基于SEC指标的电力企业无监督同业对标研究  

SEC Index Based Unsupervised Benchmarking Technology for Grid Enterprises

作　　者：田洪迅[1] 齐立忠[2] 曹利强 吴经锋 游强 丁彬 李旭 TIAN Hongxun QI Lizhong CAO Liqiang WU Jingfeng YOU Qiang DING Bin LI Xu(State Grid Corporation of China, Beijing 100031, China State Grid Economic Research Institute, Beijing 102209, China State Grid Shaanxi Electric Power Research Institute, Xi＇an 710100, China State Grid Shaanxi Electric Power Company, Xi＇an 710048, China State Power Shaanxi Economic Research Institute, Xi＇an 710065, China)

机构地区：[1]国家电网公司,北京100031 [2]国网北京经济技术研究院,北京102209 [3]国网陕西省电力公司电力科学研究院,陕西西安710199 [4]国网陕西省电力公司,陕西西安710048 [5]国网陕西省电力公司经济技术研究院,陕西西安710065

出　　处：《智慧电力》2017年第10期50-55,共6页Smart Power

基　　金：国家电网公司科技项目(B3442016K001)~~

摘　　要：为了解决传统同业对标技术对初始聚类中心与数据噪声敏感问题,提出了基于电力企业安全效能成本(SEC)指标的电力企业无监督同业对标方法。该方法采用SEC指标,结合模糊学习矢量量化与可能聚类原理,引入隶属度与典型值实现对矢量量化网络中学习速率的更新学习,实现电力企业的无监督同业对标。仿真试验验证了所提出可能模糊聚类算法的有效性与稳定性。此外,通过实例表明了所提出无监督同业对标方法可以深度挖掘其他企业与标杆企业之间技术差距,实现电力企业资产管理安全、效能方面的精益化提升。In order to solve the probtem that the traditional clustering methods are sensitive to initial cluster center and noise, the paper proposes a security efficiency cost （SEC） index based unsupervised benehmarking method for grid enterprises. Based on the SEC index, in order to implement the unsupervised benchmarking for grid enterprises, the method introduces the membership and typical values to achieve the update and learning of the leaming rate of the vector in quantization network, combining with fuzzy learn vector quantization and possibilistic clustering principle. The simulation results show that the proposed method is of effectiveness and robustness, can make the mining analysis of the difference between other enterprises and the benchmarking, and greatly enhance the safety and efficiency of asset management in electric power enterprises.

关 键 词：同业对标 电力企业 SEC指标 可能聚类 无监督学习 

分 类 号：TM731[电气工程—电力系统及自动化]