高超声速飞行器再入初始段建模与姿态控制的研究

The attitude dyanmics of hypersonic vehicles is characterised by strong coupling, high nonlinearity, large-scale parameter varying and various uncertainties, which inspires us to the research of interconnected characteristic modelling of a class of multi-input-multi-output linear time-varying and nonlinear time-varying systems. In this research, the relationship between coupling structure in the plant and interconnection structure in the characteristic model is analysed, and the interconnected characteristic models over a minimal graph and an optimal graph are concluded. Stability of the system under decentralized or distributed golden-section adaptive control law is proved. Moreover, based on model predictive control, improvements are made to the golden-section adaptive control law so as to accomandate input constraints. The interconnected characteristic model and the advanced golden-section adaptive control law are applied to the initial re-entry phase of a hypersonic vehicle with a high lift-to-drag ratio, and the overall systems under different kind of control strategies are eveluated by adaptivity, robustness and feasibility. Those experiments are carried out on an animated simulation testbed. In the two subsequential failures of the well-known hypersonic vehicle HTV-2, the anomalies were detected either before the gliding phase or when the vehicle was about to start the gliding phase. This phenominal inspires us to carry out this research, with the hope to provide the attitude system of the hypersonic vehicle with both solid theoretical basis and reliable control strategy.

针对无动力滑翔式高超声速飞行器姿态系统的三通道强耦合、非线性、参数大范围变化且难以估计、不确定性广泛存在的特点，研究多输入多输出线性时变和非线性时变系统的关联特征建模，分析耦合结构在关联特征模型中进行合理压缩的一般性规律，得到具有最小关联结构和具有最优关联结构的关联特征模型。在此基础上，给出具有分散式或分布式结构的黄金分割鲁棒自适应控制的稳定性证明。进一步结合滚动时域优化思想，解决控制受限下黄金分割鲁棒自适应控制器的设计方法。将上述控制方案应用于高超声速飞行器再入初始段姿态控制中，着重从自适应性、鲁棒性和工程易于实现的角度对不同关联结构下的控制策略进行改进和评估；设计高超声速飞行器可视化仿真演示平台。举世瞩目的美国高超声速飞行器HTV-2连续两次试飞失败都出现于再入初始段，本项目正是在这样的背景下提出的，以期为未来的高超声速飞行器姿态控制提供理论依据和可靠的控制方案。

高超声速飞行器在再入大气层过程中,要经历大范围的温度变化、大气密度变化以及速度变化，同时地面风洞实验无法提供可靠的气动数据，因此在飞行器姿态控制系统的设计时，需要考虑非线性、模型参数摄动、未建模动态和强耦合等特点，一方面需要建立尽可能精确的面向控制的数学模型，另一方面需要设计自适应性和鲁棒性强的智能控制算法。.本项目针对滑翔式高超声速飞行器再入过程横侧平面强耦合的特点，研究MIMO时变系统的特征建模和鲁棒自适应控制问题。着重从以下三方面开展理论研究：.1. 针对横侧平面强耦合特点，建立了关联特征模型。其中关联特征模型的参数界难以给出解析计算方法，因此进一步研究了黄金分割系数与模型参数界的制约关系，以闭环系统的普半径最小或对估计误差的鲁棒性最强为优化目标，得到了不同的模型参数界下黄金分割系数的一般化选取方法；.2. 为了提高闭环系统的鲁棒性，对黄金分割控制器中的可调参数\lambda进行优化，通过将绝对偏差建模为外界扰动，结合H2/Hinfty鲁棒控制思想，得到了黄金分割鲁棒自适应控制器的设计方法；.3. 为了提高关联特征模型对耦合及飞行环境的适应性，结合多模型自适应控制的思想，针对不同的机动特点，将特征模型参数区间划分为若干个合理的子区间，根据准则函数，从固定模型和可重置的自适应模型中选取合适的控制器实施控制。