Modeling of wastewater treatment plant in Hama city using regression and regression trees

Abstract

Background: Modeling of wastewater treatment plants is necessary to predict their later works. In this research, three methods were compared to predict some parameters at the outlet of wastewater treatment plant in Hama city in Syria. Methods: In this paper, three methods (linear regression, power regression, and regression trees) to model wastewater treatment plant in Hama city were compared to predict the parameters at the outlet of the plant (cBOD5out, CODout, TSSout) in terms of the parameters at the inlet of the plant (Qin, cBOD5in, CODin, TSSin). Results: When predicting cBOD5out, the values of RMSE of the test data set were 4.4105, 4.3875, and 3.8418; when predicting CODout, the values of RMSE of the test data set were 6.9325, 6.8003, and 5.3232; and when predicting TSSout, the values of root mean squared error (RMSE) of the test data set were 3.7781, 3.6936, and 3.2391 using linear regression, power regression, and regression trees (RTs), respectively. Conclusion: According to the results, the RTs outperforms in predicting cBOD5out, CODout, and TSSout because this method achieved the least RMSE of the test data set.