Modeling of fatigue life in double shear lap joints using artificial neural networks

Fatigue is one of the most important failure sources of material that is caused by repeatedly applied loads. It is a progressive and localized structural damage that occurs when a material is subjected to cyclic loading. The experimental results of fatigue tests on Al-alloy 2024-T3 in double shear lap joints were used to estimate (model) fatigue life with artificial neural networks (ANN). Artificial neural networks with experimental data processing can find the knowledge or law lies behind the data, and unlike mathematical models, it’s not necessary to determine the mathematical relation between inputs and outputs. To model by artificial neural network, one of the experimental data of fatigue life randomly selected for validation and two other were selected for testing, the rest of the data were used to find the optimal values of weights and bias. After being ensured of the model accuracy, it was used to predict the fatigue life at different loads in the working phase that had not been tested. Comparison of experimental results and the results of the model shows that a 3-layer artificial neural network with less than 10% error could be used to predict the fatigue life at different loads.