【摘要】：近年來鐵路總公司積極推進貨物運輸改革,努力提高鐵路貨運服務(wù)水平,加強鐵路物流化建設(shè),力求擴大鐵路運輸在貨物運輸市場中的份額。但是目前鐵路貨運中最突出的問題是貨物運輸?shù)臅r間具有很大的不確定性,大大制約了鐵路運輸?shù)母偁幜�。雖然鐵路部門制定了鐵路貨物運到期限,但是貨物運輸時間仍會受到車流集結(jié)、車流改編等運輸作業(yè)的影響,因此鐵路部門需要加強對車流的有效掌控,提升運輸組織效率。而準(zhǔn)確的車流預(yù)測可以實現(xiàn)對車流分布的有效掌控,可以及時避免車流擁堵的發(fā)生,保證路網(wǎng)的運輸效率。車流預(yù)測的基礎(chǔ)是車流徑路,確定合理的車流徑路,對于準(zhǔn)確地進行車流預(yù)測、高效地進行運輸組織具有重要作用。本文以確定合理的車流徑路為研究目標(biāo),借助鐵路運輸信息集成平臺提供的各業(yè)務(wù)系統(tǒng)整合的數(shù)據(jù),構(gòu)建車流運行徑路獲取模型來獲取車流運行過程中的真實走行徑路,并利用Hadoop平臺處理車流運行徑路大數(shù)據(jù)集,對車流經(jīng)由的每個車站構(gòu)建針對不同貨種的車流運行徑路模式和概率后綴樹,并在車流預(yù)測系統(tǒng)中進行驗證。具體工作如下:(1)對反映車流實際走行情況的車流運行徑路進行研究,借助鐵路運輸信息集成平臺獲取經(jīng)過整合和共享的鐵路業(yè)務(wù)數(shù)據(jù),建立徑路節(jié)點映射模型、車輛報文匹配模型、徑路序列拼接模型來獲取完整的車流運行徑路。(2)考慮到鐵路車流運行徑路的數(shù)據(jù)量會隨著時間的累積不斷增大,傳統(tǒng)數(shù)據(jù)分析方法不能實現(xiàn)對不斷增加的海量數(shù)據(jù)的有效分析,本文設(shè)計大數(shù)據(jù)分析方法利用Hadoop平臺來處理車流運行徑路數(shù)據(jù),利用Sqoop工具實現(xiàn)傳統(tǒng)關(guān)系型數(shù)據(jù)庫與分布式文件系統(tǒng)HDFS之間的數(shù)據(jù)傳輸,利用MapReduce編程模型高效處理車流運行徑路大數(shù)據(jù)集,利用變階馬爾科夫模型對車流運行徑路的徑路序列進行處理,建立運行徑路模式并構(gòu)建概率后綴樹,用以預(yù)測車流運行徑路。(3)利用Java編程實現(xiàn)基于運輸信息集成平臺的車流運行徑路獲取過程,搭建Hadoop平臺進行MapReduce編程開發(fā)來處理車流徑路大數(shù)據(jù),并實現(xiàn)車流運行徑路模式提取和概率后綴樹的構(gòu)建,以不同貨種、不同徑路模式預(yù)測車流運行徑路,并在車流預(yù)測系統(tǒng)中進行驗證。
[Abstract]:In recent years, the railway corporation has actively promoted the reform of freight transport, made great efforts to improve the level of railway freight service, strengthened the construction of railway logistics, and made every effort to expand the share of railway transportation in the freight transport market. However, the most prominent problem in railway freight transportation is the uncertainty of the time of freight transportation, which greatly restricts the competitiveness of railway transportation. Although the railway department has established the railway freight delivery deadline, but the freight transport time will still be affected by the train flow assembly, the train flow adaptation and other transportation operations, so the railway department needs to strengthen the effective control of the train flow and improve the transport organization efficiency. The accurate forecast of traffic flow can effectively control the distribution of traffic flow, avoid traffic congestion in time, and ensure the transportation efficiency of the road network. The basis of the traffic flow prediction is the traffic flow path and the determination of the reasonable traffic flow path plays an important role in accurate traffic flow prediction and efficient transportation organization. Based on the integrated data of various business systems provided by the railway transportation information integration platform, this paper constructs a model to obtain the real path of the train flow in the course of the train flow operation, aiming at the determination of the reasonable path of the vehicle flow, and with the help of the integrated data of the various business systems provided by the railway transportation information integration platform, The Hadoop platform is used to deal with the big data set of the train flow running path, and the train flow path mode and probability suffix tree for different kinds of goods are constructed for each station through which the traffic flow is processed, and verified in the traffic flow prediction system. The main works are as follows: (1) the paper studies the running path of the train flow which reflects the actual traffic flow, obtains the integrated and shared railway business data with the aid of the railway transportation information integration platform, and establishes the mapping model of the path node. Vehicle message matching model, path sequence splicing model to obtain the complete train flow running path. (2) considering that the amount of data of railway train flow running path will increase with time, The traditional data analysis method can not realize the effective analysis of the increasing mass data. In this paper, the big data analysis method is designed to use the Hadoop platform to deal with the traffic flow path data. The data transmission between traditional relational database and distributed file system (HDFS) is realized by using Sqoop tool, and the MapReduce programming model is used to efficiently deal with the big data set of train flow running path. The variable order Markov model is used to deal with the train flow path sequence, the running path pattern is established and the probability suffix tree is constructed. It is used to predict the running path of the vehicle flow. (3) the process of obtaining the running path of the train flow based on the integrated platform of transportation information is realized by using Java programming, and the Hadoop platform is built for MapReduce programming to deal with the big data of the train flow path. The train flow path pattern extraction and the construction of probability suffix tree are realized. The traffic flow path pattern is predicted by different goods and different path modes, and verified in the traffic flow prediction system.
【學(xué)位授予單位】：北京交通大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2017
【分類號】：U294.1;TP311.13

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 王龍;馬建軍;林柏梁;陳雷;溫旭紅;;鐵路網(wǎng)重空車流徑路整體優(yōu)化模型[J];北京交通大學(xué)學(xué)報;2014年06期

2 李瑞鴻;;云計算相關(guān)技術(shù)研究[J];中國科技信息;2014年22期

3 王啟東;;面向數(shù)據(jù)整合的鐵路運輸信息集成平臺方案研究[J];鐵路計算機應(yīng)用;2014年08期

4 王鐵毅;;云計算平臺系統(tǒng)運行模式設(shè)計研究[J];山東冶金;2014年03期

5 顏昌盛;高明星;趙s，

本文編號：2197048