A Neural Network Model for Prediction of Delamination in Composite Materials

We have developed an artificial neural network model (ANN) to investigate the crack propagation in composite materials. First, a finite element (FE) model is developed, to simulate the delamination in composite structures under mode I loading. The FE model is validated by comparing the load-displacement response with experimental results. A series of mode I simulations are performed, where the crack initiation force is evaluated for different specimens. This data-base is used later to develop a multilayer neural network, which is capable of predicting the stress initiation force for arbitrary initial crack lengths and specimen thicknesses. The ANN results verify the good performance of this model.

Keywords:

Delamination , Composite materials , Neural network , Finite element model