Estimation of organic matter dependent on different variables in drinking water network using artificial neural network and multiple regression methods

English (OpenAlex)

The aim of this study is to estimate of organic matter values based on chlorine and turbidity values with the help of ANN and multiple regression (MR) methods. Three different models were done with ANN, and the statistical performance of these models was evaluated with statistical parameters like; µ, SE, σ, R2, RMSE and MAPE. The R2 value of the selected best model was found to be quite high with 0.94. The relationship between the evaluation results of the ANN model and the empirical data (R2 = 0.92) showed that the model was quite successful. In the MR analysis, R2 was determined as 0.63, and a middling significant (p <0.05) relationship was found. Since the calculated F value was greater than the tabulated F value, it was concluded that there is a clear relationship between dependent and independent variables. In addition, spatial distribution maps of chlorine, turbidity, organic matter values were created with the help of the GIS. With these maps, the estimated distribution of the measured parameters in the whole city network was accomplished. This study revealed that turbidity and chlorine parameters are related to organic matter value, and by establishing this relationship, organic matter can be estimated by ANN.