[25p-61C-1~13] 23.1 Joint Session N "Informatics"

3:45 PM - 4:00 PM JST(6:45 AM - 7:00 AM UTC)Presentation by Applicant for JSAP Young Scientists Presentation Award

[25p-61C-11] Exploration of Materials Design Guideline via Fusion of Machine Learning and Domain Knowledge for Anion Exchange Membrane Polymer

〇YinKan Phua¹, Tsuyohiko Fujigaya^1,2,3, Koichiro Kato^1,2,4 (1.Grad. Sch. of Eng., Kyushu Univ., 2.CMS, Kyushu Univ., 3.WPI-I2CNER, Kyushu Univ., 4.RIIT, Kyushu Univ.)

Keywords:

Polymer,Machine Learning,Fuel Cell

Anion exchange membrane (AEM)-based polymer electrolyte fuel cells plays an essential role for the realization of hydrogen society. However, practical application is hindered by insufficient performance such as anion conductivity of AEMs. In this study, we aimed to introduce unsupervised machine learning (ML) to accelerate AEM research. Combining principal component analysis and uniform manifold approximation and projection as unsupervised ML method, a two-dimensional map of AEM chemical structures was obtained. This revealed seven distinct clusters, with one of the clusters comprise of consistently high performing AEMs. Interpreting the background of obtaining such results by leveraging domain knowledge, valuable insights that can potentially be used as material design guidelines was for next-generation AEMs were obtained.