基于混合轨迹的柔性空间机械臂神经网络控制

摘要

讨论了基于混合轨迹的漂浮基柔性空间机械臂的栽体姿态、关节协调运动及柔性振动的滑模神经网络控制器设计问题.为了对柔性空间机械臂系统的柔性振动进行主动控制,基于虚拟控制力观念生成了能同时反映姿态、关节运动的期望轨迹和柔性振动的混合轨迹.基于该混合轨迹,利用神经网络逼近系统未知部分,提出了柔性空间机械臂协调运动的滑模神经网络器控制方案,该控制方案可在保证轨迹跟踪的同时对柔性杆的振动进行主动抑制.对一平面刚柔空间机械臂系统进行了数值仿真,仿真结果证实了所设计控制器的有效性.%A sliding mode-neural network scheme based on hybrid trajectory was proposed to track the hybrid trajectory of a space flexible manipulator, meanwhile, the flexible vibration can be suppressed. In order to actively suppress the flexible vibration, the hybrid trajectory for sliding mode -neural network controller was generated by using the virtual control force concept, which reflected the flexible variables as well as the rigid desired trajectories. Based on the hybrid trajectory, the new sliding mode-neural network control can be designed. Simulations results confirm that the proposed control scheme can dominates the trajectory tracking the coordinated motion and actively suppress the vibrations in the presence of parameter uncertainty.