基于负荷预测与关联规则修正的不良数据辨识方法

发布时间：2018-01-28 23:48

本文关键词： 不良数据辨识数据存储回归分析预测模型相关性分析建模关联规则　出处：《电力系统保护与控制》2017年23期 　论文类型：期刊论文

【摘要】：随着电力系统的快速发展,使得电网需要对海量、异构和多态的数据进行分析与辨识。传统的不良数据辨识方法辨识效率较低,且不能够高效率利用已知的全部数据信息。为解决此问题,提出了一种基于负荷预测与关联规则修正的不良数据辨识方法。根据数据量之间的内在联系,给出了一种三维矩阵的数据存储方法。建立基于回归分析法的预测模型与基于灰色关联的相关性分析模型,分析节点注入功率与温度之间的变化关系,并采用关联规则与特殊断面修正法对预测值进行修正,进而完成对注入功率的辨识。在此基础上,再通过基尔霍夫定律与残差辨识法完成对支路潮流数据的辨识工作。最后应用实际系统的仿真算例证明了该方法能够在克服残差污染和残差淹没现象的前提下准确辨识出全部的不良数据。
[Abstract]:With the rapid development of power system, the power network needs to analyze and identify the massive, heterogeneous and polymorphic data. In order to solve this problem, a bad data identification method based on load forecasting and association rule correction is proposed. In this paper, a data storage method of 3D matrix is presented, and a prediction model based on regression analysis and a correlation analysis model based on grey correlation are established to analyze the relationship between node injection power and temperature. The prediction value is corrected by association rules and special section correction method, and then the injection power identification is completed. On this basis. Through Kirchhoff's law and residual identification method, the identification of branch power flow data is completed. Finally, the simulation example of practical system is used to prove that this method can overcome residual pollution and residual submergence phenomenon. Identify all the bad data.
【作者单位】：国网北京市电力公司;国网沧州供电公司;燕山大学电力电子节能及传动控制河北省重点实验室;
【基金】：国家自然科学基金(61374098) 教育部高等学校博士学科点专项科研基金(20131333110017)~~
【分类号】：TM711
【正文快照】： 3.燕山大学电力电子节能及传动控制河北省重点实验室,河北秦皇岛066004)This work is supported by National Natural Science Foundation of China(No.61374098)and Research Fund for theDoctoral Program of Higher Education of China(No.20131333110017).电力系统规模与

【参考文献】