Hybrid modeling of mechanical properties and hardness of aluminum alloy 5083 and C100 Copper with various machine learning algorithms in friction stir welding

Abstract

Knowledge of mechanical properties and hardness is essential for the cost-effective and efficient friction stir welding (FSW) in producing Cu and Al composites. Correct process management requires knowledge of consistent model predictions. Many factors affect the mechanical properties and hardness, which include pin geometry, forward speed, rotational speed, and pin angle. The purpose of this research is to investigate the use of Relevance Vector Machine (RVM), Support Vector Machine (SVM), and Least Square Support Vector Machine (LSSVM) algorithms in hybrid modeling of the mechanical properties and hardness in FSW. Then, the resulting mechanical properties are modeled to three modeling methods: the hybrid LSSVM-RVM, hybrid SVM-RVM, and hybrid SVM-LSSVM. A comparison between the experimental data and the output of these models were made. R2, RMSE, and RMSE/ymax statistical indices was used in this research. The values ​​of R2 for the results of the tensile test and hardness area were obtained as 0.9712 and 0.9126, respectively. The results showed that the hybrid LSSVM-RVM model could efficiently estimate the mechanical and hardness properties.