基于多算力融合的数字孪生黄河黄河云设计研究

Design of digital twin Yellow River Yellow River Cloud based on multi-computing power fusion

为解决数字孪生黄河建设过程中出现的算力基础设施能力不足,数据存储、处理、服务效率不高,资源服务模式不够灵活等问题,立足数字孪生黄河建设对算力基础设施的实际需求,结合水利部出台的相关技术规范,提出多算力融合黄河云重构搭建方案。开展对多元算力、多模态存储模式等关键技术的综合分析,提出多算力融合黄河云的总体框架、部署架构、资源池及资源管理设计。多算力融合黄河云针对模型计算、数据底板、智能应用、大数据处理与分析等不同场景,建设虚拟化、高性能、裸金属等算力资源,根据数据类型、特点,以及数据量匹配集中和分布式存储资源,提出适应黄河水利委员会组织架构的云管理模式,可为数字孪生黄河建设提供高效算力底座。

There are problems in the process of digital twin Yellow River construction,such as insufficient capacity of computing infrastructure,low efficiency of data storage,processing and service,and inflexibility of resource service mode.In view of that,this paper proposes the reconstruction and construction scheme of multi-computing integration Yellow River Cloud based on the actual demand on computing infrastructure,combined with the relevant technical specifications issued by the Ministry of Water Resources.A comprehensive analysis of key technologies is carried out,such as multi-computing power and multi-modal storage mode.In terms of the multi-computing integration Yellow River Cloud,the overall framework,deployment architecture,resource pool design and resource management design are put forward.Multi-computing power integration Yellow River Cloud can generate virtualization,highperformance,bare metal and other computing power for diff erent scenarios,such as model computing,data base,intelligent application,big data processing and analysis.It can match centralized and distributed storage resources,based on data types,characteristics,and data volumes.A cloud management mode adapted to the organizational structure of Yellow River Conservancy Commission Information Center is proposed,which can provide an efficient computing base for the construction of digital twin Yellow River.