Inversion of Particle Size Distribution by Spectral Extinction Technique Using the Attractive and Repulsive Particle Swarm Optimization Algorithm

The particle size distribution plays an important role in environmental pollution detection and human health protection, such as fog, haze, and soot. In this study, the attractive and repulsive particle swarm optimization algorithm and the basic particle swarm optimization were applied to retrieve the particle size distribution. The spectral extinction technique coupled with the anomalous diffraction approximation and the Lambert-Beer law were employed to investigate the retrieval of the particle size distribution. Three commonly used monomodal particle size distribution, i. e. the Rosin-Rammer distribution, the normal distribution, the logarithmic normal distribution were studied in the dependent model. Then, an optimal wavelengths selection algorithm was proposed. To study the accuracy and robustness of the inverse results, some characteristic parameters were employed. The research revealed that the attractive and repulsive particle swarm optimization showed more accurate and faster convergence rate than the basic, particle swarm optimization even with random measurement error. Moreover, the investigation also demonstrated that the inverse results of four incident laser wavelengths showed more accurate and robust than those of two wavelengths. The research also found that if increasing the interval of the selected incident laser wavelengths, inverse results would show more accurate, even in the presence of random error.