ANN Modeling of a Membrane System for Concentrating the Sugar Syrup

5-10
Ali Mehrjouyan and Abbas Khoshhal (Department of Chemical Engineering, Payamenoor University, Tehran, Iran)

In this study, artificial neural network modeling of a reverse osmosis membrane system for concentrating the sugar syrup was investigated. For this purpose, data from a pilot was used for this modeling. Network input parameters included feed temperature, feed pressure and feed brix degree and the output of the network was the brix degree of concentrated stream. The selective network structure was considered multi-layer perceptron with the back propagation algorithm. Transfer function in the hidden and output layers and the number of neurons in the hidden layers were optimized to assess the model accuracy. The transfer functions “logsig” and “purelin” were used in the hidden and output layer respectively. The optimum number of neurons in the hidden layer was 14. Two-thirds of the data set was used for training the networks and the other part was applied for the validation of the ANN models. For analyzing the accuracy of the modeling, mean square error (MSE) and correlation coefficient) R2 (were used. The values of the presented indexes are 0.00017 and 0.983 respectively. The comparison between the experimental values and the ANN predicted results showed good agreement, which implied that the adopted models are suitable for correctly predicting extraction yield.