Scientific Machine Learning for Engine Health Management and Remaining Useful Life Prediction

In three linesScientific ML framework for turbine Remaining Useful Life (RUL) prediction. Shared encoder (CNN + bidirectional LSTM + attention pooling) with task-specific heads predicts turbine gas temperature, Delta TGT, and RUL with quantified uncertainty intervals. Evaluated on heterogeneous real-world fleet data using MAE, PICP, MPIW, and coverage-width criterion metrics.

Summary generated by Claude — human-verified