This paper proposes a fault diagnosis method for electrical machine drive system by using backpropagation neural network (BPNN) optimized by chaotic adaptive gravity search algorithm (GSA) and particle swarm optimization (PSO) algorithm. In this method, an adaptive gravitational constant factor based on iteration times and chaotic mapping is introduced to balance the global search ability and local development ability of GSA. Then, it is combined with PSO algorithm to solve the problem of prematurity and local optimum of PSO algorithm. Finally, combined with BPNN, a fault diagnosis model based on chaos adaptive GSA-PSO-BPNN is established. The experimental results show that the introduction of the attenuation factor of adaptive gravitational constant and the chaotic mapping can improve the classification performance of GSA-PSO-BPNN, and the feasibility and effectiveness of the chaotic adaptive GSA-PSO Algorithm are also proved.

本文提出了一种利用混沌自适应重力搜索算法（GSA）和粒子群优化（PSO）算法优化的反向传播神经网络（BPNN）对电机驱动系统进行故障诊断的方法。该方法引入基于迭代次数和混沌映射的自适应引力常数因子来平衡GSA的全局搜索能力和局部开发能力。然后结合PSO算法解决PSO算法的早熟和局部最优问题。最后，结合BPNN，建立了基于混沌自适应GSA-PSO-BPNN的故障诊断模型。实验结果表明，引入自适应引力常数衰减因子和混沌映射可以提高GSA-PSO-BPNN的分类性能，