个人信息Personal Information
教授
博士生导师
硕士生导师
性别:男
出生日期:1972-11-18
毕业院校:大连理工大学
学位:博士
所在单位:力学与航空航天学院
学科:工程力学. 计算力学. 动力学与控制
办公地点:综合实验1号楼513
联系方式:手机号码: 13942024929; 微信号码: 13942024929;
电子邮箱:zhaogz@dlut.edu.cn
An Efficient and Robust Mixed-Gauge Scheme Based on Current-Splitting for Nonlinear Magnetostatic Field Computation
点击次数:
论文类型:期刊论文
发表时间:2020-03-01
发表刊物:IEEE TRANSACTIONS ON MAGNETICS
收录刊物:EI、SCIE
卷号:56
期号:3
ISSN号:0018-9464
关键字:Edge element method; gauge scheme; magnetic vector potential; magnetostatic problem
摘要:In the edge element discretization for magnetostatic problems, a gauge scheme, such as tree gauge, Lagrangian multiplier (LM) gauge, and auto gauge, is usually adopted to handle the singular curl-curl equation in terms of the magnetic vector potential. However, tree-gauge and LM-gauge schemes are not very efficient as a nonlinear problem with many increments has to be resolved. In order to overcome those difficulties, a mixed-gauge scheme based on current-splitting is proposed in this article. First, we make an improvement for LM-gauged formulation and adapt it to calculate the initial increment with a specially designed CG solver; then split the discrete source current density into a compatible part and incompatible part based on discrete Helmholtz decomposition; auto gauge cooperates with the compatible part to calculate the following increments. The mixed-gauge scheme combines the advantages of LM-gauge and auto-gauge schemes, which avoids the step to determine the source term, and at the same time, has high computational efficiency. The mixed-gauge scheme is tested in some examples against some other gauge schemes and found to be efficient and robust, where others possibly converge slowly or fail to produce a reliable solution.