| 학회 |
한국공업화학회 |
| 학술대회 |
2019년 봄 (05/01 ~ 05/03, 부산 벡스코(BEXCO)) |
| 권호 |
23권 1호 |
| 발표분야 |
(화학공정) 4차산업 혁명 시대의 공정시스템기술 적용 |
| 제목 |
Optimal control for aftertreatment system of vehicle using digital twin and reinforcement learning |
| 초록 |
Diesel vehicles cause environmental problems including NOx emission. To reduce NOx, various catalysts have been developed, and operation strategies have been proposed to use catalytic reactors efficiently. This study constructs a digital twin for a selective catalytic reactor which is widely used, and we design optimal control policies using a reinforcement learning method with the digital twin. For the automotive aftertreatment system with limited information to be measured, more virtual data can be obtained using a digital twin. In addition, optimal control policies reflecting the characteristics of the system are proposed using the reinforcement learning method based on the digital twin. |
| 저자 |
김연수1, 이병준1, 임산하1, 정창호2, 김창환2, 김용화2, 이종민1
|
| 소속 |
1서울대, 2현대자동차 |
| 키워드 |
Digital twin; Reinforcement learning; Diesel aftertreatment system
|
| E-Mail |
|
