神经网络自适应位置控制算法研究与实时仿真

Research on Neural Network Based Adaptive Position Control Algorithm and Realtime Simulation

在传统PID控制算法的基础上,基于神经网络的自学习特性,提出了基于小脑模型神经网络-CerebellarModelArticulationController(CMAC)和单神经元的复合控制实现前馈反馈位置控制算法。该算法特点是不需要对被控对象建模、控制参数自调整、运算速度快、适用于实时控制。利用dSPACE在松下的电动机上所做实验表明,该算法控制效果良好。

Based on the self-learning feature of neural network, a neural network based CMAC(Cerebellar Model Articulation Controller)and adaptive arithmetic was proposed for the position servo control of AC motor. The feature of this algorithm is that the identification of system model is no longer necessary. This algorithm reduces the computation amount of weight adjustment and improves the efficiency of computation. So it can be used in real-time control. Based on dSpace, the simulation results show that the new method has a good performanceand high efficiency.