Performance evaluation of machine learning based-classifiers in friction stir welding of Aa6061-T6 alloy.

Autor: Kumar, A. Kiran, Surya, Mulugundam Siva, Venkataramaiah, P.
Zdroj: International Journal on Interactive Design & Manufacturing; Feb2023, Vol. 17 Issue 1, p469-472, 4p
Abstrakt: Machine learning (ML) in the manufacturing sector shows promising results that help lower costs, time, and complexity in identifying the best parameters out of the wide range. This work studied three machine learning classification models' (decision tree, random forest, and XGBoost models) performance for a friction stir welded AA 6061-T6 aluminium alloy to determine the best classifier model. The models were created by training and testing the experimental data to check the influence of input parameters on the yield strength of the FSW AA6061-T6 alloy. The XGBoost model showed a maximum accuracy of 95.24% among the three models. The classifier models performance is evaluated using AUC (area under the curve metric) and confusion matrix. Finally, the model can be used to predict the combination of parameters that produce the required weld strength, which helps reduce the experimental cost and material wastage. [ABSTRACT FROM AUTHOR]
Databáze: Complementary Index
Externí odkaz: