基于数据挖掘的配电网故障诊断研究

发布时间：2018-07-10 07:08

本文选题：数据挖掘 + 粗糙集理论　；参考：《华北水利水电大学》2017年硕士论文

【摘要】：电力行业是我国经济发展的支柱性行业,其技术先进性与否影响着国民经济的发展,配电网中故障诊断是电力领域的研究重要内容之一。本文研究了基于粗糙集与贝叶斯网络的数据挖掘技术在配电网故障诊断中应用,并进行验证性分析。简述了国内外故障诊断领域的研究成果,分析了数据挖掘技术在该领域的应用优势,提出一种基于粗糙集与贝叶斯网络的数据挖掘技术,并进行了配电网故障诊断的应用分析。首先,根据给出的配电网模型,利用继电保护动作原理,形成原始决策表。其次,根据粗糙集理中可辨识矩阵的形成规则,建立可辨识矩阵。然后根据粗糙集中核属性求取方法,进行核属性提取。若提取结果不唯一,则引入信息论中基于信息熵的平均互信息的理论,进行最简核属性提取。最后,根据得到的最简核属性建立贝叶斯网络模型,按照贝叶斯推理模型进行概率运算,得出某种信息下,可能发生的故障区域。通过实例分析表明,基于粗糙集与贝叶斯网络的数据挖掘技术在配电网故障诊断中具有很好的应用前景。
[Abstract]:The electric power industry is the pillar industry of our country's economic development, its technical advancement or not affects the development of the national economy, the fault diagnosis in the distribution network is one of the important contents of the research in the electric power field. In this paper, the application of data mining technology based on rough set and Bayesian network in fault diagnosis of distribution network is studied and verified. This paper briefly introduces the research achievements in the field of fault diagnosis at home and abroad, analyzes the application advantages of data mining technology in this field, puts forward a data mining technology based on rough set and Bayesian network, and analyzes the application of fault diagnosis in distribution network. First, according to the given distribution network model, the original decision table is formed by using relay protection operation principle. Secondly, according to the rules of discernibility matrix in rough set theory, identifiable matrix is established. Then the kernel attribute is extracted according to the method of rough set. If the result is not unique, the theory of average mutual information based on information entropy in information theory is introduced to extract the minimum kernel attribute. Finally, the Bayesian network model is established according to the minimum kernel attribute, and the probability operation is carried out according to the Bayesian reasoning model, and the possible fault areas under some information are obtained. The analysis of examples shows that the data mining technology based on rough set and Bayesian network has a good application prospect in distribution network fault diagnosis.
【学位授予单位】：华北水利水电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.13;TM711

【参考文献】