摘要
多关节机械臂在建模过程中存在不确定干扰和建模误差,对机械臂的控制系统带来了许多不良的影响。为此,文中研究了一种反演滑模神经网络干扰观测器控制策略。文中采用非线性干扰观测器对多关节机械臂的外部不确定干扰进行观测补偿,且无需外界不确定干扰的上界先验知识。对于建模过程中存在的不确定性,采用反演滑模神经网络自适应控制,运用神经网络建模误差进行逼近,并采用自适应策略对网络权值进行自适应改变。通过李雅普诺夫稳定性理论证明了系统的稳定性,并通过MATLAB仿真,通过与传统滑模神经网络控制相比较,仿真结果表明该控制算法有效提高了机械臂末端轨迹跟踪速度和精度,降低了系统中存在的抖颤。
There are uncertain disturbance and modeling errors in the modeling process of the multi-joint manipulator,which may bring many adverse effects on the control system of the manipulator. Thus,in this article,the strategy of disturbance observer control is proposed,based on the inverse sliding mode’s neural network. The nonlinear disturbance observer is adopted to compensate the external uncertain disturbance of the multi-joint manipulator without the prior knowledge of the upper bound of the external uncertain disturbance. For the uncertainty in the modeling process,the inverse sliding mode’s neural network is subject to self-adaptive control;the neural network is used to correct the modeling error;the self-adaptive strategy is used to adaptively change the network weight. The system proves stable by means of the Lyapunov stability theory;the MATLAB simulation results show that compared with its traditional counterpart,the improved control algorithm effectively improves the tracking speed and accuracy of the manipulator’s end track on one hand and reduces the chattering of the system on the other hand.
作者
李正楠
殷玉枫
张锦
祁辰
LI Zheng-nan;YIN Yu-feng;ZHANG Jin;QI Chen(School of Mechanical Engineering,Taiyuan University of Science and Technology,Taiyuan 030024;Department of Engineering Machinery,Shanxi Traffic Vocational and Technical College,Taiyuan 030031)
出处
《机械设计》
CSCD
北大核心
2021年第3期126-131,共6页
Journal of Machine Design
基金
国家自然科学基金资助项目(U1610118)
山西省交通运输厅科技计划项目(2019-1-9)
山西省研究生创新项目(2019SY473)。
关键词
多关节机械臂
反演滑模
神经网络
自适应
干扰观测器
multi-joint manipulator
inverse sliding mode
neural network
self-adaptive
disturbance observer
