Tracking of impedance trajectory for distance protection using neural networks

Abstract

The variation in voltage and current signals during power swing is often root cause for incorrect operation of distance relay. As a result, the distance relay may initiate a trip command to the breaker to isolate transmission lines unnecessarily. This can be counteracted by incorporating an algorithm for Power Swing Blocking (PSB) in distance relay to revoke its operation during the power swing. Conversely, in case of a fault during swing, the relay must sense the fault and shall initiate an unblock function of relay to isolate the faulty line from the circuit. This paper presents a Multi-Layer Perceptron neural network (MLP) to predict voltage and current signals during power swing precisely. Subsequently, the impedance perceived by distance relay is estimated using the predicted signals by post-processing. The estimated impedance is compared with actual impedance continuously to distinguish swing and fault conditions. The proposed method is tested for a two-area four-machine power system and compared with eminent methods for performance evaluation.