一种基于数据融合和方法融合的时空综合预测算法.pdf
文本预览下载声明
计算机研究与发展 IsSN1000一1239/CN11.1777/TP
of Researchand 1255~1
JournalConlputer I)evelopment 42(7): 260,2005
一种基于数据融合和方法融合的时空综合预测算法
徐 薇1 黄厚宽1 王英杰2
1(北京交通大学计算机与信息技术学院北京100044)
2(中国铁道科学研究院电子计算技术研究所北京 100081)
(xuwe妇zy@sohu.com)
An BasedonDataFusionand
IntegratedSpatio-TemporalForecastingApprOach
MethOdFusiOn
Xu
Weil,HuangHoukuanl,andWangYin西ie2
1(&^00Z n行d 100044)
o,CD聊加£Pr J挖如r卅nfio押%^oZq纠,&巧i行gJi∞幻理g‰i傀玲泐,&巧i行g
100081)
2(j起盯i£“据o,C0埘加fi玎gTk^舢Z9到,吼i行口A∞如my盯Rniz伽ySfi硎伽,&巧i竹g
Abstract data isan researchindata inwhich
Spatio—temporalminingimportanttopic mining,andspatio—
is most used.Inordertoovercomethe
the limitationsofcurrent
temporalforecasting widely spatio—temporal
this a basedondatafusionand
forecastingmethods,paperproposesspatio—temporalforecastingapproach
methodfusion.The firstforecaststime ofthe itselfwithstatistical
approach sequence targetobject
and influencesof neuralnetwork then
principlescomputes neighboringobjectsemploying technique,
forecastsmixed the
data and
sequenceusingspatio—temporalauto—regressivemodel, finallyintegrates
individualtime forecastand forecast linear to
sequenceforecast,spatialspatio—temporalthroughregression
deliverthefinalresult.The was usedinthe of flowin
approachsuccessfully
显示全部