PREDICTION OF DELAMINATION LOCATION IN COMPOSITE STRUCTURES WITH DIFFERENT PLY ORIENTATIONS: A FRAMEWORK INTEGRATING FINITE ELEMENT SIMULATION AND DEEP LEARNING

In this paper, a framework integrating guided-wave-based finite element (FE) simulation and deep learning is developed to predict the location of internal delamination within a composite plate with different layup configurations. The data needed for training the prediction model are obtained through guided-wave-based FE simulations. Convolutional neural network (CNN) models are then built and trained to predict the location of the existing delamination. The possibility of using transfer learning is explored to account for composite plates with different layup combinations while minimizing additional data needed for CNN model training. It is found that once a baseline CNN model is trained by the data extracted from a specific composite layup case, transfer learning can significantly reduce the amount of data required to achieve similar levels of prediction accuracy for other scenarios with different layup orientations, thus greatly improving the efficiency and versatility of the proposed framework.