基于自组织小脑神经网络的挠性卫星姿态控制

Attitude Control of Flexible Satellites Based on Self-organizing CMAC

研究卫星稳定性优化控制,卫星姿态控制系统是一个耦合的不确定非线性系统,在轨运行的卫星不可避免地受到模型参数不确定性和各种干扰力矩的影响,存在使挠性卫星大角度姿态机动的控制问题进一步复杂化。为了完成姿态控制任务,需要所设计的控制律具有较高的鲁棒性。采用滑模变结构控制对挠性卫星进行姿态机动控制,用神经网络对不确定性进行补偿,改变了传统的小脑神经网络补偿时实时性差的缺点,提出用自组织小脑神经网络对不确定性进行补偿,根据输入自动增加和减少节点数,并可以更新权值。对于模型的不确定性可以实时地补偿,提高了泛化能力。通过数值仿真,验证了所设计控制方法的有效性和鲁棒性,对优化卫星姿态稳定性控制提供依据。

The attitude control system of satellites is a coupled nonlinear system of uncertainties.Satellites on the track are inevitably influenced by model parameter uncertainties that make the attitude control of large angle maneuver more complicated.This task demands that the designed controllers should be highly robust.As a result,this paper adopted Sliding Mode Variable Structure Control for the attitude of flexible satellites and Neural Networks to compensate the uncertainties.This paper proposed the methods that using self-organizing CMAC to compensate the uncertainties.This network enables real-time compensation of the uncertainties of the model.This paper carried out the theoretical analysis of the proposed methods and verified its validity through simulation.The simulation shows the designed control system is robust to varied uncertainties.