An inverse simulation for simultaneous identification of randomly oriented arbitrarily shaped particle size distribution and its degree of non-sphericity from spectral transmittance measurement

Abstract

An experimental model is suggested and integrated with a numerical scheme for simultaneous identification of arbitrary shape particle size distribution (PSD) and its degree of non-sphericity present in a system. In this spectroscopic instrumentation, the transmittance intensity of light is measured by calculating the bulk absorption coefficient of randomly oriented arbitrarily shaped particles by adopting parameterized anomalous diffraction theory. The effects of reflection and refraction of arbitrary shapes of particles are also considered. Multi-wavelength transmittance signals (estimated and measured values) are used to fit the objective functions for the identification of target parameters. The unique feature of this simulation is the simultaneous identification of the joint function PSD and the shape perimeter of arbitrary particles using the joint log-normal distribution function. A probability-density-function-based ant colony optimization (PDF-ACO) algorithm is carried out for the inverse simulation. The parameters retrieved by the multi-wavelength transmittance signals show good agreement with the set original values of the target parameters. The robustness of the PDF-ACO algorithm could successfully keep the retrieval errors of the estimated parameters within the tolerable limit (error < 10%) even at high noise in the system.