Enhancing Constitutive Description of 5A06 Aluminum Alloy During Warm Deformation Using Machine Learning-Assisted...

Area of Science:

• Materials Science
• Mechanical Engineering
• Computational Science

Background:

• Accurate characterization of warm deformation behavior is crucial for optimizing metal forming processes.
• Conventional models often lack precision in predicting complex material responses under varying thermal and strain conditions.
• The 5A06 aluminum alloy's workability requires advanced predictive tools for industrial applications.

Purpose of the Study:

• To develop and validate machine learning-assisted Johnson-Cook (ML-JC) frameworks for characterizing the warm deformation of 5A06 aluminum alloy.
• To enhance the accuracy and generalization capability of deformation behavior prediction models.
• To establish a workflow for identifying and mitigating defects during warm forming processes.

Main Methods:

• Construction of two ML-JC frameworks: parallel-decoupled (PD-ANN-JC) and multi-objective integrated (MOI-ANN-JC), utilizing artificial neural network (ANN) surrogate models.
• Quantitative validation using testing and validation datasets to assess stress prediction accuracy and generalization.
• Integration of the MOI-ANN-JC framework with finite element (FE) simulations for dynamic visualization of thermal softening and defect analysis.

Main Results:

• Both ML-JC frameworks significantly outperform the conventional Johnson-Cook model in stress prediction accuracy and generalization.
• The MOI-ANN-JC framework achieved superior performance with an average absolute relative error (AARE) of 1.424% and an R² of 0.997 on the testing set.
• Direct spatial correspondence between low thermal softening exponent (m) regions and macroscopic defects was established and verified through component forging.

Conclusions:

• The MOI-ANN-JC framework, leveraging an ANN-mnδ surrogate model, provides a highly accurate and robust method for predicting warm deformation behavior.
• This approach effectively bridges the limitations of conventional processing maps in specific temperature regimes.
• The developed workflow offers a broadly applicable solution for complex forming optimization and deformation stability evaluation in materials processing.