ISSN: 0304-128X ISSN: 2233-9558
Copyright © 2024 KICHE. All rights reserved

Articles & Issues

Conflict of Interest
In relation to this article, we declare that there is no conflict of interest.
articles This is an Open-Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/bync/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Copyright © KIChE. All rights reserved.

All issues

공정자료만을 이용한 모델링 및 최적화에서 Data Reconciliation 과 Gross Error Detection

data Reconciliation and Gross Error Detection for Data Intensive Modeling and Optimization

HWAHAK KONGHAK, October 1995, 33(5), 652-658(7), NONE
downloadDownload PDF

Abstract

공장에서 측정되는 데이터들은 random error 및 gross error와 같은 측정오차를 포함하고 있으며, 이 데이터들은 공정을 설명한는 물질수지식이나 에너지수지식을 만족시키지 못한다. 이와 같은 데이터를 이용하여 공정해석및 최적화를 하기 위하여는 측정치들을 조정하고(data reconciliation),
gross error를 제거해야 하는 문제(gross error detection)가 선행되어야 한다. 이러한 자료 처리 방법으로는 기존에는 수학적, 통계학적 방법들을 많이 이용하고 있으나, 최근에는 인공신경망을 이용하기도 한다. 본 연구에서는 autoassociative neural network(AAN)과 AAN에 견실성을 부여한 Robust AAN을 이용하여 gross error detection과 data reconciliation(GED/DR)을 동시에 수행할 수 있었다. AAN과 RAAN을 CSTR에 적용하여 비교하고, GED/DR에의 적용 가능성을 검증하였다. AAN을 이용한 GED/DR보다 RAAN을 이용하였을 경우의 GED/DR이 더욱 우수한 결과를 보였다. 공정에 대한 수학적 모델이 없이 공정자료만으로 모델링과 최적화를 행하는 data intensive모델링과 최적화기법에서 위의 방법으로 GED/DR이 가능함을 알 수 있었다.
Measured process data are usually containing random errors and gross errors. These measured data do not satisfy process constraints such as the mass and energy balances that describe a process. For the use of these error-contained data in process analysis and optimization, the preprocessing steps such as gross error identification and elimination, and data reconciliation(data rectification)are prerequisite. The existing methods are based on mathematical and statistical techniques, but recently neural networks were investigated for data rectification. In this study, autoassociative neural networks(AAN) and robust ANN(RAAN) were applied for the data rectification of process data of CSTR. The performance of RAAN proved to be superior to that of AAN in the data rectification. we conclude that the use of AAN and RAAN appears to be a promising tool data rectification.

Keywords

References

Eom TI, Kim IW, Park SW, "Distillation Process optimization with Neural Networks," in Submitted to I&EC (1995)
Ferguson RB, International Technical Symposium at SICHEM'92
Kuehn DR, Davison H, Chem. Eng. Prog., 57, 44 (1961)
Hlavacek V, Comput. Chem. Eng., 1, 75 (1977) 
Mah R, Proceedings of the Engineering Foundation Conference on Chemical Process Control (Vol. 2), New York: Engineering Foundation, 525-540 (1981)
Crowe C, AIChE J., 32, 616 (1986) 
Crowe C, Garcia-Campos Y, Hrymak A, AIChE J., 29, 881 (1983) 
Knepper J, Gorman J, AIChE J., 26, 260 (1980) 
Kim IW, Liebman MJ, Edgar TF, AIChE J., 36, 985 (1990) 
Mah R, Tamhane A, AIChE J., 28, 828 (1982) 
Stephenson G, Shewchuk C, AIChE J., 32, 247 (1986) 
Iordache C, Mah R, Tamhane A, AIChE J., 31, 1187 (1985) 
Serth R, Heenan W, AIChE J., 32, 733 (1986) 
Kramer MA, AIChE J., 37, 233 (1991) 
Leonard J, Kramer MA, Comput. Chem. Eng., 14, 337 (1990) 
Kramer MA, Comput. Chem. Eng., 16, 313 (1992) 
Serth R, Valero C, Heenan W, Chem. Eng. Sci., 51, 89 (1987)

The Korean Institute of Chemical Engineers. F5, 119, Anam-ro, Seongbuk-gu, 233 Spring Street Seoul 02856, South Korea.
Phone No. +82-2-458-3078FAX No. +82-507-804-0669E-mail : kiche@kiche.or.kr

Copyright (C) KICHE.all rights reserved.

- Korean Chemical Engineering Research 상단으로