飞机短周期运动参数的神经网络辨识

AIRCRAFT SHORT-PERIOD MOTION PARAMETER IDENTIFICATION USING NEURAL NETWORK

本文提出了一种用前馈神经网络辨识飞机短周期运动参数的新方法。该方法的特点是不需要预先假定模型 ,只需要用所获得的飞行数据训练一前馈神经网络 ,就可以用这个网络直接辨识参数的值。模拟飞行数据的仿真结果证明所提出的方法是有效的并且具有较好的鲁棒性。

A new method for estimating aircraft parameters using feedforward neural network(FNN) is proposed. The significant feature of the proposed method is that it does not require a priori postulations of an aircraft model. One may straightaway obtain the parameter estimates after the FNN has been trained using the given flight data. The method is validated and robust on the simulated flight data.