Robust state estimation for repetitive operating mode process: Application to sequencing batch reactors

The issue of this paper is related with the on-line state estimation of a class of sequencing batch reactors (SBR). This operating mode of SBR can be analyzed as a repetitive process, where some operating conditions changes from one batch to other, which leads to nominal model degradation of the plant and consequently a bad estimation performance when standard observers are employed. To avoid the problems above mentioned it is proposed a finite time convergence observer with a fractional power of the time estimation error plus a discrete integral-type contribution of the discrete estimation error in order to reach finite time convergence and compensate the modeling error which arises when each batch is processed. The proposed observer is applied to a class of simple bioreactor model experimentally corroborated, where numerical simulations show the satisfactory performance of the proposed methodology in comparison with standard nonlinear Luenberger observers. Mathematical proof of the convergence of the proposed observer is addressed.