TSVD-Tikhonov正则化多尺度动态光散射反演

发布时间：2018-08-23 10:55

【摘要】：在大反演范围内,小波正则化反演(WRIM)是一种改进动态光散射(DLS)数据反演精度的有效方法,但该方法对强噪声及双峰颗粒数据的反演精度偏低。在WRIM基础上,结合传统的单尺度Tikhonov、截断奇异值分解(TSVD)正则化在动态光散射反演中的优点,提出了一种Tikhonov-TSVD-WRIM(TTWRIM)多尺度动态光散射反演方法。该方法首先将Tikhonov用于粗尺度反演范围的自适应调整,然后将TSVD用于细尺度反演,并对反演结果进行五点三次平滑处理。在0.001、0.005及0.01三种噪声下,分别采用Tikhonov、TSVD、WRIM及TTWRIM四种方法对模拟数据进行了反演,结果表明,TTWRIM反演精度更高、抗干扰能力及双峰分辨力更强。最后,实测颗粒的反演也验证了模拟数据的结论。在[1nm,2000nm]的反演范围内,TTWRIM对300nm单峰及100~500nm双峰实测颗粒的反演峰值误差分别为0.18%和2.81%。
[Abstract]:Wavelet regularization inversion of (WRIM) is an effective method to improve the inversion accuracy of dynamic light scattering (DLS) data in large inversion range, but the inversion accuracy of this method is low for strong noise and bimodal particle data. Based on WRIM and the advantages of the traditional single-scale Tikhonov-truncated singular value decomposition (SVD) (TSVD) regularization in dynamic light scattering inversion, a multi-scale Tikhonov-TSVD-WRIM (TTWRIM) dynamic light scattering inversion method is proposed. In this method, Tikhonov is first used to adjust the range of coarse scale inversion, then TSVD is applied to fine scale inversion, and the inversion results are smoothed for 5.3 times. The simulated data are retrieved by Tikhonovan TSVDW RIM and TTWRIM respectively under three kinds of noise: 0. 001 and 0. 01 respectively. The results show that TTWRIM is more accurate, anti-jamming and bimodal resolution is stronger. Finally, the inversion of measured particles verifies the conclusion of the simulation data. In the inversion range of [1 nm], the peak error of 300nm single peak and 100~500nm double peak measured by TTWRIM is 0.18% and 2.81%, respectively.
【作者单位】：山东理工大学电气与电子工程学院;天津大学电子信息工程学院;
【基金】：国家自然科学基金(61601266) 山东省自然科学基金(ZR2014FL027,ZR2015FL034,ZR2016EL16)
【分类号】：O436.2

【参考文献】