基于神经网络驾驶员模型的PIO预测分析

在PIO研究中，拟线性的McRuer驾驶员模型，因其结构简单、能反映驾驶员基本的行为特性而被广泛采用，但在通常情况下，驾驶员的行为具有高度的非线性，而运用神经网络模型可以更精确地描述驾驶员的非线性特性。以某机为例，利用试飞数据对神经网络进行训练和验证，建立了神经网络模型。通过仿真运算，并用时域Neal-Smith（TDNS）准则对该飞机进行了PIO预测分析。结果表明，与McRuer驾驶员模型相比，神经网络模型可以更好地逼近驾驶员特性，据此运用TDNS准则可以很好地对PIO趋势进行预测。

Research on PIO Based on Artificial Neural Network Pilot Model

Quasi-linear McRuer pilot model is widely used in PIO study because it has simple structure and can reflect the basic characteristics of pilot’s behavior.But in most cases,the pilot’s behavior is highly nonlinear,and neural network model can describe the nonlinear characteristics more accurately.Take an aircraft as an example,using flight test data to train and validate the neural network,and then establish a neural network model.Through computer simulation,time domain Neal-Smith(TDNS) criterion is used for predicting the PIO tendencies.The results show that,compared with the McRuer pilot model,neural network model can be a better approximation of pilot’s characteristics,under which TDNS criterion can be available for predicting PIO trends.