Assessment of Fault Location Methods for Electric Power Distribution Networks

In this work, an attempt is made to assess and compare the performance of the methods. The methods considered in this work stem from classical impedance-based, knowledge-based and hybrid combination The power system faults were planted at different locations of the MATLAB simulated IEEE network, and the distribution substation transformer fault currents were used for the computation of fault location. The accuracy of the ANFIS was found better among the applied algorithms. The impedance-based algorithm suffers from the drawback of multi-estimation issue, since it cannot detect the faulted line-section on a multi-terminal distribution network. The fault type classification of feed-forward neural networks was more consistent than other models for the tested data. The automated computation of fault location greatly assists the maintenance crew in quickly rushing to the fault spot. Thus, fault location improves the power system reliability and economy.