SUPERVISED LEARNING BASED APPROACH FOR TURBOFAN ENGINES FAILURE PROGNOSIS WITHIN THE BELIEF FUNCTION FRAMEWORK

Recent developments in maintenance modeling, powered by data-driven approaches such as Machine Learning (ML), have enabled a wide range of applications. For example, industrial systems coming with a huge operating database make ML an ideal candidate for their predictive maintenance (PdM). PdM is the process of predicting malfunctions using data from equipment monitoring and process performance measurements. Indeed, PdM and ML have developed a very strong connection. However, it is not always easy or straightforward to perform effective predictive maintenance for several reasons such as imperfect data. Therefore, we aim, during this paper, to manage uncertainty and/or imprecision during learning, using an evidential supervised learning approach built on a powerful framework called the belief function theory. This research work is applied on NASA’s C-MAPSS dataset for turbofan engines failure prognosis.