弥散颗粒介质辐射特性参数反演

发布时间：2018-03-30 01:06

本文选题：有限元法　切入点：GMM理论　出处：《南京理工大学》2017年硕士论文

【摘要】：弥散颗粒介质与日常生活息息相关,其辐射特性被广泛应用于工业生产、国防等各个方面。本文针对弥散颗粒介质的辐射特性参数反演进行了研究,根据反问题的特点,从弥散颗粒介质的辐射正问题计算和反问题求解方法两个方面入手,主要完成了以下工作:在正问题方面,首先对颗粒的散射特性进行了研究。采用COMSOL有限元方法求解了非球形颗粒的散射特性,并采用GMM理论结合CCA模型建立了随机团聚颗粒散射求解模型。在研究范围内的计算结果显示随尺度参数的变化颗粒的形状及团聚状态都对颗粒的散射特性有影响。在求解颗粒散射特性基础上,在MATLAB平台建立了在非参与性介质中的弥散颗粒介质的辐射传输模型,包括蒙特卡洛方法模拟的半球透射率、半球反射率模型以及基于BEER定律的准直透射率模型,为反问题求解提供正向计算模型。在反问题求解方面,首先基于L-M梯度优化方法,以透射光谱为测量量,根据颗粒散射求解理论结合K-K关系在MATLAB环境下反演了颗粒材料的复折射率。对K-K关系中非测量区间外推并且对柯西主值积分采取交错离散格式避免奇点影响。分别采用COMSOL livelink with MATLAB模块和GMM可执行文件将颗粒散射求解方法与反演算法结合,反演了不同形状及团聚态飞灰颗粒复折射率。然后以半球透射、半球反射、准直透射为测量量,基于BP神经网络方法建立了弥散颗粒介质辐射特性参数反演模型和团聚颗粒复折射率反演模型,并分析了网络参数和测量误差对反演结果的影响。在散射不对称因子接近于0时,测量误差对散射不对称因子反演准确性有很大影响,对吸收系数和散射系数影响较小。
[Abstract]:Dispersion granular media is closely related to daily life, and its radiation characteristics are widely used in industrial production, national defense and other aspects. In this paper, the parameter inversion of radiation characteristics of dispersive granular media is studied, according to the characteristics of inverse problem. From the two aspects of the calculation of the radiation positive problem and the inverse problem solving method of the dispersive particle medium, the following work has been accomplished: in the positive problem aspect, Firstly, the scattering characteristics of particles are studied, and the scattering characteristics of non-spherical particles are solved by COMSOL finite element method. The scattering solution model of random agglomeration particles is established by using GMM theory and CCA model. The calculated results show that the shape and agglomeration state of particles with the change of scale parameters have an effect on the scattering characteristics of particles. On the basis of solving the scattering properties of particles, The radiation transfer model of dispersible granular media in non-participatory media is established on MATLAB platform, including the hemispherical transmittance model, the hemispherical reflectivity model and the collimation transmission model based on BEER's law, which are simulated by Monte Carlo method. In the inverse problem solving, firstly, based on L-M gradient optimization method, the transmission spectrum is taken as the measurement. Based on particle scattering theory and K-K relation, the complex refractive index of granular material is inversed in MATLAB environment. The non-measurement interval extrapolation of K-K relation and the staggered discrete scheme for Cauchy principal value integral are adopted to avoid singularities. Instead of using the COMSOL livelink with MATLAB module and the GMM executable file to combine the particle scattering solution method with the inversion algorithm, The complex refractive index of fly ash particles with different shapes and agglomerations is inversed. The measurements are hemispherical transmission, hemispherical reflection and collimation transmission. Based on BP neural network method, the inversion model of radiation characteristic parameters and the complex refractive index inversion model of agglomerated particles are established. The effects of network parameters and measurement errors on the inversion results are analyzed. When the scattering asymmetry factor is close to zero, the measurement error has a great influence on the inversion accuracy of the scattering asymmetry factor, but has little effect on the absorption coefficient and the scattering coefficient.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TB30

【相似文献】