Regulation and control of thermal processes in buildings has a large influence on the heat consumption for heating purposes. Applied solution so far uses standard algorithms like proportional derivative or proportional integrating derivative. The disadvantage of these solutions is the need for constant supervision of qualified technical services. Character of thermal processes is the variation of dynamic parameters with the change of temperature factors of heat exchanging. In order to ensure proper control quality it is necessary to change the parameters of the controller system. Persons exercising supervision over district heating systems are not adequately prepared for such actions. The result of that are often unstable heat distribution centers, and a substantial deterioration in the quality of control. This has a direct impact on the heat consumption and the durability of control devices. In this paper a new, not used yet dynamic process control solutions of heating systems, using artificial neural networks has been presented. The developed neural network algorithm, and conducted research shows that the quality of control is significantly better than for the case of conventional controllers. This has also a major impact on saving energy on heating purposes in buildings.
