A hybrid artificial bee colony algorithm and pattern search method for inversion of particle size distribution from spectral extinction data. Issue 19 (28th October 2017)

• Record Type: Journal Article
• Authors: Wang, Li; Li, Feng; Xing, Jian
• Is Part Of: Journal of modern optics. Volume 64:Issue 19(2017)
• Page Start: 2051
• Page End: 2065
• Publication Date: 2017-10-28
• Subjects: Spectral extinction technique -- particle size distribution -- artificial bee colony -- pattern search -- inversion; Optics -- Periodicals; 535
• DOI: 10.1080/09500340.2017.1337250
• Languages: English
• ISSNs: 0950-0340

Abstract

Abstract: In this paper, a hybrid artificial bee colony (ABC) algorithm and pattern search (PS) method is proposed and applied for recovery of particle size distribution (PSD) from spectral extinction data. To be more useful and practical, size distribution function is modelled as the general Johnson'sS B function that can overcome the difficulty of not knowing the exact type beforehand encountered in many real circumstances. The proposed hybrid algorithm is evaluated through simulated examples involving unimodal, bimodal and trimodal PSDs with different widths and mean particle diameters. For comparison, all examples are additionally validated by the single ABC algorithm. In addition, the performance of the proposed algorithm is further tested by actual extinction measurements with real standard polystyrene samples immersed in water. Simulation and experimental results illustrate that the hybrid algorithm can be used as an effective technique to retrieve the PSDs with high reliability and accuracy. Compared with the single ABC algorithm, our proposed algorithm can produce more accurate and robust inversion results while taking almost comparative CPU time over ABC algorithm alone. The superiority of ABC and PS hybridization strategy in terms of reaching a better balance of estimation accuracy and computation effort increases its potentials as an excellent inversion technique for reliable and efficient actual measurement of PSD.