Artificial neural network based fatigue life assessment of friction stir welding AA2024-T351 aluminum alloy and multi-objective optimization of welding parameters

• Published in: International journal of fatigue 2022-07, Vol.160, p.106840, Article 106840
• Main Authors: Masoudi Nejad, Reza, Sina, Nima, Ghahremani Moghadam, Danial, Branco, Ricardo, Macek, Wojciech, Berto, Filippo
• Format: Article
• Language: English
• Subjects: Algorithms; Alloys; Aluminum alloy; Aluminum alloys; Aluminum base alloys; Artificial neural network; Artificial neural networks; Crack propagation; Fatigue failure; Fatigue life; Fatigue life assessment; Fracture mechanics; Fracture toughness; Friction stir welding; Life assessment; Materials fatigue; Metal fatigue; Multiple objective analysis; Neural networks; Optimization; Parameter sensitivity; Sensitivity analysis; Welding parameters
• ISSN: 0142-1123; 1879-3452
• DOI: 10.1016/j.ijfatigue.2022.106840

•The FCP rate of the CT specimens of the friction stir welded 2024-T351 aluminum alloys have been studied.•The effects of rotational and traverse speeds of FSW on the FCP rate and the fracture toughness have been investigated.•An artificial neural network to predict slope of FCP rate and fracture toughness has been designed.•Multi-objective optimization algorithm has been used to obtain the best rotational and traverse speeds.•Sensitivity analysis has been applied to obtain the relationship and effect of rotational and traverse speeds. In this paper, the fracture behavior and fatigue crack growth rate of the 2024-T351 aluminum alloy has been investigated. At first, the 2024-T351 aluminum alloys have been welded using friction stir welding procedure and the fracture toughness and fatigue crack growth rate of the CT specimens have been studied experimentally based on ASTM standards. After that, in order to predict fatigue crack growth rate and fracture toughness, artificial neural network is used. To obtain the best neuron number in the hidden layer of the artificial neural network, different neuron numbers are tested and the best network based on the performance is selected. Then the fitting method is applied and the fitted surfaces that illustrate the behavior of welding are shown and the results of artificial neural network and fitting method are compared. Also, multi-objective optimization algorithm is used to obtain the best welding parameters and finally sensitivity analysis is applied to measure the effect of rotational and traverse speeds on the fracture toughness and fatigue crack growth rate.