关键词: 永磁同步电机, 径向基函数神经网络, 改进资源分配网络学习算法, 梯度下降法, PID控制器

Abstract: Aiming at the nonlinear, multi parameter variation and system disturbance of PMSM, the traditional PID control was combined with radial basis function neural network which has strong self-adaptive and self-learning ability to design a neural network speed controller of PMSM. RBF neural network was used to adaptively adjust the PID speed controller parameters to improve the system robustness and control accuracy; improved resource allocation network (Iran) and gradient descent method were used for offline learning and online learning to improve the operation speed of RBF neural network. Through MATLAB Simulation experiments, compared with the traditional PID speed control, the neural network speed controller has higher control accuracy, better speed regulation performance and robustness.

Key words: permanent magnet synchronous motor (PMSM), radial basis function (RBF) neural network, IRAN (improved resource allocating network) learning algorithm, gradient descent method, PID controller