Investigating the Application of Stress Wave Factors in Machine Learning for Delamination Location Prediction in a Composite Laminate

A framework integrating machine learning and simulation was developed to predict the existing delamination location in a composite laminate. The simulation was run in a coupled-physic finite element model to represent a guided-wave-based structural health monitoring system. A set of stress wave factors were adopted in the present study to be used as input features in the machine learning models. Support vector machine and random forest models were then used to predict the in-plane location of the delamination. It was found that both machine learning models could provide accurate prediction of the delamination location while random forest model produces better prediction accuracy. The obtained results also show that stress wave factors can be effective input features in a machine learning model used for predicting the delamination location.