基于Arnoldi算法的金属导体PEEC模型缩减方法  

PEEC Model Reduction of a Conductor Based on Arnoldi Algorithm

作　　者：潘启军[1,2] 马伟明[1,2] 赵治华[1,2] 孟进[1,2] 张磊[1,2] 张向明[1,2] 

机构地区：[1]海军工程大学电力电子技术研究所,武汉430033 [2]海军工程大学舰船综合电力技术国防科技重点实验室,武汉430033

出　　处：《电工技术学报》2008年第3期1-7,共7页Transactions of China Electrotechnical Society

基　　金：国家自然科学基金委员会创新研究群体科学基金(50421703);国家自然科学基金(50677070)资助项目

摘　　要：详细分析了渐近波形估算(AWE)方法的固有数值病态问题,提出使用基于Krylov子空间的Arnoldi算法对系统高阶模型进行缩减,将直接矩量显式计算转化为间接矩量隐式计算,将主导极点的抽取转化为主导特征值的抽取。然后给出了独立于频率之外的简化等效电路解耦模型,得到了解耦电阻电感参数,并通过计算实例验证了该方法的正确性。该研究成果克服了AWE等直接矩量显式计算方法只能提取有限几条缩减支路的不足,建立了模型缩减方法广泛使用的基础,具有重要的理论和工程应用价值。The intrinsic numerical ill-conditioning problem of asymptotic waveform evaluation（AWE） method was analyzed in this paper. The paper proposed that the Arnoldi algorithm in Krylov subspace was applied to reduce the high-order original system. The new algorithm transferred the moment direct explicit computation into indirect implicit calculation, and transferred extracting the dominant poles into picking up corresponding dominant eigenvalues. Then, the resistances and inductances of the reduced equivalent circuit decoupling model were obtained indepentdent of frequencies which was validated by examples. The achievements above avoid the defect only drawing several reduced branches by the moment direct explicit computation approaches such as AWE and settle the base of the widespread application of model reduction methodology, which is a most significant in both theory and Engineering.

关 键 词：PEEC模型 数值病态 间接矩量匹配 Amoldi算法 模型缩减 

分 类 号：TM93[电气工程—电力电子与电力传动]