?
Решение проблем редукции данных в автоматизированных системах коммерческого учёта потребления тепловой энергии для создания робастной модели
Energy saving is one of the trends in the development of modern urban economics. More than 40% of energy resources are consumed in residential buildings, and most of the consumed energy is used for heating. An urgent problem is increasing the energy efficiency of residential multi-floor buildings. It is associated with a decrease in heat loss and regulation of the heat energy supply depending on weather conditions. The design of the regulation model is in a framework of commercial and academic interest. Nevertheless, situation in modeling is often associated with high-quality data mining: it contains errors, the data are not stratified by concerning the heat supply scheme, and the size of daily data does not allow to create a model of the minimum order that should be robust but correctly reflect the dynamics of changes in heat consumption from the outside temperature. We propose approaches to data transformation that can be used to carry out a robust model of minimum order heat consumption based on spectral analysis and histogram estimates of the statistical distribution of data collected for a long time. The method for comparison houses with different heat consumption schemes was found. The average value of hot water consumption was calculated, and it was shown that data stratification is possible with the combined accounting of hot water supply and heating. As a result of the spectral analysis of the heat consumption series, it was found that the weather dependence is mainly determined by the first ten low-frequency harmonics of the heat consumption series. An algorithm for the reduction of rows is proposed. It makes it possible to design a robust model of heat consumption in all city dimestion.