改进神经网络自适应滑模控制的机器人轨迹跟踪控制

摘要

为了提高机器人轨迹跟踪控制性能，在神经网络滑模控制方法的基础上，提出了一种改进型神经网络自适应滑模控制方法。该方法将神经网络作为控制器，利用其非线性映射能力来逼近各种未知非线性，同时通过在控制律中加入鲁棒项来消除逼近误差。考虑到隐含层单元数和网络结构参数对神经网络映射有效性的影响，将降低抖振作为优化目标，采用粒子群优化算法对网络结构参数进行优化。最后在 Matlab/Simulink 环境下进行了仿真实验，并与其他控制方法进行了对比分析。仿真结果表明，基于该方法所设计的控制系统具有良好的鲁棒性和控制精确度，同时有效地削弱了抖振。%In order to improve the traj ectory tracking control performance of the robot,a modified neural network adaptive sliding mode control method is proposed on the basis of the neural network sliding mode control method.This method uses neural network as a controller,and uses the nonlinear mapping ability of neural network to approximate unknown nonlinearity.At the same time,the robust control law is added to eliminate the approximation error.Considering the influence of the hidden layer unit number and the network structure parameters on the validity of neural network mapping,reducing chattering is regarded as optimization target,and particle swarm optimization algorithm is adopted to optimize the network structure parameters.Finally,the simulation experiment is done under the environment of Matlab/Simulink,and comparative analyses with other control methods are conducted.The simulation results show that the control system designed by the proposed method has good robustness and control precision,and can reduce the chattering efficiently.