空间机器人捕获航天器操作的避撞柔顺无源神经网络H∞控制

研究了空间机器人在轨捕获非合作航天器过程避免关节受冲击破坏的避撞柔顺控制问题。为此在关节电机与机械臂之间配置了一种柔顺机构--旋转型串联弹性执行器(RSEA),可通过其内置弹簧的变形来吸收捕获过程目标航天器对空间机器人关节产生的冲击能量;结合所设计的开、关机控制策略可保证关节冲击力矩受限在安全范围内。首先利用拉格朗日方法及牛顿-欧拉法分别获得了捕获前空间机器人及目标航天器的分体系统动力学模型;之后,结合冲量定理、系统运动几何关系及力的传递规律,建立了捕获后两者形成混合体系统的动力学模型,并计算了碰撞过程的冲击力矩;最后,基于无源性理论提出了一种神经网络鲁棒H∞避撞柔顺控制策略以实现失稳混合体的镇定控制。数值仿真结果表明,配置柔顺空间机器人在捕获碰撞阶段最大可减小61.9%的关节冲击力矩,最小也可减小47.8%;而在镇定运动阶段,各关节冲击力矩均受限在安全范围内,实现了对关节有效地保护。

Passivity-based neural network H∞ avoidance compliant control of space robot capturing spacecraft

The collision avoidance compliant control for space robot on-orbit capture of noncooperative spacecraft was studied. A compliant mechanism, i.e., Rotary Series Elastic Actuator(RSEA), was mounted between the joint motor and manipulator. Its functions were as follows: first, the deformation of its internal spring could absorb the impact energy of the captured spacecraft on the joints of the space robot;second, the joint impact torque could be limited to a safe range by combining with the collision avoidance compliant control scheme. The dynamic models of the space robot and target spacecraft before capture were obtained with the Lagrange approach and Newton-Euler method. Then, based on the impulse theorem, kinematic constraints, and the law of force transfer, the dynamic model of the composite system was derived. Finally, a passivity-based neural network robust H-infinity compliant control was proposed for the post-capture composite system. The simulation results show that the space robot system with RSEA can reduce the collision impact torque of the joint during capture phase by as much as 61.9%, or by at least 47.8%. During the motion stabilization control phase, the joint impact torque is limited to a safe range to realize the effective protection of joints.