摘要
利用人工神经网络对某型航空发动机在全飞行包线内的稳态、动态特性(非加力状态)进行建模,网络的训练样本来自该发动机在高空试车台的试车数据。模型的输入量包括燃油流量、喷口面积、高度及马赫数,输出量则包括高、低压转子的转速及各截面的温度和压力。计算表明,当具有足够的训练样本时,神经网络模型不仅在整个飞行包线内具有较高的稳态、动态精度,而且具有较好的实时性。
A neural network model of an aeroengine is established based on experimental data, and steady and transient characteristics of the engine can be simulated. The model inputs include the fuel flow, the nozzle area, the altitude and Mach number, and model outputs include the rotor speed, the temperature and the air pressure. Calculational results show that the neural network model can satisfy the accuracy on whole flight envelope as long as the network is trained by sufficient samples.
出处
《南京航空航天大学学报》
EI
CAS
CSCD
北大核心
2007年第1期26-29,共4页
Journal of Nanjing University of Aeronautics & Astronautics
关键词
发动机
神经网络
模型
试车数据
aeroengine
neural network
model
experimental data
