Modeling chemical processes using prior knowledge and neural networks

Thompson, Michael L.; Kramer, Mark A.

doi:10.1002/aic.690400806

Cited by 495 publications

(264 citation statements)

References 39 publications

Supporting

Mentioning

248

Contrasting

Unclassified

Order By: Relevance

“…The complementary strengths and weaknesses of these two modelling routes are widely recognized [9,10,11], and the value of an approach that allows for their strengths to complement one another is generally acknowledged [11,12].…”

Section: -Background Information On Complex System Modellingmentioning

confidence: 99%

Development of PUNDA (Parametric Universal Nonlinear Dynamics Approximator) Models for Self-Validating Knowledge-Guided Modelling of Nonlinear Processes in Particle Accelerators \& Industry

Sayyarrodsari¹,

Schweiger²,

Hartman³

2007

View full text Add to dashboard Cite

Section: -Background Information On Complex System Modellingmentioning

confidence: 99%

Development of PUNDA (Parametric Universal Nonlinear Dynamics Approximator) Models for Self-Validating Knowledge-Guided Modelling of Nonlinear Processes in Particle Accelerators \& Industry

Sayyarrodsari¹,

Schweiger²,

Hartman³

2007

View full text Add to dashboard Cite

“…This enables one to numerically forecast the results, being the " simulated object's responses", for assumed, interesting input d ata-sets from the examined experimental range [e.g., Stephanopoulos and Han, (1996), Willis et al (1992)]. In the recent days the ANN has proved their usefulness within many different chemical engineering problems proved difficult for conventional modeling approaches, both applied independently [ e.g., Meert and Rijckaert (1998), Hoskins and Himmelblau (1988) ] or being an integral part of the complex hybrid models [e.g., Psichiogos and Ungar (1992), Thompson and Kramer (1994)]. …”

Section: Iron Oxide Reduction Processmentioning

confidence: 99%

Simultaneous influence of gas mixture composition and process temperature on Fe2O3->FeO reduction kinetics: neural network modeling

Piotrowski

Wiltowski

Mondal

et al. 2005

Braz. J. Chem. Eng.

View full text Add to dashboard Cite

-The kinetics of Fe 2 O 3 →FeO reaction was investigated. The thermogravimetric (TGA) data covered the reduction of hematite both by pure species (nitrogen diluted CO or H 2 ) and by their mixture. The conventional analysis has indicated that initially the reduction of hematite is a complex, surface controlled process, however once a thin layer of lower oxidation state iron oxides (magnetite, wüstite) is formed on the surface, it changes to diffusion control. Artificial Neural Network (ANN) has proved to be a convenient tool for modeling of this complex, heterogeneous reaction runs within the both (kinetic and diffusion) regions, correctly considering influence of temperature and gas composition effects and their complex interactions. ANN's model shows the capability to mimic some extreme (minimum) of the reaction rate within the determined temperature window, while the Arrhenius dependency is of limited use.

show abstract

“…Depending on the information available, process modelling may be performed through a number of approaches including mechanistic, black-box (or data-driven), or hybrid modelling. By bringing together both existing mechanistic knowledge and data gathered from the process, a hybrid model that fuses both components has been shown, in a number of applications, to be advantageous when compared with a model formulated from either limited mechanistic knowledge or one constructed solely from the process data (Psichogios, Ungar, 1992;Thompson, Kramer, 1994;Duarte et al, 2004;Oliveira, 2004). The advantages of hybrid models have motivated a number of applications, such as the modelling of batch polymerization reactors (Tian et al, 2001), fermentation processes (Wang et al, 2009;Saraceno et al, 2009) and boilers (Rusinowski, Stanek, 2009).…”

Section: Introductionmentioning

confidence: 99%