Neural Network-Based Identification and MPC Control of SMB Chromatography

Wang, Chaoyong; Engell, Sebastian; Hanisch, Felix

doi:10.3182/20020721-6-es-1901.01156

Cited by 10 publications

(8 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The full economical potential of the SMB can be exploited by using a proper feedback control scheme. Several approaches have been proposed 3–16. A detailed review of these different control schemes may be found in the literature 17, 18.…”

Section: Introductionmentioning

confidence: 99%

Incidence of extraovarian primary cancers in the United States, 1992-1997

Roffers

Johnson

et al. 2003

Cancer

View full text Add to dashboard Cite

show abstract

Section: Introductionmentioning

confidence: 99%

Incidence of extraovarian primary cancers in the United States, 1992-1997

Roffers

Johnson

et al. 2003

Cancer

View full text Add to dashboard Cite

show abstract

“…Model identification in MPC is also of importance where commercial implementations of MPC structures typically rely on step test data to develop input/output relationships (model identification) that are extracted to build an MPC [53]. Since the 1990s, there has been an interest in using Neural Networks to identify complex models (non-linear) for use in MPC [54]. Besides, on-line model identification using data-driven methods has been proposed to counter state-and time-varying processes [55].…”

Section: Model Predictive Control (Mpc)mentioning

confidence: 99%

Digital Twin in biomanufacturing: challenges and opportunities towards its implementation

Udugama

López

Gargalo

et al. 2021

Syst Microbiol and Biomanuf

View full text Add to dashboard Cite

The domain of industrial biomanufacturing is enthusiastically embracing the concept of Digital Twin, owing to its promises of increased process efficiency and resource utilisation. However, Digital Twin in biomanufacturing is not yet clearly defined and this sector of the industry is falling behind the others in terms of its implementation. On the other hand, some of the benefits of Digital Twin seem to overlap with the more established practices of process control and optimization, and the term is vaguely used in different scenarios. In an attempt to clarify this issue, we investigate this overlap for the specific case of fermentation operation, a central step in many biomanufacturing processes. Based on this investigation, a framework built upon a five-step pathway starting from a basic steady-state process model is proposed to develop a fully-fledged Digital Twin. For demonstration purposes, the framework is applied to a bench-scale second-generation ethanol fermentation process as a case study. It is proposed that the success or failure of a fully-fledged Digital Twin implementation is determined by key factors that comprise the role of modelling, human operator actions, and other propositions of economic value.

show abstract

“…Several approaches have been proposed. [3][4][5][6][7][8][9][10][11][12][13][14][15][16] A detailed review of these different control schemes may be found in the literature. 17,18 In general, the drawback of these approaches is the need for accurate data about the system.…”

Section: Introductionmentioning

confidence: 99%

‘Cycle to cycle’ optimizing control of simulated moving beds

Grossmann¹,

Amanullah²,

Erdem³

et al. 2007

AIChE Journal

View full text Add to dashboard Cite

in Wiley InterScience (www.interscience.wiley.com).The novel feature of the presented simulated moving bed (SMB) controller is its capability to make use of the average outlet concentration of the product streams over a cycle as feedback information, i.e., 'cycle to cycle' control. Its effectiveness is confirmed experimentally on an eight-column four-section laboratory SMB unit, which is used to separate a binary mixture of the nucleosides, uridine and guanosine. The performance of the 'cycle to cycle' SMB control scheme is also demonstrated by several SMB simulation runs that are chosen to test the robustness of the controller. Furthermore, the case where measurements have a time delay is presented. The results illustrate that the 'cycle to cycle' controller is able to meet the products' purity specifications and operate the process optimally with minimal information about the system regardless of the disturbances that might take place during the operation.Note that Y k contains the concentrations of both species for every time step n 5 0, . . ., N 2 1 of cycle k. Figure 4. Performance of the controller on the laboratory SMB plant for case studies 1 and 2. Outlet purities and cost function vs. time measured in cycles. Feed pump delivers 10% more than its set point after cycle 70. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

show abstract

Neural Network-Based Identification and MPC Control of SMB Chromatography

Cited by 10 publications

References 8 publications

Incidence of extraovarian primary cancers in the United States, 1992-1997

Incidence of extraovarian primary cancers in the United States, 1992-1997

Digital Twin in biomanufacturing: challenges and opportunities towards its implementation

‘Cycle to cycle’ optimizing control of simulated moving beds

Contact Info

Product

Resources

About