An artificial neural network combined with response surface methodology approach for modelling and optimization of the electro-coagulation for cationic dye

An artificial neural network (ANN) approach with response surface methodology (RSM) technique has been applied to model and optimize the removal process of Brilliant Green dye by batch electrocoagulation process. A multilayer perceptron (MLP) - ANN model has been trained by four input neurons which represent the reaction time, current density, pH, NaCl concentration, and two output neurons representing the dye removal efficiency (%) and electrical energy consumption (kWh/kg). The optimized hidden layer neurons were obtained based on a minimum mean squared error. The batch electrocoagulation process was optimized using central composite design with RSM once the ANN network was trained and primed to anticipate the output. At optimized condition (electrolysis time 10 min, current density 80 A/m(2), initial pH 5 and electrolyte NaCl concentration 0.5 g/L), RSM projected decolorization of 98.83% and electrical energy consumption of 14.99 kWh/kg. This study shows that the removal of brilliant green dye can be successfully carried out by a batch electrocoagulation process. Therefore, the process is successfully trained by ANN and optimized by RSM for similar applications.

Automated summary

This study models and optimizes the batch electrocoagulation process for removing brilliant green dye using an artificial neural network (ANN) and response surface methodology (RSM). The optimized conditions resulted in a dye removal efficiency of 98.83% and an electrical energy consumption of 14.99 kWh/kg.