CN106530698A - Urban road travel time estimation method considering exit turning - Google Patents

Abstract

Translated fromChinese

本发明公开了一种考虑出口转向的城市道路行程时间的估计方法，包括：步骤一、建立累计直方图修正模型；步骤二、对车辆不同出口转向的上游累计车辆数进行估算；步骤三、利用估算出的车辆不同出口转向的上游累计车辆数和车辆不同出口转向的下游累计车辆数估计路段具体出口转向的行程时间。本发明将定点检测器和探测车采集的数据与传统累积直方图模型相结合，重新建立考虑不同出口转向的行程时间累积直方图模型,本发明考虑了城市道路中相对偏差的存在，提高了城市道路行程时间的估计精度，本发明方法的估计值与探测车实测值的误差低于10％。

The invention discloses a method for estimating travel time of urban roads considering exit turning, including: step 1, establishing a cumulative histogram correction model; step 2, estimating the cumulative number of upstream vehicles with different exit turns; step 3, using The estimated cumulative number of upstream vehicles with different exit turns of vehicles and the cumulative number of downstream vehicles with different exit turns of vehicles estimate the travel time of the specific exit turns of the road section. The invention combines the data collected by fixed-point detectors and detection vehicles with the traditional cumulative histogram model to re-establish the travel time cumulative histogram model considering different exit turns. The invention considers the existence of relative deviations in urban roads and improves urban For the estimation accuracy of the road travel time, the error between the estimated value of the method of the present invention and the actual measured value of the probe car is lower than 10%.

Description

Translated fromChinese

一种考虑出口转向的城市道路行程时间的估计方法An Estimation Method for Urban Road Travel Time Considering Exit Turns

技术领域technical field

本发明属于智能交通领域，具体涉及一种考虑出口转向的城市道路行程时间的估计方法。The invention belongs to the field of intelligent transportation, and in particular relates to a method for estimating travel time of urban roads considering exit turning.

背景技术Background technique

行程时间是智能交通系统的主要参数之一，是衡量路网交通拥堵状况、出行效率的重要指标。在城市路网中，行程时间是指从路段上游点到下游点所需时间总和。由于城市道路交叉口信号控制、交通流自身的不守恒性，造成不同出口转向之间行程时间差别很大，准确估计城市路网行程时间仍然面临着挑战。行程时间的预测通常采用的方法包括时间序列法、人工神经网络模型、卡尔曼滤波等，这些方法虽然都具有一定的准确性，但存在基础数据多计算量大求解不易的问题。Travel time is one of the main parameters of the intelligent transportation system, and it is an important indicator to measure the traffic congestion and travel efficiency of the road network. In the urban road network, the travel time refers to the sum of the time required from the upstream point of the road segment to the downstream point. Due to the signal control of urban road intersections and the non-conservative nature of traffic flow itself, the travel time between different exit turns is very different, and it is still a challenge to accurately estimate the travel time of the urban road network. The methods commonly used to predict travel time include time series method, artificial neural network model, Kalman filter, etc. Although these methods have certain accuracy, they have the problem of large amount of basic data and large amount of calculation, which is not easy to solve.

发明内容Contents of the invention

为解决上述问题，本发明提供一种考虑出口转向的城市道路行程时间的估计方法，该方法包括：In order to solve the above problems, the present invention provides a method for estimating travel time on urban roads considering exit turning, the method comprising:

步骤一、建立累计直方图修正模型；Step 1. Establish a cumulative histogram correction model;

步骤二、对车辆不同出口转向的上游累计车辆数进行估算；Step 2. Estimate the cumulative number of upstream vehicles with different exit diversions;

步骤三、利用估算出的车辆不同出口转向的上游累计车辆数和车辆不同出口转向的下游累计车辆数估计路段具体出口转向的平均行程时间。Step 3: Estimate the average travel time of the specific exit turns of the road section by using the estimated upstream cumulative number of vehicles with different exit turns and the downstream cumulative number of vehicles with different exit turns.

