改进人工鱼群算法在气溶胶粒径分布反演中的应用

摘要

基于光谱消光法，利用一种改进的人工鱼群算法(improved artificial fish school algorithm，IAFSA)反演单峰和双峰气溶胶粒子粒径分布。其中，正问题的求解采用反常衍射近似(anomalous diffraction approximation，ADA)和Lambert−Beer定律。研究结果表明：相对标准人工鱼群算法而言，IAFSA能很好地避免出现后期收敛速度慢、精度低等问题。在非独立模型下反演气溶胶粒径分布时，IAFSA算法体现出很好的收敛精确度和鲁棒性。在非独立模型下，IAFSA算法可用于反演实际测量得到的哈尔滨地区气溶胶粒径分布。%Based on the spectral extinction method, an improved artificial fish school algorithm (IAFSA) was employed to retrieve the monomodal and bimodal aerosol size distributions. The direct problem was solved by the anomalous diffraction approximation (ADA) and Lambert−beer law. The results show that IAFSA can overcome low convergence rate and accuracy, compared with AFSA. Moreover, in retrieving the aerosol size distribution with the dependent model, the IAFSA shows satisfactory convergence accuracy and robustness. IAFSA can be employed to estimate the actual aerosol size distribution over Harbin with the independent model.