任务空间内空间机器人鲁棒智能控制器设计

研究了基于神经网络的自由漂浮空间机器人在任务空间内的轨迹跟踪问题。首先利用RBF神经网络来逼近自由漂浮空间机器人高度非线性的动力学模型，然后设计了鲁棒控制器对逼近误差和外部干扰进行抑制。利用Lyapunov直接方法建立的新的神经网络参数和连接权值的在线学习算法，以及利用耗散理论设计的鲁棒控制器保证了系统的稳定性，并能够使系统L2增益小于给定的指标。利用该控制器对平面二连杆空间机器人进行了仿真研究，表明该智能控制方案是有效的。

Design of Robust Intelligent Controller for Space Robot in Task Space

The problem of tracking control of free-floating space robot based on neural network in task space was studied in this paper. A RBF neural network was used to approximate the highly nonlinear model of free-floating space robot and a robust controller was designed to compress the approximation errors and external disturbances. New on-line adaptation laws of parameters and weights of RBF neural network were proposed using Lyapunov direct method. The robust controller was proposed based on dissipative theory. Above these assured the stability of the whole system, and L2 gain also was less than the index. Simulation results of a two-link planar free-floating space robot verify the validity of the proposed robust intelligent controller in the presence of uncertainties and disturbances.