Mission

Several research projects have succesfully demonstrated the need for automating building components controllers. As much as 30% of the total energy consumption in Western countries is used on buildings, much of which could be reclaimed by formulating adequate control strategies. Energy savings of up to 30% have been achieved with heating and building facade controllers, but a major obstacle to a succesful commercial implementation has always been a relatively high user rejection rate. People have a tendency to reject an imposed environment, feeling artificially controlled especially if the controller's settings disagree with their personal preferences. Unhapiness, unproductivity, and frustration result, leading sometimes to the user disconnecting the system. Previous research projects have demonstrated that user-adaptive control systems may hold the answer, i.e. controllers attentive to the user's individual preferences. Various techniques have been used to implement a learning system, often borrowed from the natural world. A recent succesful project, AdControl, uses fuzzy logic to have the user cooperate with the controller's installator to build a set of rules that attempt to satisfy both energy requirements and the user's wishes. A possible drawback with such a system is that it makes no attempt to quantify the user's visual comfort, in the same way that Fanger has quantified the thermal comfort. This present work will use statistical tools derived from Bayes' theorem to have the system learn what variables quantify best the visual comfort. Together with an accurate model of the illuminance distribution in the office, based on the Daylight Coefficients proposed by Tregenza in 1983, this controller will, it is hoped, both yield energy savings and help us understand what variables are important in determing user visual comfort.

Biography

Resume available as html or pdf.

Smart Buildings

I am the maintainer of a weblog related to the topic of computers and buildings research.