Abstract: A decentralized fault-tolerant control method was designed based on non-singular terminal sliding mode for a space robots subjected to partial losses of actuator effectiveness. Firstly, the dynamics equation of the system was established by using linear momentum conservation law and Lagrangian method. Subsequently, the system was decentralized into several subsystems based on the local informations of the carrier and joints, and the dynamics equations of subsystems were obtained. Then the effective factors were separated from the subsystems' dynamic equations, and the adaptive decentralized neural network was used to estimate the separated variables in real time. According to the estimation results, the corresponding control law was designed online to counteract the influences of the actuator faults on the stability of system and ensure good trajectory tracking performance. The asymptotic stability of the whole closed-loop system may be guaranteed under the proposed control scheme via Lyapunov function method. The simulation results demonstrate the effectiveness of the proposed control method.

Key words: space robot, actuator fault, terminal sliding mode, decentralized fault-tolerant control

摘要： 针对空间机器人在太空作业时执行器发生部分失效故障的问题，设计了一种基于非奇异终端滑模的分散容错控制方法。根据线动量守恒定律与拉格朗日法建立了系统的动力学方程，然后基于载体和关节的局部信息将系统进行分散，从而得到子系统的动力学方程;将子系统动力学方程中表示执行器故障程度的有效因子进行变量分离，再利用自适应分散神经网络对分离后的变量进行实时估计，根据估计结果在线设计控制律以消除执行器故障对系统稳定性的影响，保证良好的轨迹跟踪性能。通过Lyapunov函数法证明了该控制方案能保证整个闭环系统的渐进稳定性。仿真结果验证了控制方法的有效性。

关键词: 空间机器人, 执行器失效, 终端滑模, 分散容错控制