Predictions of isolate and normal pentene of debutanizer catalytic reforming unit by using artificial neural network

This paper presents a feed-forward Artificial Neural Network (ANN) model for prediction of isolate and normal pentene of debutanizer catalytic reforming unit. Temperature, reflux flow, and flow rate are used as input variables to the network. Isolate pentene (iC₅), and normal pentene (nC₅) are employed as the output variable. About 500 field data collected from PETRONAS Penapisan (Melaka) Sdn Bhd were used to develop the ANN model. The developed ANN model is obtained by dividing the collected data set into three different group; training, validation, and testing group. Back-propagation algorithm was used to train the network. A correlation coefficient of 0.999 was obtained with standard deviation of 0.006 for iC₅. For nC₅ a 0.999 correlation coefficient and 0.005 standard deviation obtained. © 2008 IEEE.