基于云计算的电磁问题并行计算方法

摘要
图/表
参考文献
相关文章 (15)

全文: PDF (17871 KB)
输出: BibTeX | EndNote (RIS)

摘要针对电工装备性能分析与优化所需的易用高性能计算问题,使用云计算技术搭建了弹性集群,实现了典型电磁问题在弹性集群中的并行计算。使用虚拟化技术将计算机资源整合为资源池,搭建并部署了可实现弹性计算的云平台及并行计算所需的弹性集群。计算节点通过千兆路由器相连,节点之间使用SSH通信协议。选取感应电机和变压器作为计算案例,使用Fortran编写静磁场计算程序。对云计算的并行架构OpenMpi与MapReduce进行了分析研究,使用OpenMpi并行化方法实现了数百万计算节点的并行计算。并通过与商业软件计算结果的对比验证了计算的正确性。研究结果表明,可在云计算的弹性集群上使用OpenMpi实现大规模并行计算。相比超级计算机提供的并行计算方案,本文使用云计算建立的弹性机群具有便捷、按需、可配置的优点,这为高性能计算提供了一种更易获取、易用的解决方法,为开展复杂真实机构的集成计算提供了高性能计算的理论和实践依据。

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	金亮
	邱运涛
	杨庆新
	牛萍娟
	祝丽花

关键词 ：云计算, 并行计算, 电磁场数值计算, 有限元法

Abstract：High performance and easy calculation are required for the performance analysis and optimization of electrical equipment. For this reason, a kind of elastic cluster is built using cloud computing and parallel computing, which the typical electromagnetic problems can be achieved. The virtualization technology is used to integrate the computer resources into virtual resource pool. Thus, the cloud platform for elastic computing and a kind of elastic cluster for parallel computing can be built and implemented. Computing nodes are connected by Gigabit router, and SSH communication protocol is used between nodes. Induction motor and transformer are chosen as the calculation cases. The calculation programs of static magnetic field are written by Fortran. Cloud computing parallel architecture of OpenMpi and MapReduce are analyzed. The parallel calculation including millions of meshing scales are realized based on OpenMpi. The correctness of the calculation is verified by the result comparisons from a commercial software. It is shown that larger scale of parallel computing can be realized by using OpenMpi in the elastic cluster of cloud computing. Compared with the calculation schemes of parallel supercomputers, the built elastic cluster using cloud computing in this paper is more convenient and configurable, which provides a more accessible and easy-to-use solution for high performance computing. The proposed method also provides the theoretical and practical basis for the integrated calculation theory and method in complex real model.

Key words： Cloud computing parallel computing numerical calculation of electromagnetic fields finite element method

收稿日期: 2016-03-01 出版日期: 2016-12-02

PACS:

TM153

基金资助:国家自然科学基金青年项目（51207105、51507110）,国家自然科学基金面上项目（51577132）和国家火炬计划（2015GH611592）资助项目

作者简介: 金亮男,1982年生,博士,副教授,研究方向为工程电磁场与磁技术、电磁场云计算和电磁无损检测等。E-mail:jinliang@tjpu.edu.cn;牛萍娟女,1973年生,教授,研究方向为高性能计算、第三代半导体器件、半导体光源与照明系统等。

引用本文:

金亮, 邱运涛, 杨庆新, 牛萍娟, 祝丽花. 基于云计算的电磁问题并行计算方法[J]. 电工技术学报, 2016, 31(22): 5-11. Jin Liang, Qiu Yuntao, Yang Qingxin, Niu Pingjuan, Zhu Lihua. A Parallel Computing Method to Electromagnetic Problems Based on Cloud Computing. Transactions of China Electrotechnical Society, 2016, 31(22): 5-11.

链接本文:

