Big Data in Predicting the Climatic Resistance of Building Materials. I. Air Temperature and Humidity
Abstract
The article provides a comparative analysis of the quantitative values of temperature and relative humidity of the ambient air calculated according to GOST 16350-80 for a moderate continental climate (representative location – Moscow), and obtained at the meteorological station of Ogarev State University (Saransk) for the period from 2015 to 2022. A significant discrepancy between the calculated and accumulated values of meteorological factors has been established. In the studied time interval, the spread of the difference between the calculated and accumulated values ranged from -5.8 to 10 oC for air temperature and from -36.4 to 32.5% for relative humidity. There is also a significant variation in the difference of the studied indicators depending on the calendar month, as well as a de-viation of a number of distribution curves from the normal form, which, in general, indicates the im-possibility to reliably assess distributions using only the value of the quadratic deviation recommended by GOST 16350-80. It has been established that the use of calculated distributions virtually does not allow taking into account events occurring near the boundaries of the distributions of the studied me-teorological parameters. This, in turn, irreversibly affects the accuracy of forecasting the climatic aging of building materials. When predicting the climatic resistance of materials by machine learning methods, it is proposed to use absolute humidity values instead of relative humidity as a sign that has a clear physical meaning (mass of water vapor contained in 1 m3 of air), as well as defined only in the range of non-negative values.