基于BP网络的飞行器解耦设计

摘要

根据BP神经网络在函数逼近方面的能力,将BP神经网络应用于飞行器系统解耦.在解耦过程中,通过神经网络离线训练标准数据,得到输入输出的函数关系,从而可以得到解耦中的各个参数.根据气动力矩矩阵的行半径系数的大小来验证耦合的强弱.仿真结果表明,BP网络对控制系统的解耦效果是明显的.%Based on the BP neural network's approximation theory, a BP neural network is used to be applied to the decoupling of aircraft's control system. In the process of decoupling, the network gets the nonlinear function of the input and output by training the standard data. Based on the nonlinear function, we can get the parameters of decoupling. The simulation shows the efficient decoupling.