基于軌跡數(shù)據(jù)的長距離路徑通行時間估計問題
發(fā)布時間：2018-11-23 10:18
【摘要】：導航系統(tǒng)對于解決城市交通擁堵,緩解交通壓力具有重要意義,而結(jié)合路況的路徑通行時間估計是導航中的基礎和關(guān)鍵。隨著車輛軌跡數(shù)據(jù)的大量積累,使得估計動態(tài)路網(wǎng)中路徑的通行時間變?yōu)榭赡?即針對給定的起點和終點,對不同路徑的通行時間進行預測,從而找出通行時間最短的路徑。然而通行時間最短的路徑并不一定是累積概率分布最大的路徑。當用戶需要在指定時刻前抵達的時候,獲得累積概率分布最大的路徑就能發(fā)揮很大的作用。為了找出動態(tài)路網(wǎng)中累積分布最大的路徑,就需要對路徑通行時間的概率進行估計,而非僅得到一個單一值的估計結(jié)果。在現(xiàn)有的研究方案中,研究人員將整個道路網(wǎng)劃分為以路段為基本單位的網(wǎng)絡結(jié)構(gòu),基于路段對路徑的通行時間進行估計,然而這種依賴于路段組合的方式忽略了完整路徑通行中十字路口的拐彎時間和紅綠燈的等候時間等,導致路徑較長時估計結(jié)果更不準確。與基于路段的研究方案不同,為了提高估計的準確度和效率,本文提出基于子路徑的路徑通行時間估計方案。為了提高估計的效率,本文利用歷史軌跡數(shù)據(jù)建立后綴索引樹的存儲結(jié)構(gòu),將實時獲取的軌跡通行時間存儲在后綴索引樹的節(jié)點上,對于數(shù)據(jù)稀疏的子路徑,由歷史數(shù)據(jù)提供通行時間的結(jié)果,通過這種存儲結(jié)構(gòu)可以快速地獲取查詢路徑的子路徑序列,及其相應的軌跡通行時間。為了提高估計的準確性,本文對子路徑序列采用線性插值算法和基于時空相關(guān)性的預測算法對其通行時間的概率估計進行驗證,并采用2016年及2017年1月哈爾濱市出租車的軌跡數(shù)據(jù)集驗證了算法的準確性和效率。
[Abstract]:Navigation system plays an important role in solving urban traffic congestion and relieving traffic pressure, and the estimation of road passage time combined with road condition is the basis and key of navigation. With the accumulation of vehicle track data, it is possible to estimate the passage time of the path in the dynamic road network, that is, to predict the passage time of different paths according to the given starting point and the end point, so as to find out the shortest path. However, the shortest path is not always the path with the largest cumulative probability distribution. When the user needs to arrive before the specified time, the path with the largest cumulative probability distribution can play a significant role. In order to find the path with the largest cumulative distribution in the dynamic road network, it is necessary to estimate the probability of the passage time of the path, rather than to get the result of a single value. In the existing research scheme, the researchers divide the whole road network into a network structure based on the road section, and estimate the passage time based on the road section. However, in this way, the intersection time and the waiting time of the traffic lights in the complete path are ignored, which leads to the inaccurate estimation results when the path is longer. In order to improve the accuracy and efficiency of the estimation, a subpath-based approach is proposed to estimate the passage time. In order to improve the efficiency of the estimation, the storage structure of the suffix index tree is established by using the historical track data, and the track passage time obtained in real time is stored on the node of the suffix index tree. By using the historical data to provide the result of the passage time, the subpath sequence of the query path and the corresponding path passage time can be obtained quickly by this storage structure. In order to improve the accuracy of the estimation, the linear interpolation algorithm and the prediction algorithm based on spatio-temporal correlation are used to verify the probability estimation of the passage time of the subpath sequence. The accuracy and efficiency of the algorithm are verified by using the track data set of Harbin taxis in 2016 and 2017.
【學位授予單位】：哈爾濱工業(yè)大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：U495
[Abstract]:Navigation system plays an important role in solving urban traffic congestion and relieving traffic pressure, and the estimation of road passage time combined with road condition is the basis and key of navigation. With the accumulation of vehicle track data, it is possible to estimate the passage time of the path in the dynamic road network, that is, to predict the passage time of different paths according to the given starting point and the end point, so as to find out the shortest path. However, the shortest path is not always the path with the largest cumulative probability distribution. When the user needs to arrive before the specified time, the path with the largest cumulative probability distribution can play a significant role. In order to find the path with the largest cumulative distribution in the dynamic road network, it is necessary to estimate the probability of the passage time of the path, rather than to get the result of a single value. In the existing research scheme, the researchers divide the whole road network into a network structure based on the road section, and estimate the passage time based on the road section. However, in this way, the intersection time and the waiting time of the traffic lights in the complete path are ignored, which leads to the inaccurate estimation results when the path is longer. In order to improve the accuracy and efficiency of the estimation, a subpath-based approach is proposed to estimate the passage time. In order to improve the efficiency of the estimation, the storage structure of the suffix index tree is established by using the historical track data, and the track passage time obtained in real time is stored on the node of the suffix index tree. By using the historical data to provide the result of the passage time, the subpath sequence of the query path and the corresponding path passage time can be obtained quickly by this storage structure. In order to improve the accuracy of the estimation, the linear interpolation algorithm and the prediction algorithm based on spatio-temporal correlation are used to verify the probability estimation of the passage time of the subpath sequence. The accuracy and efficiency of the algorithm are verified by using the track data set of Harbin taxis in 2016 and 2017.
【學位授予單位】：哈爾濱工業(yè)大學
【學位級別】：碩士
【學位授予年份】：2017
【分類號】：U495
【參考文獻】
相關(guān)期刊論文 前1條
1 趙新正;李夢雪;李秋平;李同f;芮e，
本文編號：2351235
本文鏈接：http://www.lk138.cn/kejilunwen/daoluqiaoliang/2351235.html
