Transformer fault diagnosis based on BP neural network optimized by the strongest crow search algorith

Abstract

In order to improve the accuracy of transformer fault diagnosis, a transformer fault diagnosis model based on the Strongest Crow Search Algorithm(SCSA) to optimize BP neural network is proposed. In the SCSA-BP fault diagnosis model, adaptive weights are introduced to coordinate the global and local search capabilities of the Crow Search Algorithm(CSA). The crazy factor is introduced to increase the diversity of the crow population and prevent individual crows from falling into the local optimum. Bidirectional Random Optimization(BRO) is added to reduce the amount of nonessential exploration by the crow and to improve the search efficiency of the crow. SCSA is used to optimize the weights and thresholds of the BP neural network to avoid the premature problem of the BP neural network. Simulation calculations show that the accuracy of the transformer fault diagnosis of the SCSA-BP model is 8.955% and 4.478% higher than that of the PSO-BP and CSA-BP diagnosis models, respectively.