Performance of Distance Protection in Presence of SVC Using Neural Network Approach

Shunt static VAR compensation has become a desired and effective solution for improving power transfer capability, voltage regulation and system stability. SVC has a significant impact on the impedance of the line, which in turn affects the accuracy of the impedance measurement. Transmission lines equipped with conventional distance relays may mal-operate in the presence of VAR compensators which in turn affect the accuracy the measured apparent impedance seen by the distance relay during a fault from the relay location to the fault point. The accuracy of the distance measurement is affected by the amount of compensated reactive power injected or absorbed by the VAR compensator. In that sense, the distance relay may under-reach or over-reach depending on the amount of compensation. In this paper, an artificial neural network approach is proposed to enhance the zone classification of distance relay. The proposed method is tested on a 400kV transmission line using Matlab/Simulink software. Extensive simulations are performed for various faults and over wide range of VAR compensation levels, fault locations, and fault types. These data sets are processed as neural network training/testing data. The simulation results showed high accuracy and speed of zone classification within less than three cycles.