表征超声衰减谱粒度的改进和声搜索算法*

针对超声衰减谱法颗粒粒径测量，提出一种改进和声搜索算法，在和声搜索算法迭代达到200次时，加入拟牛顿算法，调节和声搜索算法全局和局部搜索能力，提高算法的收敛精度。对服从三种典型粒径分布函数颗粒系进行数值模拟，改进算法反演得出R-R(Rosin-Rammler)分布参数值相对误差小于8%、正态分布参数值相对误差小于5%，对数正态分布参数值相对误差小于10%，同时用改进算法对两种硅-水悬浮液样品的实验超声衰减谱反演，与标称尺寸对比粒径相对误差为6.0%和8.2%，表明了利用改进算法表征颗粒粒径的可行性。

Application of improved harmony search algorithm in ultrasonic attenuation spectroscopy

For particle size measurement by ultrasonic attenuation spectroscopy, an improved harmony search algorithm is proposed. When the harmony search algorithm iteration reaches 200 times, the BFGS algorithm is added to adjust the global and local search capabilities of the harmony search algorithm to improve the convergence accuracy of the algorithm. Numerical simulation of the particle system subject to three typical particle size distribution functions, the improved algorithm inversion shows that deviation of the parameter values are less than 10%.Simultaneously, the experimental ultrasonic attenuation spectrum inversion of the two silicon-water suspension samples with the improved algorithm, compared with nominal size, the deviation are 6.0% and 8.2%.It is feasible to characterize particles with improved algorithms.