基于大数据技术调度端电网模型管理和分析架构  

Management and Analysis Framework of Power Grid Models Based on Big Data Technology in Dispatching Center

作　　者：赵林 张令涛 马仲佳 郭子明 张浩 李新鹏 沈国辉 

机构地区：[1]南瑞集团公司(北京),北京市海淀区100192 [2]国网冀北电力有限公司,北京市西城区100053

出　　处：《电网技术》2017年第12期3750-3756,共7页Power System Technology

基　　金：国家重点研发计划项目(2017YFB0902602)~~

摘　　要：为了实现大运行体系下的共享多标准电网模型的统计和分析,提出了一种应用于调度端的基于大数据技术的电网模型管理和分析的架构。首先分析了大数据存储结构的特点,根据电网模型特点采用文本式结构保存电网模型,设计出结构化的主键值表征电网模型的时间、空间和应用多维度的扩展。然后研究了多源异构电网模型的ETL(extract,transform,load)技术,将调度端电网模型提取转存至分布式大数据平台。采用Map Reduce并行计算框架对模型数据进行分布式计算,为了提高分布式计算的效率,降低I/O消耗,采用索引技术对模型数据进行处理,减少各节点对全表数据进行扫描的次数。最后在调度端图形编辑器中,利用所提出的模型分析技术实现了电网图形模型的搜索,利用热度图实现了大数据搜索结果的可视化展示,并通过实例证明了所提方法的有效性。A framework of power grid model management and data analysis based on big data technology is proposed for multi-source and multi-standardized data.Firstly,characteristics of big data storage structure are analyzed.According to their characteristics,the grid models are preserved with key-document structure.The structured primary key is designed to characterize time,space and multidimensional expansion of the power grid models.Then,ETL（Extract TransformLoad） tool for multi-source heterogeneous power grid models is studied,by which the power grid models are extracted and transferred to distributed big data platform,and the models are analyzed using Map Reduce parallel computing framework.In order to improve efficiency of distributed computation and reduce I/O consumption,compound index technology is used to deal with the models to reduce number of times scanning whole table data.Finally,analysis technology for the proposed grid models is applied to the collected grid models,and visualization of the big data searching results is shown with thermal graphs.The example proves effectiveness of the proposed method.

关 键 词：ETL Map REDUCE 复合索引 大数据 电网模型 

分 类 号：TM734[电气工程—电力系统及自动化]