航空发动机排气温度基线建模新方法研究

Investigation on New Method for Baseline Modelling of Aeroengine Exhaust Gas Temperature

关键词:  航空发动机  堆叠降噪自编码器  支持向量回归机  排气温度  基线模型

Key words:  Aeroengine  Stacked denoising auto encoder  Support vector regression  Exhaust gas temperature  Baseline modelling

摘要:

为实现航空发动机气路性能诊断与预测，提出一种基于堆叠降噪自编码器（Stacked denoising auto encoder，SDAE）和支持向量回归（Support vector regression，SVR）相结合的航空发动机排气温度（Exhaust gas temperature，EGT）基线建模方法。以CFM56-7B发动机实际采集的飞行数据作为原始数据样本，利用SDAE进行数据特征提取和降噪处理后，将提取到的非线性特征作为SVR网络的输入，建立排气温度基线模型。利用同型号的另一台发动机航后数据对所建立的排气温度基线模型进行验证，并与基于单一网络的基线模型进行对比。结果表明，基于SDAE-SVR融合模型的基线建模方法具有更强的鲁棒性和更高的预测精度。

Abstract:

For realizing the aim of real-time monitoring and health management of aeroengines overall performance， a combination model of stacked denoising auto encoder （SDAE） and support vector regression （SVR） was proposed to set up exhaust gas temperature （EGT） baseline model. Taking a commercial aero engine CFM56-7B as an example， the measured flight data during the aircraft operation was used as the original data， and the SDAE was employed to data features extraction and noise reduction， the learned nonlinear features were used as the input of the SVR network to further improve the fitting and prediction ability of the baseline model. At the same time， the proposed model was verified on the basis of another flight data of the same type of engine， and compared with other baseline models based on a single network. The results show that the exhaust gas temperature baseline model based on the SDAE-SVR hybrid model has stronger robustness and higher prediction accuracy.