GNSS大网双差模型并行快速解算方法

针对 GNSS大网数据采用双差模型解算时存在时效性差的问题,提出了-种改进的独立双差观 测值构建与独立基线并行解算的方法,采用并行技术实现多核并行与网络多节点并行的双层自动快速 解算策略.通过对约375个 IGS站1周的观测数据进行处理,改进的独立双差观测值选取方法比传统 路径最短方法所选的单天全网独立双差数据平均多了 53万个,E、N、U 方向坐标重复性平均提升了 14.0%、12.9%和29.2% .采用不同解算策略的计算结果表明,4台普通计算机的并行计算比传统串行 方案的计算效率提升了 14倍左右,如375个测站采用改进观测值构建方法的4节点并行方案仅需要 35.62 min,显著提高了...

Full description

Saved in:
Bibliographic Details
Published inCe hui xue bao Vol. 46; no. 7; pp. 848 - 856
Main Author 崔阳 吕志平 李林阳 陈正生 孙大双 邝英才
Format Journal Article
LanguageChinese
English
Published Beijing Surveying and Mapping Press 01.07.2017
Subjects
Algorithms
Ethyl nitrosourea
Global positioning systems
GPS
Parallel processing
Parallel programming
Vertical orientation
双差模型
多核多节点
大型GNSS网
并行计算
独立基线
Online AccessGet full text
ISSN1001-1595
1001-1595
DOI10.11947/j.AGCS.2017.20160585

Cover

More Information
Summary:针对 GNSS大网数据采用双差模型解算时存在时效性差的问题,提出了-种改进的独立双差观 测值构建与独立基线并行解算的方法,采用并行技术实现多核并行与网络多节点并行的双层自动快速 解算策略.通过对约375个 IGS站1周的观测数据进行处理,改进的独立双差观测值选取方法比传统 路径最短方法所选的单天全网独立双差数据平均多了 53万个,E、N、U 方向坐标重复性平均提升了 14.0%、12.9%和29.2% .采用不同解算策略的计算结果表明,4台普通计算机的并行计算比传统串行 方案的计算效率提升了 14倍左右,如375个测站采用改进观测值构建方法的4节点并行方案仅需要 35.62 min,显著提高了整网双差的解算效率.
Bibliography:GUI Yang1 ,LU Zhiping2 ,LI Linyang2 ,CHEN Zhengsheng3 ,SUN Dashuang2 ,KUANG Yingcai2（1. Logistical Engineering University, Chongqing 401331, China; 2. Information Engineering University,Zhengzhou 450052, China; 3. Xi＇an High-Tech Institute,Xi＇an 710025,China）
11-2089/P
In view of the existing problem of poor timeliness for larger GNSS network ddouble-difference （DD） model, a new approach for the independent DD observations selection and integrated processing of huge GNSS networks is presented by using the parallel algorithm design and parallel programming technology in thispaper. Thenew approach is implemented byusing theparallel computing strategy of multi-core parallel level and network parallel level. Validation experiment withGPSobservation data collected fromabout 375 IGS stations was carried out, which demonstrated that the new approach has an average about 530,000 more independentDD ambiguities than the traditional method. The average ENUCEast North Vertical） positioning precision of he new approach12.9% and 29.2%. In thewhole network processing of 375 stations,the efficiencyfour quad-core nodes is increased about 14 times than the traditional serial computing scheme,significantly improves the efficiency of the entire network processingwith DD model.
large GNSSnetwork; double difference model; independent baseline; multi-core multi-node; parallelcomputing
ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1001-1595
1001-1595
DOI:10.11947/j.AGCS.2017.20160585