中国空间科学技术

中国空间科学技术 ›› 2021, Vol. 41 ›› Issue (1): 64-74.doi: 10.16708/j.cnki.1000-758X.2021.0008

• 研究探讨 • 上一篇    下一篇

挠性航天器多目标鲁棒姿态控制的DPSO算法实现

王梦菲，张军   

  1. 1.北京控制工程研究所, 北京100190
    2.空间智能控制技术国家级重点实验室，北京100190
  • 发布日期:2021-02-02 出版日期:2021-02-25

Multi-objective robust attitude control via DPSO algorithm for flexible spacecraft 

WANG Mengfei，ZHANG Jun   

  1. 1.Beijing Institute of Control Engineering, Beijing100190， China
    2.National Laboratory of Space Intelligent Control, Beijing100190, China
  • Online:2021-02-02 Published:2021-02-25

PDF (PC)

476

摘要： 复杂航天器高性能姿态控制是完成现代新型空间任务的基础，需兼顾鲁棒性、快速性、精度和控制能量等多目标要求，但目前大多数控制系统只针对某单一目标设计。针对大型挠性航天器多目标姿态控制问题，提出一种基于差分粒子群优化算法和输出反馈的鲁棒控制方法。首先，推导了含参数不确定性的系统动力学模型；然后，给出了差分粒子群优化算法的定义和鲁棒D-稳定的线性矩阵不等式（LMI）表达；最后，在区域极点约束和Pareto最优原则下，利用所提算法对干扰抑制和控制能量指标进行了优化，得到反馈增益矩阵。该方法满足了系统多目标约束要求，且具有一定的振动抑制作用；可避免传统带极点配置的LMI方法在解决多目标问题时的保守性，也解决了将多目标转化为一个指标函数时加权系数的选择困难。数学仿真验证了该方法的有效性，相比于传统PID控制，干扰下姿态稳态误差可减小约54%。

关键词: 挠性航天器, 区域极点配置, 鲁棒控制, 差分粒子群优化算法, 多目标

Abstract: Attitude control system with high performance for complex spacecraft is the foundation of modern space mission, and multiple objectives such as robustness, convergence speed, accuracy and control energy are required. However, most of the current control systems are designed for a single objective. Aiming at the problem of multi-objective attitude control for large flexible spacecraft, a robust design method based on differential particle swarm optimization algorithm and output feedback was proposed. Firstly, the dynamic model with parameter uncertainty was derived. Then, the differential particle swarm optimization algorithm and the linear matrix inequality (LMI) expression of robust D-stability were given. Finally, under the regional pole constraint and Pareto optimal principle, the proposed algorithm was used to optimize the objectives about disturbance suppression and control energy. The feedback gain matrix was obtained. This method satisfies the requirement of multiobjective constraints and has certain effect on vibration suppression. In the multi-objective problem with regional pole assignment, it avoids the conservatism of the traditional LMI method, and also solves the difficulty of selecting the weighting coefficient when transforming multiple objectives into one index function. A simulation example illustrates the effectiveness of the proposed method. Compared with the traditional PID control, the steady-state error of attitude can be reduced by about 54% under disturbance.

Key words: flexible spacecraft, regional pole assignment, robust control, differential particle swarm optimization algorithm, multi-objective