高超声速飞行器自适应神经网络动态面控制

Self-adaptive NN dynamic surface control of hypersonic vehicle

提出了一种新的高超声速飞行器自适应神经网络控制方法。根据飞行器纵向模型的特点,分别设计了基于直接自适应神经网络动态面控制的高度控制器和基于神经网络动态逆的速度控制器。RBF神经网络系统用于逼近高度控制器的中间控制信号,控制器只需一个更新参数,解决了神经网络逼近模型不确定性时更新参数多的问题,计算量显著减小。通过Lyapunov定理,证明了飞行控制系统半全局稳定。仿真结果表明,所设计的控制器不仅结构简单,且能保证飞行器在气动参数不确定性存在情况下具有良好的跟踪控制性能。

A new self-adaptive neural network （NN） control method is proposed for hypersonic vehicle. Based on the characteristics of the longitudinal model, the altitude controller based on the direct self-a- daptive NN dynamic surface control （DSC） and the speed controller based on the NN dynamic inversion are designed. The RBF NN system is used to approximate the intermediate control signals of altitude con- troller, and only one parameter is required to be updated. The strategy solves the problem that the num- ber of updated parameters depends on the number of the neural network nodes when NNs approximate un- certain plant model. It is proved that the developed method can guarantee the semi-global stability of the flight control system via the use of Lyapunov theorem. Simulation results show that the controller is simple in structure, and has good tracking performance in the presence of uncertain parameters.