大型GNSS观测网的并行计算

Parallel Resolution of Huge GNSS Networks

CGCS2000精化及其全球拓展需要采用最新的模型,处理长期积累的大型GNSS观测网数据,大型GNSS网联合、快速和协同解算是空间基准精化、维持与服务的重要技术方向。在大数据技术背景下,以并行计算、云计算为代表的高性能计算技术逐渐成为大规模数据处理的首选方法。针对海量、多源、异构GNSS数据在解算处理与平差分析等方面面临的挑战,阐述了大型GNSS观测网质量控制并行计算、非差模型并行解算、双差模型并行解算、高维模糊度并行搜索固定等方法。介绍了可同时处理上千个观测站的大型GNSS观测网并行计算软件GNSSer,开展了CGCS2000精化计算的试验与验证。构建了一套适合并行计算的大型GNSS观测网解算方法,实现了CGCS2000精化及其全球拓展的多节点协同并行计算的创新。

The elaboration of CGCS2000 and its global expansion require the adoption of the latest model to deal with the long-term accumulation of massive GNSS observation networks data.The joint,rapid and collaborative solution of huge GNSS networks is a critical technical direction for the elaboration,maintenance and service of spatial reference.Under the background of big data technology,high-performance computing technology represented by parallel computing and cloud computing has gradually become the preferred method for the processing of huge-scale data.In view of the challenges faced by massive,multi-source and heterogeneous GNSS data in the aspects of solving,processing and adjustment analysis,such parallel computing methods as quality parallel control,undifference model parallel processing,double-difference parallel processing,parallel searching and fixing of high-dimensional ambiguity of huge GNSS networks are described here.A huge GNSS networks parallel computing software package,GNSSer,with processing capability of thousands of stations,is introduced,and experiments and validations of CGCS2000 refinement are carried out.Therefore,a new set of huge GNSS networks solution methods suitable for parallel computing is established,and the innovation of CGCS2000 refinement and its global expansion with multi-node collaborative parallel computing is realized.