Show simple item record

dc.contributor.advisorAldrich, C.
dc.contributor.authorBarkhuizen, Marlize
dc.contributor.otherUniversity of Stellenbosch. Faculty of Engineering. Dept. of Process Engineering.
dc.date.accessioned2011-08-22T12:42:38Z
dc.date.available2011-08-22T12:42:38Z
dc.date.issued2003-12
dc.identifier.urihttp://hdl.handle.net/10019.1/16252
dc.descriptionThesis (MScIng)--University of Stellenbosch, 2003.en_ZA
dc.description.abstractENGLISH ABSTRACT: The analysis of process data obtained from chemical and metallurgical engineering systems is a crucial aspect of the operating of any process, as information extracted from the data is used for control purposes, decision making and forecasting. Singular spectrum analysis (SSA) is a relatively new technique that can be used to decompose time series into their constituent components, after which a variety of further analyses can be applied to the data. The objectives of this study were to investigate the abilities of SSA regarding the filtering of data and the subsequent modelling of the filtered data, to explore the methods available to perform nonlinear SSA and finally to explore the possibilities of Monte Carlo SSA to characterize and identify process systems from observed time series data. Although the literature indicated the widespread application of SSA in other research fields, no previous application of singular spectrum analysis to time series obtained from chemical engineering processes could be found. SSA appeared to have a multitude of applications that could be of great benefit in the analysis of data from process systems. The first indication of this was in the filtering or noise-removal abilities of SSA. A number of case studies were filtered by various techniques related to SSA, after which a number of neural network modelling strategies were applied to the data. It was consistently found that the models built on data that have been prefiltered with SSA outperformed the other models. The effectiveness of localized SSA and auto-associative neural networks in performing nonlinear SSA were compared. Both techniques succeeded in extracting a number of nonlinear components from the data that could not be identified from linear SSA. However, it was found that localized SSA was a more reliable approach, as the auto-associative neural networks would not train for some of the data or extracted nonsensical components for other series. Lastly a number of time series were analysed using Monte Carlo SSA. It was found that, as is the case with all other characterization techniques, Monte Carlo SSA could not succeed in correctly classifying all the series investigated. For this reason several tests were used for the classification of the real process data. In the light of these findings, it was concluded that singular spectrum analysis could be a valuable tool in the analysis of chemical and metallurgical process data.en_ZA
dc.description.abstractAFRIKAANSE OPSOMMING: Die analise van chemise en metallurgiese prosesdata wat verkry is vanaf chemiese of metallurgiese ingenieursstelsels is ‘n baie belangrike aspek in die bedryf van enige proses, aangesien die inligting wat van die data onttrek word vir prosesbeheer, besluitneming of die bou van prosesmodelle gebruik kan word. Singuliere spektrale analise is ‘n relatief nuwe tegniek wat gebruik kan word om tydreekse in hul onderliggende komponente te ontbind. Die doelwitte van hierdie studie was om ‘n omvattende literatuuroorsig oor die ontwikkeling van die tegniek en die toepassing daarvan te doen, beide in die ingenieursindustrie en in ander navorsingsvelde, die navors van die moontlikhede van SSA aangaande die verwydering van geraas uit die data en die gevolglike modellering van die skoon data te ondersoek, ‘n ondersoek te doen na sommige van die beskikbare tegnieke vir nie-lineêre SSA en laastens ‘n studie te maak van die potensiaal van Monte Carlo SSA vir die karakterisering en identifikasie van data verkry vanaf prosesstelsels. Ten spyte van aanduidings in die literatuur dat SSA wydverspreid toegepas word in ander navorsingsvelde, kon geen vorige toepassings gevind word van SSA op chemiese prosesse nie. Dit wil voorkom asof die chemiese nywerhede groot baat kan vind by SSA van prosesdata. Die eerste aanduiding van hierdie voordele was in die vermoë van SSA om geraas te verwyder uit tydreekse. ‘n Aantal tipiese gevalle is ondersoek deur van verskeie benaderings tot SSA gebruik te maak. Nadat die geraas uit die tydreekse van die toetsgevalle verwyder is, is neurale netwerke gebruik om die prosesse te modelleer. Daar is herhaaldelik gevind dat die modelle wat gebou is op data wat eers deur SSA skoongemaak is, beter presteer as die wat slegs op die onverwerkte data gepas is. Die effektiwiteit van lokale SSA en auto-assosiatiewe neurale netwerke om nie- lineêre SSA toe te pas is ook vergelyk. Albei tegnieke het daarin geslaag om nie- lineêre hoofkomponente van die data te onttrek wat nie geïdentifiseer kon word deur die lineêre benadering nie. Daar is egter gevind dat lokale SSA ‘n meer betroubare tegniek is, aangesien die autoassosiatiewe neurale netwerke nie op sommige van die datastelle wou leer nie en vir ander tydreekse sinnelose hoofkomponente onttrek het. Laastens is ‘n aantal tydreekse geanaliseer met behulp van Monte Carlo SSA. Soos met alle ander karakteriseringstegnieke, kon Monte Carlo SSA nie daarin slaag om al die tydreekse wat ondersoek is korrek te identifiseer nie. Om hierdie rede is ‘n kombinasie van toetse gebruik om die onbekende tydreekse te klassifiseer. In die lig van al hierdie bevindinge, is die gevolgtrekking gemaak dat singuliere spektrale analise ‘n waardevolle hulpmiddel kan wees in die analise van chemiese en metallurgiese prosesdata.af
dc.format.extentxv, 232 leaves : ill.
dc.language.isoen_ZAen_ZA
dc.publisherStellenbosch : University of Stellenbosch
dc.subjectSpectrum analysisen_ZA
dc.subjectChemical process controlen_ZA
dc.subjectTheses -- Process engineeringen_ZA
dc.subjectDissertations -- Process engineeringen_ZA
dc.titleAnalysis of process data with singular spectrum methodsen_ZA
dc.typeThesisen_ZA
dc.rights.holderUniversity of Stellenbosch


Files in this item

Thumbnail
Thumbnail

This item appears in the following Collection(s)

Show simple item record