ABSTRACT
ABSTRACT This chapter addresses the use of artificial neural networks (ANNs) in the prediction of fatigue life of composite materials. The justification of using soft computing methods including ANNs over other existing methods is initially addressed. Section 8.2 focuses on the various ANN paradigms used in modeling the fatigue life prediction of laminae and laminates. The effect of various network structures and characteristics such as the input parameters, the number of hidden layers, and the number of neurons per hidden layer is discussed. This is followed by a comparison of the different predictions obtained using the aforementioned network architectures and to results obtained using other fatigue life prediction methods. Some ANN design and selection guidelines to follow when predicting the fatigue life of composites using ANNs are also included. The polynomial classifiers, an alternative technique that may address some of the ANN shortcomings, are then introduced. The results obtained using polynomial classifiers are compared to those obtained using the various ANN architectures previously introduced. The chapter concludes by giving a status report about the effectiveness of ANNs in predicting the fatigue life of composites, highlighting the future challenges facing their use.
