Performance analysis of a novel AI based approach to fault classification and location in an active distribution network with type 3 and type 4 wind turbine connections

Accurate fault location in an electrical power system improves the response time to short circuit faults on the system and increases the system reliability. This paper presents novel Artificial Neural Network algorithms that identify with high accuracy whether a short circuit fault lies on a feeder or on one of the spurs. These algorithms are also able to evaluate the distance to the point of fault on the feeder or a spur using only the phase currents measured at the substation. Further tests demonstrate the robustness of the proposed method to the integration of doubly fed induction generator and permanent magnet synchronous generator wind turbines into the network.