Attention-based LSTM for Remaining Useful Life Estimation of Aircraft Engines

Abstract

In a critical business sector such as the aviation industry, remaining useful life (RUL) prediction helps engineers schedule maintenance to avoid the risk of catastrophic failure in both the manufacturing and the servicing sectors. This paper attempts to review and evaluate various RUL predictive models for aircraft engines and compare their performance with a proposed Long-Short Term Memory (LSTM) method based on a data-driven machine learning approach. This study uses the C-MAPSS datasets in order to evaluate the performance and the results of each approach. The obtained outcomes show that the modified LSTM method with Attention mechanism improves the RUL prediction for aircraft engines and provides better performance.