带有前馈和神经网络补偿的机械手系统轨迹跟踪控制  

Trajectory tracking control for robotic manipulators endowed with feedforward and neural networks

作　　者：王良勇[1] 杨枭[1] 

机构地区：[1]东北大学流程工业综合自动化国家重点实验室,辽宁沈阳110819

出　　处：《电机与控制学报》2013年第8期113-118,共6页Electric Machines and Control

基　　金：国家自然科学基金重点项目(61134001);辽宁省博士启动基金(20121011);中央高校基本科研业务费专项资金(N110308001)

摘　　要：针对机械手系统的轨迹跟踪控制,提出一种带有前馈和神经网络补偿的机械手控制策略,该控制策略由比例微分(proportional-derivative,PD)控制器、前馈补偿器和神经网络补偿器组成。首先通过实验数据辨识出机械手系统的近似线性模型,并基于此线性模型设计PD控制器的参数和前馈补偿器,采用神经网络来补偿近似模型与实际模型的偏差。该方法不需要建立机械手系统复杂的动力学模型,仅需要系统的输入输出数据,既保留PD控制器的优势,又保证了系统的控制性能。最后,在机械手平台上进行物理实验研究,实验结果表明:所提的控制策略实现简单,同时保证了系统的控制精确度。To deal with the trajectory tracking control of robotic manipulators, a feedforward and neural network compensator based control strategy were prensented. In this scheme, neural network compensaor and feeforward compensaor were integrated in addition to PD controller. Firstly, a linear model was constructed from the expermential datum. The parameters of PD controller and feedforward conpensator could be designed by this linear model. Then a neural network compenstor was intrduced to deal with the modelling error between the linear model and the real model. The proposed control sheme needed only the input and output data without building the complex dynamic model of the robotic manipulator. The advantage of PD controller and good control performance could be preserved at the same time. Finally, the proposed control scheme was applied to a robotic manipulator. Experimental results demonstrates that the addressed control strategy can easily be implemented and the control precision of the robotic manioulator is guaranteed.

关 键 词：机械手 PD控制 神经网络 轨迹跟踪 前馈 

分 类 号：TP241[自动化与计算机技术—检测技术与自动化装置]