进一步地，所述累计直方图修正模型为：Further, the cumulative histogram correction model is:

式中：U_A(t)为上游路段累计车辆数；U_m(t)为车辆不同的出口转向的上游累计车辆数；ε为误差参数；用字母m表示车辆不同的出口转向，m代表T、L、R，其中T表示直行，L表示左转，R表示右转。In the formula: U_A (t) is the cumulative number of vehicles on the upstream section; U_m (t) is the cumulative number of upstream vehicles with different exit turns; ε is the error parameter; the letter m represents the different exit turns of the vehicle, and m represents T , L, R, where T means go straight, L means turn left, and R means turn right.

进一步地，所述对车辆不同出口转向的上游累计车辆数估算方法为：Further, the method for estimating the cumulative number of upstream vehicles for different exit diversions of vehicles is as follows:

(1)计算U_A(t)和D_m(t)；(1) Calculate U_A (t) and D_m (t);

所述U_A(t)和D_m(t)通过布设在停车线处的检测器数据求出，其中，D_m(t)表示下游路段累计车辆数；Described U_A (t) and D_m (t) obtain by the detector data that are arranged at the stop line place, and wherein, D_m (t) represents the accumulative number of vehicles of the downstream section;

(2)垂直缩放U_A(t)，估计U_m(t)；(2) Scale U_A (t) vertically and estimate U_m (t);

定义变量V_m,p为有效比例因子，表示一天中p时间间隔内m出口转向的比例因子；t_s,p和t_e,p分别表示p时间间隔车辆行程的起点时刻和终点时刻(本文V_m,p的取值可以根据路段各出口转向的流量比例设置各出口转向对应的比例因子值，如果V_m,p＝1，则U_m(t)＝U_A(t))；The variable V_m,p is defined as an effective scaling factor, which represents the scaling factor of m exit turning in the p time interval in a day; t_s,p and t_e,p represent the starting time and end time of the vehicle journey in the p time interval respectively (in this paper, V The value of_m,p can be set according to the flow ratio of each exit turn of the road section, and the proportional factor value corresponding to each exit turn, if V_m,p =1, then U_m (t)=U_A (t));

在p时间间隔的起始时刻t_s,p对上游m出口转向的初始累积车辆数U_m(t_s,p)进行估计，然后运用公式(2)对各时间间隔U_A(t)以相应比例因子进行垂直缩放，得出U_m(t)的估计值；Estimate the initial cumulative number of vehicles U_m (t_s,p ) turning at the upstream exit m at the initial moment t_s,p of the time interval p, and then apply the formula (2) to the corresponding time interval U_A (t) Scale factor vertically to obtain an estimate of U_m (t);

式中：Y(t)为t_s,p到t时间间隔m转向的车辆数；U_A(t)-U_A(t_s,p)为t_s,p到t时间间隔内累积车辆数；U_m(t)为t时刻进行m转向的累计车辆数；In the formula: Y(t) is the number of vehicles turning to m in the time interval from t_s,p to t; U_A (t)-U_A (t_s,p ) is the cumulative number of vehicles in the time interval from t_s,p to t; U_m (t) is the cumulative number of vehicles making m turns at time t;

(3)利用探测车数据，计算U_m′(t)；(3) Calculate U_m ′(t) using the probe car data;

引入探测车是为了使其在交叉口提供时间标记，将探测车采集到的数据定义为D′_m(t)，现定义有n辆探测车，探测车位于上游和下游的时间点分别为t_u和t_d，对U_m(t)曲线所经过的点进行定义，步骤如下：The purpose of introducing the probe car is to provide time stamps at the intersection. The data collected by the probe car is defined as D′_m (t). Now there are n probe cars, and the time points when the probe car is located upstream and downstream are respectively t_u and t_d , to define the points passed by the U_m (t) curve, the steps are as follows:

a、将t_d的数据列表按照数据值的升序(即到达的先后顺序)进行排序；a. Sort the data list of t_d according to the ascending order of data values (that is, the order of arrival);

b、将t_u的数据列表按照数据值的升序(即出发的先后顺序)进行排序；b. Sort the data list of t_u according to the ascending order of data values (that is, the order of departure);

c、则将U_m(t)经过的点定义为(t_uj,D′(t_dj))，其中t_uj和t_dj分别是t_u、t_d数据列表中第j辆探测车的数据；c. Then define the point passed by U_m (t) as (t_uj , D′(t_dj )), where t_uj and t_dj are the data of the jth probe car in the data list of t_u and t_d respectively;

