基于前向神经网络的导弹自适应控制方法

Design Method of a Kind of Missile's Flight Control System Based on Feedforward Neural Networks

结合导弹非线性动力学模型,介绍一种基于经典控制和神经网络自适应控制的系统设计方法。该方法以导弹飞行的攻角α和Ma数等为信息,应用BP网络的学习能力求出具有自适应能力的控制参数,应用于参数时变的导弹控制系统。其特点是避免了传统自适应方法所要进行的繁琐的参数辨识计算。利用Matlab仿真实验表明,所设计的控制系统性能良好,而且易于工程实现。

The classical adaptive control system is a nonlinear time-varing feedback system, in which the process is very complicated. This paper introduces a method of adaptive control based on neural networks and classical control theory. Using the information such as the missile's Mach incidence etc, neural networks are used for optimizing the parameters of control system. This way avoids estimating parameters perplexingly .By simulation and analysis with the nonlinear time-varing dynamical model of missile, it is proved that the method is feasible.