摘要
高比例新能源接入使配电网具备一定的主动电压支撑能力,可通过调节公共连接点无功功率,实现输配协同电压调控。然而,新能源出力波动及输配耦合效应加剧了电压失稳过程的复杂性,给短期电压稳定(short-term voltage stability,STVS)评估带来挑战。为此,提出计及输配协同的STVS数据驱动评估方法,首先,区别于传统评估中将配电网简化为不可控等值负荷,构建计及配电网主动电压支撑能力的系统时域仿真拓展模型,基于优化方法量化支撑能力并嵌入时域仿真,反映其对电压稳定的影响。其次,基于该模型与历史数据,构建以系统量测量为输入、稳定性状态为输出的训练数据集,训练卷积神经网络(convolutional neural network,CNN)实现STVS在线评估。相比于现有基于深度学习的STVS评估,提出了基于关键节点电压的输入-输出变量降维提取方法,可显著减少训练数据量,提升学习效率。算例仿真结果验证了所提方法在STVS评估和电压失稳程度量化方面的有效性。
High-proportional new energy penetration endows distribution networks with active voltage support through reactive power regulation at the point of common coupling(PCC),enabling coordinated transmission and distribution control.However,new energy fluctuations and intensified transmission and distribution coupling complicate voltage instability dynamics,challenging short-term voltage stability(STVS)assessment.To address this,this paper proposes a data-driven STVS assessment method considering transmission and distribution collaboration.Firstly,different from conventional approaches that simplify distribution networks as uncontrollable equivalent loads,an extended time-domain simulation model incorporating distribution-level active voltage support capability is constructed.This model quantifies voltage support capacity based on the optimization theory and embeds it into time-domain simulation to reflect its influence on voltage stability.Secondly,leveraging the proposed model and historical operational data,a training dataset using system measurement as inputs and stability indices as outputs is constructed,and a convolutional neural network(CNN)is then trained to achieve online STVS assessment.Compared to existing deep learning based STVS evaluation methods,the proposed approach introduces a dimensionality reduction technique for input and output variables based on critical node voltage features,significantly reducing training data requirements while improving computational efficiency.Case studies validate the effectiveness of the proposed method in both STVS assessment and voltage instability severity quantification.
作者
李广
阚常涛
鲍冠南
李嫣然
武诚
马锐
胡正阳
LI Guang;KAN Changtao;BAO Guannan;LI Yanran;WU Cheng;MA Rui;HU Zhengyang(Jinan Power Supply Company of State Grid Shandong Electric Power Co.,Ltd.,Jinan,Shandong 250102,China;State Grid Shandong Electric Power Co.,Ltd.,Jinan,Shandong 250001,China;School of Electrical Engineering,Shandong University,Jinan,Shandong 250061,China;School of Electrical and Electronic Engineering,Hong Kong Polytechnic University,Hong Kong 999077,China)
出处
《广东电力》
北大核心
2025年第9期4-14,共11页
Guangdong Electric Power
基金
国网山东省电力公司科技项目(52060124000G)
国家自然科学基金青年项目(52407119)。
关键词
短期电压稳定
数据驱动评估方法
配电网电压支撑能力
卷积神经网络
时域仿真
short-term voltage stability
data-driven assessment
voltage support capability of distribution network
convolution neural network
time-domain simulation
