An ensemble learning-based remaining useful life prediction method for aircraft turbine engine

Remaining useful life (RUL) prediction for aircraft turbine engines plays a significant role in ensuring aircraft safety. Many researches have been conducted on RUL prediction for aircraft turbine engines, however, few of them are based on ensemble learning method. The ensemble learning is a combination of several machine learning algorithms, which is capable of outperforming any of its member algorithms. This paper introduces an ensemble learning based RUL prognostic method with Euclidean distance weight for aircraft turbine engine. Three different deep learning algorithms, i.e. stacked autoencoder (SAE), convolutional neural network (CNN) and long short-term memory (LSTM), are included in the proposed ensemble prognostic method. The weight of each member algorithm is assigned based on the Euclidean distance between the predicted RUL from each member algorithm and the real RUL calculated from the training dataset. The effectiveness of the proposed method is validated based on an aircraft engine dataset generated from an aero-propulsion system simulator, C-MAPSS. The results have shown that the proposed ensemble prognostic method outperforms any of its member algorithms.