Synthesis of Heat Supply System Automatic Control Algorithms Using a Heat Balance Model

Abstract

The article addresses matters concerned with studying the possibility of applying additional stochastic methods and forecasting the demand in automated control systems to improve their efficiency. An analysis of statistical archive data on heat consumption of the studied object and climatic data at the object location place carried out by applying a simplified artificial neural network was used as the stochastic method. In carrying out the study, numerical experiments were performed on the basis of 675 unique types of artificial neural network configurations. Based on the obtained results, a conclusion can be drawn about the optimal parameters according to the least RMS error criterion for application at the object studied. Dependencies of the predicted amount of heat that the studied object can consume have been derived proceeding from testing and processing the results obtained. The calculated data are compared with the forecasting results obtained by the deterministic analysis method and real data on the heat consumption of the object for the period under consideration. An "idealized" step function response curve describing the object and a transient process graph have been drawn. By applying the developed methodology on the example of refining the setting of the heating point for a higher educational institution building in the city of Moscow, it becomes possible to achieve the calculated effect of 11.60% relative to the heat consumption of the object without changing the PID controller settings. The developed methodology can also be applied to various infrastructure facilities and housing stock in the city of Moscow to increase the efficiency of heat supply system automated control systems.