Prediction of mechanical properties of steel based welds with artificial neural networks

Welded structures are industrial methods frequently employed in assembling complex systems such as machines, buildings, and ships. The mechanical properties of the welding metal are determined by the elements constituting the source, and this effect is achieved by altering the microstructure of the welding metal. Neural network analysis is a relatively new methodology widely applicable in various fields, aiming to enhance efficiency and facilitate thorough analysis. The primary objective of this study is to investigate the relationships between structural and alloy elements in welding metal using neural networks and analyze them through this innovative methodology. In this research, the Levenberg-Marquardt algorithm (LM) is utilized to predict the physical properties of welding metal through artificial neural networks. The LM algorithm is an optimization algorithm used in the training process of artificial neural networks, aiming to minimize the difference between a prediction function and actual data. Due to the time-consuming and costly nature of traditional methods, the application of artificial intelligence techniques for the rapid and accurate determination of physical properties offers a significant advantage in terms of time, cost, and labor in industrial production processes. The success rates of the research are found to be 93.135% for acicular ferrite, 95.923% for hardness, 94.17% for yield point, and 96.324% for maximum tensile strength.