ANN Modeling and GA Optimization of Zinc Removal from Wash water by Electro-coagulation Process

Abstract

The present research concentrates on Artificial Neural Network (ANN) for modeling and Genetic Algorithm (GA) for optimization of zinc removal from the industrial wash water by EC process. In the EC process, the most important control independent variables are Initial Concentration (Ic � 216.5 to 866mg/L), Current Density (Cd � 0.1 to 0.6A/dm�) and Time (T � 2 to 15mins). These variables are also affecting the performance of zinc removal. ANN model was able to predict the maximum removal of zinc with two transfer functions like tan sigmoid at hidden layer with 8 neurons, purelin at output layer. Feed forward multilayered perception with Levenberg - Marquardt back propagation training algorithm was used for train the design with Mean Squared Error (MSE) of 1.18 and Correlation Coefficient (R2) of 0.9909 in ANN shows that the model was capable to predict the zinc removal. Single Objective Optimization for maximizing the zinc removal was conducted using GA over the ANN model. Using pattern search method in a GA, the best optimum conditions are recorded as 217.5 mg/L, 0.1A/dm� and 2mins for Ic, Cd and T respectively and the maximum zinc removal at the above condition as 88.71%.