Naslov
Artificial intelligence models for the prediction of NOx emissions in gas turbines

Autori
Glučina, Matko ; Mrzljak, Vedran ; Poljak, Igor ; Car, Zlatan

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Skup
International Student Scientific Conference (Ri-STEM 2022)

Mjesto i datum
Rijeka, Hrvatska, 08.06.2022. - 09.06.2022

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
Artificial intelligence ; Gas Turbine Power Plants ; Machine Learning ; NOX ; Regression Prediction

Sažetak
Gas emissions from power plants and gas turbines use fossil fuels as their energy source and have a detrimental effect on people and the environment, one of the harmful gases obtained by burning fossil fuels is Nitrogen Oxide which can have harmful effects on humans. This research boils down to the use of artificial intelligence algorithms from the scikit-learn library that can potentially predict future NOx values from a gas turbine. The research was conducted on a publicly available data set from a gas turbine in Turkey. Linear models Linear regression, Ridge, Stochastic gradient descent, ExtraTreesRegessor, GradientBoostingRegressor, and RandomForestRegressor were used, with the best result obtained using GradientBoostingRegressor, with R2 and MSE being 0.9 and 12.60. The onclusion is that a simple approach could not provide sufficient quality results for the use of the model.

Izvorni jezik
Engleski

Znanstvena područja
Elektrotehnika, Računarstvo, Temeljne tehničke znanosti

POVEZANOST RADA