Physics-Informed Transfer Learning for Predicting Engine Oil Degradation and RUL Across Heterogeneous Heavy-Duty Equipment Fleets

Publikation: Beitrag in FachzeitschriftArtikelForschungPeer-Review

Autorschaft

  • Mohamed G.A. Nassef
  • Omar Wael
  • Youssef H. Elkady
  • Habiba Elshazly
  • Jahy Ossama
  • Sherwet Amin
  • Dina ElGayar
  • Florian Pape
  • Islam Ali

Externe Organisationen

  • Egypt-Japan University of Science and Technology (E-JUST)
  • Alexandria University
Forschungs-netzwerk anzeigen

Details

OriginalspracheEnglisch
Aufsatznummer545
FachzeitschriftLubricants
Jahrgang13
Ausgabenummer12
PublikationsstatusVeröffentlicht - 16 Dez. 2025

Abstract

Predicting the Remaining Useful Life (RUL) of engine oil is critical for proactive maintenance and fleet reliability. However, irregular and noisy single-point sampling presents challenges for conventional prognostic models. To address this, a hierarchical physics-informed transfer learning (TL) framework is proposed that reconstructs nonlinear degradation trajectories directly from non-time-series data. The method uniquely integrates Arrhenius-type oxidation kinetics and thermochemical laws within a multi-level TL architecture, coupling fleet-level generalization with engine-specific adaptation. Unlike conventional approaches, this framework embeds physical priors directly into the transfer process, ensuring thermodynamically consistent predictions across different equipment. An integrated uncertainty quantification module provides calibrated confidence intervals for RUL estimation. Validation was conducted on 1760 oil samples from dump trucks, dozers, shovels, and wheel loaders operating under real mining conditions. The framework achieved an average R2 of 0.979 and RMSE of 10.185. This represents a 69% reduction in prediction error and a 75% narrowing of confidence intervals for RUL estimates compared to baseline models. TL outperformed the asset-specific model, reducing RMSE by up to 3 times across all equipment. Overall, this work introduces a new direction for physics-informed transfer learning, enabling accurate and uncertainty-aware RUL prediction from uncontrolled industrial data and bridging the gap between idealized degradation studies and real-world maintenance practices.

ASJC Scopus Sachgebiete

Zitieren

Physics-Informed Transfer Learning for Predicting Engine Oil Degradation and RUL Across Heterogeneous Heavy-Duty Equipment Fleets. / Nassef, Mohamed G.A.; Wael, Omar; Elkady, Youssef H. et al.
in: Lubricants, Jahrgang 13, Nr. 12, 545, 16.12.2025.

Publikation: Beitrag in FachzeitschriftArtikelForschungPeer-Review

Nassef, MGA, Wael, O, Elkady, YH, Elshazly, H, Ossama, J, Amin, S, ElGayar, D, Pape, F & Ali, I 2025, 'Physics-Informed Transfer Learning for Predicting Engine Oil Degradation and RUL Across Heterogeneous Heavy-Duty Equipment Fleets', Lubricants, Jg. 13, Nr. 12, 545. https://doi.org/10.3390/lubricants13120545
Nassef, M. G. A., Wael, O., Elkady, Y. H., Elshazly, H., Ossama, J., Amin, S., ElGayar, D., Pape, F., & Ali, I. (2025). Physics-Informed Transfer Learning for Predicting Engine Oil Degradation and RUL Across Heterogeneous Heavy-Duty Equipment Fleets. Lubricants, 13(12), Artikel 545. https://doi.org/10.3390/lubricants13120545
Nassef MGA, Wael O, Elkady YH, Elshazly H, Ossama J, Amin S et al. Physics-Informed Transfer Learning for Predicting Engine Oil Degradation and RUL Across Heterogeneous Heavy-Duty Equipment Fleets. Lubricants. 2025 Dez 16;13(12):545. doi: 10.3390/lubricants13120545
Nassef, Mohamed G.A. ; Wael, Omar ; Elkady, Youssef H. et al. / Physics-Informed Transfer Learning for Predicting Engine Oil Degradation and RUL Across Heterogeneous Heavy-Duty Equipment Fleets. in: Lubricants. 2025 ; Jahrgang 13, Nr. 12.
Download
@article{8a4e9393f5fc4fd88cda16b212f0e33d,
title = "Physics-Informed Transfer Learning for Predicting Engine Oil Degradation and RUL Across Heterogeneous Heavy-Duty Equipment Fleets",
abstract = "Predicting the Remaining Useful Life (RUL) of engine oil is critical for proactive maintenance and fleet reliability. However, irregular and noisy single-point sampling presents challenges for conventional prognostic models. To address this, a hierarchical physics-informed transfer learning (TL) framework is proposed that reconstructs nonlinear degradation trajectories directly from non-time-series data. The method uniquely integrates Arrhenius-type oxidation kinetics and thermochemical laws within a multi-level TL architecture, coupling fleet-level generalization with engine-specific adaptation. Unlike conventional approaches, this framework embeds physical priors directly into the transfer process, ensuring thermodynamically consistent predictions across different equipment. An integrated uncertainty quantification module provides calibrated confidence intervals for RUL estimation. Validation was conducted on 1760 oil samples from dump trucks, dozers, shovels, and wheel loaders operating under real mining conditions. The framework achieved an average R2 of 0.979 and RMSE of 10.185. This represents a 69% reduction in prediction error and a 75% narrowing of confidence intervals for RUL estimates compared to baseline models. TL outperformed the asset-specific model, reducing RMSE by up to 3 times across all equipment. Overall, this work introduces a new direction for physics-informed transfer learning, enabling accurate and uncertainty-aware RUL prediction from uncontrolled industrial data and bridging the gap between idealized degradation studies and real-world maintenance practices.",
keywords = "condition monitoring, hierarchical learning, oil degradation, physics-informed modeling, Remaining Useful Life (RUL), Transfer Learning (TL)",
author = "Nassef, {Mohamed G.A.} and Omar Wael and Elkady, {Youssef H.} and Habiba Elshazly and Jahy Ossama and Sherwet Amin and Dina ElGayar and Florian Pape and Islam Ali",
note = "Publisher Copyright: {\textcopyright} 2025 by the authors.",
year = "2025",
month = dec,
day = "16",
doi = "10.3390/lubricants13120545",
language = "English",
volume = "13",
number = "12",

