Classification of natural circulation two-phase flow image patterns based on self-organizing maps of full frame DCT coefficients

Página do item simplificado
dc.contributor.authorMESQUITA, ROBERTO N. de
dc.contributor.authorCASTRO, LEONARDO F.
dc.contributor.authorTORRES, WALMIR M.
dc.contributor.authorROCHA, MARCELO da S.
dc.contributor.authorUMBEHAUN, PEDRO E.
dc.contributor.authorANDRADE, DELVONEI A.
dc.contributor.authorSABUNDJIAN, GAIANE
dc.contributor.authorMASOTTI, PAULO H.F.
dc.coverageInternacionalpt_BR
dc.date.accessioned2018-07-18T12:50:11Z
dc.date.available2018-07-18T12:50:11Z
dc.date.issued2018pt_BR
dc.description.abstractMany of the recent nuclear power plant projects use natural circulation as heat removal mechanism. The accuracy of heat transfer parameters estimation has been improved through models that require precise prediction of two-phase flow pattern transitions. Image patterns of natural circulation instabilities were used to construct an automated classification system based on Self-Organizing Maps (SOMs). The system is used to investigate the more appropriate image features to obtain classification success. An efficient automated classification system based on image features can enable better and faster experimental procedures on two-phase flow phenomena studies. A comparison with a previous fuzzy inference study was foreseen to obtain classification power improvements. In the present work, frequency domain image features were used to characterize three different natural circulation two-phase flow instability stages to serve as input to a SOM clustering algorithm. Full-Frame Discrete Cosine Transform (FFDCT) coefficients were obtained for 32 image samples for each instability stage and were organized as input database for SOM training. A systematic training/test methodology was used to verify the classification method. Image database was obtained from two-phase flow experiments performed on the Natural Circulation Facility (NCF) at Instituto de Pesquisas Energéticas e Nucleares (IPEN/CNEN), Brazil. A mean right classification rate of 88.75% was obtained for SOMs trained with 50% of database. A mean right classificationrate of 93.98% was obtained for SOMs trained with 75% of data. These mean rates were obtained through 1000 different randomly sampled training data. FFDCT proved to be a very efficient and compact image feature to improve image-based classification systems. Fuzzy inference showed to be more flexible and able to adapt to simpler statistical features from only one image profile. FFDCT features resulted in more precise results when applied to a SOM neural network, though had to be applied to the full original grayscale matrix for all flow images to be classified.pt_BR
dc.format.extent161-171pt_BR
dc.identifier.citationMESQUITA, ROBERTO N. de; CASTRO, LEONARDO F.; TORRES, WALMIR M.; ROCHA, MARCELO da S.; UMBEHAUN, PEDRO E.; ANDRADE, DELVONEI A.; SABUNDJIAN, GAIANE; MASOTTI, PAULO H.F. Classification of natural circulation two-phase flow image patterns based on self-organizing maps of full frame DCT coefficients. <b>Nuclear Engineering and Design</b>, v. 335, p. 161-171, 2018. DOI: <a href="https://dx.doi.org/10.1016/j.nucengdes.2018.05.019">10.1016/j.nucengdes.2018.05.019</a>. Disponível em: http://repositorio.ipen.br/handle/123456789/28971.
