基座弹性的双柔杆空间机器人的神经网络动态面控制

Neural Network Dynamic Surface Control of a Two-flexible-Link Space Robot with Elastic Base

针对基座弹性的双柔杆漂浮基空间机器人系统存在外部干扰时的轨迹跟踪及柔性抑振问题,推导了系统的动力学模型,应用奇异摄动理论,将系统分解为表示刚性运动的慢变子系统和表示基座弹性、双柔杆振动的快变子系统。对于慢变子系统,设计了一种基于动态面的神经网络控制器,通过动态面的应用避免反步法带来的计算膨胀问题;通过RBF神经网络逼近了含有外部干扰在内的动力学不确定项;针对快变子系统,采用线性二次型最优控制同时抑制弹性基座与双柔杆的振动。数值仿真验证了控制方法的有效性。

The trajectory tracking and vibration suppression of a two-flexible-link and elastic base free-floating space robot system with the external disturbance were discussed.The dynamic model of the sys-tem was derived,and then a slow-subsystem describing the rigid motion and a fast-subsystem correspond-ing to vibration of elastic base and two flexible links were obtained using singular perturbation theory.For the slow-subsystem,a neural network controller based on dynamic surface was designed.Dynamic surface control scheme was adopted to avoid calculation expansion caused by back stepping method and to sim-plify calculation.The RBF neural network was applied to approximate uncertainties terms of dynamic e-quation including the external disturbance.For the fast-subsystem,an optimal linear quadratic regulator controller was adopted to damp out the vibration of the two flexible links and the elastic base.The numer-ical simulations demonstrated the effectiveness of the control method.