Artificial intelligence algorithms for prediction of the ultimate tensile strength of the friction stir welded magnesium alloys

Artificial Intelligence algorithms based on the machine learning approach finds application in manufacturing and materials industries for the prediction and optimization of mechanical and microstructure properties. In the present study, six supervised machine learning regression-based algorithms i.e., Decision Trees, XGBoost, Artificial Neural networks, Random Forests, Gradient Boosting, and AdaBoost are used for the prediction of the Ultimate Tensile Strength of the Friction Stir Welded magnesium joints. Magnesium alloy type (AM20, AZ61A, AZ31B, and AZ31), Plunge Depth (mm), Shoulder Diameter (mm), Tool Traverse Speed (mm/min), Pin Diameter (mm), Axial Force (kN), and Tool Rotational Speed (RPM) are the input parameters while the Ultimate Tensile Strength (MPa) of the Friction Stir Welded joints is an output parameter. The results showed that the Magnesium Alloy type has the highest feature importance in comparison to other input parameters. It is also observed that the XGBoost algorithms yield highest coefficient of determination of 0.81.& nbsp;