Neural controller - Warsaw University of Technology Neural controller - Warsaw University of Technology
Our partners

Welcome to the neural controller design

Aim of project

The aim of the project is to develop a prototype model of a neural network controller, called in short NN controller, designed for heat distribution centers, which will be using artificial neural networks algorithms. It will be used to control heat distribution centers for central heating, process heat and hot water, supplied from centralized heat sources.
Creating a new NN controller needs extensive research analysis, simulation and experimentation. The most commonly used method is the numerical modeling. This research project involves the regulation of heat processes in the building and includes a district heating node and heat exchange in the building. 

Simulation programs must, therefore, contain both a description of transient heat transfer for the building as well as a description of the control systems. Although there are many programs for thermal analysis of buildings, there is no such that could be used in this research project. It will be necessary to develop a numerical model of the building which will contain the real life control systems. This issue is very complex and requires extensive work. However, it is necessary to carry out research to develop a new neural controller.

During the development of the project experimental research will be performed, both in the laboratory, and also in the real objects.
Author of the project was involved for many years in similar research and owns a lot of results data, which will be used in the planned research.

Expected Results

The neural network will provide better control quality and a wider usage scope than the existing PID controllers. The phenomenon of instability of control systems caused by interference will not occur. Stable and minimized transient processes will satisfy the users, and also enable possibility of implementation of algorithms that will give significant energy savings, as compared to the existing systems of traditional PID controllers. Neural network system will provides proper control without requiring special settings both during system start time as well as possible changes in working conditions in later time.

This means that the effect of the adaptive control will provide the optimal control of the particular process. This results inenergy savings during the life time of the building. Savings depends on the intended use and can be up to 20-30% for the full year, while providing the same comfort at the any given time. This is directly related to the costs incurred by users of the facilities.


  • full adaptation to changing conditions in the non-linear processes,
  • automatic adaption to different configurations of heating equipment,
  • significant reduction of actuator movement  and as a result a significant extension of valves and actuators life,
  • reduce or even eliminate overshoot and reduce of the control time,
  • reducing the amount of the input parameters occurring in current standard systems

Expected results

It is expected that after the implementation of the Project, the existing PID controllers will be replaced by neural controllers. It will be the only Polish device used in heating control systems of heating devices, heating and ventilation systems in our country.

Website uses cookies in order to ensure access to the content (more information).
Change the settings for the storage of cookies, it is possible to directly from your browser.