道路上车流量较少处于自由流状态时，此时将累积直方图数据进行初始化处理，定义初始参考点P₀(U_m(t₀)＝D′_m(t₀)＝0)，选取道路上交通流量从自由流逐渐向缓行状态过渡的时刻作为t_r，定义P_r(t_r,U_m(t_r))为累积直方图中的一个参考点，对于直方图中其他点P_j，以前一点P_j-1作为参考进行重新计算，则重新计算得到的U_m′(t)所经过的点(t_p,Y_p)的计算公式如下：When the traffic flow on the road is less and in the state of free flow, the cumulative histogram data will be initialized at this time, and the initial reference point P₀ (U_m (t₀ )=D′_m (t₀ )=0) will be defined, and the road The moment when the traffic flow gradually transitions from free flow to slowing state is taken as t_r , and P_r (t_r , U_m (t_r )) is defined as a reference point in the cumulative histogram. For other points P_j in the histogram, The previous point P_j-1 is used as a reference for recalculation, then the calculation formula of the point (t_p , Y_p ) passed by the recalculated U_m ′(t) is as follows:

U_m′(t)＝U_m(t)+C (3)U_m '(t) = U_m (t) + C (3)

其中，in,

式中：U_m′(t)为重新计算U_m(t)得到的m出口转向的上游累积车辆数；C为修正变量；s为缩放因子。In the formula: U_m ′(t) is the cumulative number of upstream vehicles diverted from exit m obtained by recalculating U_m (t); C is the correction variable; s is the scaling factor.

进一步地，所述比例因子V_m,p，是将前一天各固定时间间隔的U_m′(t)和U_A(t)进行整合，用于对未来时刻路网交通状态进行预测，V_m,p的计算公式如下：Further, the scaling factor V_m,p is to integrate U_m ′(t) and U_A (t) of each fixed time interval of the previous day to predict the traffic state of the road network in the future, V_{m , the calculation formula of p} is as follows:

其中：in:

式中：t_e,p、t_s,p分别为p时间间隔开始和结束时刻；Y_A,p、Y_m,p分别为U_A(t)和U_m′(t)在p时间间隔内的累积计数。In the formula: t_e,p , t_s,p are the start and end moments of p time interval respectively; Y_A,p , Y_m,p are U_A (t) and U_m ′(t) in p time interval cumulative count of .

进一步地，所述利用估算出的车辆不同出口转向的上游累计车辆数和车辆不同出口转向的下游累计车辆数估计路段具体出口转向的平均行程时间方法为：将U_m′(t)和检测器数据D_m(t)代入公式(8)中，计算路段具体出口转向的平均行程时间Further, the method of estimating the average travel time of the specific exit turning of the road section by using the estimated upstream cumulative vehicle numbers of different exit turns of vehicles and the downstream cumulative number of vehicles of different exit turns of vehicles is: U_m '(t) and detector The data D_m (t) is substituted into the formula (8) to calculate the average travel time of the specific exit turn of the road section

其中：in:

N＝U′_m(t₂)-U′_m(t₁)＝D_m(t₄)-D_m(t₃) (9)N=U′_m (t₂ )-U′_m (t₁ )=D_m (t₄ )-D_m (t₃ ) (9)

式中：U_m′(t)为重新计算得出的上游m出口转向的累计车辆数；S为全部行程时间；N为时间t₁-t₂(t₃-t₄)内到达上游点(离开下游点)的累积车流量；In the formula: U_m ′(t) is the accumulative number of vehicles diverted from the upstream exit m; S is the total travel time; N is the time t₁ -t₂ (t₃ -t₄ ) to reach the upstream point ( Cumulative traffic flow leaving the downstream point);

