What question did this study set out to answer?

To optimize and predict peak temperatures during friction stir welding of Al 6061 T6 alloys using various methods.

February 11, 2026Scientific Reports0 citationsOpen Access

Optimization and prediction of peak temperature in friction stir welding of Al 6061 T6 using statistical and machine learning techniques

AAAssad AnisNED University of Engineering and Technology MSM. ShakaibNED University of Engineering and Technology MHMuhammad Sohail HanifNED University of Engineering and Technology

Key Points

To optimize and predict peak temperatures during friction stir welding of Al 6061 T6 alloys using various methods.
Utilized finite element analysis with COMSOL Multiphysics
Applied Taguchi method and ANOVA for parameter optimization
Employed a feed-forward backpropagation neural network for prediction
Identified axial force and rotational speed as significant factors affecting peak temperatures
Achieved an R² of 0.9903 and MSE of 1.2746 × 10 –7 with the BPNN
BPNN predicted peak temperatures accurately with an error of 1.01%

Abstract

This paper presents an optimized approach for predicting peak temperatures during Friction Stir Welding (FSW) of Al 6061 T6 alloys. COMSOL Multiphysics is used to perform finite element analysis (FEA) to predict peak temperatures, incorporating seven distinctive welding parameters: tool material, pin diameter, shoulder diameter, tool rotational speed, welding speed, axial force, and coefficient of friction. A novel methodology integrating the Taguchi method, Analysis of Variance (ANOVA), and machine learning was employed to optimize parameters and predict peak temperatures. The influence of these parameters was investigated through L32 Taguchi array and ANOVA, revealing axial force and tool rotational speed as the most significant parameters affecting peak temperatures. Some simulations showed temperatures exceeding the material’s melting point, indicating the need for improved thermal control. The experimental work was compared with FEA results that confirmed a strong agreement between the two. A feed-forward backpropagation neural network (BPNN) was implemented, achieving an R² of 0.9903 and a mean squared error (MSE) of 1.2746 × 10 –7 . BPNN predicted peak temperatures with an error of 1.01%, outperforming Taguchi (3.57%) and ANOVA (3.39%). These findings contribute to sustainable welding practices by minimizing excessive heat generation, preserving material properties, and enhancing weld quality.

Ask AI

Helpful

Bookmark

View Full Paper