基于数据挖掘的电网数据智能分析的研究与实现

【作者】 刘芳；

【导师】 刘晓霞；

【作者基本信息】 西北大学 ， 计算机软件与理论， 2008， 硕士

【摘要】 数据挖掘是致力于数据分析和理解、揭示数据内部蕴藏知识的技术,是从大量的含有噪声的数据中挖掘出隐含其中的知识和信息,是当前数据分析的先进手段之一。数据挖掘的模式主要包括分类模式、聚类模式、时间序列模式、关联模式、序列模式等。电网数据智能分析系统借助数据挖掘领域中的各种算法模型对电网的电力设备故障、日报数据、运行数据进行智能分析,通过对大量初始记录数据的清理,根据电网安全运行特点提炼出与分析因素有联系的记录数据,装载到数据仓库,然后对其进行相应挖掘算法的处理,得到需要的知识,为保障电网的安全运行提供理论支持。电网数据智能分析系统主要包括数据ETL、知识挖掘、数据动态更新和可视化显示四个子系统。数据ETL实现数据的清洗、整理和装载,在保证不减少数据所包含信息的前提下改善数据质量,提高数据挖掘算法的性能;知识挖掘部分是系统的核心,运用合适的挖掘算法模型对数据仓库中的数据进行挖掘,得到需要的知识;数据动态更新指随电网数据的变化实现动态知识挖掘;可视化显示实现挖掘结果的图形显示。本文首先概述数据挖掘的基本概念和数据挖掘模型,阐明了电网数据智能分析系统的设计思想和体系结构,其次详细论述系统中采用的数据ETL、数据挖掘的关联规则算法、时序预测算法、数据的动态更新等关键技术及其软件实现,然后分析了系统测试结果,最后给出结论和展望。本项目课题已通过西北电网公司的验收,并在西北电网智能数据分析中得以使用,取得良好的效果。更多还原

【Abstract】 In general, data mining is an advanced technology for data analysis,and it focuses on analyzing and understanding data and revealing the essence knowledge and information hidden in some large data sets. In other words, data mining trends to find the useful knowledge and information from some large data sets with the noise information. The patterns of data mining include: classification, clustering, time series, association, sequence, etc. The intelligent analysis of power grid data is to employ different algorithms from the data mining field to analyze intelligently the faults about the electric power equipments, the daily-report data, and the implementary data. More detailedly, the intelligent analysis for electric grid data is based on two steps: first, according to the features from the electric grid implementation and the analyzed factors, it extracts and analyzes some related data from the initial data, and then stores the related data into data storehouse; second, by applying some data mining algorithms, we can obtain some useful knowledge, which plays a theoretical foundation for the security about the electric grid implementation.The intelligent analysis for electric grid data includes four parts, such as: data ETL, knowledge mining, data dynamic update, and data visualization. In general, data ETL implements the cleanout, organization, and loading for data, and it can effectively improve data quality, leading to further boost the performance of data mining algorithms. The knowledge mining plays a significant role in the intelligent analysis for electric grid data, and it can obtain the useful knowledge by applying the data mining algorithms to the data sets from the data storehouse. The data dynamic update focuses on implementing the dynamic knowledge mining on the basis of the variety about electric grid data. The data visualization trends to visualize the finial results.In this thesis, firstly, I give a brief review about the background of data mining (such as some basic concepts, models, etc), and further present the design idea and system configuration about the intelligent analysis for electric grid data. Secondly, I detailedly address the system based on the significant data mining technologies such as data ETL, conjunction rule, time series, data dynamic update, and further I also present some appealing results on the basis of implementing the data mining algorithms via software. Finally, some conclusions are made, and some research directions are also addressed.更多还原