Show simple item record

dc.contributor.advisorCook, Diane J.
dc.creatorChen, Chao
dc.date.accessioned2014-03-07T21:54:13Z
dc.date.available2014-03-07T21:54:13Z
dc.date.issued2013
dc.identifier.urihttp://hdl.handle.net/2376/4924
dc.descriptionThesis (Ph.D.), Department of Electrical Engineering and Computer Science, Washington State Universityen_US
dc.description.abstractSociety is becoming increasingly aware of the impact that our lifestyle preferences have on energy usage and the environment. In this dissertation, we look more closely at the impact that human behavior has on energy consumption. In particular, we design and evaluate smart home and machine learning techniques to examine the relationship between behavioral patterns and resource consumption. The contribution of this research has two components. In the first part, we use smart home technologies to examine the relationship between behavior patterns and energy usage at the scale of individual homes. In particular, machine learning techniques are used to predict energy usage based on residents' activities. Data mining techniques are then introduced to identify anomalies and abnormal patterns in home power data. Lastly, a web-based tool for visualizing smart home activities and power consumption is designed. This tool is used to present the results of the previous techniques to inform users about their personal energy usage and to encourage more energy-efficient behaviors. In the second part, we focus on the creation of data mining algorithms to analyze per-home energy use at the scale of an entire community. Specifically, we analyze data collected from thousands of smart meters. Our contributions to such large-scale analysis include the design of an automated outlier detection tool for noise reduction. In addition, a web-based visualization system was developed to depict an energy usage heat map and compare residential electricity usage between neighboring homes. In order to build a more complete model of electricity usage, we designed a learning algorithm that takes into account the function of various building characteristics. Using this model, a web-based user interface was designed to estimate building energy usage. Finally, an unsupervised algorithm was designed to cluster large-scale time-series datasets in an efficient manner.We describe and evaluate each of these contributions using electricity consumption data from actual smart homes as part of the CASAS smart home project. In each case we illustrate the efficacy of these algorithms to gaining insights on human behavior and its impact on energy consumption, and offer ideas for using these insights to promote sustainable behaviors.en_US
dc.description.sponsorshipDepartment of Electrical Engineering and Computer Science, Washington State Universityen_US
dc.language.isoEnglish
dc.rightsIn copyright
dc.rightsPublicly accessible
dc.rightsopenAccess
dc.rights.urihttp://rightsstatements.org/vocab/InC/1.0/
dc.rights.urihttp://www.ndltd.org/standards/metadata
dc.rights.urihttp://purl.org/eprint/accessRights/OpenAccess
dc.subjectComputer scienceen_US
dc.subjectData Miningen_US
dc.subjectEnergy Efficiencyen_US
dc.subjectEnergy Modelingen_US
dc.subjectMachine Learningen_US
dc.titleINVESTIGATING THE HUMAN BEHAVIOR SIDE OF BUILDING ENERGY EFFICIENCY
dc.typeElectronic Thesis or Dissertation


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record