供电分区场景下基于数据驱动的负荷密度综合评估及预测方法  

A Comprehensive Evaluation and Prediction Method for Load Density Based on Big Data under Power Supply Partition Scenarios

作　　者：贾巍 雷才嘉 方兵华 刘涌 JIA Wei;LEI Caijia;FANG Binghua;LIU Yong(Guangzhou Power Supply Bureau,Guangdong Power Grid Co.,Ltd.,Guangzhou 510620,China)

机构地区：[1]广东电网有限责任公司广州供电局,广东广州510620

出　　处：《中国电力》2023年第6期71-81,共11页Electric Power

基　　金：中国南方电网有限责任公司科技项目(大数据环境下规划负荷特性及供电分区相关影响因素的分析与应用,GZHKJXM20180011)。

摘　　要：为满足配电网供电分区和网格化规划需求,提出一种基于数据驱动的负荷密度综合评估及中长期精细化预测方法,通过改进Agglomerative算法,实现了相似单元的聚类。所提方法可有效提取出各类负荷密度的典型特征,进而降低系统对数据样本的要求,为后续各类负荷的分类精细化预测提供支持。首先,基于数据思维,通过核密度估计方法对网格内地块样本进行负荷密度特征提取;其次,采用E熵权法对各类负荷密度的特征值进行赋权,实现各供电单元不同类型负荷密度的评估,并进一步对供电单元和供电网格的综合负荷密度水平进行计算;最后,通过供电单元聚类,采用最小二乘法对负荷密度S型增长曲线的参数进行分类求解,实现供电单元各类负荷密度的中长期预测。在算例部分,进行了详细分析,结合工程实例验证了该方法的可行性。In order to meet the requirements of power supply partition and grid planning,a comprehensive evaluation and mid-long term refined prediction method for load density based on big data under power supply scenarios is proposed,and similar units are clustered through the improved Agglomerative algorithm.The proposed method can effectively extract the typical features of various load densities,so as to reduce the requirement of the system for data sampling and provide support for the classified refined forecasting of various loads.Firstly,based on the data samples,the load density features of the plot samples in the grid are extracted with the kernel density estimation(KDE)method.Then,the entropy method is used to weight the eigenvalues to realize the evaluation of different types of load densities in each power supply unit,and further calculate the integrated load density level of the power supply units and power grids.Finally,the power supply units are clustered,and the parameters of the S-shaped growth curve are solved by the least square method,so as to realize the mid-long term prediction of various load densities.In case study,a detailed analysis is carried out,and the effectiveness of the method is verified by engineering examples.

关 键 词：配电网网格化 负荷密度 核密度估计 Ē熵权法 中长期预测 

分 类 号：TM715[电气工程—电力系统及自动化]