Show simple item record

dc.contributor.advisorO'Donnell, Garreten
dc.contributor.authorSEILER, DOMINIKen
dc.date.accessioned2020-09-17T09:17:23Z
dc.date.available2020-09-17T09:17:23Z
dc.date.issued2020en
dc.date.submitted2020en
dc.identifier.citationSEILER, DOMINIK, Proxy measurement and strategic metering methodologies for resource transparency of industrial systems, Trinity College Dublin.School of Engineering, 2020en
dc.identifier.otherYen
dc.identifier.urihttp://hdl.handle.net/2262/93476
dc.descriptionAPPROVEDen
dc.description.abstractFor the latest advancement of manufacturing systems, namely Industry 4.0 or Smart Manufacturing, large amounts of precise process data are required. However, the acquisition of sufficient data represents a challenge for many manufacturing plants. Developed metering frameworks from academia are often not applicable to real-world manufacturing processes due to monitoring gaps. These gaps result from missing and broken measuring devices of industrial systems and lead to an incomplete monitoring outcome that can prevent system optimisations. Therefore, this research focuses on the development of effective data gathering tools for complex industrial systems in order to provide a comprehensive process understanding. By considering industrial needs, the design and development of functional resource measurement methodologies and the study of predictive models with a reduced-order characteristic have been addressed to accelerate holistic data acquisition in industrial environments. Following from the investigations of two purified water and one steam system in an industrial case facility, a comprehensive metering strategy for detailed information gathering of technical building services has been developed. Although these systems require large shares of resources, holistic analysis methods are not comprehensively addressed in the literature. The novel four phase metering methodology in this research fills this gap by identifying available data sources, abstracting the central process steps, and mapping the resource flows within technical building services. In the last phase, the absence of functional online meters is surrogated by a basic proxy metering strategy that enables the approximation of missing parameters by combining related data sources in a regression model. This holistic strategy represents an effective data acquisition approach with high adaptability to existing conditions and constraints that has not been reported in the literature to date. The gathered process information from the holistic meter approach allowed the development of an automated calculation method for the total cost of technical building services. By focusing on the interactions of the identified resources and further added values, this cost index tracks the real value that is embedded in the analysed system. Compared to similar approaches in the literature, the cost calculation method in this research can be automated which enables the impact of system changes to be studied. Based on the potential of basic proxy metering devices (PMDs) for estimating missing process data and the industrial need for simpler implementation strategies, the development of comprehensive PMD modelling methodologies has been successfully addressed in the second part of this research. For the common case of small datasets, the regression performance of PMD algorithms for diverse dataset regression complexities has been analysed. As well as the complexity characterisation of five small datasets from an experimental rig, the estimation accuracies of the most common linear and non-linear PMD algorithms have been studied, including the impact of training sample manipulations with bootstrap and artificial noise injection. The results of this comparison highlight which PMD algorithm and training sample manipulation technique is appropriate, depending on the regression complexity of the dataset and the number of training samples. With this novel methodology, industrial users can select a targeted and straightforward PMD development approach based on their specific regression complexity and use case.en
dc.publisherTrinity College Dublin. School of Engineering. Discipline of Mechanical & Manuf. Engen
dc.rightsYen
dc.subjectMetering auditen
dc.subjectSoft sensoren
dc.subjectAdvanced process monitoringen
dc.subjectProxy meteringen
dc.titleProxy measurement and strategic metering methodologies for resource transparency of industrial systemsen
dc.typeThesisen
dc.contributor.sponsorScience Foundation Ireland (SFI)en
dc.type.supercollectionthesis_dissertationsen
dc.type.supercollectionrefereed_publicationsen
dc.type.qualificationlevelDoctoralen
dc.identifier.peoplefinderurlhttps://tcdlocalportal.tcd.ie/pls/EnterApex/f?p=800:71:0::::P71_USERNAME:SEILERDen
dc.identifier.rssinternalid220243en
dc.rights.ecaccessrightsopenAccess


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record