摘要
配电网中准确的拓扑结构辨识对运行和控制具有重要意义,针对实际配电网拓扑结构变动的情况,搭建了可智能辨识配电网拓扑结构的深度学习模型。首先,生成不同拓扑结构下的配电网量测数据并进行数据预处理。其次,构建了融合CNN(卷积神经网络)、LSTM(长短期记忆网络)和Attention(注意力机制)的拓扑结构智能辨识模型,并结合历史量测数据对模型训练并测试。最后,在IEEE 33节点和PG&E69节点配电系统仿真算例中,验证了该基于CNN-LSTM-Attention模型的拓扑辨识方法相较于传统辨识方法在辨识精度上的优越性,实现了该模型的在线应用。
Accurate identification of the topology in a distribution network is crucial for its operation and control.Addressing the dynamic changes in the actual topology of distribution networks,an intelligent deep learning model capable of recognizing distribution network topologies was developed.Firstly,measurement data for distribution networks under different topologies were generated,followed by data preprocessing.Subsequently,an intelligent topology identification model was constructed,integrating convolutional neural network(CNN),long short-term memory network(LSTM),and Attention mechanism.The model was trained and tested using historical measurement data.Finally,in simulation scenarios using the IEEE 33-node and PG&E69-node distribution systems,the superiority of this CNN-LSTM-Attention-based topology identification method over traditional approaches in terms of identification accuracy was validated,and online application of the model was achieved.
作者
凌佳凯
章逸舟
胡金峰
秦军
戴健
费有蝶
朱振
LING Jiakai;ZHANG Yizhou;HU Jinfeng;QIN Jun;DAI Jian;FEI Youdie;ZHU Zhen(Wuxi Power Supply Company of State Grid Jiangsu Electric Power Co.,Ltd.,Wuxi Jiangsu 214061,China;School of Energy and Electrical and Power Engineering,Hohai University,Nanjing 211100,China)
出处
《浙江电力》
2024年第3期84-94,共11页
Zhejiang Electric Power
基金
国网江苏省电力有限公司科技项目(J2021026)。
关键词
配电网
拓扑辨识
卷积神经网络
长短期记忆网络
注意力机制
distribution networks
topology identification
convolutional neural network
long short-term memory network
Attention mechanism
