中文摘要：

在超声衰减谱法测量颗粒粒径分布的反问题求解中,反演算法至关重要,遗传算法作为一种全局最优化算法,其参数优化对于颗粒粒径分布反演效果极其关键.从最大代数、种群尺度、交叉和变异概率等参数优化角度,对于服从3种典型分布函数的颗粒系进行数值模拟.研究发现,当最大代数为300、种群尺度为60、交叉概率范围为0.4~0.85、变异概率为0.045~0.08时,获得的反演结果与设定值较吻合.在此优化参数下的数值算例和对两种玻璃微珠-甘油悬浮液样品的实验超声衰减谱反演,进一步验证了此优化参数下遗传算法具有较好的稳定性和抗噪性.

英文摘要：

In the inversion problem of particle sizing by ultrasonic attenuation spectroscopy,the inversion algorithm plays an extremely important role.As the genetic algorithm is a global optimization algorithm, it’s highly pivotal to optimize the parameters in genetic algorithm for the inversion effects of particle size distribution.After optimizing the parameters of maximum generation number,population size and genetic operators （crossover fraction and mutation fraction）,a series of particle systems with three typical distribution functions were numerically calculated.The numerical simulation reveals that a relatively anastomotic inversion result can be obtained using ultrasonic spectroscopy,as compared to the input particle size distribution,when the maximum generation number is 300,the value of population size is 60 and the ranges for crossover fraction and mutation fraction are 0.4~0.85 and 0.045~0.08 respectively.Finally, the stability and noise immunity of the genetic algorithm based on the optimization of parameters were validated through the numerical and experimental examples of two kinds of micron-sized glass beads-glycerol suspension samples.