基于改进神经网络的空间机械臂阻抗控制方法

Impedance control of space manipulator based on improved neural network

针对环境信息不确定和碰撞模型未知情况下的空间机械臂柔顺控制问题,提出了一种基于改进型神经网络的阻抗控制方法。以空间机械臂阻抗控制系统闭环方程为基础,分析了环境信息不确定和碰撞模型未知情况下不能实现精准力控制的原因。利用粒子群优化算法调整神经网络中的权值矩阵,以提高神经网络的收敛速度和寻优性能。基于改进后的神经网络设计阻抗控制器,使改进后的神经网络能实时调整阻抗参数以达到更好的柔顺控制效果。数值结果表明,相比传统阻抗控制,该控制器能够有效减小力控制误差和位置控制误差,且对于力反馈干扰信号具有更强的抗干扰能力。

An impedance control method based on improved neural network was proposed oriented to compliance control of space manipulators under the condition of uncertain environmental information and unknown collision model.Based on the closed-loop equation of impedance control system,the reasons why precise force control can’t be achieved under the condition of uncertain environmental information and unknown collision model were analyzed.The weight matrices in the neural network were adjusted by particle swarm optimization algorithm to improve the convergence speed and optimization performance of neural network.An impedance controller based on the improved neural network was proposed,which accomplished compliance control.The improved neural network can adjust the impedance parameters on line to achieve better compliance control effect.Numerical simulation results show that the proposed controller can reduce the force control error and position control error effectively,and has a better anti-jamming capability for force feedback interference signal than traditional impedance controller.