A novel defect-based fatigue damage model coupled with an optimized neural network for high-cycle fatigue analysis of casting alloys with surface defect. (May 2023)

• Record Type: Journal Article
• Authors: Gao, Tongzhou; Ji, Chenhao; Zhan, Zhixin; Huang, Yingying; Liu, Chuanqi; Hu, Weiping; Meng, Qingchun
• Is Part Of: International journal of fatigue. Volume 170(2023)
• Publication Date: 2023-05
• Subjects: High-cycle fatigue -- Casting alloys -- Surface defect -- Damage model -- Optimized neural network; Materials -- Fatigue -- Periodicals; Materials -- Fatigue; Periodicals; 620.1122
• DOI: 10.1016/j.ijfatigue.2023.107538
• Languages: English
• ISSNs: 0142-1123

Abstract

Highlights: High-cycle fatigue behavior of casting alloys with surface defect is investigated. A novel defect-based fatigue damage evolution model is established. An optimized neural network is proposed for the high-cycle fatigue prediction. Numerical results are verified by the fatigue test data of casting alloy specimens. Abstract: A novel defect-based fatigue damage model coupled with an optimized neural network is proposed for high-cycle fatigue prediction. Based on parametric studies and continuum damage mechanics, the defect-based fatigue damage evolution equation is derived, and the numerical simulation and fatigue damage computation are then implemented and validated. After that, more computations are performed to acquire a batch of reliable fatigue data, and the database is obtained. Finally, the architecture of the optimized neural network is established, and the predicted results are verified by experimental fatigue data. The proposed methodology works well for the fatigue analysis of casting alloys with surface defect.