Enhancing Friction Stir Welding in Fishing Boat Construction through Deep Learning-Based Optimization

In the present study, authors have attempted to a novel approach for prediction, analysis, and optimization of Friction Stir Welding (FSW) process based on Deep Neural Network (DNN) model. To obtain DNN structure with high accuracy, most focus has been number hidden layers activation functions. The was developed by small database containing results tensile hardness tests welded 7075-T6 aluminum alloy. This material production method were selected application in construction fishing boat flooring, because one hand, it faces corrosion caused proximity sea water other due direct contact human food, i.e., fish etc., antibacterial issues should be considered. All major parameters FSW process, including axial force, rotational speed, traverse speed as well tool diameter hardness, considered investigate their correspondence effects strength zone. important achievement this research showed that using SAE pre-training neural networks, higher accuracy can obtained predicting responses. Finally, optimal values various welding reported speed: 1600 rpm, 65 mm/min, force: 8 KN, shoulder pin diameters: 15.5 5.75 mm, hardness: 50 HRC.