无刷直流电机的神经网络控制器的仿真研究

The Simulating on Neural-net based Controller of BLDCM

提出了一种无刷直流电机的神经网络非线性跟踪器的设计方法。在无刷直流电机的高性能速度跟踪中,若仅采用传统的PID调节器,则难以克服系统本身参数扰动所带来的转速偏差问题。本文采用双神经元网络的控制方法,一个用于辨识,一个用于控制,并在Matlab与Simulink下进行了仿真。文中提出一种易于应用自适应权值修正的方法,可以加大采样频率。仿真结果表明,这种方法可以在不影响控制精度的条件下,大大减少计算量,有利于在线实现。

In this paper, a novel nonlinear artificial neuro-controller is proposed for controlling the speed of BLDCM in the environment of high-performance drives, such as carvers. In this environment, if PID controller is used, there must be error brought by the disturb in parameters. A double neural network system is used, one for identifying the parameters of the system, and the other for regulating the terminate voltage of BLDCM. An adaptive weight update algorithm is proposed in this paper by changing the sampling frequency. The result of this control method is satisfied in simulation based on Matlab and Simulink. By reducing the amount of calculation, this method can help the online realization.