数据仓库与数据挖掘若干理论研究及在CIPS中的应用

《数据仓库与数据挖掘若干理论研究及在CIPS中的应用》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、同济大学电子与信息工程学院博士学位论文数据仓库与数据挖掘若干理论研究及在CIPS中的应用姓名：闫伟申请学位级别：博士专业：系统工程指导教师：张浩20050601摘要理论依据。采用加权聚类方法对关键设备的大量历史数据进行分析，建立了设备正常情况的聚类模型，提出了紧密(离核)指数、离群指数和综合指数的计算方法，实现了对设备有效的实时监控。蚁群算法是一种比较先进的模拟进化算法，采用蚁群算法改进了聚类分析。并根据蚁群算法中的P。O)综合考虑了簇团内数据点个数和到簇团内中心点距离的特点，创造性提出一种新的离群数据挖掘的方法，得到较好的效果。关键词：数据仓库，数据挖掘，

2、ETL框架，概念模型，模糊加权关联规则，时间序列分析，时间序列拓扑，加权聚类分析，蚁群算法，离群数据挖掘，CIPSIIAbstractABSTRACTWiththerapiddevelopmentofcomputertechnologies，networkanddatastoragetechnique，largeamountsofdataarestored．Therapidgrowthdemandformanaginggrowingmountainsofdata，extractingandunderstandingusefulknowledgefromthe

3、sedataoutpacesthetraditionaldatabaseanddataanalysismethods，whichleadstotheemergingfieldofdatawarehouseanddatamining．Thefocusofthedissertationisonsometheoriesofdatawarehouseanddatamining，whichisusefulinaprocessindustry--anPetroleumandNaturalGasCooperation．Datawarehouseneedhighqualit

4、ydatabycleaningdataanddiscardingmissingdata．ThedissertationfirstdesignsaspecialETLtoolonthedifferentdatasourceandformat．Inwhich，thetechnologyproblemsandsolvingprojectsareanalyseswhendatacleaningandtheCanopymethodarestudiedindetail．DataWarehousewiththesubjectiscost、outputofoilandgas

5、、storageofgasandbyproductsaredesignedbyStarmodel、DWERmodel、StarChainERmodelandinquiredbyBIBeans(OLAP)．Secondly,thedissertationfocusesontheresearchandapplicationofassociationrules．BecauseequipmentSparametershavethecharactersofdifferentweightandcontinuity,thelargehistorydataareanalyz

6、edbyweightedapriorialgorithmandfuzzyapriorialgorithmseparately．Then，anovelalgorithm--thefuzzyweightedassociationrolesisputforwardbytheweightedvaluethatisgotbycombiningitsimportantdegreeandsubjectfunctionvalueandtheFuzz鼍NedApriorialgorithm．Dependingonthevaluablerules，theequipmentsca

7、ncirculateinexcellentstatus．Thirdly,thedissertationaddressestheresearchandapplicationoftimesequentialanalysis．ThelargehistorydatabaseareanalyzedbyfuzzytheoryandscatteredbyTime—windowapproach．DealingwithtimeIntervals，anewtime．intervalsequentialdatabaseiSgot．Inordertofindtime-interva

8、lrules，thetime．intervalApr