弹性高超声速滑翔飞行器的状态/参数联合估计

弹性高超声速滑翔飞行器具有强非线性、强不确定性和刚体/弹性耦合的特点，对其状态和参数进行估计十分必要。为解决这一问题，提出了一种传感器布置策略和一种利用正交三角（QR）分解更新到达代价的滚动时域估计算法（MHE-QR）。首先，建立了考虑弹性的传感器观测模型并分析了传感器位置对可观性的影响，并在此基础上提出了一种反映系统可观性的性能指标。传感器布置策略以此性能指标为目标函数，将传感器布置问题转化为约束非线性优化问题并求解，即可得到最优传感器布置方案。然后提出了MHE-QR算法。在滚动时域估计的框架下，该算法利用前向动态规划原理将到达代价的计算转化为最小二乘问题，并给出了基于QR分解的到达代价更新算法。仿真结果表明该传感器布置策略和MHE-QR算法能够有效提高估计精度、收敛速度和计算速度。此外，MHE-QR算法具有实时应用的潜力。

State/parameter joint estimation for flexible hypersonic glide vehicles

It is essentially to jointly estimate the state and parameter of flexible hypersonic vehicles due to its nonlinearity, uncertainty, and rigid/elastic coupling. To solve these problems, a sensor placement strategy and a Moving Horizon Estimation with arrival cost updated by QR decomposition (MHE-QR) are proposed. First, the influence of sensor placement on observability is analyzed, based on which a performance index is proposed. Using this performance index, the sensor placement problem is transformed into a constrained optimization problem. The sensor placement scheme is obtained by solving this optimization problem. Utilizing the dynamic programming principle, the MHE-QR algorithm transforms the arrival cost update problem into a least square problem that is solved by QR decomposition in the framework of MHE. The Monte Carlo simulation results demonstrate that the sensor placement strategy and the MHE-QR algorithm can effectively improve the estimation accuracy, convergence speed, and computation rate. Additionally, the CPU time validate the real-time applicability of MHE-QR.