Predicting the Fatigue Failure of Fiber Reinforced Composite Materials Using Artificial Neural Networks

Abstract

Artificial Neural Networks (ANN) have recently been used in modeling the mechanical behavior of fiberreinforced composite materials. ANN have also been successfully used in predicting the fatigue behavior of a certain material under loading conditions other than those used for training. The use of ANN in predicting fatigue failure in composites would be of great value if one could predict the failure of materials other than those used for training the network. This would allow developers of new materials to estimate in advance the fatigue properties of their material. In this work, experimental fatigue data obtained for certain fiber-reinforced composite materials is used to predict the cyclic behavior of a composite made of another material. The effect of the various mechanical properties on the training of the network is evaluated to obtain the most suitable combination of properties resulting in the best fatigue life prediction. The resilient back-propagation with 10 to 20 neurons depending on the input parameters resulted in accurate prediction when compared to experimental ones. An introduction to the use of Polynomial classifiers (PC) to predict the fatigue behavior is also presented. Using a first order PC with additional higher order terms gave good results when compared to experimental ones.