Chemical Oxygen Demand (COD) Estimation in Petrochemical Industry Wastewater Effluent via Robusted Regression

In order to increase the quality of industrial wastewater treatment and better manage of them, their approach should be simple and accurate for estimating process. Treatment processes for black box systems are due to the influence of many factors that involved in the system. Because of problems in using physical models, the use of statistics and regression methods could be helpful. Therefore, whatever model is simpler and less input variables so the model will be more important. Influent of the proposed model includes output data of biological unit and effluent is chemical oxygen demand of the clarifier. To compare the models performance three indicators of R-square, Correlation Coefficient(R) and Mean Square Error (MSE) are used. The aim of this study is creating linear data mining model and comparing them with similar methods for quality data. Finally, a linear robust regression with MSE = 0.089054, R = 0.784727 and R-Square = 0.6096 is proposed.