Machine learning model for fatigue crack growth prediction in marine structural steel under high-low frequency superimposed loading

Abstract

Accurate fatigue life prediction in marine structures subjected to combined low-frequency (LF) and high-frequency (HF) cyclic loading is of great significance. This study develops the fatigue crack growth prediction models for marine structural steel under high-low frequency superimposed loading using three machine learning (ML) algorithms: back-propagation (BP) neural network, genetic algorithm optimized BP (GA-BP) neural network and particle swarm optimized BP (PSO-BP) neural network. The ML models are trained and validated by using the dataset of fatigue crack growth tests under various loading conditions with different load amplitude ratios, load frequency ratios and mean load levels. The predictive performance of the three ML models is systematically compared with each other as well as the modified Wheeler model and the Huang model. Results demonstrate that the ML models exhibit superior agreement with experimental data compared to the classical theoretical models, and the GA-BP neural network model achieves the best overall accuracy. These findings suggest that the neural network models, by effectively capturing the interaction effects between LF and HF load components, can provide robust and promising tools for predicting the fatigue crack growth behaviour of marine structures.