基于图卷积网络的配电网碳排放因子计算与分析

边晓燕; 刘天蔚; 李锡瑞; 高美金; 俞楚天; 张利军

doi:10.13334/j.0258-8013.pcsee.241732

欢迎来到能源电力期刊网！

English Version 期刊登录

您当前的位置：

首页 >

文章列表页 >

基于图卷积网络的配电网碳排放因子计算与分析

电力系统 | 更新时间：2026-10-12

- 基于图卷积网络的配电网碳排放因子计算与分析
- Calculation and Analysis of Carbon Emission Factor in Distribution Network Based on Graph Convolutional Network
- 中国电机工程学报 2026年46卷第5期页码：1855-1868
- 作者机构：
  
  1. 上海电力大学电气工程学院，上海市杨浦区,200090
  2. 上海交通大学电子信息与电气工程学院，上海市闵行区,200240
  3. 国网浙江省电力有限公司经济技术研究院,浙江省,杭州市,310000
- 作者简介：
- 基金信息：
- DOI：10.13334/j.0258-8013.pcsee.241732
  中图分类号：
- 收稿：2026-03-05，
  
  纸质出版：2026
- 稿件说明：
移动端阅览
边晓燕, 刘天蔚, 李锡瑞, 等. 基于图卷积网络的配电网碳排放因子计算与分析[J]. 中国电机工程学报, 2026,46(5):1855-1868.

BIAN Xiaoyan, LIU Tianwei, LI Xirui, et al. Calculation and Analysis of Carbon Emission Factor in Distribution Network Based on Graph Convolutional Network[J]. 2026, 46(5): 1855-1868.
边晓燕, 刘天蔚, 李锡瑞, 等. 基于图卷积网络的配电网碳排放因子计算与分析[J]. 中国电机工程学报, 2026,46(5):1855-1868. DOI： 10.13334/j.0258-8013.pcsee.241732.

BIAN Xiaoyan, LIU Tianwei, LI Xirui, et al. Calculation and Analysis of Carbon Emission Factor in Distribution Network Based on Graph Convolutional Network[J]. 2026, 46(5): 1855-1868. DOI： 10.13334/j.0258-8013.pcsee.241732.

摘要

碳排放计算是实现“双碳”目标、促进经济绿色转型的基本前提，当前电力系统碳排放计算主要聚焦于输电网。配电网量测信息远少于输电网，如何在配电网现有量测信息下计算碳排放是亟待解决的问题。该文基于图卷积网络(graph convolutional network，GCN) 建立配电网量测信息下的碳排放因子计算模型。首先，基于配电网拓扑连接关系与各节点量测数据，构建表征拓扑关系的邻接矩阵和表征节点电碳信息的输入矩阵；其次，通过训练GCN拟合学习碳排放因子计算函数，建立碳排放因子与电碳信息间的映射关系，从而生成配电网各节点碳排放因子数值。算例结果表明，所提碳排放因子计算模型的平均绝对误差约为0.034 2 kgCO2/(kW•h)，平均绝对百分比误差约为9.55%，验证了模型的准确性。并基于配电网碳排放流特点对误差较大的节点进行修正，使模型整体平均绝对百分比误差下降约1.54%。最后，研究该模型在网络拓扑变化下的适用性。

Abstract

Carbon emission calculation is a fundamental prerequisite for achieving dual carbon goals and promoting green economic transformation. Currently

carbon emission calculation in the power system primarily focuses on the transmission network. However

measurement information in the distribution network is significantly less than that in the transmission network

posing an urgent challenge in calculating carbon emissions based on limited measurement information available in the distribution network. A carbon emission factor calculation model for the distribution network based on the graph convolutional network (GCN) is proposed. Firstly

based on the topological connection relationship and measurement data of each node in the distribution network

an adjacency matrix representing the topological relationship and an input matrix representing the electrical and carbon information of nodes are constructed. Secondly

by training GCN to fit and learn the carbon emission factor calculation function

the mapping relationship between carbon emission factors and electrical carbon information is established

thereby calculating carbon emission factors at each node in the distribution network. The calculation results show that the mean absolute error of the carbon emission factor calculation model proposed in this paper is about 0.0342 kgCO2/(kW•h)

and the mean absolute percentage error is about 9.55 %

which verifies the accuracy of the model. Based on the characteristics of carbon emission flow in the distribution network

the nodes with large errors are corrected

and the overall mean absolute error of the model is reduced by about 1.54%. Finally

the applicability of the model under network topology changes is studied.

关键词

Keywords

references

浏览量

979

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于碳排放流的配电网广义碳势感知方法

图信号与图卷积网络相结合的局部放电模式识别方法

计及碳排放分摊的配电网分布式低碳调度策略

多光伏电源两阶段式故障穿越协调控制策略

基于动态时空适应图神经网络的电网线路参数辨识方法