Machine learning-based fatigue lifetime prediction of structural steels
Fatigue of materials stands as a prevalent cause of mechanical structure failures, which often occur suddenly, unpredictably, and catastrophically. Accurately predicting the fatigue lifespan of materials is crucial, especially given the potential for fatigue failure to occur within a short design li...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-06-01
|
| Series: | Alexandria Engineering Journal |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1110016825004818 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849329170908184576 |
|---|---|
| author | Konstantinos Arvanitis Pantelis Nikolakopoulos Dimitrios Pavlou Mina Farmanbar |
| author_facet | Konstantinos Arvanitis Pantelis Nikolakopoulos Dimitrios Pavlou Mina Farmanbar |
| author_sort | Konstantinos Arvanitis |
| collection | DOAJ |
| description | Fatigue of materials stands as a prevalent cause of mechanical structure failures, which often occur suddenly, unpredictably, and catastrophically. Accurately predicting the fatigue lifespan of materials is crucial, especially given the potential for fatigue failure to occur within a short design life. While traditional methodologies based on S-N curve models remain prevalent in industry, there is a contemporary shift towards employing Artificial Intelligence and Machine Learning techniques to significantly refine the accuracy of fatigue lifetime predictions. In this study, a dataset containing experimental data from various structural steels is used. Through preprocessing and feature selection, four techniques are explored: Polynomial Regression, Support Vector Regression (SVR), XGB Regression and Artificial Neural Network (ANN), aiming to identify the most effective algorithm. The implementation of these methodologies for fatigue lifetime prediction yields substantial outcomes. All models exhibit satisfactory performance, with XGB Regression demonstrating superior effectiveness. Furthermore, Polynomial Regression provides highly satisfactory results, almost identical to the Artificial Neural Network. Notably, it requires significantly less computational power, making it a practical alternative in cases of restricted computational resources or limited implementation time. Overall, the proposed methodology effectively leverages material preprocessing details, mechanical properties and experimental conditions to provide accurate predictions of the remaining fatigue lifespan of structural steels. |
| format | Article |
| id | doaj-art-9fa429aba5c64f699349a97fa5e0ee32 |
| institution | Kabale University |
| issn | 1110-0168 |
| language | English |
| publishDate | 2025-06-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Alexandria Engineering Journal |
| spelling | doaj-art-9fa429aba5c64f699349a97fa5e0ee322025-08-20T03:47:20ZengElsevierAlexandria Engineering Journal1110-01682025-06-01125556610.1016/j.aej.2025.04.014Machine learning-based fatigue lifetime prediction of structural steelsKonstantinos Arvanitis0Pantelis Nikolakopoulos1Dimitrios Pavlou2Mina Farmanbar3Department of Mechanical Engineering and Aeronautics, University of Patras, Patras, GreeceDepartment of Mechanical Engineering and Aeronautics, University of Patras, Patras, Greece; Corresponding author.Department of Mechanical and Structural Engineering and Materials Science, University of Stavanger, Stavanger, NorwayDepartment of Electrical Engineering and Computer Science, University of Stavanger, Stavanger, NorwayFatigue of materials stands as a prevalent cause of mechanical structure failures, which often occur suddenly, unpredictably, and catastrophically. Accurately predicting the fatigue lifespan of materials is crucial, especially given the potential for fatigue failure to occur within a short design life. While traditional methodologies based on S-N curve models remain prevalent in industry, there is a contemporary shift towards employing Artificial Intelligence and Machine Learning techniques to significantly refine the accuracy of fatigue lifetime predictions. In this study, a dataset containing experimental data from various structural steels is used. Through preprocessing and feature selection, four techniques are explored: Polynomial Regression, Support Vector Regression (SVR), XGB Regression and Artificial Neural Network (ANN), aiming to identify the most effective algorithm. The implementation of these methodologies for fatigue lifetime prediction yields substantial outcomes. All models exhibit satisfactory performance, with XGB Regression demonstrating superior effectiveness. Furthermore, Polynomial Regression provides highly satisfactory results, almost identical to the Artificial Neural Network. Notably, it requires significantly less computational power, making it a practical alternative in cases of restricted computational resources or limited implementation time. Overall, the proposed methodology effectively leverages material preprocessing details, mechanical properties and experimental conditions to provide accurate predictions of the remaining fatigue lifespan of structural steels.http://www.sciencedirect.com/science/article/pii/S1110016825004818FatigueLifetime PredictionMachine LearningArtificial IntelligenceRegression |
| spellingShingle | Konstantinos Arvanitis Pantelis Nikolakopoulos Dimitrios Pavlou Mina Farmanbar Machine learning-based fatigue lifetime prediction of structural steels Alexandria Engineering Journal Fatigue Lifetime Prediction Machine Learning Artificial Intelligence Regression |
| title | Machine learning-based fatigue lifetime prediction of structural steels |
| title_full | Machine learning-based fatigue lifetime prediction of structural steels |
| title_fullStr | Machine learning-based fatigue lifetime prediction of structural steels |
| title_full_unstemmed | Machine learning-based fatigue lifetime prediction of structural steels |
| title_short | Machine learning-based fatigue lifetime prediction of structural steels |
| title_sort | machine learning based fatigue lifetime prediction of structural steels |
| topic | Fatigue Lifetime Prediction Machine Learning Artificial Intelligence Regression |
| url | http://www.sciencedirect.com/science/article/pii/S1110016825004818 |
| work_keys_str_mv | AT konstantinosarvanitis machinelearningbasedfatiguelifetimepredictionofstructuralsteels AT pantelisnikolakopoulos machinelearningbasedfatiguelifetimepredictionofstructuralsteels AT dimitriospavlou machinelearningbasedfatiguelifetimepredictionofstructuralsteels AT minafarmanbar machinelearningbasedfatiguelifetimepredictionofstructuralsteels |