邵中華 李竹 崔艷
摘 要: 針對(duì)目前主干道車輛出行時(shí)間估計(jì)考慮道路狀況隨機(jī)性,但在時(shí)效性和精確性方面有所欠缺,提出一種新的估計(jì)方法。以視頻檢測(cè)器為工具,采集車輛在某一區(qū)間的行駛時(shí)間,計(jì)算其平均速度并將該區(qū)間的道路劃分為三種狀態(tài),考慮到估計(jì)的時(shí)效性和計(jì)算的數(shù)據(jù)量,利用滑動(dòng)窗口選取一定量的狀態(tài)數(shù)據(jù)加入遺傳因子構(gòu)建轉(zhuǎn)移概率矩陣,獲知下一時(shí)刻所有出現(xiàn)的狀態(tài)及其對(duì)應(yīng)的概率,而這些狀態(tài)對(duì)應(yīng)的出行時(shí)間的數(shù)學(xué)期望就是主干道出行時(shí)間的估計(jì)值。在山西省臨汾市的主干道上,應(yīng)用浮動(dòng)車法對(duì)模型的準(zhǔn)確性和時(shí)效性進(jìn)行驗(yàn)證。實(shí)驗(yàn)結(jié)果表明模型具有較高的估計(jì)精度。
關(guān)鍵詞: 城市交通; 出行時(shí)間; 馬爾科夫鏈; 主干道; 多狀態(tài); 滑動(dòng)窗口
中圖分類號(hào): TN911.1?34; TP393.07 文獻(xiàn)標(biāo)識(shí)碼: A 文章編號(hào): 1004?373X(2018)13?0092?03
Abstract: The randomness of road conditions is considered in current travel time estimation of main road, which has the defects of timeliness and accuracy. Therefore, a new travel time estimation method of main load is proposed. The video detector is taken as the tool to acquire the travel time of vehicle in a certain interval. The average speed of vehicle in the traveling interval is calculated, and the road in this interval are divided into three states. Considering the timeliness of estimation and data size of calculation, a certain amount of state data is selected by the sliding window, and added with genetic factor to construct the transfer?probability matrix, so as to obtain all the states appearing in the next moment and their corresponding probabilities. The mathematical expectation of the travel time corresponding to the states is defined as the travel time estimation value of main road. The floating car method is used to verify the accuracy and timeliness of the model in the main road of Linfen City of Shanxi Province. The experimental results show that the model has high estimation precision.
Keywords: urban traffic; travel time; Markov chain; main road; multi?state estimation; sliding window
3.3 主干道出行時(shí)間估計(jì)及分析
通過(guò)臨汾市路段兩端設(shè)置的交通狀況視頻監(jiān)測(cè)器,記錄車輛進(jìn)入和離開該路段時(shí)視頻所顯示的時(shí)間點(diǎn),獲取任意車輛在該路段的行程時(shí)間,將其算術(shù)平均數(shù)作為該路段當(dāng)前時(shí)刻的行程時(shí)間估計(jì)值。
實(shí)時(shí)統(tǒng)計(jì)所得的路段各狀態(tài)下出行時(shí)間的估計(jì)值,如表1所示。
表1中,數(shù)據(jù)147.4表示路段1在A狀態(tài)時(shí)所需經(jīng)歷的時(shí)間為147.4 s,其他數(shù)據(jù)含義類似。
按3.1節(jié)中各狀態(tài)的定義,將表1所得子路段時(shí)間加權(quán)求和,得到主干道各狀態(tài)下出行時(shí)間的估計(jì)值。由式(5)分別選取滑動(dòng)窗口數(shù)據(jù)[k]為5,10,15,20,25,30,遺傳因子δ為0.84,0.85,…,0.90,得到曲線圖如圖1所示。
圖1中實(shí)測(cè)值275.8 s為浮動(dòng)車法測(cè)得的車輛實(shí)際出行時(shí)間,由于數(shù)據(jù)的原因,[k=20]和[k=25]兩條曲線重合。由圖1可知,遺傳因子大小和數(shù)據(jù)量的多少都對(duì)估計(jì)的準(zhǔn)確度產(chǎn)生影響??紤]算法運(yùn)算量,實(shí)際估計(jì)中采用數(shù)據(jù)量為20個(gè)時(shí),估計(jì)值已經(jīng)與實(shí)測(cè)值有交點(diǎn),滿足誤差較小的需求;根據(jù)交點(diǎn)橫坐標(biāo)選取遺傳因子為0.88。
本文提出的算法主要采用多狀態(tài)、遺傳因子、滑動(dòng)窗口數(shù)據(jù)處理三種措施,通過(guò)在山西省臨汾市某一主干道上的驗(yàn)證,結(jié)果表明該算法在主干道上的出行時(shí)間估計(jì)方面具有實(shí)時(shí)性、精確性。
本算法根據(jù)車輛速度將道路劃分為三種狀態(tài),若忽略運(yùn)算數(shù)據(jù)量的影響而追求估計(jì)的精確度,可將道路劃分為更多的狀態(tài)。另外,本算法未考慮十字路口的調(diào)度情況,下一步會(huì)針對(duì)這一環(huán)節(jié)進(jìn)行研究,使理論的適用范圍更廣。
[1] BRUZELIUS N. The value of travel time: theory and measure?ment [M]. US: Taylor & Francis, 1979: 1?20.
