Fatigue crack growth rate prediction under single peak overload based on WOA-BP neural network

Abstract

This study is aimed at investigating the crack growth behavior of 2A12 aluminum alloy under constant amplitude loads and various single peak overload conditions, with a focus on the effects of different load ratios and overload ratios on the crack growth rate. Due to the complexity and numerous parameters involved in traditional physical models, we proposed a whale optimization algorithm-backpropagation neural network-based model for predicting crack growth rate. By comparing results on datasets of 2A12 aluminum alloy and QSTE340TM steel, including Wheeler, Huang, WOA-SVM, and WOA-RBF models, our study demonstrates that our model achieves higher predictive accuracy. Finally, the paper calculated the crack growth life using the cycle-by-cycle method and conducted a detailed comparison and analysis of the prediction errors of various models. This research holds significant theoretical and practical value for enhancing understanding of material crack behavior and developing more accurate models for predicting crack growth rates.