Development of Software Tools for Calculating and Forecasting the Integral Index of the Heat Supply System of Typical Buildings Based on Neural Network Methods

The article addresses an urgent problem of developing methods and software tools for calculating and predicting the integral index of building energy systems based on their energy efficiency and reliability, which has been solved only partially in view of the complexity of collecting all data about the building in the currently available complex mathematical models. It is proposed to make a shift to calculating the energy system integral index using artificial neural networks. This will make it possible to significantly simplify the procedure of formulating the system of initial equations by using statistical data on the object under study. The architecture of the developed prototype software system is presented together with a database scheme, which describes the correlation of available building data, based on which the integral index components (energy efficiency and reliability indicators) are calculated. For solving the problem of predicting the demand for building energy systems' energy resources on the basis of data obtained from open-access sources, it is proposed to apply a method on the basis of the specially organized artificial neural network described in this article. The applicability of the study results to the of "green construction" objects and office buildings of business centers in Moscow is shown.