高精度伺服系统的CMAC学习控制研究  

CMAC network based learning control of high-precision servo systems

作　　者：李秀娟[1] 张媚[1] 

机构地区：[1]南京航空航天大学,江苏南京210016

出　　处：《电机与控制学报》2003年第4期344-347,共4页Electric Machines and Control

摘　　要：根据神经网络的特点,提出了基于CMAC网络的伺服系统在线学习控制方案,并给出了其再励学习算法。该方法无需离线训练,经在线学习得到系统的道动态特性,并据此实时修改控制参数。引入鲁棒控制项,增强了学习算法的鲁棒性。在某单轴速率/位置转台上的应用结果表明,该方法可有效地改善伺服系统的跟踪精度和抗扰性能。An online CMAC adaptive learning controller is presented, and its algorithm is discussed. The inverse dynamics of the system is obtained through online learning, and then the control parameters are adjusted. A robust term ensured the robustness of the learning algorithm. The proposed controller is used in a one-axes velocity/position simulator and test shows that the tracking performance is greatly improved.

关 键 词：伺服系统 CMAC 学习控制 神经网络 鲁棒性 学习算法 

分 类 号：TM921.541[电气工程—电力电子与电力传动]