您好,欢迎访问三七文档
当前位置:首页 > 行业资料 > 交通运输 > 基于BP神经网络的交通流量预测
分类号:U4910710-2009132036硕士学位论文基于BP神经网络的交通流量预测曹虹导师姓名职称许宏科教授申请学位级别硕士学科专业名称交通信息工程及控制论文提交日期2012年5月4日论文答辩日期2012年5月28日学位授予单位长安大学TrafficFlowPredictionBasedonBPNeuralNetworkADissertationSubmittedfortheDegreeofMasterCandidate:CaoHongSupervisor:Prof.XuHongkeChang’anUniversity,Xi’an,Chinai摘要交通信息预测是智能交通控制、交通诱导、交通信息服务等智能交通系统(ITS)实现的重要基础,是ITS领域的重要理论之一,而交通流预测问题又是交通信息预测的核心问题。因此,进行交通流量预测理论体系的研究,是开发实用、智能化的交通量预测系统的前提,对于改善我国交通拥堵问题,具有十分重要的学术价值和现实意义。多年来,交通预测者一直将提高交通信息预测的可靠性作为研究重点。论文将神经网络技术研究与交通信息预测研究紧密结合,将神经网络技术应用于交通流量预测;结合实际数据,采用BP神经网络、RBF神经网络、小波神经网络,使用Matlab平台实现预测,并使用Matlab提供的图形界面开发环境GUIDE,实现了BP神经网络的图形化界面仿真。论文首先通过新浪网的调查数据,说明我国面临的严重交通问题;其次,参考国内外解决交通问题的措施,总结共同点,引出了论文的研究重点交通流预测;接着,在总结国内外研究成果的基础上,对已有的交通流预测方法进行了研究,通过分析将现有的方法分为常规预测和智能预测,并对每种研究的基本方法进行了具体介绍;再次,重点研究了智能预测中的人工神经网络理论,介绍了神经网络的发展、特点、结构以及学习理论,分析了BP、RBF、小波神经网络的具体步骤,为交通流预测的实现奠定理论基础;然后,结合实际交通流量观测数据,分析了将BP、RBF、小波神经网络理论应用于交通流量预测的过程,详细描述了网络参数的选取过程,并使用Matlab平台实现预测,通过对比预测结果,认为BP神经网络的预测效果较好;最后,采用Matlab提供的图形化界面开发环境,设计实现了BP神经网络的图形界面仿真。关键词:交通流预测、智能预测方法、BP神经网络、RBF神经网络iiiAbstractTrafficinformationpredictionisanimportantfoundationforintelligenttrafficcontrol,trafficguidance,trafficinformationservicesandotherITSsubsystems,andalsoisoneoftheimportanttheoryoftheintelligenttransportationsystem(ITS)field.Trafficflowpredictionhasplayedakeyroleinthetrafficinformationprediction.Therefore,theresearchontrafficflowpredictionispremisetodeveloppracticalandintelligenttrafficforecastingsystem,whichhasaveryimportantacademicvalueandpracticalsignificancetoimprovethetrafficcongestionproblemsinChina.Overtheyears,trafficscholarshavebeentoimprovethereliabilityofthepredictedtrafficinformationasaresearchfocus.Thepurposeofthepaperistocombineadvancedneuralnetworktechnologywithtrafficflowpredictionclosely,andalsoistopredicttrafficflowwithneuralnetwork.ThepredictionsweredoneusingtheactualdatawithBPneuralnetwork,RBFneuralnetworkandwaveletneuralnetworkthroughtheMatlabplatform.Ontheend,thegraphicalinterfaceoftheBPneuralnetworksimulationwasdesignedthroughthegraphicalinterfacedevelopmentenvironment(GUIDE)ofMatlab.First,accordingtoSinasurveydata,serioustrafficproblemsinourcountrywereshowed.Onthebasisofsummingupthemeasurestosolvetrafficproblemsathomeandabroad,thepaperfocusedonthepointoftrafficflowforecasting.Onthebasisofresearchachievementsathomeandabroad,theexistingtrafficflowpredictionmethodswereanalyzed,whichweredividedintoconventionalprediction,intelligentpredictionandcombinationforecastinginthispaper.Thebasicmethodsofeachpredictionmethodwereintroduced.Secondly,focusedontheartificialneuralnetworktheory,weintroducedtheneuralnetworkdevelopment,characteristics,structure,andlearningtheory.AndthenweanalyzedthespecificstepsofBPandRBFandwaveletneuralnetwork.Thirdly,usingMatlabtotraintheactualtrafficflowobservationaldata,theBPneuralnetwork,RBFneuralnetworkandWaveletneuralnetworkwereappliedtotrafficflowforecastingandthepredictedresultswerecompared.Finally,thesimulationofthegraphicalinterfaceoftheBPneuralnetworkwasdesignedthroughtheivMatlabGUIDEdevelopmentenvironment.Keywords:trafficflowprediction;intelligentprediction;BPneuralnetwork;RBFneuralnetworkv目录第一章绪论..............................................................................................................................11.1研究背景........................................................................................................................11.2国内外研究现状............................................................................................................11.2.1国外交通解决措施..............................................................................................11.2.2交通流预测研究现状..........................................................................................21.3研究内容........................................................................................................................6第二章交通流预测方法研究..................................................................................................82.1交通流预测综述............................................................................................................82.1.1研究内容..............................................................................................................82.1.2预测流程..............................................................................................................82.1.3研究方法分类....................................................................................................102.2常规预测方法..............................................................................................................112.2.1移动平均法........................................................................................................112.2.2指数平滑法........................................................................................................122.2.3趋势曲线法........................................................................................................142.3智能预测方法..............................................................................................................152.3.1基于灰色系统理论的预测方法....
本文标题:基于BP神经网络的交通流量预测
链接地址:https://www.777doc.com/doc-1849433 .html