有源配电网电压时空分布特性的数据驱动感知方法  

Data-driven Sensing Method for Voltage Temporal and Spatial Distribution Characteristics in Active Distribution Network

作　　者：王磊 臧谦 苏灿 张润涛 张波[2] 周昊 WANG Lei;ZANG Qian;SU Can;ZHANG Runtao;ZHANG Bo;ZHOU Hao(Research Institute of State Grid Hebei Electric Power Co.Ltd.,Shijiazhuang 050021,China;School of Electrical and Electronic Engineering,North China Electric Power University,Baoding 071003,China)

机构地区：[1]国网河北省电力有限公司电力科学研究院,河北石家庄050021 [2]华北电力大学电气与电子工程学院,河北保定071003

出　　处：《华北电力大学学报(自然科学版)》2025年第2期22-31,共10页Journal of North China Electric Power University:Natural Science Edition

基　　金：国网河北省电力有限公司电力科学研究院科技项目(kj2022-021).

摘　　要：电压时空分布特性精准感知是有源配电网电压越限抑制、优化调控的前提,本文提出一种有源配电网电压时空分布特性的数据驱动感知方法。首先,利用光伏电池出力模型和电压灵敏度分析,揭示数值气象数据对光伏出力以及光伏出力对配电网节点电压分布的影响规律;然后,通过数值气象预报获取配电网气象分布数据样本,建立配电网电压分布XGBoost感知模型,表征气象数据与配电网节点电压间的映射关系,并利用概率密度函数对原始数据样本进行抽取、扩充,避免原始样本数据不足引起的局部密度偏差和预测误差问题;最后,利用克里金插值法对预测电压数据进行可视化处理,实现了典型工况和源荷波动情况下配电网电压时空分布特性的图谱分析,并由IEEE 33节点典型算例验证了所提方法的有效性。Accurate sensing of voltage spatial and temporal distribution characteristics is the premise of voltage off-limit suppression and optimal regulation.We propose a data-driven sensing method for the spatial and temporal distribution characteristics of active distribution network voltage in this paper.Firstly,we reveal the influence of numerical meteorological data on photovoltaic output,as well as the influence of photovoltaic output on node voltage of the distribution network,by using a photovoltaic cell output model and voltage sensitivity analysis.Secondly,we obtain the meteorological distribution data samples of the distribution network through numerical weather prediction.Based on these samples,we establish an XGBoost perception model of distribution network voltage distribution to capture the mapping relationship between meteorological data and node voltage of the distribution network.Furthermore,we use the probability density function to extract and expand the original data samples to avoid the local density deviation and prediction error caused by the lack of original sample data.Finally,we apply the Kriging interpolation method to visualize the predicted voltage data,and realize the map analysis of the spatial and temporal distribution characteristics of the distribution network voltage under typical operating conditions and source-load fluctuations.The effectiveness of the proposed method is validated through a typical IEEE 33-node case study.

关 键 词：电压分布特性 数据驱动 XGBoost 概率密度估计 有源配电网 

分 类 号：TM711[电气工程—电力系统及自动化]