RLED机器人神经网络在线辨识及控制器设计

对考虑模型误差和外界干扰信号等不确定性因素在内的电驱动刚性机器人系统,运用神经网络和基于Lyapunov函数的逆向递推法提出一类自适应跟踪控制器.神经网络用于辨识机器人系统中包括电流一阶导数在内的复杂非线性项和不确定性因素;而基于Lyapunov函数的逆向递推法则保证了控制系统的鲁棒稳定性和干扰抑制性能.通过系统稳定性分析可以得到控制器参数的选取原则,以进一步提高控制系统品质.数值仿真结果表明所设计的控制器具有良好的鲁棒性和跟踪精度.

On -line neural network identification and control of RLED robot

A neural network controller is proposed with respect to rigid link electrically driven robot with uncertainties including modeling errors and exogenous disturbances.In designing the controller,one presents a hybrid control strategy which involves Lyapunov functional backstepping method and neural networks,which aim to identify the complex nonlinear functions and the uncertainties in the robot system.Simulations prove the good robust and ability of suppressing disturbance under the controller.Also,the control performance of system can be more effective,given reasonable parameters in the controller.