Predicting Occupant Locations Using Association Rule Mining
Heating, ventilation, air conditioning (HVAC) systems are significant consumers of energy, however building management systems do not typically operate them in accordance with occupant movements. Due to the delayed response of HVAC systems, prediction of
- PDF / 304,043 Bytes
- 15 Pages / 439.37 x 666.142 pts Page_size
- 89 Downloads / 224 Views
Abstract Heating, ventilation, air conditioning (HVAC) systems are significant consumers of energy, however building management systems do not typically operate them in accordance with occupant movements. Due to the delayed response of HVAC systems, prediction of occupant locations is necessary to maximize energy efficiency. In this paper we present two approaches to occupant location prediction based on association rule mining which allow prediction based on historical occupant movements and any available real time information, or based on recent occupant movements. We show how association rule mining can be adapted for occupant prediction and evaluate both approaches against existing approaches on two sets of real occupants.
1 Introduction Office buildings are significant consumers of energy: buildings typically account for up to 40 % of the energy use in industrialised countries [1], and of that, over 70 % is consumed in the operation of the building through HVAC and lighting. A large portion of this is consumed under static control regimes, in which heating, cooling and lighting are applied according to fixed schedules, specified when the buildings were designed, regardless of how the buildings are actually used. To improve energy efficiency, the building management system should operate the HVAC systems in response to the actual behaviour patterns of the occupants. However, heating and cooling systems have a delayed response, so to satisfy the needs of the occupants, C. Ryan (B) · K. N. Brown Cork Constraint Computation Centre, Department of Computer Science, University College Cork, Cork, Ireland e-mail: [email protected] K. N. Brown e-mail: [email protected] M. Bramer and M. Petridis (eds.), Research and Development in Intelligent Systems XXX, DOI: 10.1007/978-3-319-02621-3_5, © Springer International Publishing Switzerland 2013
63
64
C. Ryan and K. N. Brown
the management system must predict the occupant behaviour. The prediction system should be accurate at both bulk and individual levels: the total number of occupants of a building or a zone determine the total load on the HVAC system, while knowing the presence and identity of an occupant of an individual office allows us to avoid waste through unnecessary heating or cooling without discomforting the individual. We believe that in most office buildings, the behaviour of occupants tends to be regular. An occupant’s behaviour may relate to the time of day, the day of the week or the time of year. Their behaviour on a given day may also depend on their location earlier on that day or on their most recent sequence of movements. We require a system which is able to recognize these time and feature based patterns across different levels of granularity from observed data. Further, many office users now use electronic calendars to manage their schedules, and information in these calendars may support or override the regular behaviour. The reliability of the calendar data will depend on the individual maintaining it, so the prediction system needs to be able
Data Loading...
