A Tree-Driven Ensemble Learning Approach to Predict FS Welded Al-6061-T6 Material Behavior

Abstract

This paper proposes a machine learning approach to forecast the mechanical behavior of an aluminum alloy, Al6061-T6, in the case of friction stir welding. Essentially, we investigate the performance of the bagged trees regression (BT) in forecasting the stress-strain curve of an aluminum alloy. This choice's motivation is due to BT's ability to improve the performance of machine learning models by combining multiple learners versus single regressors. Actual data was gathered by performing uniaxial tensile testing on both base material and joined using FSW at a deformation speed of 10 -3 s-1 . Then, the performance of the BT model is compared to that of the Support Vector regression, and it proved to be more accurate.

Machine learning predictions and benchmarking of non-linear mechanical behavior of polymer composites

“…This method involves costly and extensive experiments for the microstructure characterization, which can be influenced by even minor changes in the microstructure, and depends on the choice of the representative area. Dorbane et al 10 conducted a study and predicted the mechanical properties of an aluminum alloy in the friction stir welding (FSW) using the stress-strain data constructed as time-series. The time-series estimation cannot predict entirely new stress-strain curves and is greatly influenced by the noise which is undeniable in experiments.…”