Fixed-bed reactor kalman filtering and optimal control—I Computational comparison of discrete vs continuous time formulation

Abstract

A comparison is performed between the discrete and continuous implementation of Kalman filtering and optimal control for a stationary process. Theoretical results for a pilot-plant fixed-bed reactor are provided as an example. It is demonstrated that the computational effort for the two implementations differs only slightly, and that the discrete version provides more realistic results at low sampling rates. On the other hand, it is also shown that the continuous formulation may be used down to relatively low sampling rates, provided the filter parameters are adjusted appropriately. It is demonstrated that the discrete formulation may give an oscillatory behavior of the axial temperature variance profile. Some experimental results are provided in the adjoining paper [1].