A load-disaggregation framework to sense personalized energy-use information in commercial buildings

Abstract

Studies indicate that providing building occupants with personalized energy-use feedback effectively triggers energy-use behavior modifications. However, gathering such personalized information in a commercial building using conventional techniques is currently extremely expensive. Accordingly, this study proposes a novel framework that disaggregates building-wide energy data down to the level of individual occupants by harnessing recurring patterns in occupants’ energy-use behaviors. To achieve such disaggregation, the framework utilizes a density-based clustering algorithm that deciphers patterns amidst occupants’ sensed entry/exit events and the building's corresponding changes in energy-load magnitudes, load-change timings, and energy-use locations. Experimental results of two commercial buildings with an average F-measure of 0.807 and Accuracy of 0.958 demonstrate the feasibility and accuracy of the framework in generating personalized information. By gathering such data in an economically feasible manner, the framework can provide a cost-effective means for individualizing feedback, which has been shown to yield long-term decreases in commercial buildings’ energy consumption.