摘要
为实现机器人关节位置镇定和轨迹跟踪控制,控制律的设计须针对确定的机器人动力学模型,由于机器人结构参数、作业环境的外界干扰及结构振动等不确定性因素的存在,会造成机器人动力学模型不确定.为此,设计3个RBF神经网络分别对不确定机器人模型中的3个不确定项进行分块建模,得到机器人估计模型,神经网络的权值采用自适应算法.针对机器人估计模型设计PI鲁棒滑模控制律.将所设计的控制器用于三关节机器人的三个关节的力矩控制,研究结果表明:三关节均约在1 s时达到期望位置和跟踪期望轨迹,镇定误差和跟踪误差也快速、稳定地趋于零.通过定义基于积分型的Lyapunov函数,利用Lyapunov稳定性理论证明了控制系统是全局渐近稳定的.
In order to realize position stabilizing and trajectory tracking for joints of joint robot,control law must be designed for the determined dynamic model of joint robot.But generally,robot＇s dynamic model is undetermined because of uncertain structure parameter,interfere with working environment and uncertain resonant mode for joint robot.Three designed RBF neural networks are used to respectively model undetermined robot dynamic model＇s three undetermined terms.That is to say,partition modeling.Then,joint robot＇s estimation model is obtained.The neural networks＇ weights are obtained through adaptive algorithm.PI robust sliding mode controller is designed based on robot＇s estimation model.The PI robust sliding mode controller is used to control joint＇s torque for robot with three joints.Simulation experiments test and verify that three joints can trace ideal trajectory and reach ideal position in 1s,and stabilization error and tracking error can fast and stably approach zero.The control system＇s global asymptotical stability is proved through defining integral-type Lyapunov function and using Lyapunov stability theory.
出处
《辽宁工程技术大学学报（自然科学版）》
CAS
北大核心
2016年第11期1305-1311,共7页
Journal of Liaoning Technical University (Natural Science)
基金
国家自然科学基金项目(51005246)
中国人民武装警察部队工程大学基础研究基金项目(WJY201509)
关键词
关节机器人
不确定模型
分块建模
位置镇定
轨迹跟踪
joint robot
undermined dynamic model
partition modeling
position stabilizing
trajectory tracking
作者简介
马莉丽（1980-），女，江苏如东人，博士，主要从事机电系统智能控制及其自动化、机器人控制、军事装备理论及其应用等方面的研究．
