English   español  
Please use this identifier to cite or link to this item: http://hdl.handle.net/10261/141384
Share/Impact:
Statistics
logo share SHARE   Add this article to your Mendeley library MendeleyBASE
Visualizar otros formatos: MARC | Dublin Core | RDF | ORE | MODS | METS | DIDL
Exportar a otros formatos:
Title

Towards predictive models in food engineering: Parameter estimation dos and don'ts

AuthorsBalsa-Canto, Eva ; Vilas Fernández, Carlos ; Arias-Méndez, Ana ; García, Miriam R. ; Alonso, Antonio A.
KeywordsModel calibration
Optimal experimental design
Parameter uncertainty
Food processes
Predictive models
Issue Date2015
CitationEFFoST International Conference (2015)
AbstractRigorous, physics based, modeling is at the core of computer aided food process engineering. Models often require the values of some, typically unknown, parameters (thermo-physical properties, kinetic constants, etc). Therefore, parameter estimation from experimental data is critical to achieve desired model predictive properties. Unfortunately, it must be admitted that often experiment design and modeling are fully separated tasks: experiments are not designed for the purpose of modeling and models are usually derived without paying especial attention to available experimental data or experimentation capabilities. When, at some point, the parameter estimation problem is put on the table, modelers use available experimental data to ``manually'' tune the unknown parameters. This results in inaccurate parameter estimates, usually experiment dependent, with the implications this has in model validation. This work takes a new look into the parameter estimation problem in food process modeling. First the common pitfalls in parameter estimation are described. Second we present the theoretical background and the numerical techniques to define a parameter estimation protocol to iteratively improve model predictive capabilities. This protocol includes: reduced order modeling, structural and practical identifiability analyses, data fitting with global optimization methods and optimal experimental design. And, to finish, we illustrate the performance of the proposed protocol with an example related to the thermal processing of packaged foods. The model was experimentally validated in the IIM-CSIC pilot plant
Description1 póster.-- 29th EFFoST International Conference, 10-12 November 2015, Athens, Greece
URIhttp://hdl.handle.net/10261/141384
Appears in Collections:(IIM) Comunicaciones congresos
Files in This Item:
File Description SizeFormat 
Poster_Towards_predictive_models_2015.pdfPóster10,84 MBAdobe PDFThumbnail
View/Open
Towards_predicitve_models_2015.pdfAbstract821,93 kBAdobe PDFThumbnail
View/Open
Show full item record
Review this work
 


WARNING: Items in Digital.CSIC are protected by copyright, with all rights reserved, unless otherwise indicated.