ReViCEE: A recommendation based approach for personalized control, visual comfort & energy efficiency in buildings

Abstract

Built-environment, especially open-plan workplaces are often not tailored to meet individual visual comfort needs. Therefore, meeting the need for personalized visual comfort whilst achieving energy efficiency in open-plan office environment has been an open challenge. However, recent technological advancements in distributed sensing, pervasive computing, context-awareness and machine learning is progressively closing this gap. This article introduces ReViCEE–a simple recommender systems based approach to learn both individual and collaborative user-preferences from historical data and offer recommendations for intelligent building lighting controls. The intelligence in this case is achieved by being able to derive set-points to control task lights such that it balances personalized visual comfort without compromising on energy savings. The proposed approach has been developed using Python and implemented on a real test-bed in an university campus office building in National University of Singapore. The evaluation of the proposed approach is carried out for two months using field experiments involving distributed wireless sensor actuator network (WSAN) and multiple occupants having varied visual sensation. The novelty lies in proposing a new inter-disciplinary approach that supports smart and intelligent buildings paradigm by learning and predicting optimum individual user-preferences towards energy efficient control of personalized light. The results obtained from field experiments present a potential energy savings upto 72% when compared to the conventional lighting systems used.