Fatigue Life Prediction of GLARE Composites Using Regression Tree Ensemble-Based Machine Learning Model

With the advancement of machine learning in leading technologies, it is perceived that machine learning is a new and effective alternative for the classic fatigue life prediction. This paper provides a regression tree ensemble-based machine learning approach to predict the fatigue life of GLARE composites. In the model, mechanical, geometrical properties and fatigue loading stresses are selected as the training parameters (so-called features), and the GLARE fatigue life is predicted as the output of the model. Experimental data of a total of 98 pieces of GLARE specimens with nine different layups are used for the training and validation of the machine learning model. Results show that the model can provide good fatigue life prediction accuracy and model stability. The most correlated, either positively or negatively, parameters to the fatigue life span are the stress developed in the aluminum layer, the maximum cyclic stress, alternating stress, and mean fatigue stress.