Acquisition of Dynamic Material Properties in the Electrohydraulic Forming Process Using Artificial Neural Network

Area of Science:

• Materials Science
• Mechanical Engineering
• Computational Mechanics

Background:

• Electrohydraulic forming (EHF) is a high-velocity metal deformation process.
• Safety concerns limit direct observation of EHF in closed systems.
• Accurate material properties are crucial for simulating EHF.

Purpose of the Study:

• To develop a method for obtaining accurate material properties for EHF simulations.
• To utilize surrogate modeling combining reduced order models and artificial neural networks.
• To estimate optimal material parameters for Al 6061-T6 sheet metal.

Main Methods:

• Numerical estimation of material properties using a surrogate model.
• Application of the Cowper-Symonds constitutive equation with unknown strain rate parameters.
• Utilizing LS-DYNA for numerical simulations and principal component analysis for data reduction.
• Employing artificial neural networks to predict weighting coefficients.

Main Results:

• A surrogate model was developed using artificial neural networks and reduced order models.
• Optimal strain rate parameters for Al 6061-T6 were estimated by minimizing the root mean square error.
• The surrogate model demonstrated high reliability in predicting EHF behavior.

Conclusions:

• The combined approach of surrogate modeling and numerical simulation effectively determines material properties for EHF.
• The estimated material parameters enhance the accuracy of virtual simulations for Al 6061-T6.
• This methodology provides a reliable framework for analyzing complex forming processes.