基于混合神经网络的鲁棒自适应飞行控制器的设计  

Design of a Robust Adaptive Flight Controller Based on Hybrid Neural Network

作　　者：王丽[1,2] 刘春生[1] 龚华军[1] 叶青[1] 

机构地区：[1]南京航空航天大学自动化学院,南京210016 [2]南航金城学院,南京211156

出　　处：《电光与控制》2016年第11期6-8,61,共4页Electronics Optics & Control

基　　金：国家自然科学基金(61074063)

摘　　要：针对带有不确定项和干扰项的变形翼飞行器模型,提出了一种非线性控制方案。设计在线自适应补偿算法逼近干扰项,消除环境干扰对系统的影响,提高了系统的鲁棒性。然后,通过在线自适应调整神经网络的隐藏层的权重参数,使神经网络的输出逼近系统的不确定项。利用李亚普诺夫稳定理论证明了神经网络的鲁棒控制器能使系统的跟踪误差最终一致渐近稳定,并设计出了控制器参数的在线更新律。最后,仿真结果表明了该方法的有效性。A novel nonlinear robust control scheme is proposed for a morphing vehicle with unknown uncertainties and external disturbance. An online adaptive compensation algorithm is designed for approaching the disturbance item, and the parameter updating law is used for providing automatic disturbance rejection ability in real time, which improves the robustness of the system. Then, by adaptively adjusting the weighting parameters of the hidden layer of neural network, the output of the neural network is made to approach the uncertain item. It is proved by using the Lyapunov stabilization theory that the robust controller can make the tracking error asymptotically stable. An online updating law is also designed for parameters of the controller. Simulation results show the effectiveness and feasibility of the approach.

关 键 词：自适应神经网络 变形翼飞行器 飞行控制 不确定项 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]