本发明的有益效果：通过将定点检测器和探测车采集的数据与传统累积直方图模型相结合，重新建立考虑不同出口转向的行程时间累积直方图模型并进行实例分析，新建累积直方图模型的估计值与探测车实测值的误差在10％左右，且只需要少量交通流数据就能够得出不同出口转向的行程时间数据，能够简单有效地判断路口各个方向是否处于交通饱和状态，对后续控制策略的制定及防止交通拥挤的加剧具有较高的应用价值。Beneficial effects of the present invention: by combining the data collected by the fixed-point detector and the probe car with the traditional cumulative histogram model, re-establishing the cumulative histogram model of travel time considering different exit turns and carrying out case analysis, and creating a new cumulative histogram model The error between the estimated value and the measured value of the probe car is about 10%, and only a small amount of traffic flow data is needed to obtain the travel time data of different exit turns, which can simply and effectively judge whether the intersection is in a state of traffic saturation in all directions, and is useful for subsequent control. It is of high application value to formulate strategies and prevent the aggravation of traffic congestion.

附图说明Description of drawings

图1为本发明垂直缩放U_A(t)示意图；Fig. 1 is a schematic diagram of vertical scaling U_A (t) of the present invention;

图2为本发明累积直方图修正模型建立流程图；Fig. 2 is the flow chart of building up the cumulative histogram correction model of the present invention;

图3为本发明利用探测车数据计算U_m′(t)示意图；Fig. 3 is the present invention utilizes probe car data to calculate U_m '(t) schematic diagram;

图4为本发明平均行程时间计算示意图；Fig. 4 is the calculation schematic diagram of average travel time of the present invention;

图5为本发明实施例研究区域图；Fig. 5 is the research area map of the embodiment of the present invention;

图6为本发明实施例的左转行程时间图；Fig. 6 is a left-turn travel time diagram of an embodiment of the present invention;

图7为本发明实施例的左转行程时间误差分析图；Fig. 7 is an analysis diagram of the left-turn travel time error of the embodiment of the present invention;

图8为本发明实施例的直行行程时间图；Fig. 8 is a straight travel time diagram of an embodiment of the present invention;

图9为本发明实施例的直行行程时间误差分析图；Fig. 9 is an analysis diagram of the straight travel time error of the embodiment of the present invention;

图10为本发明实施例的右转行程时间图；Fig. 10 is a right-turn travel time diagram of an embodiment of the present invention;

图11为本发明实施例的右转行程时间误差分析图；Fig. 11 is a right-turn travel time error analysis diagram of an embodiment of the present invention;

图12为本发明实施例的相同交通流情况下方法准确率对比图；Fig. 12 is a comparison chart of method accuracy under the same traffic flow situation of the embodiment of the present invention;

具体实施方式detailed description

以下结合附图对本发明的优选实施案例进行说明，应当理解，此处所描述的优选实施案例仅用于解释和说明本发明方案，并不用于限制本发明。The preferred embodiments of the present invention will be described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described here are only used to explain and illustrate the solutions of the present invention, and are not intended to limit the present invention.

具体实施例：请参见图1、2、3，一种考虑出口转向的城市道路行程时间的估计方法，该方法包括：Specific embodiments: please refer to Fig. 1, 2, 3, a kind of estimation method of the urban road travel time that considers exit turning, this method comprises:

步骤一、建立累计直方图修正模型：Step 1. Establish a cumulative histogram correction model:

式中：U_A(t)为上游路段累计车辆数；U_m(t)为车辆不同的出口转向的上游累计车辆数；ε为误差参数；用字母m表示车辆不同的出口转向，m代表T、L、R，其中T表示直行，L表示左转，R表示右转。In the formula: U_A (t) is the cumulative number of vehicles on the upstream section; U_m (t) is the cumulative number of upstream vehicles with different exit turns; ε is the error parameter; the letter m represents the different exit turns of the vehicle, and m represents T , L, R, where T means go straight, L means turn left, and R means turn right.

