A Novel Physics-Guided Neural Network for Predicting Fatigue Life of Materials
Source abstract
Abstract
A physics-guided neural network (PGNN) is proposed to predict the fatigue life of materials. In order to reduce the complexity of fatigue life prediction and reduce the data required for network training, the PGNN only predicts the fatigue performance parameters under a specific loading environment, and calculates the fatigue life by substituting the load into the fatigue performance parameters. The advantage of this is that the network does not need to evaluate the effect of numerical changes in the load on fatigue life. The load only needs to participate in the error verification, which reduces the dimension of the function that the neural network needs to approximate. The performance of the PGNN is verified using published data. Due to the reduction in the complexity of the problem, the PGNN can use fewer training samples to obtain more accurate fatigue life prediction results and has a certain extrapolation ability for the changes in trained loading environment parameters. The prediction process of the PGNN for fatigue life is not completely a black box, and the prediction results are helpful for scholars to further study the fatigue phenomenon.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Industrial Vision Systems and Defect Detection
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
Mechanical Engineering College, Xinjiang University, Urumqi 830063, China
References (27)
Show source-provided references
1. Fatigue Mechanical Life Design—A Review
2. Gerber, H. (1874). Bestimmung der Zulässigen Spannungen in Eisen-Constructionen, Wolf.
3. Goodman, J. (1919). Mechanics Applied to Engineering, Longmans, Green.
4. Factor of safety and working stress10.1115/1.4057307 ↗
5. The exponential law of endurance tests
6. Fatigue modeling using neural networks: A comprehensive review10.1111/ffe.13640 ↗
7. Ross, C.T. (1993). Best Practice Guidelines for Developing Neural Computing Applications—An Overview, Ministry of Defense Procurement Executive.
8. Use of neural networks for the prediction of fatigue lives of composite materials10.1016/S1359-835X(99)00027-5 ↗
9. Fatigue life predictions of metal matrix composites using artificial neural networks10.2478/amm-2014-0016 ↗
10. Comparing repeatability and reproducibility of topographic measurement types directly using linear regression analyses of measured heights10.1016/j.precisioneng.2024.02.009 ↗
11. Rubber fatigue life prediction using a random forest method and nonlinear cumulative fatigue damage model10.1002/app.48519 ↗
12. A machine-learning fatigue life prediction approach of additively manufactured metals10.1016/j.engfracmech.2020.107508 ↗
13. A novel method of multiaxial fatigue life prediction based on deep learning10.1016/j.ijfatigue.2021.106356 ↗
14. Predicting remaining useful life of rotating machinery based artificial neural network10.1016/j.camwa.2010.03.065 ↗
15. Fatigue damage characterization for composite laminates using deep learning and laser ultrasonic10.1016/j.compositesb.2021.108816 ↗
16. A neural network approach to describing the scatter of S–N curves10.1016/j.ijfatigue.2005.08.002 ↗
17. Adaptive neuro-fuzzy inference system in modelling fatigue life of multidirectional composite laminates10.1016/j.commatsci.2008.02.028 ↗
18. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations10.1016/j.jcp.2018.10.045 ↗
19. Probabilistic physics-guided machine learning for fatigue data analysis10.1016/j.eswa.2020.114316 ↗
20. Physics-informed neural networks for missing physics estimation in cumulative damage models: A case study in corrosion fatigue10.1115/1.4047173 ↗
21. Dourado, A., and Viana, F.A.C. (2019, January 21–26). Physics-informed neural networks for corrosion-fatigue prognosis. Proceedings of the Annual Conference of the PHM Society, Scottsdale, AZ, USA.10.36001/phmconf.2019.v11i1.814 ↗
22. Dourado, A., and Viana, F.A.C. (2020, January 6–10). Physics-informed neural networks for bias compensation in corrosion fatigue. Proceedings of the AIAA Scitech 2020 Forum, Orlando, FL, USA. Part F.10.2514/6.2020-1149 ↗
23. Plate, J.D., van de Leur, R.R., Leenen, L.P., Hietbrink, F., Peelen, L.M., and Eijkemans, M.J. (2019). Incorporating repeated measurements into prediction models in the critical care setting: A framework, systematic review and meta-analysis. BMC Med. Res. Methodol., 19.10.1186/s12874-019-0847-0 ↗
24. The determination of fatigue limits under alternating stress conditions
25. Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., and Dubourg, V. (2012). Scikit-learn: Machine Learning in Python. arXiv.
26. Paszke, A., Gross, S., Chintala, S., Chanan, G., Yang, E., DeVito, Z., Lin, Z., Desmaison, A., Antiga, L., and Lerer, A. (2017, January 4–9). Automatic differentiation in pytorch. Proceedings of the 31st Conference on Neural Information Processing Systems (NIPS 2017), Long Beach, CA, USA.
27. Rice, R.C., Jackson, J.L., Bakuckas, J., and Thompson, S. (2003). Metallic Materials Properties Development and Standardization, US Department of Commerce.