| 引用本文: | 谷海彤.地区电力调度数据仓库建模与数据处理[J].电力自动化设备,2006,(10):49-53,106 |
| .Dimensional modeling and data processing in data warehouse for regional power dispatch[J].Electric Power Automation Equipment,2006,(10):49-53,106 |
|
|
| 摘要: |
| 分析电力调度中心各业务系统中的数据,将系统数据分为设备台帐、周期采样和事件动作3种类型,后2种数据中大部分为事实数据,其数据量占数据仓库的90%以上。提出其共享维度建立数据模型的建模思路,数据仓库的数据来源是任何与需要分析主题有关的地区电力调度数据,在数据抽取、转换、清洗、装载(ETL)的过程进行数据过滤、重新选择粒度、统一编码等清洗与转换后存入数据仓库中,形成按照主题定义的数据子集或数据子立方体结构。数据仓库维度建模有第三范式(3NF)和星型2种模式,3NF模型适用于地调数据仓库中心建模;星型模型适用于数据集市中建模,利于信息检索。论述了数据仓库维度模型中的基本表——事实表和维度表,并对地区电力调度中心数据仓库的测点、日期与时间维度表进行了设计,说明了ETL数据仓库中的相关数据是操作环境对应数据,首先将逻辑数据模型转换为物理数据模型后,进行扩展性标识语言(XML)描述实现异构系统的数据共享。提出要考虑故障发生与天气、温度、湿度等气象条件和负荷变化之间的联系,给出了通过人工神经网络等特定数据挖掘方法的基本流程。 |
| 关键词: 电力系统调度,数据仓库,维度建模 |
| DOI： |
| 分类号:TM73 TP392 |
| 基金项目: |
|
| Dimensional modeling and data processing in data warehouse for regional power dispatch |
|
GU Hai-tong
|
| Abstract: |
| Based on the analysis,the data of operational systems for electric power dispatch are divided into three types:equipment description,periodical sample and events.More than 90 percent of warehouse data belong to the latter two types,most of them are real data.The method of sharing dimensional modeling is proposed.The source of data warehouse is the regional dispatch data relating to analysis.After data is processed in ETL(Extract,Transform,cleansing,Load) stage,they are saved into data warehouse,in the formation of data subset or data cube according to certain theme.There are 3NF(third Normal Form) scheme and star scheme for data warehouse dimensional modeling.The 3NF scheme is suitable for the core of data warehouse,while the star scheme is for data market to make query easier.The real data table and dimension table of dimensional model are described,and tables of sensor point dimension,date dimension and time dimension are designed.The data in ETL stage is the corresponding data of operational environments.After logical data model is transformed into physical data model,data is described in XML(eXtensible Markup Language) for data sharing between isomerous systems.It is pointed out that relations between fault and weather(temperature,humidity,etc.) or load should be considered.The basic dataflow of data mining by artificial neural networks is given. |
| Key words: electric power dispatch,data warehouse,dimensional modeling |
