Adaptive distance protection of double-circuit lines using artificial neural networks

Abstract

Because of the zero sequence mutual coupling of parallel transmission circuits, the distance calculation performed by a ground distance relay is incorrect. This error is influenced by the actual power system condition. Although accounted for by using a large safety margin in the zone boundaries, unexpected overreach can still occur and the operation speed is decreased. Adaptive protection offers an approach to compensate for the influence of the variable power system conditions. By adapting the relay settings to the actual power system condition, the relay will respond more accurately to power system faults. The selectivity of the protection system is increased, as is the power system reliability. In this paper, an adaptive distance relaying concept is presented. In order to minimize the required communication, local measurements are used to estimate the entire power system condition. An artificial neural network is used to estimate the actual power system condition and to calculate the appropriate tripping impedance. Application of this concept to the model of the Dutch 380 kV power system has resulted in an enormous increase in relaying accuracy. The relaying error is reduced substantially. Most importantly, the standard deviation, indicating the relay's sensitivity to power system condition variations, is reduced to nearly zero. The zone boundary is kept nearly constant, which facilitates the relay coordination. The selectivity of the entire power system protection system is improved. It is shown that adaptive protection improves the protection system selectivity, and that artificial neural networks can very well be used to estimate the actual power system condition.