步骤二、对车辆不同出口转向的上游累计车辆数进行估算，该估算方法为：Step 2. Estimate the cumulative number of upstream vehicles with different exit diversions. The estimation method is:

(1)计算U_A(t)和D_m(t)；(1) Calculate U_A (t) and D_m (t);

所述U_A(t)和D_m(t)通过布设在停车线处的检测器数据求出，其中，D_m(t)表示下游路段累计车辆数；Described U_A (t) and D_m (t) obtain by the detector data that are arranged at the stop line place, and wherein, D_m (t) represents the accumulative number of vehicles of the downstream section;

(2)垂直缩放U_A(t)，估计U_m(t)，请参见图1；(2) Scale U_A (t) vertically to estimate U_m (t), see Figure 1;

定义变量V_m,p为有效比例因子，表示一天中p时间间隔内m出口转向的比例因子；t_s,p和t_e,p分别表示p时间间隔车辆行程的起点时刻和终点时刻(本文V_m,p的取值可以根据路段各出口转向的流量比例设置各出口转向对应的比例因子值，如果V_m,p＝1，则U_m(t)＝U_A(t))；The variable V_m,p is defined as an effective scaling factor, which represents the scaling factor of m exit turning in the p time interval in a day; t_s,p and t_e,p represent the starting time and end time of the vehicle journey in the p time interval respectively (in this paper, V The value of_m,p can be set according to the flow ratio of each exit turn of the road section, and the proportional factor value corresponding to each exit turn, if V_m,p =1, then U_m (t)=U_A (t));

在p时间间隔的起始时刻t_s,p对上游m出口转向的初始累积车辆数U_m(t_s,p)进行估计，然后运用公式(2)对各时间间隔U_A(t)以相应比例因子进行垂直缩放，得出U_m(t)的估计值；Estimate the initial cumulative number of vehicles U_m (t_s,p ) turning at the upstream exit m at the initial moment t_s,p of the time interval p, and then apply the formula (2) to the corresponding time interval U_A (t) Scale factor vertically to obtain an estimate of U_m (t);

式中：Y(t)为t_s,p到t时间间隔m转向的车辆数；U_A(t)-U_A(t_s,p)为t_s,p到t时间间隔内累积车辆数；U_m(t)为t时刻进行m转向的累计车辆数；In the formula: Y(t) is the number of vehicles turning to m in the time interval from t_s,p to t; U_A (t)-U_A (t_s,p ) is the cumulative number of vehicles in the time interval from t_s,p to t; U_m (t) is the cumulative number of vehicles making m turns at time t;

(3)利用探测车数据，计算U_m′(t)，请参见图2、图3；(3) Calculate U_m ′(t) using the probe car data, please refer to Fig. 2 and Fig. 3;

引入探测车是为了使其在交叉口提供时间标记，将探测车采集到的数据定义为D′_m(t)，现定义有n辆探测车，探测车位于上游和下游的时间点分别为t_u和t_d，对U_m(t)曲线所经过的点进行定义，步骤如下：The purpose of introducing the probe car is to provide time stamps at the intersection. The data collected by the probe car is defined as D′_m (t). Now there are n probe cars, and the time points when the probe car is located upstream and downstream are respectively t_u and t_d , to define the points passed by the U_m (t) curve, the steps are as follows:

a、将t_d的数据列表按照数据值的升序(即到达的先后顺序)进行排序；a. Sort the data list of t_d according to the ascending order of data values (that is, the order of arrival);

b、将t_u的数据列表按照数据值的升序(即出发的先后顺序)进行排序；b. Sort the data list of t_u according to the ascending order of data values (that is, the order of departure);

c、则将U_m(t)经过的点定义为(t_uj,D′(t_dj))，其中t_uj和t_dj分别是t_u、t_d数据列表中第j辆探测车的数据；c. Then define the point passed by U_m (t) as (t_uj , D′(t_dj )), where t_uj and t_dj are the data of the jth probe car in the data list of t_u and t_d respectively;

