适用于全包线的航空发动机BP网络模型的动态辨识

为克服传统的发动机动态模糊辨识中存在的辨识精度低，辨识模型应用范围窄等不足，把对非线性系统具有高度逼近能力的神经网络应用于航空发动机动态特性的辨识，从而为发动机动态辨识开辟更为广阔的道路，采用均方差归一法的处理方法和BP算法的改进算法－输出端动量BP地，以某型发动机在飞行包线内某一飞行条件下的数据作为学习样本，辨识了发同的神经网络模型，在全包线范围内对该模型进行检验，结果表明，所得的发动机动态模型在全包线内都有很高的逼近精度，而且对噪声有很强的抑制能力。

Dynamic Identification with Neural Networks for Aircraft Engines in the Full Envelope

To overcome the shortages of traditionally dynamic identification for aircraft engines, a dynamic identification method with neural network is introduced. The neural networks are trained by improved back-propagation algorithm. Standard variance normalization is used for weight-training data. An improved momentum back-propagation algorithm is developed to train the weights of the neural network in order to provide a better solution. Experimental result shows that the model can trail the aircraft engines with high accuracy in the full envelope. The research proves that neural networks identification is a suitable and promising method for aircraft engine modeling.