Crack Propagation Rate Prediction Using ML

Abstract

The key to predicting structure fatigue lifetime is determining the fatigue crack propagation rate. Models such as the nine-parameter fatigue crack propagation rate and the McEvily model are widely used today, but realizing these models is difficult because partial derivatives must be calculated and there is a large deviation between the fitted static parameter and the actual value, and the physical conception is unclear. Because of this, we developed an optimal common machine learning algorithm (LSSVM—least squares support vector machine) for predicting fatigue crack propagation rates based on optimal parameter selection using grid search and cross-validation. It's both complicated and confusing.