道路上车流量较少处于自由流状态时，此时将累积直方图数据进行初始化处理，定义初始参考点P₀(U_m(t₀)＝D′_m(t₀)＝0)，选取道路上交通流量从自由流逐渐向缓行状态过渡的时刻作为t_r，定义P_r(t_r,U_m(t_r))为累积直方图中的一个参考点，对于直方图中其他点P_j，以前一点P_j-1作为参考进行重新计算，则重新计算得到的U_m′(t)所经过的点(t_p,Y_p)的计算公式如下：When the traffic flow on the road is less and in the state of free flow, the cumulative histogram data will be initialized at this time, and the initial reference point P₀ (U_m (t₀ )=D′_m (t₀ )=0) will be defined, and the road The moment when the traffic flow gradually transitions from free flow to slowing state is taken as t_r , and P_r (t_r , U_m (t_r )) is defined as a reference point in the cumulative histogram. For other points P_j in the histogram, The previous point P_j-1 is used as a reference for recalculation, then the calculation formula of the point (t_p , Y_p ) passed by the recalculated U_m ′(t) is as follows:

U_m′(t)＝U_m(t)+C (3)U_m '(t) = U_m (t) + C (3)

其中，in,

式中：U_m′(t)为重新计算U_m(t)得到的m出口转向的上游累积车辆数；C为修正变量；s为缩放因子。In the formula: U_m ′(t) is the cumulative number of upstream vehicles diverted from exit m obtained by recalculating U_m (t); C is the correction variable; s is the scaling factor.

如图3所示，在时间t_r之前不对原有直方图进行缩放，在t_r-t_p时间间隔对直方图进行垂直缩放，超过t_p之后由于修正变量C为定值，对其进行垂直转换，从而使得U_m′(t)是连续的，且在时刻t_p平行于原有U_m(t)。最后利用U_m′(t)与D_m(t)计算考虑出口转向的城市路段行程时间平均行程时间。As shown in Figure 3, the original histogram is not scaled before the time t_r , and the histogram is vertically scaled at the time interval t_r -t_p , and after t_p is exceeded because the correction variable C is a fixed value, it is vertically scaled transformation so that U_m ′(t) is continuous and parallel to the original U_m (t) at time t_p . Finally, U_m ′(t) and D_m (t) are used to calculate the average travel time of urban road sections considering exit turning.

进一步地，所述比例因子V_m,p，是将前一天各固定时间间隔的U_m′(t)和U_A(t)进行整合，用于对未来时刻路网交通状态进行预测，V_m,p的计算公式如下：Further, the scaling factor V_m,p is to integrate U_m ′(t) and U_A (t) of each fixed time interval of the previous day to predict the traffic state of the road network in the future, V_{m , the calculation formula of p} is as follows:

其中：in:

式中：t_e,p、t_s,p分别为p时间间隔开始和结束时刻；Y_A,p、Y_m,p分别为U_A(t)和U_m′(t)在p时间间隔内的累积计数。In the formula: t_e,p , t_s,p are the start and end moments of p time interval respectively; Y_A,p , Y_m,p are U_A (t) and U_m ′(t) in p time interval cumulative count of .

请参见图4，步骤三、利用估算出的车辆不同出口转向的上游累计车辆数和车辆不同出口转向的下游累计车辆数估计路段具体出口转向的平均行程时间：将U_m′(t)和检测器数据D_m(t)代入公式(8)中，计算路段具体出口转向的平均行程时间Please refer to Figure 4, step 3, using the estimated cumulative number of upstream vehicles with different exit turns of vehicles and the cumulative number of downstream vehicles with different exit turns of vehicles to estimate the average travel time of the specific exit turns of the road section: combine U_m ′(t) and detection Substitute the device data D_m (t) into the formula (8) to calculate the average travel time of the specific exit turn of the road section

其中：in:

N＝U′_m(t₂)-U′_m(t₁)＝D_m(t₄)-D_m(t₃) (9)N=U′_m (t₂ )-U′_m (t₁ )=D_m (t₄ )-D_m (t₃ ) (9)

式中：U_m′(t)为重新计算得出的上游m出口转向的累计车辆数；S为全部行程时间；N为时间t₁-t₂(t₃-t₄)内到达上游点(离开下游点)的累积车流量；In the formula: U_m ′(t) is the accumulative number of vehicles diverted from the upstream exit m; S is the total travel time; N is the time t₁ -t₂ (t₃ -t₄ ) to reach the upstream point ( Cumulative traffic flow leaving the downstream point);

