A gene expression programming-enabled prediction for hot flow stress of 5A06 aluminum alloy used in large structures

Hot rolling of 5A06 aluminum alloy is crucial for manufacturing large structural components in the aerospace and shipbuilding industries. Accurate prediction of the alloy’s hot flow stress is essential for optimizing the rolling process and ensuring product quality. Hot compression experiments were conducted to obtain the experimental data using the Gleeble-3500 in this study. Research was performed at several temperature levels for deformation (specifically 573 K, 623 K, etc., up to 773 K), utilizing diverse strain rates (from 0.01 s −1 to 10 s −1 , increasing by 10 times each), concluding with a peak deformation percentage of 60 %. It established Johnson-Cook (J-C), Zerilli-Armstrong (Z-A), and Gene Expression Programming (GEP) constitutive models, evaluating accuracy via statistics and DEFORM-2D finite element simulation (FES). Statistics showed GEP was best: R²= 0.98 (J-C:0.96, Z-A:0.95), lowest RMSE= 11.41(J-C:26.51, Z-A=30.43), MAE= 8.62(J-C:18.83, Z-A: 15.86), AARE= 9.47 % (J-C: 31.40 %, Z-A: 11.64 %). FES further confirmed the GEP model’s superiority, as it exhibited the smallest load deviation from experimental values. Furthermore, deformation stress predictions of non-experimental conditions were conducted under two sets of conditions: a strain rate of 0.1 s −1 with deformation temperatures of 593 K and 643 K, and a deformation temperature of 573 K with strain rates of 0.05 s −1 , 0.5 s −1 , and 2.5 s −1 . The results show that the stress curves at 593 K and 643 K align with the trend of experimental curves at 573 K, 623 K, and 673 K and lie within their corresponding intervals; similarly, the stress curves at 0.05 s −1 , 0.5 s −1 , and 2.5 s −1 conform to the trend of experimental curves at 0.01 s −1 , 0.1 s −1 , and 1 s −1 , with their positions following the rule that higher strain rates lead to greater deformation stresses. The GEP model can not only effectively predict the flow stress under experimental conditions but also forecast the flow stress under non-experimental deformation conditions, thus providing a valuable tool for the numerical simulation and process optimization of the hot rolling process of 5A06 aluminum alloy.