Neural network-based modelling and prediction for crack growth process in alloys

The work in this thesis focuses on the development of advanced methodologies for the study of crack growth with artificial neural networks, to develop accurate prediction and diagnostic algorithms for crack growth under various loading environments. The developed models show better performance in comparison with typical existing crack growth models. It is particularly useful to replace some of the burdensome fatigue tests in practical engineering applications. The research results of this thesis may contribute to the intelligent prediction of crack growth in real-time prognosis and health management systems.