基于自适应神经网络的导弹制导与控制一体化反演设计

为使导弹具有更小的脱靶量和更快的响应时间,研究了导弹俯仰通道的制导与控制一体化设计问题。由于导弹和目标的相对运动方程以及导弹自身运动方程通常是在不同的坐标系下建立的,因此,制导与控制的一体化模型很难建立。此外,在导弹飞行过程中,很难获得全弹道的精确导弹气动参数,这使得控制器的设计也很困难。本文首先建立了一种较为合适的、紧凑的一体化模型,并进一步将其化为具有一般形式的含不确定量的级联系统。采用反演控制设计方法,保证了系统全部状态的稳定性。然后,设计了一体化自适应神经网络控制器,神经网络用来在线逼近系统中由于导弹气动参数摄动引起的匹配和非匹配不确定量。仿真结果表明了这种导弹制导与控制一体化设计方案的有效性。

Integrated Backstepping Design of Missile Guidance and Control via Adaptive Neural Network

To obtain smaller miss distance and shorter response time, the problem of integrated guidance and control design of missiles in the pitch plane is considered. As the relative motion equations and the missile dynamics equations are usually built in different coordinates, the integrated guidance and control model is difficult to buiht. During the flight, the accurate full-envelope aerodynamic parameters of the missile are not available, which makes the controller design difficult. In this paper, the proper and compact integrated model is formulated and further changed into a common form of cascade systems with uncertainties. By adopting the backstepping control approach, the stability of the entire states is guaranteed. Then an integrated adaptive neural network controller is designed and the network is employed to approxi- mate the matched and unmatched uncertainties on line brought by the variety of missile aerodynamic pa- rameters. The numerical simulation results demonstrate the validity of the design scheme.