資源描述:
《od估計(jì)雙層規(guī)劃模型擴(kuò)展及求解》由會(huì)員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫(kù)。
1、長(zhǎng)沙交通學(xué)院碩士學(xué)位論文OD估計(jì)雙層規(guī)劃模型擴(kuò)展及求解姓名:王正武申請(qǐng)學(xué)位級(jí)別:碩士專業(yè):交通運(yùn)輸規(guī)劃與管理指導(dǎo)教師:黃中祥20030425中文摘要[摘要]oD流是交通運(yùn)輸規(guī)劃與管理的重要依據(jù)之一,是道路網(wǎng)絡(luò)交通流預(yù)測(cè)的基礎(chǔ)。獲得OD流有兩條途徑:一是進(jìn)行大規(guī)模的∞調(diào)查,一是通過(guò)觀測(cè)路段交通量來(lái)估計(jì)。在現(xiàn)有oD估計(jì)方法中,由于雙層規(guī)劃具有良好的結(jié)構(gòu),能嵌入出行者的路線選擇行為,因而具有極其重要的地位。在分析現(xiàn)有雙層規(guī)劃模型的不足之處后,本文對(duì)現(xiàn)有模型進(jìn)行了擴(kuò)展:對(duì)于上層目標(biāo)函數(shù),引入Bayes估計(jì)和多元正態(tài)分布,能有效地消除觀測(cè)交通量的隨機(jī)誤差
2、和時(shí)間影響;對(duì)于下層目標(biāo)函數(shù),基于用戶均衡或隨機(jī)用戶均衡建模考慮了用戶路線選擇行為。擴(kuò)展后的模型能有效解決我國(guó)城市混合交通、彈性需求的oD估計(jì)難題,而且具有很強(qiáng)的適應(yīng)能力,既能適用于非擁擠網(wǎng)絡(luò),也能適用于擁擠網(wǎng)絡(luò)。本文研究了雙層規(guī)劃的求解算法,討論了兩種求解算法:一是基于凸規(guī)劃求解,一是基于靈敏度分析求解。將模型的下層目標(biāo)函數(shù)用交分不等式表示,進(jìn)行靈敏度分析,得到關(guān)于oD流的靈敏度函數(shù),利用此函數(shù)開(kāi)發(fā)了一種基于靈敏度分析的求解算法。將此算法應(yīng)用于較大規(guī)模交通網(wǎng)絡(luò)進(jìn)行oD估計(jì),得到了較為滿意的結(jié)果。[關(guān)鍵詞]OD估計(jì)混合交通彈性需求用戶均衡雙層規(guī)
3、劃靈敏度分析變分不等式ABSTRACT【ABSTRACT】OD(Origin—Destination)flowisnotonlyoneofthefoundation011traffictransportationPlanning,andmanagement,butalsothebasisofforecastingtrafficvolumeontheroadnetwork.TherearetwowaystoobtainODflow,oneisalargescaleODsurvey,theotherisestimationbythe1inkacco
4、untsontheroadnetwork.AfteranalyzingandevaluatingalmostallmethodsonODestimationinthepresent,thispaperdiscussesthebilevelprogramming,whichnotonlyhasgoodstructure,butalsocantakeintoaccounttheroutechoicebehavior.Onthebasisofpresentachievementonthebilevelprogramming,thenewbilevel
5、programmingisextended.Theextendedbilevelprogramminghasmanyadavantages.Firstlytheupperoptimizationmodel,whichcombinesBayesandmuti—variablenormalmodel,caneliminatetherandomerrorsandtimeeffectsonlinkaccountsandhassimplystructure.SecondlyintheloweroptimizationmodelfollowingWardr
6、opequilibriumorStochasticUserEquilibrium,theroutechoicebehaviorisconsidered.ThirdlytheloweroptimizationmodelissuitabletosuchODestimationproblemthatthetrafficismixedandthedemandiselastic.Fourthlythisbilevelprogrammingissuitabletonotonlycongestedroadnetwork,butuncongestedroadn
7、etwork.Thispaperdiscussestwokindsofarithmeticofsolvingthebilevelprogramming,oneisbasedonconvexprogramming,theotherisbasedonsensitivityanalysis.Afterthelowerprogrammingisexpressedbyvariationalinequation,thesensitivityanalysisiscarriedout.Then,thesensitivityanalysisfunctiononO
8、Ddemand,whichisusedtosolvethebilevelprogramming,isfoundout.Atthesametime,th