基于代理模型技术的支架-球囊系统优化设计  

Optimization of stent-balloon system based on surrogate modeling technique

作　　者：李红霞[1] 高月华[2] 王希诚[1] 

机构地区：[1]大连理工大学运载工程与力学学部,工业装备结构分析国家重点实验室,大连116024 [2]大连交通大学交通运输工程学院,大连116028

出　　处：《医用生物力学》2013年第2期210-215,共6页Journal of Medical Biomechanics

基　　金：国家自然科学基金资助项目(11072048);国家重点基础研究发展计划973项目基金(2009CB918501)

摘　　要：目的将改进的Kriging优化算法引入支架-球囊系统的优化设计,从而给其配以适当长度的球囊,使支架能够沿轴向扩张均匀。方法基于有限元计算结果,采用与拉丁超立方(latin hypercube sampling,LHS)抽样方法、期望提高(expected improvement,EI)函数相结合的Kriging优化算法,以减小支架扩张时的狗骨头效应为目的,对球囊长度进行优化设计。结果 Kriging代理模型能近似建立支架狗骨头率与球囊长度间的函数关系,来替代高成本的计算。LHS抽样方法可以确保产生的样本点代表向量空间的所有部分。EI函数能有效地用于平衡搜索,从而找到最优解。优化长度的球囊可以使支架均匀扩张。结论该优化算法能有效地应用于支架—球囊系统的优化设计。Objective To make the stent expand uniformly along the axial, an improved kriging optimization algo- rithm is applied to the optimal design of stent-balloon system to find the optimal length of the balloon. Methods Based on finite element method results, Kriging surrogate model combining with Latin hypercube sampling （LHS） approach and expected improvement （El） function was employed for the optimization of balloon length to reduce stent dog-boning effect during its dilation. Results The Kriging surrogate model could approximately establish the relationship between stent dog-boning rate and balloon length and so to replace the expensive reanalysis of stent dilation. Sample points from LHS could perfectly represent the vector space. El function could be used to effectively find out the global optimal solution with high probability. The optimal length of balloon could make the stent expand uniformly. Conclusions This adaptive optimization method based on Kriging surrogate model can optimize the design of stent-balloon system effectively.

关 键 词：支架 有限元分析 Kriging代理模型 期望提高函数 优化 

分 类 号：R318.01[医药卫生—生物医学工程]