Delta法和Zero法参数辨识的应用研究

近年来，使用神经网络进行飞机系统辨识研究成为了研究的热点。本文研究了基于神经网络的Delta法和及其改进型--Zero法估计纵向气动参数的算法。使用加噪的飞机纵向数据仿真实验发现，Delta法和Zero法估计纵向气动参数不需要建立数学模型和假设参数初值，在合理选择参数的情况下能较好的辨识气动参数，但是实验发现，应用Delta法和Zero法估计参数对于飞行数据的质量、数量要求较高，且辨识的精度并不是很理想，在应用的时候应该慎重。

Analysis on Application Problems of Delta and Zero Method

Neural network(NN) is drawing more and more attention in aircraft system identification recently.A research on NN based Delta method and its modified form-Zero method is made.The simulation works on aircraft longitude system identification illustrate the good performance of the algorithm.What’s more,without making any mathematical models in advance and exempt from initial value postulate,the two methods indeed show their superiorities.However,the network used in the two methods is vulnerable to the data,and has unsatisfying estimation accuracy.Hence,the algorithms should be used in caution.