广东电力2025，Vol.38Issue(9)：36-43,8.DOI:10.3969/j.issn.1007-290X.2025.09.004
基于多元特征筛选技术的公共建筑需求侧资源特征构建方法
Method for Constructing Public Building Demand-side Resource Features Based on Multivariate Feature Selection Technology
摘要
Abstract
As a major component of energy consumption in urban power systems,public buildings possess considerable potential for peak load regulation due to their stable electricity usage patterns.It is able to accurately evaluate the peak shaving potential and optimize demand response strategies through constructing load feature representations of the public buildings and analyzing their correlation with peak shaving potential.At present,commonly used indicators such as daily maximum load and load factor are limited by insufficient temporal granularity and single-method analysis,which restrict their effectiveness in accurately evaluating peak shaving potential.To address the aforementioned issues,this paper analyzes the peak shaving potential of nine typical building types from four categories of public buildings in different typical days.It proposes the fine-grained peak period features including the maximum-minimum load difference of peak load,the standard deviation of peak load,the coefficient of variation during the peak period,and the peak load factor,as well as constructs a comprehensive feature set.The paper applies Pearson correlation coefficient,mutual information,and random forest importance to rank and evaluate the relevance between the comprehensive feature set and the peak shaving potential of public buildings.Finally,four highly correlated peak shaving features of public buildings are selected.关键词
调峰潜力/公共建筑能耗/特征筛选/相关性分析Key words
peak-shaving potential/public building energy consumption/feature selection/correlation analysis分类
信息技术与安全科学引用本文复制引用
蒋思宇,惠红勋,解戴旭,陈烨..基于多元特征筛选技术的公共建筑需求侧资源特征构建方法[J].广东电力,2025,38(9):36-43,8.基金项目
广东省基础与应用基础研究基金项目(2025A1515011531) （2025A1515011531）
国家自然科学基金项目(52407075) （52407075）
澳门特别行政区科学技术发展基金项目(001/2024/SKL) （001/2024/SKL）
