Receding horizon state estimation with measurements at different sampling intervals

State or parameter estimation is often part of control or optimization strategies in chemical processes. One of the more advanced techniques which is especially applicable to nonlinear state estimation is the moving horizon estimator (MHE). Muske et al. (1994) reformulated the batch-least-square estimation problem as the recursive MHE. Applications are seldom reported, as it is difficult to implement and requires a lot of online computing power. Nonetheless, this approach automatically suggests the inclusion of infrequently occurring measurements, as it utilizes the process history over a user defined horizon. This approach is different from earlier approaches as no interpolation between the infrequent measurements is used, which supplies "fake" measurements. We investigate this approach by the application to the example of the batch polycondensation of polyethyleneterephthalate.