Prediction of Fatigue Crack Length in Aircraft Aluminum Alloys using Radial Basis Function Neural Network

Fatigue crack growth calculation exclusively for aircraft materials is a serious issue due to life risk involved besides economic deficit. Empirical methods are not that much flexible to deal with the problem of non-linearity while predicting the data. Precise prediction of crack length a, that is progressed as a result of fatigue, with respect to the number of cycles, N is imperative to predict fatigue life of materials used in aircraft industry. As compared to the empirical and mathematical techniques, Machine Learning Algorithms (MLA) satisfy non-linearity problem reasonably as they have admirable learning ability besides robust nature. In this research paper, MLA based technique for crack length and subsequently fatigue life prediction is introduced that uses Radial Basis Function Neural Network (RBF-NN). The method recommended below is verified on two dissimilar alloys of aluminum specifically used in airplane industry. The comparison of result displays a commendable correspondence to the experimental information. From the two alloys of aluminum used in the experimentation phase, D16 aluminum alloy displays better results with a Mean Squared Error of 2.1853×10
^-2
.