ERT新型拓扑结构有限元模型优化  

Optimization of finite element model with novel topology in ERT

作　　者：肖理庆[1,2] 王化祥[2] 邵晓根[1] 

机构地区：[1]徐州工程学院虚拟现实与多维信息处理实验室,徐州221111 [2]天津大学电气与自动化工程学院,天津300072

出　　处：《仪器仪表学报》2014年第6期1298-1305,共8页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(61201350;61302122);江苏省高校自然科学研究面上项目(13KJD510007);中央高校基本科研业务费中国民航大学专项(3122013C007)资助项目

摘　　要：为了提高电阻层析成像正问题计算精度,提出一种新型拓扑结构有限元模型:模型内部区域采取非均匀分布形式,模型外部区域模拟敏感场均匀分布时电流线的分布形式与分布密度,并以模型外部区域的起始层数和模型各层半径为变量(最外层除外),利用改进遗传算法优化新型拓扑结构有限元模型。实验结果表明:敏感场均匀分布时,优化后新型拓扑结构有限元模型相比传统按等间隔原理剖分的有限元模型及其改进模型,模型均方根值分别降低了84.175%、23.643%、20.943%,有效提高了正问题计算精度,降低了图像相对误差。In order to improve the solution accuracy of the forward problem in electrical resistance tomography （ERT） , a finite element model with novel topology is proposed. The inner region of the finite element model is subdivided non-uniformly, and the outer region simulates the distribution and density of the current lines in uniformly distributed sensitivity field. Besides, the initial number of layers in the outer region and the radius of each layer （except for the outermost layer） are taken as variables, the modified genetic algorithm （GA） is adopted to optimize the finite element model with novel topology. Experiment results show that for homogeneous sensitivity field, the optimized finite element model with novel topology reduces the root mean square （RMS） value of the model by 84. 175% , 23. 643% and 20.943% compared with the conventional equally-spaced finite element model and its improved version, thus enhances the solution accuracy of the forward problem effectively, and reduces the relative image error.

关 键 词：电阻层析成像 正问题 拓扑结构 遗传算法 图像相对误差 

分 类 号：TH701[机械工程—仪器科学与技术]