| 학회 |
한국화학공학회 |
| 학술대회 |
2022년 봄 (04/20 ~ 04/23, 제주국제컨벤션센터) |
| 권호 |
28권 1호, p.154 |
| 발표분야 |
[주제 2] 기계학습 |
| 제목 |
Using machine learning with explainable artificial intelligence tool to develop catalyst for dry reforming of methane |
| 초록 |
Reaction conversion of catalysts for dry reforming of methane (DRM) is highly dependent on the composition and components of active metals, supports, and promoters, which are the design variables of the catalyst. Nevertheless, the conversion of DRM is governed by the reaction temperature due to the nature of the endothermic reaction, making the design difficult. Therefore, it is necessary to develop a model that evaluates and fully reflects the effects of operating conditions such as reaction temperature and preparing conditions such as calcination. To this end, we created a model that can predict the conversion of DRM catalysts with high accuracy (R2>0.97) using machine learning (ML) and explainable artificial intelligence (XAI) tools. Furthermore, we used the ML model and XAI tool to recommend catalyst candidates with the best performance for each reaction temperature (600, 700, 800℃) and analyze the prediction results. This study can be used not only for DRM catalyst development but also as an effective tool for deriving optimal candidates for material design with many variables to consider and with different feature importance and analyzing the impact of each variable. |
| 저자 |
노지원1, 권혁원1, 주종효1, 박현도1, 조형태1, 문일2, 노인수3, 김정환1
|
| 소속 |
1한국생산기술(연), 2연세대, 3서울과학기술대 |
| 키워드 |
촉매(Catalyst) |
| E-Mail |
|
| 원문파일 |
초록 보기 |
