面向电力边缘数据分析的图分类算法研究进展

Survey of Graph Classification Algorithms for Edge Data Analysis in Smart Grid

随着智能电网等技术的发展,关于图数据的研究应运而生。图分类的目的是将图数据中表示的目标进行分类。在智能电网领域,从电网边缘终端设备产生的电力数据可以通过构图的形式表示成图数据进行处理。根据基础模型的不同,将现有图分类的方法归纳为四类,分别是嵌入方法、基于图的特征挖掘方法、核函数方法和深度学习方法。介绍图分类中常用的数据集,总结图分类方法的相关开源代码,可有效促进智能电网领域边缘数据分析技术的发展。

With the development of information technology and applications in smart grid,many researches on graph data have emerged,and the pur⁃pose of the graph classification is to classify the targets represented in the graph data.In the field of smart grids,power data generated from grid edge terminals can be processed in the form of a map representation.According to the different basic models,classifies the existing methods of graph classification into four categories,namely embedding method,graph-based feature mining method,kernel function meth⁃od and deep learning method.We introduced the data sets commonly used in graph classification and summarized the relevant open source code for graph classification methods.This paper hopes to introduce the existing classification methods of graphs and promote the develop⁃ment of edge data analysis technology in smart grid.