基于神经网络自适应算法的机械臂位置/力控制

Mechanical Arm Position/Force Control Based on the Neural Network Adaptive Algorithm

针对与柔性环境相互作用的机械臂位置/力控制中的建模误差和工作过程中存在的外界干扰的问题,设计一种基于神经网络的自适应控制策略。首先,对机械臂系统进行从关节空间变换到工作空间的坐标变换,结合环境刚度将坐标分解为切向位置空间和法向接触力空间;然后,考虑到外界干扰和建模误差等不确定因素,采用神经网络对不确定项进行逼近,加入鲁棒项减少逼近误差;最后,经过仿真验证得出在位置跟踪和力控制上能够快速收敛,说明了该方法的实用性和有效性。

An adaptive control strategy based on the neural network is designed to address the modeling error and the external interference in the mechanical arm position/force control interacting with flexible environment.Firstly,the mechanical arm system transforms from joint space to workspace and decomposes the coordinates into tangential position space and normal contact force space combined with the environmental stiffness.Then,considering the uncertain factors such as external interference and modeling errors,the neural network is used to approximate the uncertain term,and the robust term is added to reduce the approximation error.Finally,the simulation results illustrate that it can converge quickly in position tracking and force control,which shows the practicability and effectiveness of this method.