自由漂浮空间机器人神经网络自适应控制律设计

发布时间：2018-05-29 14:24

本文选题：控制器 + 李雅普诺夫函数　；参考：《西北工业大学学报》2017年01期

【摘要】：自由飘浮空间机器人是一种复杂的非线性系统,当考虑各种因素引起的模型不确定性和外部干扰时,传统控制已不能满足要求。为此,文章考虑不确定性干扰对系统的影响,基于神经网络在线建模技术,将系统误差和误差变化率作为状态量,利用李雅普诺夫方程设计网络权值变化率,通过训练权值对不确定部分进行逼近,在此基础上设计神经网络自适应控制项,用于修正基于名义模型的控制律,补偿不确定项造成的控制误差,并证明了控制系统的稳定性。7关节空间机器人的数值仿真结果表明,文中所设计的自适应神经网络控制器,在机械臂末端有干扰和负载的情况下,也能向期望轨迹靠拢,并在一定误差范围内保持收敛,相比于传统PID控制器,性能更优。
[Abstract]:The free-floating space robot is a complex nonlinear system. The traditional control can not meet the requirements when considering the model uncertainty and external disturbance caused by various factors. Therefore, considering the influence of uncertainty disturbance on the system, based on the on-line modeling technology of neural network, the system error and error change rate are taken as the state variables, and the network weight change rate is designed by using Lyapunov equation. The uncertain parts are approximated by the training weights, and the adaptive control terms based on neural networks are designed, which are used to modify the control law based on the nominal model and compensate the control errors caused by the uncertainties. The numerical simulation results of the stability of the joint space robot show that the adaptive neural network controller designed in this paper can approach the desired trajectory under the condition of interference and load at the end of the manipulator. The performance of the controller is better than that of the traditional PID controller.
【作者单位】：西北工业大学航天学院;航天飞行动力学技术国家级重点实验室;
【基金】：国家自然科学基金(11472213)资助
【分类号】：TP242

【相似文献】