Dye Removal Probing by Electrocoagulation Process: Modeling by MLR and ANN Methods

The present study performed to investigate dye removal efficiency (DR%) of solutions containing direct blue 71 (DB71) using electrocoagulation (EC) process. applied voltage (VEC), Initial pH of the solution (PH0), time of electrolysis (tEC) and initial dye concentration (C0) considered as more effective operational parameters. The experimental data obtained in a laboratory batch reactor. The achieved DR% of 4.4-99.3 gained under experimental conditions. The multiple linear regression (MLR) and non linear artificial neural network (ANN) models utilized to EC modeling and DR% predicting. By applying best MLR and ANN models to predict the test set, Q2ext and RMSE determined 0.79 and 13.7 for MLR and 0.93 and 8.01 for ANN. Further tests and data treatments were done for more validation and introduce model applications and also to clarify other aspects of EC, such as Leave-n-Out (n=1, 43-44, 74) cross-validation, energy consumption calculation, graphical prediction of the optimum experimental conditions and diversity test. The experimental results proved that EC is an effective way to treat dye solutions containing DB71. VEC, pH0, tEC and C0 parameters influenced DR% and the ANN and MLR have been successfully used to modeling EC.

AFSHIN MALEKI, HIUA DARAEI, LOGHMAN ALAEI, LEILA ABASI AND ANISE IZADI