Experts from the Perm National Research Polytechnic University (PNIPU) have developed an intelligent algorithm for controlling industrial gas burners. The press service of the scientific institution told socialbites.ca that the equipment is able to automatically select the optimal operating mode.
Apparatuses with gas burners are widely used in production, for example, when rapid heating of raw materials or drying of bulk materials is required. Their operations are guided by advanced control algorithms that require accurate data about device parameters. Because such information is difficult to obtain and update under continuous production conditions, researchers have turned to the capabilities of neural networks.
As scientists discovered, if historical measurement data and standardized values are added to the current calculation model, the calculation results will contain serious errors. Thus, the program for estimating heat loss in the stove showed an error of almost 6%, corresponding to approximately 33 °C.
The polytechnicians then trained the neural network based on the data sample, thanks to which artificial intelligence (AI) was able to refine the computational model’s information. The application reduced the error by a factor of 16, from 5.9% to 0.35%. In this case, the algorithm’s processing cycle was 0.03 seconds.
“The use of a neural network will make it possible to quickly determine changes in the calorific value of the fuel gas, fouling, clogging of the combustion chamber walls with carbon deposits and changes in the amount of heat released during the combustion of gas.” PNIPU noted.
Previously in Moscow presented Projects based on productive artificial intelligence technologies.