基于融合神经网络的航空发动机剩余寿命预测

Remaining Useful Life Prediction of Aeroengine Based on Fusion Neural Network

关键词:  航空发动机  剩余寿命  卷积神经网络  长短期记忆网络  深度学习

Key words:  Aeroengine  Remaining useful life  Convolutional neural network  Long-short-term memory networks  Deep learning

摘要:

航空发动机的性能退化是影响飞机飞行安全的重要因素。准确预测发动机的退化过程，对于飞机安全飞行具有重要意义。针对航空发动机剩余寿命预测问题，提出了一种将卷积神经网络和长短期记忆网络相融合的数据驱动模型。与常规使用单一的神经网络不同，所提出的融合模型结合了两种神经网络的优点，利用卷积神经网络提取数据中的空间特征并采用长短期记忆网络提取时间特征。实验结果证实，在寿命预测中，将提出的数据驱动模型与已有的方法相比，评分和均方根误差分别下降了32%和8.3%。可见，所提出的数据驱动模型可对数据中所包含的信息进行充分挖掘，其对航空发动机寿命预测精度较高，并具有良好的稳定性。

Abstract:

The performance degradation of the aeroengine is an important factor that affects the flight safety of the aircraft. Accurately predicting the degradation process of engines is of great significance for the safe flight of the aircraft. Aiming at the remaining useful life prediction of the aeroengine， this paper proposes a data-driven model that combines convolutional neural networks and long-short-term memory networks. Different from the single neural network， the proposed fusion model combines the advantages of the two neural networks. The convolutional neural network can be used to extract the spatial features in the data and the long short-term memory network is used to extract the temporal features. The experimental results show that， in the life prediction， compared with the existing methods， the score and the root mean square error of the proposed data-driven model have been reduced by 32% and 8.3%， respectively. Therefore， the proposed data-driven model can fully mine the information contained in the data， and it has high accuracy and good stability for the life prediction of the aeroengine.