Fatigue life prediction in presence of mean stresses using domain knowledge‐integrated ensemble of extreme learning machines

Abstract

An accurate and stable data-driven model is proposed in this work for fatigue life prediction in presence of mean stresses. Multiple independent extreme learning machines are integrated into the model with distinct neural network configurations to simulate the complex correlations among mean stress levels, material properties, and fatigue lives. Meanwhile, the theoretical prediction, as a representation of domain knowledge, is used to optimize the data-driven processes of model training and prediction. Extensive experimental data of 13 metallic materials with different mean stress levels are collected from the open literatures for model training and evaluation. The results demonstrate that the proposed model can achieve high accuracy and good stability in fatigue life prediction under mean stress loading conditions, even with a small training dataset, showing great applicability for fatigue life prediction.

Evaluation of the influence of mean stress on the fatigue behavior of notched and smooth medium carbon steel components through an energetic local approach

“…For instance, as first pointed out by Morrow, 22 the effectiveness of mean‐stress models such as previous studies 10–12 provide poor correlations with high mean stress values (R ≥ 0.6) or in case of compressive mean stress (R < −1). To this end, it is worth mentioning that, while some revised formulations have been recently proposed for the case of σ m < 0, 23 the general topic of mean‐stress effect remains an open field of research, with an increasing interest in complementing analytical methods with machine learning techniques 24 . Moreover, predictive accuracy of mean‐stress models further reduces in presence of strong stress gradients and/or multi‐axial stress fields, for example, when estimating the fatigue life of notched components 15 .…”

A physics‐informed generative adversarial network framework for multiaxial fatigue life prediction

“…Deep learning is also a method that researchers use to predict fatigue life 18–26 . Deep learning is a tool for learning the inherent laws of sample data aimed at fitting an optimal analytical function between input and output.…”

“…Deep learning is also a method that researchers use to predict fatigue life. [18][19][20][21][22][23][24][25][26] Deep learning is a tool for learning the inherent laws of sample data aimed at fitting an optimal analytical function between input and output. Unlike life prediction models based on fatigue test results, deep learning is a potential method that is not limited by loading conditions and materials.…”

Application of machine learning methods in multiaxial fatigue life prediction

“…They found that multiple researchers used many ANN architectures to solve problems classified as fatigue life prediction, fatigue crack, fatigue damage diagnosis, fatigue strength, and fatigue load. Gan et al 16 applied a variant of the ANN model, called extreme learning machine, for fatigue life prediction considering the mean stress effect. Zhu et al 17 used an ML‐based method combined with a physics‐based parameter to evaluate the fatigue life of four materials in the very high‐cycle fatigue regime.…”