| 학회 | 한국화학공학회 |
| 학술대회 | 2022년 봄 (04/20 ~ 04/23, 제주국제컨벤션센터) |
| 권호 | 28권 1호, p.123 |
| 발표분야 | [주제 2] 기계학습 |
| 제목 | Product yield prediction model of Naphtha Cracking Center using Deep Neural Networks |
| 초록 | The naphtha cracking center (NCC) is an indispensable process for manufacturing ethylene (EL) and propylene (PL). Among operating conditions of furnace in NCC, coil outlet temperature (COT) is a crucial parameter for product yield. Until now, the decision of optimal COT has been decided by the simulation program at the plant. However, this process is very inefficient because the complex simulation model that according to inlet compositions and product prices makes it time-consuming problem. Here, this study suggests a product yield prediction model of NCC using Deep Neural Networks (DNN) to find the optimal COT in a short time. 784 simulation data were used for DNN model and 24 inlet compositions of NCC furnace and COT were used as input variables. The model shows high performance with R2 0.9696 and 0.9474 for EL and PL, respectively. Therefore, this study is expected to propose the optimal COT condition rapidly by case study using the model. For the future work, we would propose the optimal inlet composition and COT conditions considering profit as well as energy consumption using machine learning. Keywords: naphtha cracking center, deep neural network, machine learning |
| 저자 | 주종효1, 권혁원1, 이창용2, 이유민2, 박성문2, 조형태1, 김정환1 |
| 소속 | 1한국생산기술(연), 2한화토탈 |
| 키워드 | 공정시스템(Process Systems Engineering) |
| 원문파일 | 초록 보기 |
