DAYLIGHT GLARE PROBABILITY PREDICTION FOR AN OFFICE ROOM

Glare assessments are currently made from High Dynamic Range (HDR) images taken from the Point Of View (POV) and viewing direction of a user. This paper analyses the feasibility to estimate the Daylight Glare Probability (DGP) at the user-level based on machine-learning techniques, sun position and a downward-pointing camera sensor mounted at the ceiling of a simulated office environment. Three different office cases have been considered: an empty room, an empty room with venetian blinds and a furnished room without venetian blinds. The influence of the sun direction has been considered as a parameter to predict the observer DGP. Subsequently, the best parameters have been selected to build a black box model using Artificial Intelligence (AI). Results show that, by using the DGP of the ceiling camera and the sun position, it is possible to accurately predict the DGP for an observer’s POV.