An ATPSO-SVM prediction model for flow stress investigation of lightweight materials: a case study of 6181/6016H18 aluminum alloys

Al–Mg–Si (6xxx series) aluminum alloys are widely employed in the automotive industry for lightweight applications, but crack formation during thermal forming remains a common issue. To address the problem and enhance the mechanical properties of these materials, an in-depth study of the flow stress is crucial. This work introduces the active target particle swarm optimization (ATPSO) algorithm into the improvement of the support vector machine (SVM), forming the ATPSO-SVM prediction model. The flow stress of two extensively used lightweight materials, specifically 6181H18 and 6016H18 aluminum alloys, are predicted and comparatively analyzed. Numerical results indicate that both the SVM and ATPSO-SVM models are effective in predicting the flow stress. Error analysis results show that the ATPSO-SVM model improves prediction accuracy by over 24% compared to the original model, demonstrating its superior accuracy. This investigation reflects the impact of various thermal deformation parameters on flow stress, providing significant technical references for the development of lightweight materials.