Soft Sensor for On-Line Monitoring of Particle Sizes in Miniemulsion Polymerization

Abstract

This paper proposes a soft-sensor (SS) for on-line monitoring the droplet/particle size distribution (PSD) in miniemulsion polimerizations. The SS utilizes turbidity measurements and a global particle refractive index (PRI) estimated on the basis of the instantaneous polymer conversion and the PRIs of the monomer and the polymer. The proposed method requires solving an illconditioned inverse problem (ICIP). Two different approaches are proposed for solving the ICIP: i) a Tikhonov regularization (TR), and ii) a General Regression Neural Network (GRNN). Both approaches are evaluated on the basis of simulated examples corresponding to a styrene miniemulsion polymerization. For unimodal PSDs, both TR and GRNN produce acceptable estimates of the average diameters of the PSD along the polymerization. The TR method produces PSDs with spurious peaks, while the GRNN allows better estimates. For bimodal PSDs both methods produce acceptable average diameters but erroneous PSDs. Simulation results suggest that the proposed method is a robust tool for on-line monitoring of the average diameters in miniemulsion polymerizations.