Probabilistic Fault Identification Based on Maximum Likelihood Method in Distribution Networks

Fault identification is a vital function of advanced applications in distribution networks. This paper proposes a method for clustering fault types using von Mises-Fisher(VMF) distribution and using Maximum Likelihood Estimation(MLE) for fault identification. First, when a fault occurs, sampling is performed at a fixed update time, the three-phase voltage phasors are calculated by the DFT algorithm, and the typical fault vectors are extracted to form a vector set. The obtained vector set will be used as a training set for the vMF model, According to different fault types, a variety of single vMF models are formed, and then multiple single models are formed into a mixture of vMF model. The MLE method is used to solve the parameters of the model, and fault identification is realized by using the method of probability. Finally, the simulation results show that the method proposed in this paper can obtain reliable identification results under different fault conditions.