Lyapunov-based online parameter estimation in continuous fluidized bed spray agglomeration processes. Issue 1 (2019)

• Record Type: Journal Article
• Authors: Golovin, Ievgen; Otto, Eric; Dürr, Robert; Palis, Stefan; Kienle, Achim
• Is Part Of: IFAC-PapersOnLine. Volume 52:Issue 1(2019)
• Page Start: 329
• Page End: 334
• Publication Date: 2019
• Subjects: Adaptive identification -- distributed parameter systems -- population balance equations -- partial differential equations; Automatic control -- Periodicals; 629.805
• DOI: 10.1016/j.ifacol.2019.06.083
• Languages: English
• ISSNs: 2405-8963

Abstract

Abstract: Fluidized bed spray agglomeration is a particle formation process in many industrial applications, e.g. pharmaceutical and food processing. The properties of the formed agglomerates, like characteristic volume, significantly affect the product quality and can be affected by variation of certain operating parameters. Mathematical modeling not only provides an abstract characterization of the effects of those on the product properties but also supports thorough understanding of the underlying physical and chemical mechanisms. Moreover, it enables application of advanced process analysis, control and intensification schemes. As characteristic properties underlie variations within the ensemble of agglomerates the process can be described as a distributed parameter system, where the resulting model equations are partial differential equations. Adaption to experimental data requires the solution of inverse problems, which tend to be ill-conditioned. As an alternative approach, in this contribution an adaptive identification procedure is presented. Therefore, a modified plant model is run in parallel to the process and adaption rates are chosen based on a Lyapunov-function. The approach is validated in a parametric study for two scenarios: In the first, it is assumed that the structure of the dynamics is fully known, while in the second, this assumption does not hold. It is shown that the proposed approach allows to reconstruct unknown kinetic information of the process dynamics.