高超声速飞行器自适应神经网络控制

针对一类高超声速飞行器,在充分考虑其非线性模型包含未建模动态、气动参数变化、弹性形变等产生的未知非线性不确定函数以及外界扰动的情况下,设计了一种基于自适应神经网络的非线性逆控制器。首先,将系统的动态特性分为标称部分和不确定部分,采用非线性逆的思想设计标称部分的控制器,利用神经网络逼近不确定部分,将神经网络的最优权值采用自适应律进行调节,提高神经网络的在线逼近能力。利用改进的变结构控制来消除神经网络逼近误差的影响,最终使跟踪误差收敛为零,并保证闭环系统的信号有界。通过仿真验证了设计方法的正确性。

Adaptive neural network controller design for hypersonic vehicle

A nonlinear inversion-based controller with adaptive neural network compensation was designed for a class of hypersonic vehicle in the case that vehicle's nonlinear model comprises of nonlinear unknown uncertain function caused by unmodeled dynamics,aerodynamic parameter variation and unknown external disturbances were fully considered.Firstly the dynamic characteristics of system were separated as two components:a nominal component and an uncertain component.The controller for the nominal component was designed based on nonlinear inversion control scheme,and uncertain component was approximated by the neural network.The optimal weight matrix was adjusted by adaptive learning law to improve the online approximated capability of neural network.The modified variable structure controller was designed for eliminating the effects of the neural network approximated errors.The controller can guarantee that the tracking error converges to zero and all signals of closed loop system are bounded.The simulation results demonstrate the rightness of the proposed schemes.