基于随机森林的电网GIS数据分布式存储方法  

Distributed storage method of GIS data in power grid based on random forest

作　　者：杨秋勇 王建欣 符飞虎 罗政 YANG Qiuyong;WANG Jianxin;FU Feihu;LUO Zheng(China Southern Power Grid Co.,Ltd.,Guangzhou 510663,China;China Southern Power Grid Digital Grid Research Institute Co.,Ltd.,Guangzhou 510663,China)

机构地区：[1]中国南方电网有限责任公司,广东广州510663 [2]南方电网数字电网研究院有限公司,广东广州510663

出　　处：《电子设计工程》2024年第17期27-30,35,共5页Electronic Design Engineering

基　　金：2022年南网数研院平台安全分公司数据中心管理体系研究项目(0002200000091292)。

摘　　要：内部电网地理信息系统(Geographic Information Systern,GIS)数据体量增加,对电网数据存储性能造成了极大的困难,为此,提出一种基于随机森林的电网GIS数据分布式存储方法。以跨域资源共享(Cross-Origin Resource Sharing,CORS)技术在电网GIS空间信息服务平台中获取的电网GIS数据为基础,根据类区分度数值选择电网GIS数据特征,引入随机森林算法分类处理电网GIS数据,将其合理分发给不同的服务器,采用并行处理手段存储分类数据,从而实现了电网GIS数据的分布式存储。实验数据显示:应用所提方法后,电网GIS数据分类精度达到了96.8%,电网GIS数据分布式存储时间最小值为5.2 s,充分证实了所提方法数据存储性能更佳。The increase of the data volume of the internal power grid Geographic Information System(GIS)has caused great difficulties to the power grid data storage performance.Therefore,a distributed storage method of power grid GIS data based on random forest is proposed.Based on the grid GIS data obtained from the grid GIS spatial information service platform using the Cross⁃Origin Resource Sharing(CORS)technology,the grid GIS data characteristics are selected according to the class differentiation value,and the random forest algorithm is introduced to classify and process the grid GIS data,which is reasonably distributed to different servers,and the parallel processing method is used to store the classified data,thus realizing the distributed storage of the grid GIS data.The experimental data shows that after applying the proposed method,the classification accuracy of grid GIS data reaches 96.8%,and the minimum distributed storage time of grid GIS data is 5.2 s,which fully proves that the proposed method has better data storage performance.

关 键 词：数据分类 电网GIS数据 并行处理 分布式存储 随机森林算法 类区分度 

分 类 号：TP399[自动化与计算机技术—计算机应用技术] TN8[自动化与计算机技术—计算机科学与技术]