Power disaggregation in commercial buildings considering unmonitored facilities and multiple routines

Abstract

Estimating energy conservation for commercial buildings is challenging due to unreliable power predictions. Previously proposed disaggregation technologies require site surveys to decompose a single meter into multiple elements and address commercial buildings’ unknown power consumption/equipment. We propose a novel energy disaggregation method for the energy consumption of commercial buildings with limited facility status monitoring points. The method accurately estimates and identifies the energy consumption of each monitored/unmonitored facility using the statuses of the monitored facilities. The unmonitored facility consumption is evaluated using the linear regression residual of the monitored facilities and is clustered by daily routine. Daily variations in the magnitude of the power consumption of unmonitored facilities are estimated by assigning a linear basis function model with different diurnal periodicities to each cluster. The monitored facility regression parameters and basis functions are determined according to the total energy consumption of the building and the operational statuses of the monitored facilities. When applied to a commercial building, our method estimated the power consumption of each facility more accurately than the conventional method, without accessing the building’s operational calendar or the equipment power rating. Thus, the proposed method can be used to help energy managers make informed decisions regarding commercial buildings.