需要说明的是，为验证本发明可行性，申请人以青岛香江路从江山南路至武夷山路一段为实验区域，请参见图5。这是一段由西向东的城市主干道，路段全长2.1km，双向四车道，交通检测器布设在停车线附近，选用2015年7月20日由探测车在上述区间采集到的数据代入神经网络模型和累积直方图修正模型进行实验模拟，分别模拟从S点到L、T、R三点进行左转、直行、右转的车辆。将探测车采集的时间数据定义为实际行程时间t_a，利用上述两种模型仿真所得出口转向行程时间定义为预测行程时间t_e，评价指标如下：It should be noted that, in order to verify the feasibility of the present invention, the applicant took the section of Xiangjiang Road from Jiangshan South Road to Wuyishan Road in Qingdao as the experimental area, see Figure 5. This is a section of urban arterial road from west to east, with a total length of 2.1km, two-way four-lane, and traffic detectors arranged near the stop line. The data collected by the probe car in the above section on July 20, 2015 was selected and substituted into the neural network. The model and the cumulative histogram correction model are used for experimental simulations, respectively simulating vehicles turning left, going straight, and turning right from point S to points L, T, and R. The time data collected by the probe car is defined as the actual travel time t_a , and the exit steering travel time obtained by using the above two models simulated is defined as the predicted travel time t_e , and the evaluation indicators are as follows:

A_M＝(1-MAPE) (12)A_M = (1-MAPE) (12)

式中：APE为行程时间绝对百分率误差；MAEP为行程时间平均百分率误差；A_M为行程时间估计有效性百分率。In the formula: APE is the absolute percentage error of travel time; MAEP is the average percentage error of travel time; A_M is the percentage of validity of travel time estimate.

利用上述公式对路段运用上述两种模型进行左转、直行、右转三个方向行程时间误差分析。从图6、图8、图10行程时间预测结果可以看出，累积直方图曲线总体上与实际行程时间曲线的契合度较高，能够对行程时间进行估计；从图7、图9、图11的误差分析可知，运用累积直方图方法进行预测时，误差一般情况下控制在10％以内，并且相对神经网络模型而言准确度稍高，能够较准确的对行程时间进行预测。Use the above formula to analyze the travel time error in the three directions of left turn, straight ahead and right turn by using the above two models on the road section. From the travel time prediction results in Figures 6, 8, and 10, it can be seen that the cumulative histogram curve generally has a high degree of fit with the actual travel time curve, and the travel time can be estimated; from Figures 7, 9, and 11 The error analysis shows that when the cumulative histogram method is used for prediction, the error is generally controlled within 10%, and the accuracy is slightly higher than that of the neural network model, which can predict the travel time more accurately.

另外，为了证明该方法与神经网络模型相比具有简单易行的优势，申请人利用公式(10)-(12)分别对累积直方图模型和神经网络模型在不同交通流量时单位时间间隔内的百分率误差进行分析，请参见图12，从图中可以发现：In addition, in order to prove that this method has the advantages of simplicity and ease compared with the neural network model, the applicant uses formulas (10)-(12) to analyze the cumulative histogram model and the neural network model in the unit time interval of different traffic flows Percentage error analysis, please refer to Figure 12, from the figure can be found:

(1)采用累积直方图模型进行道路行程时间估计时，估计精度能够达到90％左右，并且随着交通流数据的增加估计精度会有所提高。(1) When the cumulative histogram model is used to estimate the road travel time, the estimation accuracy can reach about 90%, and the estimation accuracy will increase with the increase of traffic flow data.

(2)本文所提出方法只需要少量探测车数据就可以做出相对准确的预测，但是神经网络模型只有在交通流量较大时检测精度才会相对准确。所以运用累积直方图模型预测行程时间更为简单易行。(2) The method proposed in this paper only needs a small amount of probe car data to make relatively accurate predictions, but the detection accuracy of the neural network model is relatively accurate only when the traffic flow is large. Therefore, it is simpler and easier to use the cumulative histogram model to predict travel time.

