hig.sePublications
Change search
ExportLink to record
Permanent link

Direct link
BETA

Project

Project type/Form of grant
Project grant
Title [sv]
Flexibla modeller för smart underhåll
Title [en]
Flexible Models for Smart Maintenance
Abstract [sv]
Syfte och mål: Underhåll i befintliga anläggningar blir allt viktigare, där prediktivt underhåll har blivit en framväxande teknik. Användningen av digitala verktyg för beslutsstöd bidrar till en miljömässigt och ekonomiskt hållbar produktion. Inom detta projekt har olika typer av digitala tvillingar utformats och utvärderats. Specifikt har nya prediktiva modelltyper testats i två olika industriella fallstudier. Den modell som visat bäst generella resultat är LAVA-modellen. Dessutom har en teknisk plattform identifierats för att implementera metoden i befintliga anläggningar. Förväntade effekter och resultat: Målsättningen har varit att hitta generella metoder för smart underhåll i befintliga industrianläggningar. Metoderna har utvärderats både avseende användbarhet i specifika tillämpningar och hur väl de kan generaliseras för olika typer av anläggningar. Den modell som visat bäst generella resultat är LAVA-modellen som är en Black box modell. Fördelen med blackbox modeller är att de är generell, men fortfarande är processkännedom nödvändigt för implementering. Resultaten från projektet är lovande, men en längre testperiod krävs för att utesluta säsongsvariationer. Upplägg och genomförande: De två fallstudierna är en värmeväxlare på SSAB och en profilhyvel på Svenska fönster AB. Dessutom har en laborativ miljö på Högskolan i Gävle byggts upp då det inte varit möjligt att prova olika metoder för att detektera (och kanske även initiera) fel i industrianläggningar som var i drift. Mätdata från anläggningarna har inhämtats från befintliga styrsystem samt med ett kompletterande mätsystem. Modelleringsarbetet har skett off-line och analys av resultaten har gjorts gemensamt av modellutvecklare och personal med processkännedom.
Abstract [en]
Purpose and goal: Maintenance in existing plants is becoming increasingly important, where predictive maintenance has become an emerging technology. The use of decision support tools contributes to environmentally and economically sustainable production. Within this project, different types of digital twins have been designed and evaluated. Specifically, new predictive model types have been tested in two different industrial case studies. The model showing best overall results is the LAVA model. In addition, a technical platform has been identified for implementation in existing plants. Expected results and effects: The aim has been to find general methods for smart maintenance in existing industrial plants. The methods have been evaluated for both usability in specific applications and how well they can be generalized for any industrial plant. The model showing best overall results is the LAVA model, which is a Black box model. The advantage of black box models is that they are general; however, process knowledge is still necessary for implementation. The results from the project are promising, but a longer test period is required to rule out e.g. seasonal variations. Approach and implementation: The two case studies are a heat exchanger on SSAB and a profiled header on Svenska Fönster AB. In addition, a laboratory environment at the University of Gävle has been built since it has not been possible to try different methods to detect (and perhaps even initiate) errors in industrial plants that were in operation. Measured data from the plants have been acquired from existing control systems and with a complementary measurement system. The modeling work has been done offline and analysis of the results has been done jointly by model developers and staff with process knowledge.
Publications (1 of 1) Show all publications
Mattsson, P., Zachariah, D. & Björsell, N. (2019). Flexible Models for Smart Maintenance. In: Proceedings 2019 IEEE International Conference on Industrial Technology (ICIT): . Paper presented at 20th IEEE International Conference on Industrial Technology (ICIT), 13-15 February 2019, Melbourne, Australia (pp. 1772-1777). IEEE
Open this publication in new window or tab >>Flexible Models for Smart Maintenance
2019 (English)In: Proceedings 2019 IEEE International Conference on Industrial Technology (ICIT), IEEE, 2019, p. 1772-1777Conference paper, Published paper (Refereed)
Abstract [en]

Smart maintenance strategies are becoming increasingly important in the industry, and can contribute to environmentally and economically sustainable production. In this paper a recently developed latent variable framework for nonlinear-system identification is considered for use in smart maintenance. A model is first identified using data from a system operating under normal conditions. Then the identified model is used to detect when the system begins to deviate from normal behavior. Furthermore, for systems that operate on separate batches (units), we develop a new method that identifies individual models for each batch. This can be used both to detect anomalous batches and changes in the system behavior. Finally, the two methods are evaluated on two different industrial case studies. In the first, the purpose is to detect fouling in a heat exchanger. In the second, the goal is to detect when the tool in a wood moulder machine should be changed.

Place, publisher, year, edition, pages
IEEE, 2019
Keywords
Applications, Modeling, Nonlinear systems
National Category
Robotics
Identifiers
urn:nbn:se:hig:diva-30445 (URN)10.1109/ICIT.2019.8754932 (DOI)000490548300287 ()2-s2.0-85069039435 (Scopus ID)
Conference
20th IEEE International Conference on Industrial Technology (ICIT), 13-15 February 2019, Melbourne, Australia
Funder
Vinnova, 2017-04807Swedish Energy AgencySwedish Research Council FormasEuropean Regional Development Fund (ERDF)Swedish Agency for Economic and Regional Growth
Available from: 2019-07-22 Created: 2019-07-22 Last updated: 2019-11-27Bibliographically approved
Principal InvestigatorBjörsell, Niclas
Co-InvestigatorKolluri, Sowjanya
Co-InvestigatorMattsson, Per
Coordinating organisation
University of Gävle
Funder
Period
2017-11-10 - 2018-06-30
National Category
Robotics
Identifiers
DiVA, id: project:295Project, id: 2017-04807_Vinnova

Search in DiVA

Robotics

Search outside of DiVA

GoogleGoogle Scholar