We have developed a self-learning controller for technical equipment in buildings

Proper control of HVAC systems is crucial for efficient and economical building operation. However, mistakes are often made that unnecessarily increase costs. Their solution is diagnostics based on standardised labels and self-adjusting controllers capable of automatically optimising building operation and minimising human error. As part of the RoboPID project, UCEEB CTU, in collaboration with ENERGOCENTRUM PLUS, developed a new controller and successfully tested it on more than thirty installations in real buildings.
 
In the RoboPID project, we have developed and implemented a self-learning proportional-integral-derivative (PID) controller that operates similarly to a Metering and Control specialist. Based on the transient response to small changes in the desired reference, it gradually adjusts the parameters so that the resulting response is fast, smooth, and oscillation-free. The great advantage of automatic algorithmic processing is that the observation of transient responses, which can take many hours, is handled by a machine instead of skilled personnel.

The implementation of the self-learning controller is available in the programming languages C++, Python and especially in ST code for running on industrial computers (PLC). RoboPID has been successfully validated on more than 30 real installations.

If you are interested in more information, please do not hesitate to contact Ing. Jiří Dostál at the following e-mail address: jiri.dostal@cvut.cz. 

The project RoboPID - diagnostics and automatic tuning of controller parameters for industrial use in buildings was supported by the European Regional Development Fund under the Operational Programme Enterprise and Innovation for Competitiveness.