Full metadata record
DC FieldValueLanguage
dc.contributorDepartment of Building Services Engineeringen_US
dc.contributor.advisorXiao, Fu (BSE)en_US
dc.creatorWang, Na-
dc.identifier.urihttps://theses.lib.polyu.edu.hk/handle/200/11256-
dc.languageEnglishen_US
dc.publisherHong Kong Polytechnic Universityen_US
dc.rightsAll rights reserveden_US
dc.titleInvestigation of feature selection schemes for building energy modelingen_US
dcterms.abstractThe increase in energy consumption, especially the burning of fossil fuels, has caused severe environmental pollution, increased carbon emissions, and led to the greenhouse effect. Building energy consumption accounts for 40% of total global energy consumption, and more than 30% of carbon emissions come from buildings. Much research related to building energy consumption has been carried out to reduce building energy consumption. Building energy consumption prediction plays a vital role in energy planning, energy conservation, and management because it can help evaluate the performance of buildings and optimize structural operation procedures. The traditional method of energy consumption prediction is to use physical models, which requires researchers to have relevant domain knowledge. Also, the physical model requires an exemplary description of the research object and requires a large number of complex formulas to calculate. The data-driven model provides a more practical way to predict energy consumption. However, in building energy consumption forecasting, most researchers pay more attention to the optimization of data-driven model algorithms and the improvement of model performance but seldom pay attention to the impact of input features on model accuracy. This study uses Python programming to analyze and compare eight feature selection methods. This study uses data without feature selection as a control group to explore the feasibility of feature selection methods in practice. This research preprocesses the data from the building automatic control system of Hong Kong ICC. Autocorrelation Function (ACF) and Partial Autocorrelation Function (PACF) are used in this study to explore the impact of historical time cooling load data on the current time building energy consumption forecast. The historical moment cooling load data that impacts the current energy consumption prediction is added as new features to the candidate feature set. The data in the candidate feature set is input into eight feature selection methods for calculation. The calculation results are sorted according to the degree of influence on the target feature(total cooling load). Finally, these features are added to the ANN model in order of importance for prediction. The loss function evaluation index is used to evaluate the results of the model prediction.en_US
dcterms.extentv, 103 pages : color illustrationsen_US
dcterms.isPartOfPolyU Electronic Thesesen_US
dcterms.issued2021en_US
dcterms.educationalLevelM.Eng.en_US
dcterms.educationalLevelAll Masteren_US
dcterms.LCSHBuildings -- Energy consumptionen_US
dcterms.LCSHBuildings -- Energy conservationen_US
dcterms.LCSHHong Kong Polytechnic University -- Dissertationsen_US
dcterms.accessRightsrestricted accessen_US

Files in This Item:
File Description SizeFormat 
5739.pdfFor All Users (off-campus access for PolyU Staff & Students only)13.54 MBAdobe PDFView/Open


Copyright Undertaking

As a bona fide Library user, I declare that:

  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.

By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

Show simple item record

Please use this identifier to cite or link to this item: https://theses.lib.polyu.edu.hk/handle/200/11256