Fault analysis of controllable series compensated transmission line with Wavelet Transform and Support Vector Machine

Abstract

For power system monitoring and relaying purpose, fault analysis is considered as an important task. The fault classification problem becomes difficult with introduction controllable compensation. This paper presents an approach for fault type identification with the help of Discrete Wavelet Transform (DWT) and Support Vector Machine (SVM). The scheme utilizes only half cycle post fault data of three phase currents for classification. The change in current signal features during fault has been considered as discriminatory measure. The developed method is extensively tested on a two-area 300 km, 400 kV transmission line compensated at the mid-point with TCSC. Fault cases has been generated with PSCAD/EMTDC with wide variation of system parameters like source impedance, fault resistance, fault inception angle, loading angle, types of faults at different TCSC firing angles. With extensive testing with 19,200 test cases, the algorithm has been proven to be accurate for implementation on transmission lines with TCSC.