Application of artificial neural network for predicting strain-life fatigue properties of steels on the basis of tensile tests

The applicability of artificial neural networks (ANN) in predicting the strain-life fatigue properties using tensile material data for 73 steels was investigated by conducting four separate neural networks for individual fatigue properties. The fatigue data of these steels extracted from available literatures were used in the formation of training set of ANN. Results of neural network modelling indicated that fatigue strength coefficient and fatigue ductility (strain) coefficient values, which primarily characterize the curves of the strain amplitude versus life reversals, were predicted with high accuracy of approximately 99 and 98%, respectively. It was concluded that predicted fatigue properties by the trained neural network model seem more reasonable compared to approximate methods, which were formerly suggested based on tensile material data. It is possible to claim that, ANN is fairly promising prediction technique if properly used.