GNSS大网双差模型并行快速解算方法

发布时间：2018-04-08 16:13

本文选题：大型GNSS网　切入点：双差模型　出处：《测绘学报》2017年07期

【摘要】：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,提出了一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35.62min,显著提高了整网双差的解算效率。
[Abstract]:In order to solve the problem of time-efficiency difference in the solution of GNSS data using double-difference model, an improved method for constructing independent double-difference observations and calculating independent baselines in parallel is proposed in this paper.The parallel technique is used to realize the strategy of multi-core parallel and network multi-node parallelism.By processing the observation data of about 375 IGS stations for one week,Compared with the traditional path shortest method, the improved independent double-difference observation method has an average of 530000 more independent double-difference data than the traditional path shortest method. The average repeatability of the coordinates in the direction of the single day network is increased by 14.0% and 29.2% respectively.The calculation results using different calculation strategies show that the parallel computing efficiency of four common computers is about 14 times higher than that of the traditional serial scheme.For example, the 4-node parallel scheme of 375 stations using the improved method of constructing observations only needs 35.62 mins, which greatly improves the efficiency of the two-difference solution of the whole network.
【作者单位】：后勤工程学院;信息工程大学;火箭军工程大学;
【基金】：国家重点研发计划(2016YFB0501701) 国家自然科学基金(41674019)~~
【分类号】：P228.4

【参考文献】