基于图计算的营配数据融合模型构建及应用  

Construction and Application of Marketing-distribution Data Fusion Model Based on Graph Computing

作　　者：肖凯 李道兴 王晓辉 郭鹏天 XIAO Kai;LI Daoxing;WANG Xiaohui;GUO Pengtian(China Electric Power Research Institute Co.,Ltd.,Haidian District,Beijing 100192,China)

机构地区：[1]中国电力科学研究院有限公司,北京市海淀区100192

出　　处：《电力信息与通信技术》2024年第1期61-69,共9页Electric Power Information and Communication Technology

基　　金：中国电力科学研究院有限公司研究开发项目“面向电力数据知识融合的图生成及计算技术研究”(AI83-21-001)。

摘　　要：文章首先分析了营配领域业务发展、营配贯通及数据应用现状,基于图理论,提出“营配一张图”数据融合模型构建方法,从提升营配图数据质量的角度,提出基于热重启随机梯度下降算法和余弦退火算法的户变拓扑辨识校正算法;结合营配数据规模及分析处理需求,提出营配图计算算法框架。在此基础上,通过在某省建设营配设备运维知识图谱应用,验证模型构建方法及计算框架的有效性和实用性。This paper first analyzes the current situation of business development,the transfixion and data application in the field of marketing and distribution,proposes the construction method of data fusion model of“one graph of marketing and distribution”based on the graph theory,from the perspective of improving the quality of marketing and distribution data,proposes a topology identification and correction algorithm based on hot restart stochastic gradient descent algorithm and cosine annealing algorithm.Combined with the scale of marketing data and the demand of analysis and processing,the framework of marketing and distribution graph computing is proposed.Finally,the validity and practicability of the model construction method and the computing framework are verified by the application of the operation and maintenance knowledge graph of the equipment in a province.

关 键 词：营配贯通 营配一张图 图数据 图计算 知识图谱 

分 类 号：TM73[电气工程—电力系统及自动化]