Diagnosis and on-line parametric estimation of simulated moving bed

In this paper, a recently developed graphical signatures generation tool is used for diagnosis and on-line parametric estimation of simulated moving bed (SMB). The underlying diagnosis problem corresponds to variations affecting four of system’s parameters, namely, the two Langmuir coefficients, the extract and the feed flow rates. It is shown that this diagnosis method enables detection, isolation and parameter estimation even under simultaneous faults. Comparison with a simple least squares approach shows that the proposed approach needs drastically less iterations and therefore corresponds to a fast estimation results.