表面响应算法及其在MRI主磁体优化设计中的应用  

Response surface methodology and its application to optimization design of main magnet for MRI

作　　者：孙晓文[1] 谢德馨[1] 

机构地区：[1]沈阳工业大学电气工程学院,沈阳110023

出　　处：《沈阳工业大学学报》2007年第6期623-626,共4页Journal of Shenyang University of Technology

基　　金：国家自然科学基金资助项目(50377027)

摘　　要：通过将基于径向基函数的表面响应模型和改进的遗传算法相结合,提出了一种电磁场逆问题优化策略的新方法;该方法一方面利用表面响应模型能够大大节省目标函数评价的时间,另一方面通过改进遗传算法能加速收敛,从而节省了寻优时间.这一方法已应用于永磁完全开放式磁共振成像仪主磁体结构的优化设计.此外,给出了一种产生薄片型磁场的永磁全开放式磁共振成像仪主磁体结构.计算实例表明,所提出的算法能实现多变量寻优,且能大大节省计算时间,易于编程实现.Through the combination of response surface methodology （RSM） based on radial basis function （RBF） with improved genetic algorithm （GA）, a new optimization method of magnetic field inverse problem was presented. This method uses RSM to reduce the computation time of object function value, and simultaneously uses an improved GA to accelerate convergence. The optimization design of main magnet for completely open magnetic resonance imaging （MRI） devices was discussed. Furthermore, a kind of structure of permanent magnetic main magnet for completely open MRI devices was proposed. Numerical results on practical design problem indicate that the optimization method is high efficiency and easy programmable.

关 键 词：表面响应 模型 遗传算法 多变量 优化设计 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]