Research and Simulation Analysis on Fault Early - Warning Methodology for Distribution Station Area Based on Improved Fuzzy Neural Networks

In order to further improve the power supply reliability of distribution station are under the background of increasing penetration of distributed photovoltaic, this paper studies a fault early warning method of distribution station area based on improved fuzzy neural network. Firstly, the fault feature information of the operation data of the distribution station area with high permeability distributed photovoltaic is extracted, which provides the data basis for the early warning. Then, the FCM clustering algorithm with improved membership function is used to determine the fault boundary conditions, and the LSTM prediction feature information of the improved loss function is used. On this basis, the membership degree of the predicted feature quantity is calculated again, and whether it belongs to the fault abnormal data is judged according to the determined fault boundary conditions, so as to realize the fault early warning. Finally, the simulation verifies that the proposed method can accurately realize the fault early warning of the station area.