基于数据驱动的弹性高超声速飞行器控制方法

高超声速飞行器存在气动非线性强、复杂振动干扰等特点，参数不确定性大条件下传统依赖于精确模型的控制方法品质下降明显，需要进一步提高控制系统在线适应能力。本文针对弹性高超声速飞行器过载跟踪性能在线优化和弹性振动影响下的控制参数优化问题，提出了一种基于数据驱动的自学习控制方法，首先将高超声速飞行器输出反馈控制问题转化为状态反馈形式，采用鲁棒自适应动态规划算法设计了适用于过载跟踪问题的无模型控制参数在线优化方法，然后针对飞行器复杂弹性振动干扰的问题，提出了基于陷波滤波器的自适应动态规划控制方法，从而保证了振动影响下的控制参数在线优化效果。仿真结果表明，在不依赖于准确模型参数的条件下，本文所提的方法能够有效实现弹性振动干扰下的控制参数在线优化，并提高过载跟踪控制品质。

Data-Driven Method of Flexible Hypersonic Vehicle Control

The hypersonic vehicle has strong aerodynamic nonlinearity and complex vibration disturbance, making the control quality of the traditional model-based control methods decrease significantly. It is necessary to improve the online adaptability of the control system. In this paper, a data-driven self-learning control method is presented to optimize the acceleration tracking performance under the influence of elastic vibration. First, the output feedback control problem of the hypersonic vehicle is transformed into a state feedback form. A model-free robust adaptive dynamic programming algorithm is used to design an online optimization method for control parameters. Then, an adaptive dynamic programming control method based on a notch filter is presented to solve the complex elastic vibration problem, which ensures the accuracy of online optimization. The simulation results show that the method proposed in this paper can effectively optimize the control parameters online under the influence of elastic vibration and improve the quality of acceleration tracking control without depending on the exact model parameters.