Modeling of the process of biological treatment of wastewater of the town of Donji Miholjac

Otpadna voda sadrži organska i anorganska onečišćenja koja se uspješno uklanjaju biološkim procesima na uređajima za pročišćavanje otpadnih voda. Zbog složenosti procesa i nelinearne dinamike proces uklanjanja onečišćenja prikazuje se matematičkim modelima kojima se omogućava bolje razumijevanje i predviđanje procesa. U ovom radu, za opis obrade otpadne vode na uređaju za pročišćavanje grada Donjeg Miholjca korištena su dva modela: viševarijantni linearni model (MLR) i model lokalne linearne regresije (PLR). Izlazne vrijednosti kemijske potrošnje kisika (KPK) i biokemijske potrošnje kisika (BPK5) dobivene su za oba modela na osnovi različitog broja ulaznih parametara. Najbolje vrijednosti R2 dobivene su za modele s 13 varijabli, a model lokalne linearne regresije (PLR) bolje opisuje eksperimentalne podatke što ukazuje na nelinearnost procesa obrade otpadne vode.Wastewater contains all sorts of organic and inorganic pollutants which indirectly affect drinking water, which is of utmost importance for human beings Filtration systems are used to remove pollutants from wastewater. Therefore, it’s important to mathematically model the wastewater treatment to have a better understanding and more accurate and precise prediction of wastewater treatment plant. In this study, two mathematical models are used: Multivariate Linear Regression (MLR) and Piecewise Linear Regression (PLR). For model creation, data from wastewater treatment plant in city of Donji Miholjac was used. The purpose of the model was to predict the output values of chemical oxygen demand (COD) and biochemical oxygen demand (BOD) depending on influence of other process variables. The analysis showed that the PLR model was more successful which proved non-linearity of wastewater treatmen