Demonstrating Aleatoric Uncertainty in Remaining Useful Life Prediction Using LSTM with Probabilistic Layer

A Remaining Useful Life prediction with Aleatoric uncertainty is presented in this paper.A Long Short-Term Memory (LSTM) architecture with probabilistic layer is employed where a normal distribution layer is incorporated to produce the predicted Health Index (HI) distribution of turbofan engines.Compared to the performance of other point estimates techniques in the literature, the probabilistic LSTM achieved a competitive performance in predicting the turbofan�s RUL and RUL sequence and have the advantage to express the level of uncertainty along its sequence prediction.This work is important as it reflect a real-world deep learning application where uncertainty indication is needed to evaluate prediction for important decision-making process. © 2023, Institute of Technology PETRONAS Sdn Bhd.