基于RBF的机械手无需模型自适应控制研究

Research of self-adaptive control for robotic manipulator without model based on RBF

针对机械手存在的扰动等未知模型,提出了基于RBF神经网络的自适应控制策略。采用RBF神经网络对机械手动力学模型在线自学习,并根据Lyapunov稳定性理论建立了网络权值自适应学习律,确保了网络逼近误差的收敛及系统的稳定。以平面转动双臂机械手轨迹跟踪为例进行仿真,结果表明该方法能够有效地补偿建模误差,实现了无需模型的机械手自适应控制,提高了系统的控制性能及对外部不确定扰动的鲁棒性,对实际工业机械手的自适应控制具有一定的可操作性。

According to the known modeling as perturbation and etc.with the robotic manipulators,a new self-adaptive control strategy based on RBF neural network has been proposed.The dynamical model of robotic manipulators are learned online by RBF neural network,and the adaptive learning law of network weights is developed based on Lyapunov stability theory,therefore the convergence of the network approximated error and stability of the system are guaranteed by weights tuning on-line.Simulation was carried out by taking the track following of the double arm manipulators as an example.Simulation results show that this method can effectively compensate the modeling error,allowing the self adaptive control of the manipulators without know model,improving the control performance and the robustness of external uncertain perturbation of the system,having certain operability for the self adaptive control of the actual industrial manipulators.