Physics-Informed Machine Learning for Predicting Remaining Useful Life of Aircraft Engines: A Robust Framework for Risk-Aware Maintenance

Aircraft components degrade over time, directly impacting reliability and performance. This project aims to develop a machine learning framework for predicting the Remaining Useful Life (RUL) of aircraft engines based on comprehensive life cycle data, to optimize maintenance strategies and mitigate risks associated with component failure. A diverse set of regression and classification models-including K-Nearest Neighbors (KNN), Naïve Bayes, Random Forest, and Support Vector Machine (SVM)-are implemented and evaluated on NASA’s C-MAPSS dataset to estimate engine lifetimes. The key focus is to predict the accurate RUL for turbofan engines, particularly under High-Pressure Compressor (HPC) failure scenarios, and specifically targeting low RUL values to prevent critical risks. Physics-informed machine learning approach incorporates enhanced predictions by improving the robustness of maintenance planning.