}

Download

TY - JOUR

T1 - Physics-Informed Transfer Learning for Predicting Engine Oil Degradation and RUL Across Heterogeneous Heavy-Duty Equipment Fleets

AU - Nassef, Mohamed G.A.

AU - Wael, Omar

AU - Elkady, Youssef H.

AU - Elshazly, Habiba

AU - Ossama, Jahy

AU - Amin, Sherwet

AU - ElGayar, Dina

AU - Pape, Florian

AU - Ali, Islam

N1 - Publisher Copyright: © 2025 by the authors.

PY - 2025/12/16

Y1 - 2025/12/16

N2 - Predicting the Remaining Useful Life (RUL) of engine oil is critical for proactive maintenance and fleet reliability. However, irregular and noisy single-point sampling presents challenges for conventional prognostic models. To address this, a hierarchical physics-informed transfer learning (TL) framework is proposed that reconstructs nonlinear degradation trajectories directly from non-time-series data. The method uniquely integrates Arrhenius-type oxidation kinetics and thermochemical laws within a multi-level TL architecture, coupling fleet-level generalization with engine-specific adaptation. Unlike conventional approaches, this framework embeds physical priors directly into the transfer process, ensuring thermodynamically consistent predictions across different equipment. An integrated uncertainty quantification module provides calibrated confidence intervals for RUL estimation. Validation was conducted on 1760 oil samples from dump trucks, dozers, shovels, and wheel loaders operating under real mining conditions. The framework achieved an average R2 of 0.979 and RMSE of 10.185. This represents a 69% reduction in prediction error and a 75% narrowing of confidence intervals for RUL estimates compared to baseline models. TL outperformed the asset-specific model, reducing RMSE by up to 3 times across all equipment. Overall, this work introduces a new direction for physics-informed transfer learning, enabling accurate and uncertainty-aware RUL prediction from uncontrolled industrial data and bridging the gap between idealized degradation studies and real-world maintenance practices.

AB - Predicting the Remaining Useful Life (RUL) of engine oil is critical for proactive maintenance and fleet reliability. However, irregular and noisy single-point sampling presents challenges for conventional prognostic models. To address this, a hierarchical physics-informed transfer learning (TL) framework is proposed that reconstructs nonlinear degradation trajectories directly from non-time-series data. The method uniquely integrates Arrhenius-type oxidation kinetics and thermochemical laws within a multi-level TL architecture, coupling fleet-level generalization with engine-specific adaptation. Unlike conventional approaches, this framework embeds physical priors directly into the transfer process, ensuring thermodynamically consistent predictions across different equipment. An integrated uncertainty quantification module provides calibrated confidence intervals for RUL estimation. Validation was conducted on 1760 oil samples from dump trucks, dozers, shovels, and wheel loaders operating under real mining conditions. The framework achieved an average R2 of 0.979 and RMSE of 10.185. This represents a 69% reduction in prediction error and a 75% narrowing of confidence intervals for RUL estimates compared to baseline models. TL outperformed the asset-specific model, reducing RMSE by up to 3 times across all equipment. Overall, this work introduces a new direction for physics-informed transfer learning, enabling accurate and uncertainty-aware RUL prediction from uncontrolled industrial data and bridging the gap between idealized degradation studies and real-world maintenance practices.

KW - condition monitoring

KW - hierarchical learning

KW - oil degradation

KW - physics-informed modeling

KW - Remaining Useful Life (RUL)

KW - Transfer Learning (TL)

UR - http://www.scopus.com/inward/record.url?scp=105025965191&partnerID=8YFLogxK

U2 - 10.3390/lubricants13120545

DO - 10.3390/lubricants13120545

M3 - Article

AN - SCOPUS:105025965191

VL - 13

JO - Lubricants

JF - Lubricants

SN - 2075-4442

IS - 12

M1 - 545

ER -

Von denselben Autoren