基于CMAC在线自学习模糊自适应控制的机器人力/位置鲁棒控制  

Fuzzy Adaptive Control Based on CMAC On-Line Learning and It’s Application on Robot

作　　者：魏立新[1] 李二超[1] 王洪瑞[1] 

机构地区：[1]燕山大学电气工程学院,秦皇岛066004

出　　处：《电工技术学报》2005年第5期40-44,共5页Transactions of China Electrotechnical Society

基　　金：河北省自然科学基金资助项目(F2004000260)。

摘　　要：针对机器人系统的未知、不确定、时变和非线性特性,在传统误差学习法的基础上,提出一种CMAC在线自学习模糊自适应控制结构,利用模糊推理机产生的分目标学习误差代替反馈控制器的输出信号训练CMAC,使CMAC的学习与系统的实际跟踪过程相适应,避免了控制器的输出产生振荡或进入饱和状态。仿真结果表明这种控制方案实现了对未知不确定非线性机器人系统的高精度实时力/位置控制。In this paper, a new control strategy based on CMAC online learning fuzzy adaptive control is presented. The uncertainties, time-varying, and nonlinear performances of robot system are considered. It makes use of object learning error which is produced by fuzzy reasoning to substitute feedback controller's output, it also makes CMAC to adapt to the practical tracking process. So the control strategy can avoid controller's output producing surge or entering saturation condition. Simulation shows that the presented method can improve the system's robust and real time performance.

关 键 词：机器人 力/位置控制 小脑模型联接控制 模糊推理机 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]