An occupant-centered approach to improve both his comfort and the energy efficiency of the building

The accelerating depletion of fossil fuel reserves and the growing awareness about climate issues have put forth a plethora of interesting approaches that attempt to tackle the crucial problem of energy saving, specifically in buildings, known as a major energy consumer. Existing approaches tackle the problem of energy efficiency in buildings by proposing model-based approaches such as knowledge models (thermal, CO2, cost, etc.) and regressive models. However, different factors make the building of knowledge models particularly challenging, including the very complicated interaction between several heterogeneous phenomena that can impact the use of energy in buildings like the buildings envelope characteristics, their positions, the weather conditions, but also the occupant's behavior is a critical issue in the process. More Recently, techniques from machine learning (ML) to support energy saving in buildings gained increased interest. They learn from collected historical data a model that forecasts the future energy behavior of the building. Although occupant's behavior to save energy in the building is far from trivial, has received less attention from these studies. This paper takes on this challenge and proposes an energy management system based on historical data thanks to case based reasoning approach. We guide the occupant by proposing an action plan (opening/closing of doors/windows, etc.) to help him in the process of improving his indoor comfort (thermal, air quality, luminosity, etc.) without using more energy if not using less. To encourage the occupant to trust the inference mechanism learnt and cooperate with the energy management system, this approach generates explanations arguing the proposed action plan. We assess the performance of our energy management technique on real-word data collected from a research building at the University of Grenoble, France.(C) 2022 Elsevier B.V. All rights reserved.