A weighted Bayesian inversion method is proposed for estimation of unimodal particle size distribution in multi-angle dynamic light scattering measurements. Particle size informiation distribution in autocorrelation function is used as base and adjustment parameters as exponent in the weight coefficient. Traditional Bayesian inversion method is recovered. Simulation and experimental data at different noise levels show that weighted Bayesian inversion method obtains inversion results with smaller distribution error. It weakens effectively influence of data noise and improves accuracy of particle size distribution.

A self-optimizing cumulative inversion for particle size distribution with photon correlation spectroscopy is proposed.It improves precision and stability of measurement.From a detailed fundamental of photon correlation spectroscopy and conventional cumulative inversion,we analyze instability of cumulative inversion.A self-optimizing method is presented by using the minimum mean-square error.It inverts particle size distribution stably with scattered autocorrelation curve with random noise.Experimental results illustrate validity and feasibility of the approach.

In the Sagaut mixed-scale model,an improved subgrid-scale(SGS) characteristic length,representing the scale of local energy-containing SGS motions,is proposed.Large-eddy simulations(LES) at temporal mixing layer and neutral atmospheric boundary layer indicate that by the improved SGS characteristic length,satisfactory spatial distribution of the SGS motions can be obtained and are closer to the results of direction numerical simulations(DNS).Compared with the basic model,the modified model with the improved SGS characteristic length produces more accurate mean quantities and turbulent statistics.The results demonstrate its valuable and wide application to LES.

According to the ENO scheme on structured grids, a class of weighted ENO finite volume scheme on unstructured mesh is developed. On every control volume, it constructs a new weighted quadratic reconstruction polynomial which can save computational costs. It also uses a method which can resolve the overdetermined systems and do not affect the accuracy of the schemes. Besides, the selection of interpolation points and the construction of weight are presented, Third order TVD Runge Kutta time discretization is used. In order to accelerate the convergence, local time step is introduced. The numerical experiments show the scheme effective.

It presents a class of the second order accurate explicit Gauss schemes with staggered grids for the computation of solutions of single hyperbolic conservation laws in two dimernsions, these schemes are Riemann solver-free and Maximum and Minimum Bounds under the restriction of CFL,and have been extended to system of hyperbolic conservation laws. Because these schemes are constructed under staggered grids and Riemann solver-free,the advantages of these schemes compared to other TVD schemes such as Harten's are:no complete set of eigenvectors is needed and hence weakly hyperbolic system can be solved,faster and programming is much simpler.The numerical solutions obtained in computation Riemann problem are satisfactory.