Occupancy schedules learning process through a data mining framework

Abstract

Building occupancy is a paramount factor in building energy simulations. Specifically, lighting, plug loads, HVAC equipment utilization, fresh air requirements and internal heat gain or loss greatly depends on the level of occupancy within a building. Developing the appropriate methodologies to describe and reproduce the intricate network responsible for human-building interactions are needed. Extrapolation of patterns from big data streams is a powerful analysis technique which will allow for a better understanding of energy usage in buildings. A three-step data mining framework is applied to discover occupancy patterns in office spaces. First, a data set of 16 offices with 10 minute interval occupancy data, over a two year period is mined through a decision tree model which predicts the occupancy presence. Then a rule induction algorithm is used to learn a pruned set of rules on the results from the decision tree model. Finally, a cluster analysis is employed in order to obtain consistent patterns of occupancy schedules. The identified occupancy rules and schedules are representative as four archetypal working profiles that can be used as input to current building energy modeling programs, such as EnergyPlus or IDA-ICE, to investigate impact of occupant presence on design, operation and energy use in office buildings.

Effects of Building Occupancy on Indicators of Energy Efficiency

“…Operational energy consumption is affected by lighting, plug loads, heating, ventilation and air conditioning equipment utilization, fresh air requirements, and internal heat gain/loss, which depend on the number of occupants and their behavior. However, the latter are not well known in advance and are difficult to capture during the operation [21]. It is therefore not surprising that a significant discrepancy between the predicted and actual energy consumption is often observed (on average, a 34% increase in a study [23] consisting of 62 case study buildings, with the dominant root causes for the performance gap being specification uncertainty in modeling, occupant behavior, and poor operational practices).…”

“…Therefore, there is no linear correlation between the technical calculated energy efficiency in kWh/m 2 and the actual measured energy efficiency encompassing the effects of occupants in the building. In the design phase, the operational energy consumption is typically simulated by using standard occupancy schedules [21]. However, those only provide a poor estimate of the real occupancy measured in the operation (there is a 46% difference between the American Society of Heating, Refrigerating and Air-Conditioning Engineers ASHRAE standardized occupancy used in the simulation and the real occupancy according to Duarte et al [22]).…”

The human dimensions of energy use in buildings: A review

“…In parallel, the behavior research community has developed and critically evaluated several behavior modeling approaches that would be useful to building designers [10,29,[101][102][103][104][105][106]. Such reviews have focused on methods to assess the robustness and accuracy of proposed models, to establish the scope of their effective application.…”

Data mining of space heating system performance in affordable housing

“…Motta-Cabrera and Zareipour [28] used data association mining to identify the lighting energy waste patterns in educational institutes. D'Oca and Hong [29] used data mining approaches to develop individualized occupancy schedules and to discover patterns of window opening and closing behavior in office buildings [30]. Zhao et al [31] used office appliance power consumption to investigate different occupant behavior patterns and schedules for use in the modeling of building energy simulation.…”