Please use this identifier to cite or link to this item: https://scholarbank.nus.edu.sg/handle/10635/221483
Title: THE PLAUSIBILITY OF A DATA-DRIVEN THERMAL COMFORT MODEL WITH A BAYESIAN NETWORK APPROACH WITHIN A NET-ZERO ENERGY BUILDING
Authors: LAI WEN XUAN
Keywords: Degree of B.Sc. (Project and Facilities Management)
Project and Facilities Management
2020/2021 PFM
PFM
Building
Clayton Miller
Issue Date: 5-Jan-2021
Citation: LAI WEN XUAN (2021-01-05). THE PLAUSIBILITY OF A DATA-DRIVEN THERMAL COMFORT MODEL WITH A BAYESIAN NETWORK APPROACH WITHIN A NET-ZERO ENERGY BUILDING. ScholarBank@NUS Repository.
Abstract: The implementation of Green Building Strategies by the Building and Construction Authority has transformed the way how thermal comfort is achieved in Singapore. With the introduction of the concept of Net-Zero Energy Building (NZEB) in Singapore, questions were raised pertaining to the validity of the static thermal comfort models used to determine the thermal comfort setpoints for Singapore’s thermal comfort standard (SS 553:2016). This is because the thermal conditions within a NZEB is dynamic and unpredictable, which is significantly different from the nature of the industry thermal comfort models. This highlights the need for a potential improvement on occupant thermal comfort in Singapore. Therefore, this study shall develop a data-driven thermal comfort model with a Bayesian Network approach to account for its dynamic conditions, by supporting continuous data learning for model updates based on changes recorded from the thermal environment. The accuracy of the model will be compared against standard metrics and its supplement model, and subsequently evaluated for its plausibility of practical application and prediction accuracy. The data-driven model will be illustrated and inferred using the pgmpy Python library using the dataset from the research done by Jayathissa et al. (2020). The output of this paper is the average micro-F1 accuracy score derived from the confusion matrix that represents the correlation of actual and predicted thermal sensation values. In comparison to industry standard model, the data-driven model curated by this study is a better representative for the occupants’ thermal acceptance setpoint in Singapore’s NZEB environment. Despite that, the model is unable to forecast discomfort of occupants due to the insufficient size of the dataset. However, the model allows the accumulation of operational data into its database for data learning and subsequently for inference, creating a cycle for its continuous accuracy improvement.
URI: https://scholarbank.nus.edu.sg/handle/10635/221483
Appears in Collections:Bachelor's Theses

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
Lai Wen Xuan 2020-2021.pdf2.16 MBAdobe PDF

RESTRICTED

NoneLog In

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.