| 会议名称： |
Proceeding of 11th International Conference on Computers in Urban Planning and Urban Management |
| 全部作者： |
Jiasong Zhu*,Anthony Gar-On |
| 出版年份： |
2009 |
| 会议地址： |
Hong Kong, China |
| 页 码： |
16-18 |
| 查看全本： |
|
Areliableandaccurateshort-termtrafficforecastingsystemiscrucialforthesuccessfuldeploymentofanyintelligenttransportationsystem.Alotofforecastingmodelshavebeendevelopedinrecentyearsbutnoneofthemcouldconsistentlyoutperformtheothers.Inreal-worldapplications,trafficforecastingaccuracycanbeaffectedbyalotoffactors.Impactsoflong-termchangestotrafficpatternstoshort-termtrafficforecastingareprofoundandthiscaneasilymakeanexistingforecastingsystemoutdated.Therefore,itisveryimportantforforecastingsystemstodetectlong-termchangesintrafficpatternsandmakeupdatesaccordingly.Thispaperpresentsanewforecastingmechanism,inwhichadynamichybridapproachistakenandself-learningabilityisenhanced.Resultsofacasestudyshowtheproposedapproachisfeasibleinenhancingtheadaptabilityoftrafficforecastingsystems.
