Predicting Room-Level Occupancy Using Smart-Meter Data

Occupancy information in buildings is a crucial information to enable automated load controlling resulting in significant energy savings. Unfortunately, current methods obtain occupancy data by using additional infrastructure, which can be expensive and inefficient. In this paper, we propose a method to predict room-level occupancy by utilizing only smart-meter data. Several classifiers are used to estimate room-level occupancy information. We identify the best feature set consisting of appliances energy data, appliances state, and house-level occupancy data. The features are obtained using only smart meter data along with non-intrusive load monitoring and house-level occupancy prediction. We show that the proposed methods can achieve up to 90% accuracy for room-level occupancy prediction using only smart meter data.