摘要：针对无刷直流电机（BLDCM）气隙磁密非线性分布的特性，利用粒子群优化的反向传播（PSOBP）神经网络对气隙磁密波形进行逼近，从而建立较为精确的BLDCM模型。在此模型基础上设计了一种双闭环控制系统：转速环采用单神经元PID控制，通过改进的Hebb学习规则调整权值，以实现PID三个参数的自适应；电流环采用滞环电流控制，以实现电流调节。在Matlab/Simulink中搭建仿真模型，仿真结果验证了该模型的正确性。

关键词：无刷直流电机；气隙磁密；PSOBP神经网络；单神经元PID；仿真

Abstract: For the non-linear distribution characteristic of air gap flux density of brushless DC motor，using particle swarm optimization back propagation neural network to approximate the flux density waveform in order to establish a more precise BLDCM model. A double-loop control system was designed based on this model：speed loop used the single neuron PID control，through improved Hebb learning rule to adjust the weights in order to realize the self-adaptation of three PID parameters；current loop used the hysteresis current control to realize the current regulation. The simulation model based on Matlab/Simulink was built，and the results verified the correctness of the model.

Key words: brushless DC motor；flux density；PSO-BP neural network；single neuron PID；simulation