前向散射颗粒粒径测量中的多参数正则化算法  

Multi-Parameter Regularization Algorithm in Particle Size Measurement of Forward Light Scattering

作　　者：林承军[1] 沈建琪[1] 王天恩[1] Lin Chengjun Shen Jianqi Wang Tianen(College of Science, University of Shanghai for Science and Technology, Shanghai 200093, Chin)

机构地区：[1]上海理工大学理学院,上海200093

出　　处：《中国激光》2016年第11期165-173,共9页Chinese Journal of Lasers

基　　金：国家自然科学基金(51476104)

摘　　要：反演算法是光散射颗粒测试技术中的关键问题之一,以Tikhonov正则化方法为代表的单参数正则化算法被广泛应用于激光粒度仪颗粒粒径分布函数(PSD)的反演计算中。该算法的缺点之一是所得到的反演解呈现出振荡特征,并伴随负值。为改善这一状况,提出了一种多参数正则化算法。通过构建一个由多个参数控制的带通滤波函数,分别控制正则化解的振荡程度和解的高度,并对正则化解进行非负约束。模拟计算和实验研究结果表明,对多参数进行优化后能够降低正则化算法带来的振荡和负值。此外,所提出的算法具有较好的多峰识别能力,可实现颗粒粒径分布的有效重建。Inverse algorithm is one of the key issues in particle measurement technology of light scattering. Single parameter regularization algorithm, especially Tikhonov regularization algorithm is widely used for back calculation of particle size distribution （PSD） in laser particle analyzer. One of the disadvantages oI such algorithm is that the inversion solution is usually oscillatory and contains negatives values. To improve the situation, a multi-parameter regularization algorithm is proposed. By constructing a band-pass filter function that is controlled by multiple parameters, oscillations and the height of the regularization solution can be controlled, respectively. Moreover, the regularization solution is nonnegative constrained. Simulated results and experimental evidences show that, with the optimized parameters, the algorithm can eliminate the oscillations and negative values brought by regularization algorithm. Meanwhile, the proposed algorithm has better ability to recognize the multimodal system, and exhibits high ability to effectively re-construct the PSD.

关 键 词：测量 颗粒粒径 正则化 反演计算 

分 类 号：O436[机械工程—光学工程]