[2] HESS S, BIERLAIRE M, POLAK J W. Estimation of value of travel?time savings using mixed logit models [J]. Transportation research part A: policy and practice, 2005, 39(2): 221?236.
[3] SKABARDONIS A, GEROLIMINIS N. Real?time estimation of travel times on signalized arterials [R]. College Park Maryland, United States, 2005: 1?21.
[4] SKABARDONIS A, GEROLIMINIS N. Real?time monitoring and control on signalized arterials [J]. Journal of intelligent transportation systems, 2008, 12(2): 64?74.
[5] GEROLIMINIS N, SKABARDONIS A. Identification and analysis of queue spillovers in city street networks [J]. IEEE transactions on intelligent transportation systems, 2011, 12(4): 1107?1115.
[6] LIU H X, MA W. A virtual vehicle probe model for time?dependent travel time estimation on signalized arterials [J]. Transportation research part C: emerging technologies, 2009, 17(1): 11?26.
[7] LIU H X, MA W, WU X, et al. Real?time estimation of arterial travel time under congested conditions [J]. Transportmetrica, 2012, 8(2): 87?104.
[8] 張俊婷,周晶,徐紅利,等.基于離散時(shí)間馬爾科夫鏈的主干道出行時(shí)間估計(jì)模型[J].系統(tǒng)工程,2014(5):98?104.
ZHANG J T, ZHOU J, XU H L, et al. An arterial travel time estimation model based on discrete time Markov chains [J]. Systems engineering, 2014(5): 98?104.
[9] LAWLOR S, RABBAT M G. Time?varying mixtures of Markov chains: an application to road traffic modeling [J]. IEEE transactions on signal processing, 2017, 65(12): 3152?3167.
[10] FILEV D P, KOLMANOVSKY I. Generalized Markov models for real?time modeling of continuous systems [J]. IEEE transactions on fuzzy systems, 2014, 22(4): 983?998.
[11] KUKHTAREV N V, MARKOV V B, ODULOV S G, et al.Holographic storage in electrooptic crystals: I. steady state [J]. Ferroelectrics, 1978, 22(1): 949?960.
[12] CASTILLO R G, CLEMPNER J B, POZNYAK A S. Solving the multi?traffic signal?control problem for a class of continuous?time Markov games [C]// 2015 International Conference on Electrical Engineering, Computing Science and Automatic Control. Mexico City: IEEE, 2015: 1?5.
[13] 陸建,孫祥龍,戴越.普通公路車速分布特性的回歸分析[J].東南大學(xué)學(xué)報(bào)(自然科學(xué)版),2012(2):374?377.
LU J, SUN X L, DAI Y. Regression analysis on speed distribution characteristics of ordinary road [J]. Journal of Southeast University (natural science edition), 2012(2): 374?377.
[14] THIANNIWET T, PHOSAARD S, PATTARAATIKOM W. Classification of road traffic congestion levels from GPS data using a decision tree algorithm and sliding windows [J]. Lecture notes in engineering & computer science, 2009, 2176(1): 1?6.