Highly nonlinear process model using optimal artificial neural network

Abstract

The paper deal with modeling of highly nonlinear chemical process using the artificial neural network approach. The non-linear process is represented by polymerization plant. The data set used for an identification of the artificial neural network model is a real input and output data received from an existing polypropylene plant. The identified model is a nonlinear auto regressive neural network with the exogenous input. Performance of a trained network has been verified using the real process data and the ability of the production rate prediction is shown in the case study section. The obtained artificial neural network model is used for predictive control of the polypropylene reactor.