Optimization-based state estimation — A comparative study for the batch polycondensation of polyethyleneterephthalate

Due to the improvements in computational speed and the development of effective solvers for nonlinear optimization problems, optimization-based state estimation has become an interesting alternative to common approaches such as the Extended Kalman Filter. Nevertheless the computational effort to solve the optimization problem may cause problems in the online application to chemical processes, especially if the sampling rate is high. The method which is used to solve the dynamic optimization problem has also a great influence on the computational effort. We investigate a simultaneous and a sequential implementation of the Moving Horizon Estimator [9,8,10] and compare it to the Constrained Extended Kalman Filter using the example of the polycondensation of PET. From the results we conclude that for the process considered here the advantages of the moving horizon approach are marginal and therefore do not justify the high effort needed for the design and the calculation of the estimates compared to the relatively simple design and realization of the CEKF.