Development of Predictive tool for measuring concentration of Fe and Mg in Fluidized Bed of Greensand
Volume Title: 1
1Department of chemical engineering ,Guilan university,Rasht,Iran
2Department of Chemical Engineering Faculty of Engineering, University of Guilan,Rasht,Iran
3گروه پژوهشی مهندسی آب و محیط زیست پژوهشکده حوزه آبی خزر، دانشگاه گیلان، رشت، ایران
The purpose of the present study was to investigate the output parameter of Artificial Neural Network (ANN) model, which is the concentration of Fe and Mg in the output of fluidized bed. Greensand was used as the predictor of the concentration of Fe and Mg. Time, flow rate, mass of the adsorbent, and presence or absence of KMnO4 were considered as the input parameters for the model. An ANN model with 10 neurons was selected to study the influence of transfer functions and training algorithms. Sigmoid and Purlin transmission operators and feed-forward training algorithm were the most accurate predictors of the output concentration. Output experimental data and ANN prediction values on Root Mean Square Error (RMSE) were studied by introducing small random errors within a range of ± 8 %.