The applicability of machine learning in polyurethane design

Abstract

Abstract. Polyurethanes (PU) are used in wide range of products (e.g. construction materials). The properties of polyurethane-based materials can be modified and fine-tuned by using additives (e.g. fillers). New synthetic recipes are developed to create better materials is almost exclusively based on trial-anderror cycles which is a time and material intensive process. By using machine learning (ML) algorithms the process can be significantly accelerated. Therefore, to develop new polyurethane types, we are proposing to combine the strength of computational tools with experimental methods and data.