The Use Of Neural Networks To Detect Damage In Sandwich Composites

Composite materials fail in complex failure modes that are difficult to detect. No single NDE technique is capable of detecting all damages. The ability to detect and asses the state of the damage is a key issue in order to improve service life of these materials. A Neural Network (NN) was chosen as a means to interpret and classify the information such that the type of damage, severity and location could be identified. The work describes the implementation of a NN based approach which combines thermal damage detection and vibration signatures in order to detect location and extent of damage in sandwich composites consisting of two carbon fiber/epoxy matrix face sheets laminated onto a urethane foam core. The approach analytically characterized and experimentally validated models for both thermal and vibration response. The numerical models were then used to train the neural networks. This approach is significant as it combines two techniques as opposed to just one as generally performed. Results demonstrated that the multi-component neural network approach successfully detected damage in scenarios in which using just a single method would have failed.