Optimal Personal Comfort Management Using SPOT+

Abstract

We present SPOT+, a system that allows office workers to optimally balance between heating energy consumption and personal thermal comfort. In prior work, we described SPOT: a smart personal thermal control system based on reactive control [8]. In contrast, the SPOT+ system performs predictive control. Specifically, SPOT+ uses the k-nearest-neighbour algorithm to predict room occupancy and learning-based model predictive control (LBMPC) to predict future room temperature and to compute the optimal sequence of control inputs. This allows the system to schedule future temperature setpoints to optimize an objective function expressed as a linear combination of thermal comfort and energy consumption. We have deployed SPOT+ as well as four other alternative control schemes in an office workspace. We find that SPOT+ reduces energy usage by 60% compared to a fixed-temperature setpoint and reduces personal thermal discomfort from 0.36 to 0.02 (in the ASHRAE comfort scale) compared to SPOT.