Optimization and Uncertainty Quantification of High-Fidelity Material Model Parameters for Dynamic Loading Simulation

Systematic construction, optimization, and validation of high-fidelity material models are crucial for dynamic load simulations. This study details a methodology for building and validating such models on the Dayu Digital Platform. First, a parameterized equation of state (EOS) framework is constructed, integrating all available experimental data with associated uncertainties. Global optimization methods are then employed to determine the optimal EOS parameters. Second, the optimized EOS is coupled with a constitutive model containing undetermined parameters. One-dimensional or two-dimensional numerical simulations are conducted, reproducing experimental conditions. Optimization algorithms iteratively adjust the constitutive model parameters to achieve a globally optimal match between simulated waveforms and experimental waveforms, thereby precisely calibrating the constitutive parameters. Finally, the optimized EOS and calibrated constitutive model are integrated to form a complete material model, and standardized interfaces are developed for both in-house and commercial simulation software. The validation of material models is accomplished by comparing simulated predictions under new experimental conditions with experimental results. Within this process, the optimization of theoretical model parameters constrained by experimental data is achieved using the self-developed novel importance cross-optimization (ICON) algorithm. The uncertainty in material model parameters and its propagation to computed physical quantities are rigorously quantified using a self-developed Bayesian uncertainty quantification (UQ) program.