MODELING, OPTIMIZATION AND CONTROL OF SMB PROCESSES

摘要

This article deals with the identification and control of a simulated moving bed (SMB) process for the separation of enantiomers 1-1'-bi-2-naphthol subject to bi-Langmuir isotherm. The first principles model is considered as the actual SMB unit, which is then identified by the method of subspace identification to give the input/output data-based prediction model. For this, the average purities of rich component over one switching period in extract and raffinate streams are selected as the output variables while the flow rate ratios in sections 2 and 3 are chosen as the input variables. From simulation results the identified model is found to show an excellent prediction capability. On the basis of the identified model the input/output data-based predictive controller is designed and applied to the MIMO control problem for the SMB process. It turns out that the proposed controller performs quite satisfactorily for the setpoint tracking as well as for the disturbance rejection.