神经网络在旋翼／机身气动干扰模型中的应用

摘要

Due to the characters of helicopter itself, the interaction effects of rotor wake on fuselage show obviously nonlinearity, and are correlate with several factors. Adopting a neural network is a better way to resolve a nonlinear problem. A neural network model can be built with experimental data which are constructed as a pattern set for training of net parameters.The neural network model may be used to research aerodlynamie characters of the interaction effects of rotor wake on fuselage, and the trained model can be directly used in the design of aerodynamics and real-time simulation for helicopter. In addition to analyze the neural netrnwork modeling, the experiment about aerodynamic interaction effects of rotor wake on fuselage were simply introduced.%由于直升机自身的特点，旋翼／机身气动干扰呈现非线性，且受多种因素的影响，用神经网络来解决这一非线性问题是一个很好的办法。将旋翼／机身气动干扰试验数据构造的学习样本，对网络参数进行学习，可以得到旋翼／机身气动干扰神经网络模型，进而可以用该神经网络模型研究直升机机身受旋翼气流干扰时的空气动力特性。参数训练好的旋翼／机身气动干扰神经网络模型可直接用于直升机气动设计和实时仿真。笔者在对旋翼／机身气动干扰神经网络模型的建立进行论述外，还简要介绍了旋翼／机身气动干扰试验。