Date of Award
Master of Science in Engineering Management
Department of Systems Engineering and Management
Andrew j. Hoisington, PhD
Occupants in the built environment impact facility energy consumption and indoor air quality. Predicting the presence of occupants within the built environment can therefore be used to manage these factors while providing additional benefits in terms of emergency management and future space utilization. Detecting occupancy requires a combination of sensors and models to accurate assess data collected within facilities to predict occupancy. This thesis investigated occupancy detection through a non-invasive data collection sensors and model. Specifically, this thesis sought to answer two research questions examining the ability of a radial basis function to accurately predict occupancy when generated from data collected from two facilities. Generated models were evaluated on the data from which they were derived, self-estimation, as well as applied to other areas within the same facility, cross-estimation. The motivation, sensors and models, were discussed to establish a framework. The principle implications of this research is to reduce energy consumption by knowing when the built environment is occupied through the use of non-invasive data collection sensors supplying inputs into a model. The resulting accuracy rates of the derived models ranged from 48% - 68% when using three collected parameters: temperature, relative humidity and carbon dioxide.
DTIC Accession Number
Tyhurst, James C., "Non-Intrusive Occupancy Detection Methods and Models" (2019). Theses and Dissertations. 2355.