Fault perception analysis of distribution network based on FAGA-BP neural network algorithm

In response to the unstable accuracy rate of fault location in distribution network lines and the significant impact of communication with distribution automation devices, the fault perception analysis of distribution network based on fusion alarm and genetic algorithm-BP neural network algorithm (FAGA-BP) is proposed in this paper. A fault localization and perception process is designed. Based on the alarm signal of the distribution automation device, a power loss alarm signal of the distribution transformer is introduced. A fusion alarm fault perception and localization model for the active distribution network is established, including the overcurrent alarm signal matrix of the protection action of the distribution automation device and the power loss alarm fault matrix of the distribution transformer. The selection of fault feature quantities and the fusion alarm rules are defined. The protection overcurrent alarm matrix and power loss alarm matrix are used to generate a fusion alarm fault feature quantity positioning model matrix using the definition rules. A fault type coding table is developed to characterize the fault type of the distribution network section. A neural network model based on fusion alarm and genetic algorithm is used to train the fault feature quantity positioning model matrix and perform fault discrimination in the distribution network section. Through case analysis, it is proven that the proposed algorithm can reduce the error of fault diagnosis and improve the fault tolerance rate.