แจ้งเอกสารไม่ครบถ้วน, ไม่ตรงกับชื่อเรื่อง หรือมีข้อผิดพลาดเกี่ยวกับเอกสาร ติดต่อที่นี่ ==>
หากไม่มีอีเมลผู้รับให้กรอก thailis-noc@uni.net.th
หากไม่มีอีเมลผู้รับให้กรอก thailis-noc@uni.net.th
Chutithep Rochpuang. Utility consumption prediction for vinyl chloride monomer process with a performance degradation in a cracking furnace. Master's Degree(Chemical Engineering). Kasetsart University. Office of the University Library. : Kasetsart University, 2021.
Utility consumption prediction for vinyl chloride monomer process with a performance degradation in a cracking furnace
a hybrid deep learning approach
Name: Chutithep Rochpuang
Classification :.LCCS: TP1180.V48
LCSH:
Vinyl chloride
LCSH:
Furnaces
Abstract:
Energy efficiency is an essential tool for analyzing the process and determining the best operating conditions to reduce excess utility consumption, energy costs, and greenhouse gas emissions. However, because the petrochemical process is complex, with a recycle stream, reaction, and phase separation, determining utility consumption using a first-principles model is difficult. The utility acts as a medium for heat to be transferred from the source to the equipment. Although utility can represent energy consumption, it is difficult to calculate because some processes, such as coking in the thermal cracking furnace, have performance degradation in unit operation. As a result, this study proposes an integrating framework for estimating the operating period of the equipment before maintenance and predicting process utility consumption in the presence of performance degradation in unit operation. As an example, a vinyl chloride monomer process with coke formation in the pyrolysis furnace is considered. The proposed framework consists of two focused works: the classification of the furnace operating period and the utility consumption regression. In the classification section, a dual machine learning (ML) classifier is developed to estimate how long the process will run. The classification criteria are divided into three levels: oneweek, two-week, and four-week criteria. Measured VCM process information is used to perform feature selection via random forest-based recursive feature elimination. For the best performance, various combinations of the ML-ML model are established and tested using a cross-validation data set. The results show that the dual K-nearest neighbor (KNN) classifier with 4-week criteria performs the best in predicting the operating periods. In the regression section, the recurrent neural network (RNN) model with feature results from the dual KNN classifier is used to capture the dynamic behavior of the studied VCM process. The results show that a Gated Recurrent Unit (GRU) RNN outperforms a single RNN framework in utility consumption prediction.
Kasetsart University. Office of the University Library
Address:
Bangkok
Email:
tdckulib@ku.ac.th
Name: Chanin Panjapornpon
Role:
Thesis Advisor
Name: Paweena Prapainainar
Role:
Thesis CO-Advisor
Created:
2021
Modified:
2025-07-22
Issued:
2025-07-22
วิทยานิพนธ์/Thesis
application/pdf
URL: https://www.lib.ku.ac.th/KUthesis/2564/chutithep-roc-all.pdf
CallNumber:
TP1180.V48 .C48
eng
DegreeName: Master of Engineering
Level: Master's Degree
Descipline: Chemical Engineering
Grantor: Kasetsart University
©copyrights Kasetsart University
RightsAccess:
ใช้เวลา
0.026838 วินาที
Chutithep Rochpuang
| Title | Contributor | Type |
|---|---|---|
| Utility consumption prediction for vinyl chloride monomer process with a performance degradation in a cracking furnace
มหาวิทยาลัยเกษตรศาสตร์ Chutithep Rochpuang | Chanin Panjapornpon | วิทยานิพนธ์/Thesis |
| A LSTM-based utility consumption prediction for a process with performance degradation : a case study of a vinyl chloride monomer plant
มหาวิทยาลัยเทคโนโลยีพระจอมเกล้าพระนครเหนือ Chutithep rochpuang;Chanin Panjapornpon;Phantisa Limleamthong | บทความ/Article |
Chanin Panjapornpon
Paweena Prapainainar