Prediction of fatigue crack growth rate in aircraft aluminum alloys using radial basis function neural network

Abstract

Fatigue crack propagation in aircraft structures is a critical problem as life risk is involved besides financial loss. The relationship between stress intensity factor, ΔΚ and fatigue crack growth rate, da/dN is non-linear even in Paris region (region II). Analytical techniques are not much flexible to handle non-linearity. Accurate prediction of crack growth rate developed as a result of fatigue is important to evaluate fatigue life of engineering structures. Machine learning algorithms cater for non-linarites satisfactorily because of their excellent learning capability and flexible nature. This paper presents MLA based technique for prediction of crack growth rate using Radial Basis Function Neural Network (RBF-NN). The proposed technique is tested on different aluminum alloys used for aircraft structures. The minimum predicted MSE was achieved as 1.1315 × 10−9 for 7055-T7511 Aluminum Alloy. The results show an excellent agreement to experimental data.