根據(jù)公安部交通管理局官方微博和公安部網(wǎng)站發(fā)布的數(shù)據(jù),2018年全國新注冊(cè)登記機(jī)動(dòng)車3,172萬輛,機(jī)動(dòng)車保有量已達(dá)3.27億輛,其中汽車2.4億輛,小型載客汽車首次突破2億輛;機(jī)動(dòng)車駕駛?cè)诉_(dá)4.09億人,其中汽車駕駛?cè)?.69億人。截至2018年底全國汽車保有量達(dá)2.4億輛,比2017年增加2285萬輛,增長(zhǎng)10.51%。面對(duì)如此多的汽車保有量,尤其是私家車的數(shù)量,雖然帶給了人們諸多便利,但也加重了城市交通道路的負(fù)擔(dān),使得本來就高負(fù)荷運(yùn)轉(zhuǎn)的城市公路更加擁堵不堪,尤其是上下班高峰時(shí)期,大量因?yàn)槎萝嚩诘缆飞系能囕v尾氣的排放也加重了環(huán)境的污染。面對(duì)以上問題,目前的解決方法主要有以下幾個(gè)方法:第一,控制汽車的數(shù)目,限制汽車的購買;第二,加強(qiáng)城市基礎(chǔ)道路建設(shè),擴(kuò)建道路;第三,增加公共交通或修建地鐵輕軌等;第四,搭建智能交通系統(tǒng)。前三種方法可以在一定程度上緩解交通的壓力,但是他們不能從根本上解決城市交通擁堵的問題。隨著互聯(lián)網(wǎng)+熱潮的興起,近幾年智能交通領(lǐng)域受到交通管理部門和相關(guān)企業(yè)越來越多的重視,交通智能領(lǐng)域迎來了新的發(fā)展機(jī)遇。交通預(yù)測(cè)是智能交通系統(tǒng)（ITS）的重要組成部分,在交通網(wǎng)絡(luò)規(guī)劃、... 

【文章來源】：華中師范大學(xué)湖北省211工程院校教育部直屬院校

【文章頁數(shù)】：73 頁

【學(xué)位級(jí)別】：碩士

【文章目錄】：
Abstract
Nomenclature
1 Introduction
    1.1 Research Background And Significance
    1.2 Research Status at Home And Abroad
        1.2.1 Prediction Methods Based on Statistical Theory
        1.2.2 Prediction Methods Based on Neural Network
        1.2.3 Prediction Methods Based on Hybrid Model
    1.3 The Research Content And Innovation of This Thesis
        1.3.1 The Main Research Content of This Thesis
        1.3.2 The Main Innovation of This Thesis
        1.3.3 Structure of This Thesis
2 Related Theoretical Research
    2.1 Related Concepts of Traffic Flow
        2.1.1 The Main Characteristic Parameters of Traffic Flow
        2.1.2 Traffic Flow Characteristics And Influencing Factors
        2.1.3 The Basic Flow of Traffic Flow Velocity Prediction
    2.2 Time Series Theory
    2.3 Support Vector Machine Regression (SVR)
    2.4 A Brief Introduction to The Neural Network Correlation Model
        2.4.1 Mathematical Model of Neuron
        2.4.2 Recurrent Neural Network
        2.4.3 LSTM Neural Network
3 Traffic Speed Prediction Algorithm Based on LSTM Neural Network And SVRCombination Model
    3.1 Preamble
    3.2 The Concept of Combined Forecasting Models
    3.3 Principles of Combined Forecasting Models
    3.4 Prediction Algorithm Based on LSTM-SVR Hybrid Model
    3.5 Evaluation Index of The Model
    3.6 Data Preprocessing
    3.7 Experiment And Result Analysis
        3.7.1 Model Training
        3.7.2 Result Analysis
    3.8 Summarizes of This Chapter
4 Fusion Model Based on New Information Source And Seq2Seq-LSTM-SVR
    4.1 Preamble
    4.2 A Discovery Algorithm For Extracting New Information Sources
    4.3 LSTM Neural Network Based on Seq2Seq+Attention Model
        4.3.1 Introduction of Seq2Seq Model With Attention Mechanism
        4.3.2 Build The Seq2Seq-LSTM Neural Network With Integrate NewInformation Sources
    4.4 Experimental Analysis
    4.5 Summarizes of This Chapter
5 Summary And Future Work
    5.1 Summarize
    5.2 Future Work
References
Appendices
摘要
Acknowledgements


【參考文獻(xiàn)】：
期刊論文
[1]改進(jìn)人工蜂群算法優(yōu)化RBF神經(jīng)網(wǎng)絡(luò)的短時(shí)交通流預(yù)測(cè)[J]. 黃文明,徐雙雙,鄧珍榮,雷茜茜.  計(jì)算機(jī)工程與科學(xué). 2016(04)
[2]一種基于非參數(shù)回歸的交通速度預(yù)測(cè)方法[J]. 史殿習(xí),丁濤杰,丁博,劉惠.  計(jì)算機(jī)科學(xué). 2016(02)
[3]基于動(dòng)態(tài)學(xué)習(xí)策略的群集蜘蛛優(yōu)化算法[J]. 王艷嬌,李曉杰,肖婧.  控制與決策. 2015(09)
[4]遞推SOM神經(jīng)網(wǎng)絡(luò)在短時(shí)交通流預(yù)測(cè)中的應(yīng)用[J]. 黃杰,李軍,郭翔.  公路. 2015(04)
[5]DE優(yōu)化T-S模糊神經(jīng)網(wǎng)絡(luò)的交通流量預(yù)測(cè)[J]. 侯越.  計(jì)算機(jī)工程與設(shè)計(jì). 2013(09)
[6]基于Adaboost的BP神經(jīng)網(wǎng)絡(luò)改進(jìn)算法在短期風(fēng)速預(yù)測(cè)中的應(yīng)用[J]. 吳俊利,張步涵,王魁.  電網(wǎng)技術(shù). 2012(09)
[7]基于影響模型的短時(shí)交通流預(yù)測(cè)方法[J]. 丁棟,朱云龍,庫濤,王亮.  計(jì)算機(jī)工程. 2012(10)
[8]粒子群優(yōu)化RBF神經(jīng)網(wǎng)絡(luò)的短時(shí)交通流量預(yù)測(cè)[J]. 馮明發(fā),盧錦川.  計(jì)算機(jī)仿真. 2010(12)
[9]非線性短時(shí)交通流的一種神經(jīng)網(wǎng)絡(luò)預(yù)測(cè)方法[J]. 華冬冬,陳森發(fā).  現(xiàn)代交通技術(shù). 2004(01)



本文編號(hào)：3630598