SciELO - Scientific Electronic Library Online

vol.12 issue1The influence of the incidence angle on the heat gains through transparent materialsProcedures for estimating the potential use of solar chimneys for natural ventilation enhancement in low-rise buildings author indexsubject indexarticles search
Home Pagealphabetic serial listing  

Services on Demand




Related links


Ambiente Construído

On-line version ISSN 1678-8621


FONSECA, Raphaela Walger da; DIDONE, Evelise Leite  and  PEREIRA, Fernando Oscar Ruttkay. Models for predicting the reduction of energy consumption in buildings using daylighting through a multivariate linear regression and artificial neural networks. Ambient. constr. [online]. 2012, vol.12, n.1, pp.163-175. ISSN 1678-8621.

Many studies have confirmed users' preference for daylight as a light source in buildings. In addition to the health benefits attributed to its influence on human circadian rhythms, high quality colour reproduction and other aspects, daylight has a known potential for energy savings when replacing or supplementing artificial lighting. Another factor to be considered is that the availability of daylight coincides with the working hours of commercial office buildings. In this context, the objective of this paper is to draw a comparison between two types of approximation methods used to estimate the potential energy savings through the use of daylight in office buildings. These approximation methods models are: Linear and Nonlinear Multivariate Regression, also known as Artificial Neural Network (ANN). The results show that ANNs are particularly suitable for this type of problem due to their learning aptitude, which allows significantly better extrapolation of the learning data than Multivariate Linear Regression.

Keywords : Daylight; Energy efficiency; Neural networks.

        · abstract in Portuguese     · text in Portuguese     · Portuguese ( pdf epdf )


Creative Commons License All the contents of this journal, except where otherwise noted, is licensed under a Creative Commons Attribution License