State estimation for processes with discrete-time and time-delayed measurements

Several sensors, such as analyzers, only provide discrete-time and time-delayed measurements with non neglectable sampling and delay times compared to dominant time constants of the process. This contribution describes a new model-based method for continuous-time estimation and predictive compensation of the time delay. The proposed method is applied to a polymerization reactor. An essential product quality of the polymerization product is measured by an analyzer. Based on these discrete-time and time-delayed data and the process model predictions, time-continuous and synchronized state estimates of the process can be calculated online. The method is verified by simulation of a grade change.