考虑隐私保护的综合能源负荷联合聚类方法  

Privacy preserving integrated energy load joint clustering method

作　　者：胡欣月 黄奇峰 张理寅 段梅梅 盛举 李更丰[1] HU Xinyue;HUANG Qifeng;ZHANG Liyin;DUAN Meimei;SHENG Ju;LI Gengfeng(School of Electrical Engineering,Xi’an Jiaotong University,Xi’an 710049,China;Marketing Service Center,State Grid Jiangsu Electric Power Company Limited,Nanjing 220019,China)

机构地区：[1]西安交通大学电气工程学院,陕西西安710049 [2]国网江苏省电力有限公司营销服务中心,江苏南京220019

出　　处：《电工电能新技术》2024年第7期60-69,共10页Advanced Technology of Electrical Engineering and Energy

基　　金：国家电网公司总部科技项目(5108⁃202218280A⁃2⁃243⁃XG)。

摘　　要：能源系统多元化发展的背景下,深度挖掘综合能源用户的价值,提高能源系统灵活性是构建综合能源系统的关键路径。用户负荷聚类是提炼多类型用户的共性用能特性、提取用户用能模式的基础。与传统电力用户相比,综合能源用户聚类存在多个问题:用户的用能需求不局限于单一能源,不同类型能源的数据存在耦合关系;综合能源系统结构关系复杂,能源数据分属多运营主体,各主体不会向其他运营主体披露隐私数据。本文提出一种基于参数共识的综合能源联合聚类方法。该方法考虑用户数据的隐私性,适用于数据被分散存储且不同负荷聚合商之间存在信息隔离的情况,并能得到和集中式聚类相同的结果。通过多种能源联合聚类,聚类结果能够表征多能耦合下用户的用能习惯,为综合能源需求侧管理策略和系统运行优化提供决策依据。通过美国旧金山建筑能耗数据集验证了本文所提方法的有效性。With the diverse development of the energy system,a thorough exploration of the potential inherent in integrated energy costumer and the enhancement of the energy system adaptability are important.The foundation for this endeavor lies in costumer load clustering,which enables the precise delineation of shared energy usage traits a⁃mong various costumer types and the extraction of distinct energy consumption patterns.Contrasting traditional pow⁃er costumers,integrated energy costumers present multiple challenges for clustering due to their diverse energy re⁃quirements and the interconnection of data from different energy types.Additionally,the intricate structural rela⁃tionships within integrated energy systems further complicate the matters,with energy data being the purview of multiple operational entities,each of which maintains strict data privacy protocols.This paper introduces an inte⁃grated energy load joint clustering method based on parameter consensus.This approach takes into account costu⁃mer data privacy concerns and is especially suited for decentralized data storage scenarios marked by information i⁃solation among distinct load aggregators,producing results equivalent to those achieved via centralized clustering.By collectively clustering data from multiple energy sources,the resultant clustering outcomes can effectively cap⁃ture user energy consumption habits within the context of multi⁃energy interplay.These findings offer a valuable foundation for formulating integrated energy demand⁃side management strategies and optimizing system operations.The efficacy of the proposed methodology is substantiated through experimentation with the San Francisco building energy consumption dataset.

关 键 词：综合能源负荷 负荷联合聚类 隐私保护 综合需求响应 

分 类 号：TM743[电气工程—电力系统及自动化]