Application of Selected Methods of Artificial Intelligence to Activated Sludge Settleability Predictions
Bartosz Szeląg, Jarosław Gawdzik
More details
Hide details
Faculty of Environmental, Geomatic and Energy Engineering, Kielce University of Technology,
Al. Tysiąclecia Państwa Polskiego 7, 25-314 Kielce, Poland
Publish date: 2016-07-22
Submission date: 2015-12-09
Final revision date: 2016-02-16
Acceptance date: 2016-03-18
Pol. J. Environ. Stud. 2016;25(4):1709–1714
In the study, the results of measurements of inflow (Q), wastewater temperature in the chamber (T), a degree of external (RECext) and internal (RECint) recirculation in the biological-mechanical wastewater treatment plant in Cedzyna near Kielce, Poland were used to make predictions of settleability of activated sludge. Three methods, namely genetic programming, the Support Vector Machines method and artificial neural networks were employed to compute activated sludge settleability. The results of analyses indicate that artificial neural networks demonstrate the best predictive abilities. That is confirmed by the values of parameters that describe simulation fit to sludge settleability measurement data for inputs of concern.