Location of the Fault in TCSC-based Transmission Line Using SVR

In this paper we inspect support vector regression (SVR) based fault position in a TCSC (thyristor controlled series capacitor) based long transmission line. This technique uses 1 cycle post faulty current signal from the transmission line and decomposed by wavelet packet transform. From the decomposed signal entropy and energy are extracted and fed to the forward feature selection method to eliminate the redundant data set. Then optimal future data set is normalized. Taking different simulation situation like fault type, resistance path, inception angle, and distance train and test data are produced. By using particle swarm optimization technique SVR parameters are optimized. Then normalized data set is fed to SVR to locate the fault position in TCSC based long transmission line. It is noticed that fault position error is less, than 0.29 percentages.