OPTIMIZATION OF RESIDUAL-BASED SHOCK CAPTURING IN SOLIDS USING AI

Residual-based shock capturing methods play a crucial role in accurately simulating shockwaves in solid materials. These methods aim to capture and resolve the sharp gradients and discontinuities that arise during high-speed impact events. However, determining an optimal shock capturing strategy that balances accuracy and computational efficiency remains a challenge. In this study, we propose an innovative approach to optimize residual-based shock capturing in solids using artificial intelligence (AI). The AI-based optimization framework leverages the power of machine learning algorithms to learn and adaptively adjust the shock capturing parameters in real-time. By analyzing the residuals generated during the simulation, the AI model identifies areas where additional shock capturing is required and intelligently adjusts the capturing parameters accordingly. This adaptive approach ensures that the shock capturing is selectively enhanced in regions with significant shockwave propagation, while minimizing unnecessary computational overhead in less critical areas. A large dataset of simulated shockwave events is generated with varying shock capturing parameters. The training dataset consists of paired inputoutput examples, where the inputs represent the simulation conditions and the outputs are the optimized shock capturing parameters. By training the AI model on this dataset, it learns the complex relationships between the simulation conditions and the corresponding optimal shock capturing settings. The trained AI model is then integrated into a numerical simulation framework for real-time shockwave simulations in solid materials. During the simulation, the model continuously monitors the evolving residuals and dynamically adjusts the shock capturing parameters based on the learned patterns. This adaptive process ensures that the shock capturing remains effective throughout the simulation, adapting to changing shockwave characteristics and enhancing accuracy. We evaluate the performance of the AIoptimized shock capturing approach through extensive numerical experiments and comparisons with traditional methods. The results demonstrate that the AI-based approach consistently improves the accuracy of shockwave simulations while reducing computational costs. Furthermore, the adaptability of the approach allows it to handle complex shockwave phenomena, such as multi-material interactions and evolving shockwave fronts.