Implementation of neural network algorithms for solving the problem of heat supply regulation

The article discusses a strategy for managing the supply of thermal energy, in which self-learning based on neural networks is used to predict the thermal regime of a building. Examples of the use of neural network technologies to improve the energy efficiency of technological and power plants at the stage of making a decision on their design and their operation are given. A further increase in the efficiency of heat supply provides for the transition to smart grids, which have qualitatively new characteristics of reliability and controllability. Smart grids make it possible to take into account specific consumer needs at any given time. To implement a scientific task, it is required, in particular, to reach the world average indicators of these characteristics. One of the ways to achieve them is accurate forecasting of heat energy consumption and construction of consumer profiles, which affects both technological processes and its economic efficiency of heat supply.