一种基于神经网络的非线性飞行控制系统设计方案研究

在全状态反馈的前提下，设计了一种基于在线神经网络和反馈线性化的非线性直接自适应控制器。本文首先利用多重尺度摄动与动态逆技术结合，设计了无人驾驶飞机的解析动态逆控制器；然后引入一个单隐层在线神经网络来修正各种因素引起的状态误差，并证明了控制器的稳定性。最后对在线网络的实现做了详细描述。仿真分析表明，该方案具有很强的鲁棒性和对故障状态的适应性。

Design Scheme of a Nonlinear Flight Control System Based on Neural Network

A nonlinear direct adaptive controller based on online neural network and feedback linearization is designed by precondition of all state feedback. Firstly, an analytic dynamic inversion control law of uninhabited aerial vehicle (UAV) is investigated by the combination of two time-scale separation and dynamic inversion method. Secondly, in order to eliminate the state error, an online single hidden layer neural network is introduced and the stability of the whole system is proved. And then, the structure of the online neural network is described in detail. Finally, the robustness and the reconfigurable ability to fault of the scheme are verified by the simulation.