New method for fault classification in TCSC compensated transmission line using GA tuned SVM

Abstract

Presence of TCSC (Thyristor-Controlled Series Compensator) compensated transmission lines is increasing in modern power systems due to their benefits like increased power flow capacity but these benefits come at the cost of difficulty in protection of the transmission line. This paper presents a new method using SVM (Support Vector Machine) for fault classification in such line. This method is compared with existing SVM based methods and higher classification accuracy has been achieved. The improved accuracy is achieved by changing the architecture and input of the classifier. Genetic Algorithm (GA) is used to search globally optimum value of SVM parameters. Effect of sampling frequency and data window length on proposed scheme is also analyzed.