https://dgjsxb.ces-transaction.com/CN/Y2016/V31/I22/5

[1] 侯淑萍, 杨庆新, 陈海燕, 等. 新型高性能电工材料应用特性建模的模块化设计[J]. 电工技术学报, 2008, 23(10): 1-5.
Hou Shuping, Yang Qingxin, Chen Haiyan, et al. Modularity design of modeling for new high performance materials applied to electrical engin- eering[J]. Transactions of China Electrotechnical Society, 2008, 23(10): 1-5.
[2] 谢德馨, 程志光, 杨仕友, 等. 对当前计算电磁学发展的观察与思考-参加COMPUMAG2011会议有感[J]. 电工技术学报, 2013, 28(1): 136-141.
Xie Dexin, Cheng Zhiguang, Yang Shiyou, et al. Observation and consideration on the current deve- lopment of computational electromagnetics-in perspec- tive of COMPUMAG2011[J]. Transactions of China Electrotechnical Society, 2013, 28(1): 136-141.
[3] 袁中胜. 新能源汽车驱动用永磁同步电机的设计[D]. 哈尔滨: 哈尔滨工业大学, 2013.
[4] 杨亮. 云计算与电磁仿真结合的可行性分析[J]. 电子技术与软件工程, 2014(2): 29-30.
Liang Yang. Feasibility analysis for the combination of cloud computing and electromagnetic simula- tion[J]. Electronic Technology & Software Engin- eering, 2014(2): 29-30.
[5] 唐任远, 吴东阳, 谢德馨. 单元级别并行有限元法求解工程涡流场的关键问题研究[J]. 电工技术学报, 2014, 29(5): 1-8.
Tang Renyuan, Wu Dongyang, Xie Dexin. Research on the key problem of element by element parallel FEM applied to engineering eddy current analysis[J]. Transactions of China Electrotechnical Society, 2014, 29(5): 1-8.
[6] 管建和. 电磁场有限元法解释分布式并行计算的研究[D]. 北京: 中国地质大学, 2006.
[7] 叶茂, 缪纶, 王志璋, 等. 基于LINUX和MPICH2的高性能科学计算集群搭建及其性能评测[J]. 中国水利水电科学研究院学报, 2009, 7(4): 302-306.
Ye Mao, Miao Lun, Wang Zhizhang, et al. The establishment and performance evaluation of high performance scientific computing cluster based on LINUX and MPICH2[J]. Journal of China Institute of Water Resources and Hydropower Research, 2009, 7(4): 302-306.
[8] 王海涛, 刘淑芬. 基于Linux集群的并行计算[J]. 计算机工程, 2010, 36(1): 64-66.
Wang Haitao, Liu Shufen. Parallel computing based on Linux cluster[J]. Computing Engineering, 2010, 36(1): 64-66.
[9] Mateescu Gabriel, Gentzsch Wolfgang, Ribbens Calvinj. Hybrid computing-where HPC meets grid and cloud computing[J]. Future Generation Computer Systems-The International Journal of Grid Com- puting and Escience, 2011, 27(5): 440-453.
[10] Ari Ismail, Muhtaroglu Nitel. Design and implement- ation of a cloud computing service for finite element analysis[J]. Advances in Engineering Software, 2013(60-61): 122-135.
[11] 符伟. 云计算环境下的线性有限元方法研究[D]. 武汉: 华中科技大学, 2013.
[12] 陈国良, 毛睿, 蔡晔. 高性能计算及其相关新兴技术[J]. 深圳大学学报(理工版), 2015, 32(1): 25-31.
Chen Guoliang, Mao Rui, Cai Ye. High performance computing and related new technologies[J]. Journal Shenzhen University Science and Engineering, 2015, 32(1): 25-31.
[13] 张少敏, 赵硕, 王保义. 基于云计算和量子粒子群算法的电力负荷曲线聚类算法研究[J]. 电力系统保护与控制, 2014, 42(21): 93-98.
Zhang Shaomin, Zhao Shuo, Wang Baoyi. Research of power load curve clustering algorithm based on cloud computing and quantum particle swarm optimization[J]. Power System Protection and Control, 2014, 42(21): 93-98.
[14] Peterson B, Baumgartner G, Qingyang Wang. A hybrid cloud framework for scientific computing[C]// IEEE 8th International Conference on Cloud Com- puting, 2015: 373-380.
[15] 王意洁, 孙伟东, 周松, 等. 云计算环境下的分布存储关键技术[J]. 软件学报, 2012, 23(4): 962-986.
Wang Yijie, Sun Weidong, Zhou Song, et al. Key technologies of distributed storage for cloud com- puting[J]. Journal of Software, 2012, 23(4): 962-986.
[16] 李成华, 张新访, 金海, 等. MapReduce: 新型的分布式并行计算编程模型[J]. 计算机工程与科学, 2011(3): 129-135.
Li Chenghua, Zhang Xinfang, Jin Hai, et al. MapReduce: a new programming model for distri- buted parallel computing[J]. Computing Engineering & Science, 2011(3): 129-135.
[17] 郭本俊, 王鹏, 陈高云, 等. 基于MPI的云计算模型[J]. 计算机工程, 2009, 35(24): 84-86.
Guo Benjun, Wang Peng, Chen Gaoyun, et al. Cloud computing model based on MPI[J]. Computing Engineering, 2009, 35(24): 84-86.
[18] 赵俊华, 文福拴, 薛禹胜, 等. 云计算:构建未来电力系统的核心计算平台[J]. 电力系统自动化, 2010, 34(15): 1-8.
Zhao Junhua, Wen Fushuan, Xue Yusheng, et al. Cloud computing: build the core computing paltform for the future electric power system[J]. Automation of Electric Power Systems, 2010, 34(15): 1-8.
[19] 孙超. 高性能计算平台的设计与应用[D]. 天津: 天津大学, 2013.
[20] 丁凡. 云环境中高性能计算应用的关键问题研究[D]. 兰州: 兰州大学, 2014.
[21] 宋亚奇, 周国亮, 朱永利, 等. 云平台下并行总体经验模态分解局部放电信号去噪方法[J]. 电工技术学报, 2015, 30(18): 213-222.
Song Yaqi, Zhou Guoliang, Zhu Yongli, et al. Research on parallel ensemble empirical mode decomposition denoising method for partial discharge signals based on cloud platform[J]. Transactions of China Electrotechnical Society, 2015, 30(18): 213- 222.