基于幂律特性的企业用电量网络构建与中心企业分析  

Network Modeling and Central Node Analysis of Enterprise Correlations in Terms of Electricity Consumption Based on Power-Law Distribution

作　　者：许荣华 胡仁杰 綦方中[1] 马庆国[1] XU Ronghua;HU Renjie;QI Fangzhong;MA Qingguo(School of Management, Zhejiang University of Technology, Hangzhou 310023,China)

机构地区：[1]浙江工业大学管理学院,杭州310023

出　　处：《复杂系统与复杂性科学》2021年第1期38-47,共10页Complex Systems and Complexity Science

基　　金：国家自然科学基金(71904174);国网浙江省电力有限公司科学技术项目(5211JY18000X)。

摘　　要：基于杭州市2017年工业企业用电量数据,提出了在用电量相关性网络中选择中心企业的方法,并论证了其满足的网络特性及核心作用。首先使用条件Pearson相关性模型消除季节等宏观因素影响,计算企业之间的真实相关性;其次设定相关系数阈值过滤出中心结构;最后利用头重分布、同配混合、小世界等性质对网络进行实证分析。研究结果表明,度大顶点及其连接的分布满足二八定律;所选中心企业能够有效促进其他非中心企业之间的联系,为决策者提供重要启示。Based on the 2017 Hangzhou electricity consumption data of industrial enterprises,this paper aims to establish correlation networks to filter out the central enterprises and their connections in the city.First,we exclude the seasonal effect on electricity consumption and calculate the pure correlation of these enterprises by using the conditional Pearson correlation model.Next,we adjust the correlation coefficient thresholds by making the node degree distribution approximately satisfy the power-law distributions.Then,the empirical analysis on network properties is carried out,including the heavy-head distribution,the assortative mixing,and the small-world property.Based on these properties,we propose an approach for identifying central enterprises in the electricity consumption correlation network.The study shows that the central enterprises are closely related and can promote the connection between other non-central enterprises.Decision-makers can regulate the central enterprises detected by our approach in order to affect the overall development of the city.

关 键 词：幂律特性 用电量网络 二八定律 同配混合 小世界特性 

分 类 号：N949[自然科学总论—系统科学]