基于自适应迭代学习的小行星绕飞容错控制

Adaptive iterative learning based fault tolerant control for asteroid orbiting

对于小行星绕飞任务的探测器姿态控制问题,已有方法大都考虑了干扰力矩和参数不确定等因素,而忽视了执行器故障情况。针对执行器故障条件下的小行星探测器姿态控制问题,提出了一种基于自适应迭代学习的容错控制方法。所设计的控制器包括两部分:其一针对执行器故障,设计了自适应迭代学习控制器,采用类滑模的思想和自适应迭代学习算法对控制器参数进行调整,进而补偿执行器故障带来的影响,保证系统在控制输出不足情况下的高精度姿态稳定性;其二针对探测器参量变化、外部环境干扰等不确定情况,设计了基于自适应神经网络的迭代学习控制器,采用径向基函数(Radial Basis Function,RBF)神经网络对系统非线性部分进行逼近,同时对控制器参数进行自适应迭代学习调整,进而保证系统在不确定情况下的动态性能。数值仿真结果表明该控制器能够有效抑制外部环境干扰和内部参数变化带来的不利影响,在执行器部分失效甚至完全失效故障情况下,仍能保证系统的鲁棒性并实现误差在10-2数量级内的较高姿态控制精度。

Considering the probe actuator failures,parametric uncertainties and external disturbances,an adaptive iterative learning based fault tolerant control method was designedfor asteroid orbiting.The controller was divided into two parts：robust iterative learning component and neural network iterative learning component.For the robust iterative learning component,a sliding-mode-like strategy with adaptive iterative learning law was applied to maintain the stability and improve the attitude tracking accuracy in case of actuator failures.For the neural network iterative learning component,a radial basis function（RBF）neural network based adaptive approximation was introduced to estimate the system uncertainty,with parameters adapted online to maintain dynamic performance.In addition,numerical simulations show that the method achieved the error in the order of 10-2 magnitude under the actuator failure conditions,highlights the robust and high precision attitude tracking performance.