| Abstract Scope |
Here we report new design approaches for various materials, such as nanocomposite materials, biomaterials, and bioinspired structural materials, using artificial intelligence (AI). AI can substantially improve computational ability, especially in multiscale modeling. Facilitated by a generative neural network trained with a dataset of Protein Data Bank (PDB) to generate de novo proteins with the desired ratio of secondary structures without running conventional simulations. We extend the capability of physical simulations beyond property predictions to optimize the design process through the algorithm. We also develop an algorithm consists of a machine learning predictor conjoined with an AI improved genetic algorithm, applied to discover biomaterials in a vast space of possible solutions. Our AI model generates the solutions at a dramatically lower computational cost compared to brute-force searching methods. These AI approaches can be easily applied to other nanocomposites, biomaterials, and other material classes and provides a transferrable and reliable design. |