A Method for Fast and Accurate Fault Classification in TCSC Compensated Transmission Line Using RVM

Fault classification in series compensated line compensated by Thyristor-Controlled Series Compensator (TCSC) has been a difficult task due to nonlinearities introduced by its protective equipment and variable level of compensation. This paper presents method for fault classification in TCSC compensated line using Relevance Vector Machine (RVM). Very high accuracy of fault classification is achieved at very fast speed while maintaining memory requirements very small. This makes it suitable for protection purpose which is a real time application. Effect of sampling frequency, data window length, and number of training examples on the accuracy of the classifier (RVM) is studied and rationale behind their performance is explained.