基于ICA与深度学习的发动机气路系统状态监测

为提升航空发动机气路系统状态监测的有效性，提出一种采用深度学习并结合独立分量分析（independent component analysis,简称ICA）的新方法，对航空发动机气路系统的健康状态进行了监测技术研究。首先，对实际采集的航空发动机气路系统健康监测参数进行预处理，对预处理后的参数数据采用独立分量分析方法进行处理，提取代表当前状态的特征系数矩阵；其次，由提取的特征矩阵创建深度学习状态监测模型；最后，由创建的状态监测模型对航空发动机气路系统健康状态进行监测。为验证所提方法的有效性，采用典型神经网络与支持向量机分别对由主成分分析（principal components analysis,简称PCA)和ICA构建的特征矩阵进行了状态监测研究。结果表明，采用ICA和深度学习相结合的状态监测方法，可以更好地实现对航空发动机气路系统的状态监测，有良好的应用与推广前景。

Condition Monitoring of Aero-engine Gas Path System Based on Independent Component Analysis and Deep Learning

The effective monitoring of aero-engine gas path system status has always been one of the technical bottlenecks plaguing the industry. To improve the effectiveness of aero-engine gas path system condition monitoring, this paper proposes a new method combining to monitor the health of the aero-engine gas path system, the independent component analysis (ICA) and deep learning phase. Firstly, the actual collected health monitoring parameters of aero-engine gas path system are preprocessed. The preprocessed parameter data is processed by independent component analysis method to extract the characteristic coefficient matrix representing the current state. Secondly, the deep learning state monitoring model is designed and established by the extracted feature matrix. Finally, the established state monitoring model is used to monitor the health status of the aero-engine gas path system. In order to show the effectiveness of the proposed method, the traditional neural network and support vector machine are used to monitor the state of the principal components analysis (PCA) and the ICA feature extraction matrix. The research shows that the state monitoring method combining independent component analysis and deep learning can be used and it is very suitable to realize the condition monitoring of the aero-engine gas path system. The effectiveness of condition monitoring is obviously better than other methods, and it has a good application prospect.