双臂空间机器人捕获航天器后的镇定运动分块滑模自适应神经网络控制

讨论了双臂空间机器人捕获航天器后闭链混合体系统姿态和关节的受扰运动镇定控制及双臂协调操作问题。根据冲量定理及闭环约束条件，分析了双臂空间机器人捕获操作结束后受到的冲击效应，建立了闭链混合体系统动力学方程。以此为基础，针对星载计算机运算能力有限的问题，基于神经网络控制理论设计了分块滑模自适应控制方案。将混合体系统动力学方程考虑为多个非线性关联子系统的集合，因此可同时并行地利用神经网络来分别逼近各子系统未知动力学模型，即通过并行处理来提高计算效率。设计了自适应滑模控制器来抵消交联项及神经网络逼近误差的影响。为了保证各臂杆的协同操作，基于最小权值范数法分配了各臂关节控制力矩。通过系统数值仿真验证了所提控制方案的有效性。

Partitioned Sliding Mode Adaptive Neural Network Control of Calm Movements of Dual-arm Space Robot after Capturing a Spacecraft

The calm control for attitude and joints disturbed motions and the coordinate operation problems were discussed. At first, according to the theorem of impulse and the closed-loop constraints, the impact effects of dual-arm space robot capturing a spacecraft were analyzed and the dynamics equations of composite closed chain system were obtained simultaneously. Then, in order to satisfy the compute capacities of spaceborne computer, a partitioned sliding mode adaptive control scheme was designed based on neural network. The dynamics of combined system was considered as a set of nonlinear interconnected subsystems. Concurrent neural networks were applied to approximate the unknown dynamics of the subsystems, the computational efficiency was improved by concurrent processes. The partitioned sliding mode adaptive controller was designed to eliminate both interconnection terms and approximation errors. The weighted minimum-norm theory was introduced to distribute torques guarantee that the cooperative operation between manipulators. At last, numerical examples demonstrated the effectiveness of the proposed control scheme.