Journal of Applied Sciences1812-56541812-5662Asian Network for Scientific Information10.3923/jas.2010.2558.2564OsmanMohanadRamasamyM.1220101021Distillation column is one of the most widely used unit operations in process industries and the operation and control of this unit was always a challenging task due to the complexity of operational parameters interactions and the lack of a real time measurement of the products composition. Time delay involved with GC measurement of the product composition prevented the utilization of effective closed-loop control. To overcome this problem an inferential control scheme based on soft sensor estimation of the products composition is now widely adopted. In this work, a neural network based soft sensor is developed to be used in an inferential control scheme of a pilot-scale binary distillation column. Data were collected from the distillation column under different operating conditions with forced disturbances in number of operation variables. The collected data is pre-processed for the removal of outliers, normalized and segmented into training and test data subsets. The most important process variables in the model and their lags have been chosen systematically. Different neural networks have been trained using the preprocessed data. Trial and error method is used to find the optimum number of neuron in the hidden layer for each network. The performance of different networks is discussed. The developed soft sensor can be utilized in an inferential control scheme on the distillation column.]]>De Canete, J.F., S. Gonzalez-Perez and P. Saz-Orozco,2008Sit, C.W.,2005Zamprogna, E., M. Barolo and D.E. Seborg,2005Lin, B., B. Recke, J.K.H. Knudsen and S.B. Jørgensen,2007Hoskins, J.C. and D.M. Himmelblau,1988Eriksson, L., E. Johansson, N. Kettaneh-Wold and S. Wold,2001Fortuna, L., S. Grazaini, A. Rizzo and M.G. Xibilia,2007Wang, X., R. Luo and H. Shao,1996Fortuna, L., S. Graziani, M.G. Xibilia and G. Napoli,2006Hagan, M.T., H.B. Demuth and M.H. Beale,1996Picton, P.D.,2000