An Automated Model Discovery Framework

The topic aims to develop an advanced framework for simulating quasi-brittle materials behavior using autonomous model learning techniques. The idea is to integrate Physics-Informed Neural Networks with a mechanical material model to overcome traditional limitations in constitutive modeling. Utilizing machine learning and experimental data, the framework seeks to dynamically adapt and optimize models, enhancing the accuracy and efficiency of civil engineering structure simulations. The interdisciplinary approach combines insights from mechanics, machine learning, and material science to create robust, interpretable models.