Comparative study of different regression models on predicting wastewater treatment plant effluent characteristics

ISEAIA2017: Fifth International Symposium on Engineering, Artificial Intelligence and Applications,

1-3 November 2017, Girne, KKTC

In this study, various regression models including bayesian linear regression, neural network regression, boosted decision tree regression, decision forest regression and linear regression are employed in order to predict effluent characteristics of wastewater treatment plant (WWTP) e.g. biological oxygen demand and total suspended solids. The data was collected over a two-year period from a major conventional WWTP in Bursa, Turkey, having an average flow rate of 240 thousand m3/day. Influent water characteristics, metrological data and operational parameters were used as inputs and the performance of models was evaluated in terms of coefficient of determination (R2), mean absolute error (MAE), root mean squared error (RMSE) and relative squared error (RSE). Results show that regression models are efficient and robust tools in predicting WWTP performance.