An artificial neural network for the fatigue study of bonded FRP–wood interfaces

The objective of this study is to explore the development of an artificial neural network (ANN) method for the analysis of load ratio effects on fatigue of interfaces for phenolic fiber reinforced polymer (FRP) composite bonded to red maple wood. Experiments were performed with a contoured double cantilever beam (CDCB) specimen under load control, and the crack propagation rate was obtained by the compliance method. Using linear elastic fracture mechanics, the influence of load ratio on fatigue crack growth rate was studied; leading to a modified Paris Law equation based on strain energy release rate range, ΔG, and mean value of strain energy release rate, Gmean. By constructing suitable network architectures, an ANN can be defined and trained using existing experimental data sets, to provide in turn output fatigue data sets for new input parameters. The crack growth rate as predicted by the ANN approach is compared with the experimental output and theoretical prediction from a modified Paris Law equation. It is shown that the proposed neural network model is able to predict valuable fatigue responses, such as crack growth rate, that would facilitate the development of design guidelines for hybrid material bonded interfaces.