Research On The Optimization Method Of Inversion Algorithm In Polydisperse Nanoparticle Size Based On Regularization

Compared with the ordinary particles, nanoparticles with larger specific surface area and surface energy, the unusual features such as light, electricity, magnetic, force and other properties are widely used in medical, aerospace industry, environment protection, etc. These special characteristics have a close relation with nanoparticles size and distribution, so nanoparticles granularity measurement is one of the hot topics in the study of particle properties. Nanparticles granularity inversion is to solve the first kind Fredholm integral equation; it is the most difficult to inverse particle size and distribution by photon correlation spectroscopy (PCS).Therefore, in allusion to the emphasis and difficulty in nanoparticle size inversion, this paper mainly studies the granularity inversion of polydisperse nanoparticles, in order to improve the inversion accuracy and inversion speed, by optimizing the parameters of regularization to optimize the regularization method, and experimental results show that the proposed methods have reached the expected results and have an important significance in theory research.The main work of this paper is as follows:1. According to the characteristics of scattering light signal, this paper used AR model and L-D recursion method to study and simulate the dynamic light scattering signal. The analog signal intensity autocorrelation function compared with the theoretical value, reflected well consistence and verified the reliability and feasibility of the simulation signal by Tikhonov regularization algorithm. The interface was designed using VC++by Matlab experimental results. The scattering light signal analog system can be build.2. Monodisperse and polydisperse particles were inversed respectively by iterative regularization method under different noise level. The experimental results showed that, in the noiseless and the low-noise, using regularization and iterative regularization method, monodispersion and polydispersion were inversed very well; when the noise level was high, iterative regularization method in noise immunity and precision of inversion was superior to regularization method.3. According to the principle of Morozov deviation inversed nanoparticle size. The experimental results suggested that, for unimodal distribution of particles, when the noise level is less than 0.01, Morozov fast algorithm and Morozov algorithm could inverse good particle size distribution; when the noise level is equal to 0.01, only Morozov fast algorithm could get better particle size distribution. For multi-peak distribution of particles, when the noise level is less than 0.005, Morozov fast algorithm and Morozov algorithm could get better inversion of particle size distribution; when the noise level is equal to 0.005, only Morozov fast algorithm inversed the particle size distribution.4. To study the genetic algorithm in the application of nanoparticle size inversion combined with Morozov regularization method, improvement of regularization and L-curve method and so on, fitness functions were designed, searching for optimal method of multi-peak distribution inversion by selection, crossover and mutation.