关键词: 高压断路器, BP神经网络, 麻雀搜索算法

Abstract: As a device that can widely play a role in security protection in various power system applications, high voltage circuit breaker has great significance for its fault diagnosis research. Aiming at the problems of slow convergence speed and insufficient convergence accuracy of traditional BP neural network, an improved sparrow search algorithm is proposed to optimize the BP neural network. The model obtains more reasonable initial parameters by introducing Logistic chaotic map when the sparrow algorithm initializes the population, and introduces dynamic adaptive weight, Cauchy mutation strategy and reverse learning strategy in position update and optimal solution update respectively, so that the average accuracy of the model for circuit breaker fault classification diagnosis is close to 100%, which indicates that the improved BP neural network has higher accuracy.

Key words: high voltage circuit breaker, BP neural network, sparrow search algorithm