@article{article_1822721, title={Enhancing Aircraft Safety through Predictive Monitoring of Engine Health}, journal={Ege Üniversitesi Ulaştırma Yönetimi Araştırmaları Dergisi}, volume={2}, pages={21–42}, year={2025}, url={https://izlik.org/JA79MA37DY}, author={Deniz, Mehmet}, keywords={Kalan faydalı ömür (RUL), Topluluk öğrenmesi, Kestirimci bakım, NASA C-MAPSS veri seti, Turbofan motor}, abstract={This study proposes an RMSE-calibrated (Root Mean Square Error) ensemble framework for Remaining Useful Life (RUL) prediction of turbofan engines using the NASA C-MAPSS dataset. The approach integrates CatBoost, XGBoost, and Random Forest regressors within a weighted ensemble optimized for bias-variance trade off. A 30-cycle sliding window and z-score normalization were employed to construct a robust feature space from multivariate sensor streams. Additionally, a cross-validated RMSE-based bias correction was applied to mitigate systematic underestimation at longer RUL horizons. Experimental results on the FD001 subset demonstrate that the proposed ensemble achieves an RMSE of 13.72 and an R² of 0.8909, outperforming individual learners and several reported benchmarks. The model exhibits stable generalization and a well-balanced error distribution, confirming its suitability for real-time prognostic health management (PHM) applications. The findings highlight that carefully calibrated ensemble learning can match or exceed deep learning approaches in accuracy while retaining full interpretability and computational efficiency.}, number={Aviation Technologies and Applications Conference (ATAConf’25) Special Issue}