Parameters Identification of Rubber-like Hyperelastic Material Based on General Regression Neural Network

Area of Science:

• Computational mechanics
• Materials science
• Machine learning

Background:

• Hyperelastic materials, such as rubber, require accurate constitutive models for engineering applications.
• Identifying material parameters for these models is crucial but often challenging.
• Existing methods may lack universality or precision for complex rubber-like materials.

Purpose of the Study:

• To develop a systematic and precise scheme for identifying material parameters in constitutive models of hyperelastic materials.
• To leverage the combined strengths of general regression neural networks (GRNN), experimental data, and finite element analysis (FEA).
• To validate the proposed approach using silicone rubber material parameter identification.

Main Methods:

• Utilizing finite element analysis (FEA) to generate learning samples for the general regression neural network (GRNN).
• Employing uniaxial tensile test data as the target values for training the GRNN model.
• Implementing a GRNN-based approach for the inverse problem of material parameter identification.

Main Results:

• The proposed GRNN-based scheme effectively identifies material parameters for hyperelastic models.
• Validation using silicone rubber demonstrated high universality and good precision.
• The method successfully determined parameters for complex rubber-like hyperelastic material constitutive models.

Conclusions:

• The developed GRNN-based approach offers a robust and accurate method for hyperelastic material parameter identification.
• This technique shows significant potential for application to a wide range of rubber-like materials.
• The combination of FEA and GRNN provides an efficient pathway for constitutive model calibration.