| 胡亚,林珲,徐丙立,朱军,胡明远.基于PC机群的多点源高斯大气污染扩散模型的并行计算研究[J].高技术通讯(中文),2010,20(4):436~440 |
| 基于PC机群的多点源高斯大气污染扩散模型的并行计算研究 |
| A study on PC cluster based parallel computation of Gauss plume model for multi point sources |
| |
| DOI： |
| 中文关键词: 个人计算机机群，多点源，高斯污染扩散模型，并行计算 |
| 英文关键词: PC cluster, multi point sources, Gauss plume model, parallel computation |
| 基金项目:863计划(2009AA12Z207,2006AA12Z207)和香港特区研究资助局基金(447807)资助项目 |
| 作者 | 单位 | | 胡亚 | 西南交通大学土木工程学院测量工程系，成都；香港中文大学太空与地球信息科学研究所，香港 | | 林珲 | 香港中文大学太空与地球信息科学研究所，香港 | | 徐丙立 | 香港中文大学太空与地球信息科学研究所，香港 | | 朱军 | 西南交通大学土木工程学院测量工程系，成都 | | 胡明远 | 香港中文大学太空与地球信息科学研究所，香港 |
|
| 摘要点击次数: 3119 |
| 全文下载次数: 2403 |
| 中文摘要: |
| 针对基于单机的多点源高斯大气污染扩散模型计算效率较低的问题，设计了高斯模型并行算法，该算法是基于计算输出层设计的，它将计算输出层尽量平均地分配到各个计算节点上。针对并行算法，构建了分布式并行计算平台。该平台采用三层分布式体系结构，即用户层、控制层和计算资源层。实验结果表明，基于该算法的高斯并行计算大大提高了模型计算效率，分布系统结构合理，可以为准实时甚至实时的环境管理业务提供支持。 |
| 英文摘要: |
| To solve the problem that the Gauss plume model (GPM) widely used for air pollution modeling is of low efficiency when applied to large number of point sources, the study presented in this paper tried to improve GPM computation efficiency by designing a parallel algorithm and developing a distributed system based on personal computer (PC) cluster. The parallel algorithm was designed according to the levels of output, which allocates multiple levels to multiple computers evenly as far as possible. The architecture of the distributed system based on PC cluster was designed to have the three tiers of computation sources, task allocation and user interface for command generation. The test results show that the algorithm is reasonable and the computation time can be significantly decreased to one tenth of that of one computer. The tests also prove that the PC cluster based distributed system is reliable and rational. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
