چكيده به لاتين
In educational spaces where the learning process takes place, the use of natural light has a great impact on the health and creativity of its users. The quality of natural light received by users, due to its effect on the mental and physical health of students and also its effect on reducing energy consumption in classrooms where students and teachers spend a lot of time, is very important. Students in the classroom usually sit in certain places and are not able to adapt to the environment, which means that if a student is sitting in a not suitable place in the classroom, where there is not enough light or strong sunlight causes discomfort to his eyes, Has little ability to move and select a more suitable location. In this case, the user provides comfort by changing the environment, for example, using artificial lighting or using dynamic shades; this is usually accompanied by an increase in energy consumption. For this reason, when designing a classroom, in addition to examining the distribution of daylight in the space, the comfort of users should also be considered. By maximizing user comfort, energy consumption can be minimized and a healthy environment can be provided for students. With the increasing importance of daylight use and the daily progress of science in this field, several metrics have been introduced to evaluate a space in terms of daylight distribution and visual comfort of users. These metrics, which initially evaluated space as instantaneous and static, with the advancement of computers and their speed in processing and calculation, gave way to dynamic metrics that measure space during all occupied hours throughout the year. Today, the study of the quality of light in space is done by computer simulations using simulation engines that estimate the state of space with very high accuracy. But these simulations are time-consuming and complex, even with the use of modern computers to examine all design options. For this reason, in many architectural offices, the assessment of space in the early stages of design is ignored.
Machine learning algorithms are used in many sciences for estimation and prediction. These algorithms have been widely used in estimating energy consumption in buildings, but their use in estimating the quality of space in terms of light distribution and visual comfort has been evaluated to a very limited extent. Using models obtained from machine learning algorithms, it is possible to estimate the visual quality of space in a very short time and use these models as an alternative to time-consuming simulations.
In this study, in the first step, we evaluate the efficiency of multiple linear regression to estimate the four metrics: sDA, UDI-a, ASE and SVD. In the next step, the effect of independent variables on these metrics is studied, and using the data obtained from the simulation, the change in the amount of metrics relative to the independent variables is evaluated, and in the last step the optimal design option is found. Based on the results, mathematical models obtained from multiple linear regression estimate the values of the metrics with great accuracy. In all metrics, the window size is considered the most influential variable among the independent variables. The data show that ASE shows very little change when the window dimensions are the same and the glass material changes, while SVD changes as the glass material changes. The data also show that, in general, classrooms with north-facing windows are much more efficient in terms of visual comfort and energy consumption than south-facing classrooms.
