Development of a Chemically Explainable Machine Learning Model that Allows Membrane Materials Exploration for Anion Exchange Membrane Fuel Cell

Development of a Chemically Explainable Machine Learning Model that Allows Membrane Materials Exploration for Anion Exchange Membrane Fuel Cell

Beginning from functional polymeric materials such as anion exchange membranes (AEM) used in AEM fuel cells (AEMFC), this study plans on representing complete monomer structure of functional polymers in the form of graphs, and use graph neural networks with attention mechanism to improve prediction interpretability and accuracy of prediction model. Through this study, a prediction model that can feedback useful polymer structure design guidelines to experimentalist will be built, thereby realizing a model that is highly interpretable, generalizable and carries the ability to act as a materials design guideline for experimentalists. This study also helps to accelerate the realization of decarbonized society through contributing to the research and development of AEMFC, a type of FC that is essential to achieve the decarbonization goal.

Beginning from functional polymeric materials such as anion exchange membranes (AEM) used in AEM fuel cells (AEMFC), this study plans on representing complete monomer structure of functional polymers in the form of graphs, and use graph neural networks with attention mechanism to improve prediction interpretability and accuracy of prediction model. Through this study, a prediction model that can feedback useful polymer structure design guidelines to experimentalist will be built, thereby realizing a model that is highly interpretable, generalizable and carries the ability to act as a materials design guideline for experimentalists. This study also helps to accelerate the realization of decarbonized society through contributing to the research and development of AEMFC, a type of FC that is essential to achieve the decarbonization goal.