Machine Learning for Composites

Our group started investigating the use of Machine Learning models for the surrogate modelling of composite materials in 2018, with the first work by Tom Gulikers and subsequently by Karthik Venkatesan in their master theses. 

In Tom's thesis, we showed how a Neural Network can be constructed and trained to represent highly nonlinear stress-strain curves for open-hole composite laminates.

In Karthik's work, we showed how a combination of feedforward Neural Network and a Convolutional Neural Network can be combined and trained to effectively predict damage patterns with respect to boundary loadings in an open-hole composite laminate.