基于神经网络的故障飞机仿真

An Application Of Neural Network For Fault Aircraft Simulation

传统气动系数模型中 ,拟合法精度较差 ,插值法计算速度慢 ,且占内存多。利用神经网络一致逼近任意非线性连续函数的特性 ,训练具有一个三输入六输出的神经网络模型 ,建立故障飞机仿真系统。仿真结果和故障飞机自修复应用表明 ,文中所采用的神经网络建模方法是可行的。在自修复飞行控制系统研究中 ,为故障飞机建模所需大量故障状态气动系数数据处理提供一种新思路。

In the conventional model of aerodynamic force and moment coefficient, the precision of calculating is low by the method of polynomial fitting, and the speed of calculating is slow by the method of data interpolation by which a lot of memory is needed. Using the property of approximation any nonlinear continuous function, an artificial neural network with three inputs and six outputs was trained to mapping nonlinear relationship between the input of flight parameters and the output of aerodynamic coefficients. Using trained neural network model, a flight simulation of aircraft with failure effectors was built. With analysis of flight simulation, the neural network model work well. It is a new scheme that process a great deal aerodynamic datum for modeling of aircraft with fault effectors in the research of the reconfigurable flight control system.