Premium
Statistical batch process monitoring using gray models
Author(s) -
Van sprang E. N. M.,
Ramaker H.J.,
Westerhuis J. A.,
Smilde A. K.,
Wienke D.
Publication year - 2005
Publication title -
aiche journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.958
H-Index - 167
eISSN - 1547-5905
pISSN - 0001-1541
DOI - 10.1002/aic.10348
Subject(s) - fault detection and isolation , gray (unit) , batch processing , process (computing) , computer science , process modeling , consistency (knowledge bases) , statistical model , process analytical technology , work in process , artificial intelligence , engineering , radiology , programming language , operating system , medicine , operations management , actuator
A complete strategy for monitoring industrial batches processes using gray models is presented including fault detection and fault diagnosis tools. The use of gray models is a novel concept in batch process modeling and monitoring. A gray model is a hybrid model, intermediate between hard (white) process models and soft (black) models, combining the advantages of both approaches. The principles of gray models are explained and it is shown how these models can be constructed. For this purpose an industrial batch process is available that is spectroscopically monitored, and an explanation is provided as to how the spectroscopic measurements are combined with prior process knowledge. To show the versatility of the strategy, two types of gray models are constructed and used for statistical batch process monitoring. The two models are compared and validated for both on‐line monitoring and post‐batch analysis. For the latter, the batch consistency number (BCN) is introduced to have a fast and simple post‐batch analysis. The results show how these models help to detect and diagnose process upsets. The use of gray models for batch process monitoring results in a fast detection of process upsets and a good fault diagnosis. © 2005 American Institute of Chemical Engineers AIChE J, 51: 931–945, 2005