Bifurcated Remaining Useful Life Prediction: A Hybrid Approach for Realistic Uncertainty Characterization
The study proposes a novel hybrid framework for estimating Remaining Useful Life (RUL) in turbofan engines using C-MAPSS data. It combines an LSTM autoencoder for engine state classification with Conditional Weibull survival analysis and a Probabilistic Neural Network to capture uncertainty. This method generates physically consistent uncertainty bands, providing high-confidence predictions near failure while accurately reflecting early operational variability.