摘要
传统电网无功容量计算方法往往需要耗费大量时间,为此提出了一种基于图卷积网络(Graph Convolutional Network,GCN)的无功容量计算方法。该算法通过特征矩阵以及邻接矩阵刻画电网节点的关键信息及节点间的连接关系,训练得到能够快速计算电网无功容量的模型。算例分析表明,使用GCN计算电网所需的无功容量误差较小,有良好的精度且计算速度快于传统的无功计算方法。
Traditional reactive power capacity calculation methods often take a lot of time.Therefore,a reactive power capacity calculation method based on graph convolutional network(GCN)is proposed.The algorithm describes the key information of power grid nodes and the connection relationship between nodes through characteristic matrix and adjacency matrix,and a model that can quickly calculate the reactive power capacity of power grid is trained.The example analysis shows that the reactive power capacity error required by GCN is small,has good accuracy,and the calculation speed is faster than the traditional reactive power calculation method.
作者
刘思杨
夏梓源
纪思
纪雨新
屈逸阳
田琦超
LIU Siyang;XIA Ziyuan;JI Si;JI Yuxin;QU Yiyang;TIAN Qichao(Nanjing Institute of Technology,Nanjing 211167,China;Yunnan Power Grid Co.,Ltd.,Kunming 650011,China)
出处
《电工技术》
2022年第12期145-146,150,共3页
Electric Engineering
基金
南京工程学院2021年度大学生科技创新基金项目(编号TB202104028)。
关键词
图卷积网络
深度学习
无功容量
graph convolutional network
deep learning
reactive power capacity
