Research On Selection Of Regularization Parameters In Regularization Inversion Of Dynamic Light Scattering Of Flowing Particles

Dynamic light scattering technology is an effective method to measure the particle size of submicron and nano particles.It has been widely developed and applied in the field of particle measurement.Regularization method is a commonly used inversion method in dynamic light scattering particle measurement technology.Regularization parameter selection is the focus of regularization inversion.The scattering intensity of dynamic light scattering by flowing particles is affected by Brownian motion and directional translation motion at the same time,and its inversion is more complex.In order to improve the accuracy and stability of flow particle inversion results,the selection criteria of regular parameters are studied in this paper.The main research contents include:1.Explore the scope of application of GCV(Generalized Cross-Validation)criterion and L-curve criterion.Taking the flowing aerosol as the experimental object,the GCV criterion and L-curve criterion are compared and analyzed by Tikhonov regularization method for the distribution of different widths,different noise levels and different velocities.The inversion results show that for narrow and medium distribution unimodal / bimodal aerosol particles,when the noise is small,the inversion results obtained by using the regularization parameters selected by GCV criterion are better than L-curve criterion.When the noise is large,the inversion effect of L-curve criterion is better than GCV criterion.The inversion effect of Lcurve criterion in wide distribution particle inversion is better than GCV criterion.The influence of the increase of velocity on the measurement is similar to adding signal noise.With the velocity increase,the relative error and peak position error of the two criteria show an increasing trend.2.Aiming at the problem of low inversion accuracy caused by improper selection of GCV criterion,a method for determining circulation selection GCV(C-GCV)is proposed.Based on the GCV criterion of weighted residual norm optimization and considering the influence of relative error,the iterative interval is generated according to the GCV regular parameters.Through the cyclic inversion of multiple regular parameters in the interval,the weak symmetric relationship between relative error and norm is explored,and the optimal regular parameters are determined by the principle of maximum solution norm.This method takes into account the influence of weighted residual norm and solution error,and overcomes the problem of large inversion error caused by single weighted residual norm optimization.Simulation and experimental results show that C-GCV can effectively reduce the relative error of GCV criterion inversion results.3.Aiming at the peak width adaptability of different regularization parameter selection criteria,an adaptive regularization parameter selection method based on peak width is proposed.This method considers that C-GCV and second-order multi parameter criterion(SM)are suitable for narrow peak and wide peak distribution respectively.Based on the judgment of wide and narrow peak of attenuation rate,C-GCV and SM are adaptively selected to optimize the regularization parameters.This method considers the adaptability of the criterion and avoids the problem of low inversion accuracy caused by the fixed regular parameters in the process of particle size inversion.Simulation and experimental results show that the adaptive regularization parameter selection method can effectively improve the accuracy of particle size inversion.For the regularization inversion of dynamic light scattering from flowing particles,the selections of regularization parameters is the key factor affecting the accuracy of the inversion.So far,there are few studies on the flowing particle size distribution inversion by dynamic light scattering regularization method,and can not give satisfactory results.In this paper,the selection method of dynamic light scattering regularization parameters of flowing particles is studied,which provides a reference for the subsequent measurement of dynamic light scattering flowing particles.