Wastewater treatment plant modelling and performance parameters evaluation: The case of Bolu

Within the scope of this thesis, first of all, statistical analyzes such as ADF Unit Root Test and Johansen Cointegration Test were conducted in the Ewiews 10 program for the data obtained from the Bolu Central Waste Water Treatment Plant (BAAT). Later, it was modeled with 1D Convolutional Neural Network with the data of BAAT. Modeling was carried out in Python programming language. The data set used for modeling consists of 400 days of data obtained from BAAT between the years 2016-2018. In the Convolutional Neural Network (KSA) model, the input layer variables are defined as pH, Biological Oxygen Demand (BOD5), Chemical Oxygen Demand (COD) and Suspended Solids (AKM). Output data of pH, BOD5, COD and AKM parameters were estimated in the output layer. In the KSA model, 3 different optimization techniques were used. These optimization techniques are Adam, Rmsprob and SGD. Adam optimization technique gave the best results in terms of number of losses and iterations. The accuracy rate obtained in the study was calculated as 97.6%.