模糊CMAC及其在机器人轨迹跟踪控制中的应用  

Fuzzy cerebellar model articulation controller and its application on robotic tracking control

作　　者：孙炜[1] 王耀南[1] 

机构地区：[1]湖南大学电气与信息工程学院,湖南长沙410082

出　　处：《控制理论与应用》2006年第1期38-42,共5页Control Theory & Applications

摘　　要：小脑模型关节控制器(CMAC)具有结构简单,学习快速的优点,但是它的空间划分方式不能在线进行调整,影响了其自适应能力的提高.本文将模糊理论引入CMAC,提出了一种能够反映人类小脑认知的模糊性和连续性的模糊小脑模型关节控制器(FCMAC).该控制器对CMAC的空间划分方式进行了模糊化处理,可通过BP学习算法对CMAC的空间划分方式进行在线调整,大大提高了CMAC的自适应能力.所提出的FCMAC被应用于机器人的轨迹跟踪控制系统以克服机器人系统中非线性和不确定性因素的影响.仿真实验结果表明,所提FCMAC与传统的CMAC相比性能上有了很大的改善.Cerebellar model articulation controller （CMAC） has simple structure and rapid learning speed, but its space division way can not be adapted on line. This hinders the improvement of its self-adaptive ability. In this paper, fuzzy theory is introduced to CMAC, and a fuzzy cerebellar model articulation controller （FCMAC） is proposed. By fuzzifying the space division way of CMAC and adapting it on line through BP learning algorithm, the proposed FCMAC can reflect the fuzziness and continuity of human cerebella, and greatly improve the self-adaptive ability of CMAC. The proposed FCMAC is applied on robotic tracking control system to counteract the disadvantageous influences of nonlinearities and uncertainties in robotic system. Simulation results show that the performance of proposed FCMAC is much better than that of traditional CMAC.

关 键 词：小脑模型关节控制器 模糊小脑模型关节控制器 机器人 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]