Prediction of BOD from Wastewater Characteristics and Their Interactions Using Regression Neural Network: A Case Study of Naidu Wastewater Treatment Plant, Pune, India

Abstract

AbstractAnalysis of variance (ANOVA) results were used to analyze the behavior of various water quality parameters like Total Suspended Solids, Biochemical Oxygen Demand, and Turbidity. The turbidity is not a recommended water testing parameter for wastewater testing, still it was measured on experimental basis. The datasets used in this work were derived from a detailed experimental investigation of inlet and outlet water parameters of Naidu wastewater treatment plant, a major conventional treatment plant in Pune City, India. It has an average flow rate of 115 MLD. The samples were collected daily for over three months. Statistical data analysis methods such as correlation coefficient, regression analysis, and graphical representation of the data were used to find the interrelations between selected parameters. The regression neural network model obtained from the analysis of data was used to predict BOD parameter considering standard error. The p-value obtained by ANOVA analysis was observed to be significant at p < 0.05. The ANN model predicted the results at maximum accuracy of 96% and average accuracy of 94%. Good interrelation between the selected parameters was observed and it was observed that BOD can be predicted using suspended solids and turbidity of water sample.KeywordsBOD predictionTurbidityTSSANN