Modeling and identification of heat exchanger process using least squares support vector machines

Abstract

In this paper, Hammerstein model and non-linear autoregressive with eXogeneous inputs (NARX) model are used to represent tubular heat exchanger. Both models have been identified using least squares support vector machines based algorithms. Both algorithms were able to model the heat exchanger system with-out requiring any a priori assumptions regarding its structure. The results indicate that the blackbox NARX model outperforms the NARX Hammerstein model in terms of accuracy and precision.