涡喷发动机风车启动工况的神经网络建模

Turbojet modeling in windmilling based on radial basis function networks

弹用涡喷发动机的风车启动工况是复杂的非线性过程 ,由于此时压气机处于非设计工况 (膨胀 )而造成机理建模的困难。神经网络对于非线性映射具有任意逼近能力 ,应用径向基函数神经网络 (RBFN)对涡喷发动机风车启动阶段进行了实验建模 ,通过适当地选取网络参数及训练样本 ,达到了很高的精度 。

The windmilling process of missile turbojet is such a complex nonlinear process that to obtain its dynamic model theoretically is very difficult because the compressor works in expending mode (non-normal operating mode) in this condition. Considering the great capacity of handling nonlinearity of the neural network, an experimental model of the windmilling process using radial basis function networks (RBFN) was established and a good precision through selecting the parameters and the training samples of the network properly was gained. The neural network model is of great value for computing the point of ignition or simulating the windmilling process.