基于神经网络的7-DOF机械臂力位跟踪算法

7-DOF Manipulator Force Position Tracking Algorithm Based on Neural Network

为解决7-DOF机械臂在运动过程中受到外界扰动及接触力所产生的抖振问题,给出了机械臂带有约束力的动力学方程,利用7-DOF机械臂的冗余解特性建立了约束方程以简化约束参数,提出了具有时变约束状态的神经网络自适应力位控制方法,由神经网络训练求出最优权重;设计控制律,建立李雅普诺夫函数方程和不对称项推导其收敛性;利用Simulink仿真并与无约束项的RBF神经网络控制相比较,根据机械臂的力位跟踪结果分析了角位移、角速度、约束力、控制力矩以及扰动拟合。系统仿真结果表明在有扰动和接触力的情况下,该方法可以更有效地抑制抖振现象并完成力位跟踪。

In order to solve the chattering problem of 7-DOF manipulator caused by external disturbance and contact force in the process of motion, the kinetic equation with constraint force is given. The constraint equation is established by using the redundant solution characteristics of 7-DOF manipulator, so as to simplify the constraint parameters. A neural network adaptive force-position control method with time-varying constraint state is proposed. And, the optimal weight is obtained by neural network training. Control law is designed,Lyapunov function equation and asymmetric term are established and deduce its convergence. By using Simulink simulation and comparing with RBF neural network control without constraints, angular displacement,angular velocity, constraint force, control torque and disturbance fitting are analyzed according to force position tracking results of manipulator. The system simulation results show that the method can be effectively suppressed and the force-position tracking can be completed under the condition of disturbance and contact force.