Portal de Periódicos da CAPES

Bayesian estimation of visual discomfort

2007; Taylor & Francis; Volume: 36; Issue: 1 Linguagem: Inglês

10.1080/09613210701544061

1466-4321

D. Lindelöf, Nicolas Morel,

Color perception and design

Abstract A method is proposed to estimate objectively occupants' visual discomfort in small office rooms as a function of the illuminance at one or more locations. Expressed as a visual discomfort probability, it is based on an analysis of the past history of the user's interactions with the blinds and lighting controls. A Bayesian formalism is applied to infer the probability that any illuminance distribution should be considered by the user as visually uncomfortable. This method is applied to the interactions of the experimental building's users of the Solar Energy and Building Physics Laboratory (LESO), Ecole Polytechnique Fédérale de Lausanne (EPFL), and a discomfort probability is derived as a function of the horizontal workplane illuminance. This discomfort probability is very high (0.5–1.0) for an illuminance below 200 lx; it reaches a global minimum (about 0.3) at 500 lx, and then increases gradually for larger illuminances until it reaches 1.0 again close to 3000 lx. La méthode proposée a pour objectif d'estimer objectivement l'inconfort visuel ressenti par les occupants de bureaux de petites dimensions en fonction du niveau d'éclairement en un ou plusieurs endroits. Exprimée comme probabilité d'inconfort visuel, elle est basée sur une analyse de l'historique des interactions entre l'utilisateur et les stores et les commandes d'éclairage. Un formalisme bayésien est appliqué pour inférer la probabilité que toute distribution de niveau d'éclairement devrait être considérée par l'utilisateur comme visuellement inconfortable. Cette méthode est appliquée aux interactions des utilisateurs du bâtiment expérimental du Laboratoire d'énergie solaire et de physique du bâtiment (LESO) de l'Ecole Polytechnique Fédérale de Lausanne (EPFL); une probabilité d'inconfort est dérivée comme fonction du niveau d'éclairement du plan de travail horizontal. Cette probabilité est très élevée (0,5–1,0) pour un niveau d'éclairement inférieur à 200 lx; elle atteint un minimum global (environ 0,3) à 500 lx puis augmente progressivement pour des niveaux supérieurs jusqu'à atteindre de nouveau 1,0 à proximité de 3000 lx. Mots clés: confort, mécontentement, éblouissement, niveau d'éclairement, occupants, commande intelligente, inconfort visuel Keywords: comfortdissatisfactionglareilluminanceoccupantssmart controlvisual discomfort Acknowledgements The authors wish to thank Dr Sylvain Sardy of the Chair of Statistics at EPFL for recommending the taut-string algorithm for density estimation. The authors also thank Dr Arne Kovac, one of the authors of the R implementation of the taut-string algorithm, for helping solve certain issues with the implementation. Also thanked is the Office Fédéral de l'Education et de la Science for funding the authors' contribution to the ECCO-Build project. The contribution to the funding of the ECCO-Build project by the Office Fédéral de 1'Energie is gratefully acknowledged. The authors also thank the Ecole Polytechnique Fédérale de Lausanne for its funding and infrastructure. Jessen Page, Dr Darren Robinson and Professor Jean-Louis Scartezzini, as well as the anonymous referees, are thanked for reviewing the material in the paper and for providing their helpful comments and suggestions. Notes 1. Discomfort glare is defined as glare that causes discomfort without necessarily impairing the vision of objects. It is distinct from disability glare, which is defined as glare that impairs the vision of objects without necessarily causing discomfort. 2. To be absolutely rigorous, Pr(C = F) is the probability that, presented with a given visual situation, and explicitly prompted whether that situation is judged uncomfortable, a user would answer affirmatively. The assumption is made that this interpretation is equivalent to saying that the situation is visually comfortable. Contrary to what might be intuitively believed, it is not the probability that the user is about to adjust the visual environment. 3. Siemens brightness sensors GE 252, which are actually ceiling-mounted luminance sensors shielded from the window's luminance, are used. The conversion from the workspace's luminance to its illuminance is a programmable feature of the sensor, so that it can be calibrated to determine illuminance, assuming a constant reflectance in its visual field. Each sensor has been calibrated with a reference sensor by Guillemin (2003) Guillemin, A. 2003. “Using genetic algorithms to take into account user wishes in an advanced building control system”. Lausanne: Ecole Polytechnique Fédérale de Lausanne. PhD thesis [Google Scholar]. The sensors' output is linear only up to about 500 lx, and has to be corrected by software in order to be accurate up to 3500 lx. 4. In which case it would be tempting also to use Guth's Discomfort Glare Rating, available in Radiance, but this would be a step back towards a system without user adaptation. 5. This can be power applied to either heating elements or active cooling systems. The LESO building is passively cooled, and in these situations this term could be left untouched, just as if a cooling system existed, or it could be replaced with an estimation of a user's thermal discomfort resulting from excessive solar gains. 6. Several variations on this cost function have been considered for the experimental set-up at LESO-PB. In particular, a ‘View’ term has been added to take into account a positive cost when the blinds occlude too much of the window, affecting the user's view to the outside. 7. The answer 0.5 is the right answer even if you are explicitly told that the coin is loaded one way or the other, but without specifying which. This is an example where bayesian probabilities flatly contradict ‘intuitive’ probabilities.