[Back]

F. Tahmasebi, M. Schuss, A. Mahdavi:
"Monitoring-based dynamically updated occupancy models for predictive building systems control";
Talk: NSB2014 - 10th Nordic Symposium on Building Physics, Lund, Schweden; 2014-06-15 - 2014-06-19; in: "NSB2014 - 10th Nordic Symposium on Building Physics", J. Arfvidsson, L.E. Harderup, A. Kumlin, B. Rosencrantz (ed.); (2014), ISBN: 978-91-88722-51-5; 550 - 557.

Knowledge of occupants´ presence and behavior in buildings is of central importance to the
implementation efforts concerning predictive building systems control. Specifically, prediction of
occupants´ presence in office buildings represents a necessary condition for predicting their
interactions with building systems. Implementation of occupancy prediction models in existing
buildings can benefit from available occupancy monitoring data. In the present contribution, we
focused on the evaluation of monitoring-based probabilistic occupancy models of a single-occupancy
office, which are intended to be used for predictive room control. Thereby, we examined various
scenarios to process the monitoring occupancy data obtained from this office toward developing
stochastic occupancy models. These scenarios differ in terms of the duration and horizon (moving
versus static) of the training intervals, as well as the grouping of the week days. To evaluate the
stochastic occupancy models, we performed a Monte-Carlo study. Toward this end, a number of
evaluation statistics were defined and the monitored and predicted daily profiles of occupancy were
compared for individual days as units of observation to obtain the statistics. The results facilitate a
discussion of the potential and limitations of occupancy models intended for incorporation in the
control logic of buildings.

none - see english version

Probabilistic occupancy models, predictive building systems control, evaluation statistics