The applied investigation of exit waste stream from Petrochemical industries

It’s notable that qualitative information coming from investigation of details of subject. So, the influence of majors in pretreatment of wastewater such as amounts of NaOH, , composition of mixture of coagulants and mixing rate of first pretreatment reactor on the amount of turbidity, total hardness, content and electrical conductivity of effluent wastewater are investigated experimentally and the optimum values are presented. Also, sensitivity analyzing shows the importance of these four majors on the performance of pretreatment process Experimental data are normalized and preprocessed then suitable architecture of ANN, the number of neurons in hidden layer and transfer function, the training algorithm are optimized. Also, two types of ANFIS are trained and the results of ANN and ANFIS predictive models are compared with each other using statistical criteria (RMSE, R and MAE). Results demonstrate that ANN can predict more effectively and afford high accuracy for forecasting the performance of pretreatment unit

Keywords:

artificial neural network , ANFIS , turbidity , total hardness , wastewater pretreatment