联合采用熵权和灰色系统理论的电力大数据质量综合评估  

Comprehensive Evaluation of Big Data Quality in Power Systems with Entropy Weight and Grey System Theory

作　　者：李刚[1] 焦亚菲[1] 刘福炎[2] 俞敏[2] 宋雨[1] 文福拴[3,4] 

机构地区：[1]华北电力大学控制与计算机工程学院,河北省保定市071003 [2]国网浙江省电力公司经济技术研究院,杭州市310008 [3]浙江大学电气工程学院,杭州市310027 [4]文莱科技大学电机与电子工程系

出　　处：《电力建设》2016年第12期24-31,共8页Electric Power Construction

基　　金：国家自然科学基金项目(51407076);中央高校基本科研业务费专项资金资助项目(2015ZD28);河北省自然科学基金项目(F2014502050);河北省高等学校科学研究项目(Z2013007);国网浙江省电力公司经济技术研究院研究项目(JY02201403)~~

摘　　要：随着电力系统规模的增大、量测技术的发展与成本下降,电力系统的数据量呈现快速增长趋势,逐步具备了大数据特征。充分利用大数据来改善电力系统的规划、运行与控制已受到越来越广泛的重视,如何评估大数据的质量是一个值得研究的重要问题。在数据质量提高技术如数据清洗、数据整合、相似记录检测等方面,已有相当多的研究报道。然而,在数据质量评估方面的研究还处于起步阶段。在此背景下,针对电力系统特征和电力大数据质量特性,提出一种电力大数据质量综合评估方法。首先,构建电力大数据质量评估指标体系;接着,针对大数据处理的时效性问题,利用MapReduce并行化K-means聚类算法来实现大数据样本集的快速预处理。之后,利用熵权法计算各类数据集的客观权重,采用灰色评估法判断数据质量所属等级,在此基础上实现对样本数据集的综合评价。最后,以某市电力公司所采集的用户用电负荷数据为例对所提出的方法做了说明。With the continuous expansion of power systems,as well as ever-developing technology and reduced costs of measurement devices,the recorded data in power systems have been increasing significantly and progressively exhibit the feature of big data.Much attention has been paid to the full use of big data for improving the planning,operation and control of power system,and hence how to evaluate the quality of big data is becoming an important problem to be examined.Some research publications are available on data quality improvement,such as data cleaning,data integration,and the detection of similar records,but the existing research work is still preliminary in data quality evaluations.Given this background,considering the characteristics of power systems and associated big data,this paper proposes a comprehensive method for evaluating the quality of big data in power systems.Firstly,we construct an index system for big data quality evaluations.Then aiming at the timeliness of big data,we adopt the K-means clustering algorithm in parallel with MapReduce for fast preprocessing of the big data sample set.Secondly,we use entropy weight method to calculate the objective weight of each dataset and grey evaluation method to determine the data quality level.On this basis,the comprehensive evaluation of the sample data set is carried out.Finally,the recorded electric load historical data in a city power company are employed to demonstrate the proposed method.

关 键 词：电力系统 大数据 数据质量 评估指标 K-MEANS聚类 熵权法 灰色理论 

分 类 号：TM72[电气工程—电力系统及自动化]