Modelling of automotive steel fatigue lifetime by machine learning method

2025-01-19

International Workshop on Information Technologies: Theoretical and Applied Problems

AI Summary

Conventional Paris’ law exhibits limited accuracy in predicting fatigue crack growth in automotive QSTE340TM steel, particularly under multi-parameter coupled nonlinear conditions involving stress ratio (R) and overload ratio (R_ol). Method: This study proposes a novel three-layer multilayer perceptron (MLP) neural network (3–75–1 architecture), trained on cycle count (N), R, and R_ol as inputs to predict crack length. Crucially, feature engineering incorporates fatigue mechanics–based prior knowledge to enhance physical interpretability and model robustness. Contribution/Results: The model achieves high-precision nonlinear modeling across broad R and R_ol ranges, with mean absolute percentage error (MAPE) of only 0.02%–4.59%, substantially outperforming Paris’ law. It demonstrates strong generalization capability and engineering applicability, establishing a new paradigm for intelligent, multi-condition fatigue life prediction of advanced high-strength steels.

Technology Category

Machine Learning: Life-Long and Continual LearningNatural Language Processing: Language Grounding & Multi-modal NLPCognitive Modeling & Cognitive Systems: Computational Creativity

Application Category

Web Mining and Content Analysis: Large pretrained models with web dataSystems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applicationsGraph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

In the current study, the fatigue life of QSTE340TM steel was modelled using a machine learning method, namely, a neural network. This problem was solved by a Multi-Layer Perceptron (MLP) neural network with a 3-75-1 architecture, which allows the prediction of the crack length based on the number of load cycles N, the stress ratio R, and the overload ratio Rol. The proposed model showed high accuracy, with mean absolute percentage error (MAPE) ranging from 0.02% to 4.59% for different R and Rol. The neural network effectively reveals the nonlinear relationships between input parameters and fatigue crack growth, providing reliable predictions for different loading conditions.

Problem

• Automotive steel
• QSTE340TM
• Fatigue life prediction

Innovation

• Deep Learning
• Crack Propagation Life Prediction
• Generalization Ability