采用神经网络的工业机器人双臂鲁棒控制方法

Robust control method for double arms of industrial robot using neural network

为了克服摩擦、干扰以及模型误差等不确定性因素对工业机器人双臂的影响,利用神经网络设计了反步鲁棒控制律。首先建立了工业机器人双臂协同控制模型,然后利用神经网络估计出干扰,并通过不确定性补偿设计出了反步鲁棒控制律,最终实现了对工业机器人双臂空间运动的精确控制。对比仿真得到的结果表明,所设计的反步鲁棒控制律对工业机器人双臂具有更高的控制精度,空间运动指令跟踪的最大误差仅为0.2 cm,不确定性估计的最大误差仅为0.2 N·m。测试实验验证了所设计的反步鲁棒控制律具有更优的工程实用性,空间定位的平均误差为0.18 cm,最大误差仅为0.24 cm,有效降低了各种干扰因素对工业机器人双臂控制精度的影响。

To overcome the influence of uncertainties such as friction,disturbance and model error on double arms of industrial robot,a back-stepping robust control law using neural network was designed.Firstly,the cooperative control model of double arms of industrial robot was established,then the neural network was used to estimate the disturbance,and a back-stepping robust control law was designed by the uncertainty compensation.Finally,the precise control of the spatial motion of the double arms of the industrial robot was realized.The results of comparative simulation show that the designed back-stepping robust control law has better control accuracy for double arms of industrial robot.The maximum error of spatial motion command tracking is only 0.2 cm,and the maximum error of uncertainty estimation is only 0.2 N·m.The test experiments verify that the designed back-stepping robust control law has better engineering practicability,the average error of spatial positioning is 0.18 cm,and the maximum error is only 0.24 cm,which effectively reduces the influence of various interference factors on the control accuracy of the double arms of industrial robot.