Nonlinear identification of multicomponent batch distillation processes

摘要

In this paper, a combined identification scheme is proposed, where a reduced order nonlinear Hammerstein model of the process is estimated using data generated by simulation of a rigorous model implemented with the commercial software HYSYS~circleR, which is tuned using real data. The proposed scheme for the identification of the Hammerstein model does not involve any nonlinear optimization procedure, requiring only linear least squares estimation and singular value decomposition. The performance of the proposed method is compared to that of a subspace method (4SID method), that delivers a state-space model of the column.