基于量子粒子群算法的激光粒度检测的研究  

Study on Laser Particle Size Detection Based on Quantum-behaved Particle Swarm Optimization Algorithm

作　　者：曹丽霞[1] 赵军[1] 单良[2] 郭天太[1] 孔明[1] 

机构地区：[1]中国计量学院计量测试工程学院,浙江杭州310018 [2]中国计量学院信息工程学院,浙江杭州310018

出　　处：《应用激光》2015年第3期380-386,共7页Applied Laser

基　　金：国家自然科学基金资助项目(项目编号:51476154;51404223);浙江省自然科学基金资助项目(项目编号:LQ14E060003;LY13E060006);能源清洁利用国家重点实验室开放基金资助项目

摘　　要：在工业许多领域上,对于颗粒粒度的快速、准确的在线测量的需求越来越紧迫。这对反演算法的质量如鲁棒性、运行效率、重复性及精度等提出了更高的要求。将正则化的目标函数与量子粒子群算法相结合,并用于光散射法的非独立模式下的颗粒粒径的反演,将量子粒子群算法与基本粒子群算法、模拟退火算法进行对比。在仿真方面,在不同随机噪声下,对均匀球形单峰颗粒粒径进行了模拟仿真。当随机噪声为5%时,反演误差在10%以内。在实验方面,搭建了基于CCD为探测器的激光粒度检测系统,对国家标准颗粒进行了反演计算,仿真及实验结果表明量子粒子群算法具有全局性、收敛速度快、鲁棒性好等优点,且反演时间约为1s,适合快速、高精度的在线颗粒粒度测量。The need of the fast and accurate online measurement of particle size is more and more urgent in many industrial fields. It is need to put forward higher requirements about the quality of the inversion algorithm,such as robustness,efficiency,repeatability and accuracy,etc. Quantum-behaved Particle Swarm Optimization（QPSO） is proposed to retrieve particle size distribution（PSD）in the dependent mode. Comparison is made between the proposed algorithm and the conventional algorithm,Simulated Annealing algorithm. The Simulation is performed to verify the effectiveness of QPSO,in which the spheroidal particle of unimodal distribution is retrieved at different levels of random noise. Inversion errors are within 10% when 5% random noise is added. Experiment is also performed to verify the practicability of QPSO,in which the laser particle size detection system based on CCD is built. Standardized polystyrene microsphere is tested and calculated. Both simulation and experiment results indicate that QPSO has good global superiority,speed and good robustness. Moreover,the inversion time is about 1 second,which is suitable for fast,accurate online particle size measurement.

关 键 词：反演 颗粒粒径 鲁棒性 量子粒子群算法 激光粒度检测系统 

分 类 号：O436[机械工程—光学工程]