The Effect of Kernels in SVM on the Fault Classification Accuracy of a Transmission Line Compensated with TCSC

Abstract

In a Thyristor Controlled Series Capacitor (TCSC) compensated transmission line the dynamic variation of line impedance during various operating modes of the TCSC affects the performance of the fixed impedance setting-based distance relay. The non linear variation of impedance during fault caused by the operation of the metal oxide varistor (MOV) protecting the TCSC also further affects the performance of the distance relay. This paper presents an improved fault detection and classification approach for TCSC compensated transmission lines by processing the superimposed components of the measured three-phase currents at the relay location through the support vector machine (SVM). The performance of the proposed method is tested for a wide variety of fault cases simulated on a 400 kV TCSC compensated two-machine two-bus system through EMTDC/PSCAD software. Fault classification accuracy is evaluated by considering single end as well as both end current data through SVM classifier with linear, polynomial and Gaussian Kernel functions in RStudio software. The obtained result confirms that with Gaussian Kernel function a fault classification accuracy of 98.94% and 99.60% can be achieved by using single end and both end current data, respectively.