Chemical Reactor Identification: a Comparative Study

This paper proposes a comparative study for the identification results of a chemical system. It consists in estimating the dynamic behavior of the process using PieceWise AutoRegressive eXogenous (PWARX) models, multimodel and classical ARX approaches. The determination of a PWARX hybrid model is carried out by the Outer Bounding Ellipsoid (OBE) algorithm. The latter is applied for system identification in cases where the disturbances are bounded. So, this work deals with the experimental application of OBE method in order to estimate the behavior of studied system with piecewise affine hybrid model. The obtained results are then compared with two existing approaches, showing that modeling the nonlinear chemical process with PWARX systems provides the best results in terms of precision.