基于多数据源的电力用户信用评价研究  

Research on Credit Evaluation of Power Users Based on Multiple Data Sources

作　　者：周渊 刘茹 马国瀚 胡飞虎[2] 王玉旭 Zhou Yuan;Liu Ru;Ma Guo-han;Hu Fei-hu;Wang Yu-xu

机构地区：[1]国网兰州供电公司,甘肃兰州730070 [2]西安交通大学电气工程学院,陕西西安710049

出　　处：《电力系统装备》2020年第2期122-124,共3页Electric Power System Equipment

摘　　要：现有电力用户信用研究所需的数据,基本都来源于电力公司的用采或营销系统,无法反映用户的综合信用。为了获得电力用户的综合信用特征,必须扩充信用评价的数据源。文章扩充的数据源主要源于权威部门发布的公开数据,并设计了注册资本、企业性质、税务评级、行政处罚、经营异常、行业前景、红名单信息和黑名单信息8个指标,结合用户用电指标,对914个电力用户的18个信用评价指标数据进行K-means聚类,并对聚类结果使用Topsis综合评价法,将聚类结果划分为不同的信用等级,实现对不同信用等级电力用户的信用评价。通过与电力用户实际状况比较,发现分类结果与实际具有较好的一致性。The data required for the credit research of existing power users are basically derived from the power company’s procurement or marketing system,which cannot refl ect the user’s comprehensive credit.In order to obtain the comprehensive credit characteristics of power users,the data source of credit evaluation must be expanded.The data source expanded in this article is mainly from public data released by authoritative departments,and eight indicators of registered capital,corporate nature,tax rating,administrative penalties,business abnormalities,industry prospects,red list information and black list information are designed.Electricity index,K-means clustering of 18 credit evaluation index data of 914 power users,and the clustering results are classifi ed into different credit levels using Topsis comprehensive evaluation method to achieve different credit level power users Evaluation of credit.By comparing with the actual situation of power users,it is found that the classifi cation results are in good agreement with the actual situation.

关 键 词：电力用户信用评价 多数据源 K-MEANS聚类 Topsis综合评价法 

分 类 号：U492.2[交通运输工程—交通运输规划与管理]