dc.identifier.doi10.1016/j.nucengdes.2018.05.019pt_BR
dc.identifier.issn0029-5493pt_BR
dc.identifier.orcidhttps://orcid.org/0000-0003-2445-1298
dc.identifier.orcidhttps://orcid.org/0000-0001-9544-4509
dc.identifier.orcidhttps://orcid.org/0000-0002-6689-3011
dc.identifier.orcidhttps://orcid.org/0000-0002-2887-0759
dc.identifier.orcidhttps://orcid.org/0000-0002-5355-0925
dc.identifier.percentilfi77.94en
dc.identifier.percentilfiCiteScore67.67
dc.identifier.urihttp://repositorio.ipen.br/handle/123456789/28971
dc.identifier.vol335pt_BR
dc.relation.ispartofNuclear Engineering and Designpt_BR
dc.rightsopenAccesspt_BR
dc.subjectnatural convection
dc.subjecttwo-phase flow
dc.subjectcooling systems
dc.subjectclassification
dc.subjectbrazilian cnen
dc.subjectdiagrams
dc.subjectmaps
dc.subjectcoolant loops
dc.titleClassification of natural circulation two-phase flow image patterns based on self-organizing maps of full frame DCT coefficientspt_BR
dc.typeArtigo de periódicopt_BR
dspace.entity.typePublication
ipen.autorLEONARDO FERREIRA CASTRO
ipen.autorMARCELO DA SILVA ROCHA
ipen.autorPAULO HENRIQUE FERRAZ MASOTTI
ipen.autorGAIANE SABUNDJIAN
ipen.autorDELVONEI ALVES DE ANDRADE
ipen.autorPEDRO ERNESTO UMBEHAUN
ipen.autorWALMIR MAXIMO TORRES
ipen.autorROBERTO NAVARRO DE MESQUITA
ipen.codigoautor11591
ipen.codigoautor7992
ipen.codigoautor219
ipen.codigoautor202
ipen.codigoautor1258
ipen.codigoautor754
ipen.codigoautor188
ipen.codigoautor1375
ipen.contributor.ipenauthorLEONARDO FERREIRA CASTRO
ipen.contributor.ipenauthorMARCELO DA SILVA ROCHA
ipen.contributor.ipenauthorPAULO HENRIQUE FERRAZ MASOTTI
ipen.contributor.ipenauthorGAIANE SABUNDJIAN
ipen.contributor.ipenauthorDELVONEI ALVES DE ANDRADE
ipen.contributor.ipenauthorPEDRO ERNESTO UMBEHAUN
ipen.contributor.ipenauthorWALMIR MAXIMO TORRES
ipen.contributor.ipenauthorROBERTO NAVARRO DE MESQUITA
ipen.date.recebimento18-07pt_BR
ipen.identifier.fi1.541pt_BR
ipen.identifier.fiCiteScore3.0
ipen.identifier.ipendoc24758pt_BR
ipen.identifier.iwosWoSpt_BR
ipen.range.fi1.500 - 2.999
ipen.range.percentilfi75.00 - 100.00
ipen.type.genreArtigo
relation.isAuthorOfPublication79a7ba12-7ac9-4bf6-8547-2a3907caab47
relation.isAuthorOfPublication8f88995a-927a-4491-8ced-f5be1360d3ec
relation.isAuthorOfPublication468cfa6e-810e-4cf4-9fc3-e524265e60fc
relation.isAuthorOfPublication915b9343-9eaf-4b03-b709-85699f77b71c
relation.isAuthorOfPublication0eeb4436-68e5-4573-a603-35f5fc912178
relation.isAuthorOfPublicationc1502b5a-764f-4f7b-bae0-b04b2a264aae
relation.isAuthorOfPublication5e952755-d8a5-427d-9298-6d6b76897aac
relation.isAuthorOfPublication1975aa9b-be26-4f48-8196-96eeb4c2c0c3
relation.isAuthorOfPublication.latestForDiscovery1975aa9b-be26-4f48-8196-96eeb4c2c0c3
sigepi.autor.atividadeMESQUITA, ROBERTO N. DE:1375:420:Spt_BR
sigepi.autor.atividadeCASTRO, LEONARDO F.:11591:-1:Npt_BR
sigepi.autor.atividadeTORRES, WALMIR M.:188:450:Npt_BR
sigepi.autor.atividadeROCHA, MARCELO DA S.:7992:420:Npt_BR
sigepi.autor.atividadeUMBEHAUN, PEDRO E.:754:420:Npt_BR
sigepi.autor.atividadeANDRADE, DELVONEI A.:1258:420:Npt_BR
sigepi.autor.atividadeSABUNDJIAN, GAIANE:202:420:Npt_BR
sigepi.autor.atividadeMASOTTI, PAULO H.F.:219:420:Npt_BR
Pacote Original
Agora exibindo 1 - 1 de 1
Imagem de Miniatura
Nome:
24758.pdf
Tamanho:
2.8 MB
Formato:
Adobe Portable Document Format
Descrição:
Baixar
Licença do Pacote
Agora exibindo 1 - 1 de 1
Nenhuma Miniatura disponível
Nome:
license.txt
Tamanho:
1.71 KB
Formato:
Item-specific license agreed upon to submission
Descrição:
Baixar
Coleções
Periódicos