabh Jadhao, and Dwarakadas Kothari

Keywords

Teager energy operator (TEO), support vector machine (SVM),signal processing

Abstract

Transformer protection is an important issue in electrical power system. This article presents a new approach to detect and classify amongst the inter-turn fault current and inrush current in a transformer using the teager energy operator (TEO) and support vector machine (SVM) algorithm. The inrush and inter-turn current signals are captured from the trailer made experimental set-up in the laboratory. The detection and classification algorithms are implemented using PC with MATLAB and Python script. The features are extracted using the TEO threshold-based algorithm. Those are used for training SVM. In this article, the linear kernel, polynomial kernel, radial basis function kernel, and sigmoid kernel are used for classification. All these kernels are compared based on the accuracy of classification with different values of regularization factors and changing the size of training and testing cases. The GridSearchCV optimization algorithm with a support vector classifier is also used to validate the best accuracy kernel. The computation time of TEO and classification of events is less than 10 ms.

Important Links:

Go Back