Research Thrust II
This thrust aims to prove that physics-informed AI models can integrate multiscale physics into composites and can extend the limited experimental data to precisely predict material responses to loadings, which will significantly accelerate inverse design of composite materials and corresponding manufacturing methodologies. The rationale is based on the team’s previous success of using physics-informed AI models to tackle multiscale and multi-fidelity challenges in forward and inverse engineering problems. Physics-informed AI models for composites will be developed to enable efficient training and accurate prediction/generation for the inverse design of thermoplastic composites and their manufacturing processes.
Objectives of Thrust II
-
Develop PINN models for composites by encoding conservation laws and prior physical knowledge.
​
-
Develop multiscale deep neural operators (DNOs) to act as surrogates for expensive models, providing the capability to integrate coupled physics for on-the-fly prediction of composite properties.
​​
-
Develop multi-fidelity Bayesian DNN as an integrated material platform to blend computational data and experimental characterizations with noise and quantify how uncertainty propagates.
​
-
Develop MaterialGEN and ProcessGAN to generate the material architecture and sequential combination of manufacturing steps to optimize the manufacturing processes for targeted materials.