最后应说明的是：以上所述仅为本发明的较佳实施方式，并不用于限定本发明，凡在依据本发明的技术实质所作的简单的任何修改、等同替换、修饰等，均应包括在本发明的保护范围之内。Finally, it should be noted that: the above is only a preferred embodiment of the present invention, and is not intended to limit the present invention. Any simple modifications, equivalent replacements, modifications, etc. made in accordance with the technical essence of the present invention shall include Within the protection scope of the present invention.

Claims (5)

1. A method for estimating the travel time of an urban road taking into account exit turns, characterized in that it comprises:

step one, establishing an accumulative histogram correction model;

estimating the number of vehicles at different exit turning upstream;

and thirdly, estimating the average travel time of the specific exit turning of the road section by using the estimated upstream accumulated vehicle number of the different exit turning of the vehicle and the estimated downstream accumulated vehicle number of the different exit turning of the vehicle.

2. The method for estimating urban road travel time taking account of exit turns as claimed in claim 1, wherein said cumulative histogram modification model is:

$U_A\left(t\right)=\underset{\∀m}{\Σ}{U}_m\left(t\right)+\mathrm{\ϵ}---\left(1\right)$

in the formula: u shape_A(t) the accumulated number of vehicles on the upstream road section; u shape_m(t) accumulating vehicle counts upstream for different exit turns of the vehicle; is an error parameter; the different exit turns of the vehicle are indicated by the letter m, m representing T, L, R, where T represents straight, L represents left turn, and R represents right turn.

3. The method for estimating urban road travel time taking exit turns into consideration according to claim 1, wherein the method for estimating the number of vehicles accumulated upstream for different exit turns of a vehicle is as follows:

(1) calculate U_A(t) and D_m(t)；

The U is_A(t) and D_m(t) is determined from detector data arranged at the stop line, where D_m(t) represents the cumulative number of vehicles on the downstream link;

(2) vertical zooming U_A(t), estimating U_m(t)；

Defining variable V_m,pIs an effective scale factor, and represents the scale factor of m outlet turning in p time intervals in a day; t is t_s,pAnd t_e,pRespectively represent p time intervalsStarting and ending times of the trip of the motor vehicle (text V)_m,pThe value of (A) can be set according to the flow proportion of each outlet steering of the road section, if V is_m,p1, then U_m(t)＝U_A(t))；

At the start time t of the p time interval_s,pInitial cumulative number of vehicles U turning to m outlets upstream_m(t_s,p) Estimate each time interval U using equation (2)_A(t) vertical scaling with the corresponding scale factor to give U_m(t) an estimate of;

in the formula: y (t) is t_s,pThe number of vehicles turning to t time interval m; u shape_A(t)-U_A(t_s,p) Is t_s,pAccumulating the number of vehicles in the time interval t; u shape_m(t) the cumulative number of vehicles steering m at time t;

(3) calculating U by using data of probe car_m′(t)；

The probe vehicle is introduced to provide a time mark at the intersection, and the data collected by the probe vehicle is defined as D'_m(t), n probe vehicles are defined, and the time points of the probe vehicles at the upstream and the downstream are t respectively_uAnd t_dTo U, to U_m(t) points through which the curve passes are defined by the following steps:

a. will t_dThe data list of (1) is sorted according to the ascending order (namely the arrival sequence) of the data values;

b. will t_uThe data list is sorted according to the ascending order (namely the starting sequence) of the data values;

c. then U will be_mThe point where (t) passes is defined as (t)_uj,D′(t_dj) In which t) is_ujAnd t_djAre respectively t_u、t_dThe data of the jth probe vehicle in the data list;

when the traffic flow on the road is less and in a free flow state, the accumulated histogram data is initializedDefining an initial reference point P₀(U_m(t₀)＝D′_m(t₀) 0), selecting the time when the traffic flow on the road gradually transits from the free flow to the slow running state as t_rDefinition of P_r(t_r,U_m(t_r) For one reference point in the cumulative histogram, for other points P in the histogram_jA previous point P_j-1Recalculating as a reference, the recalculated U_m' (t) passing through point (t)_p,Y_p) The calculation formula of (a) is as follows:

U_m′(t)＝U_m(t)+C (3)

wherein,

$s=\left\{\begin{array}{cc}\frac{Y_p-U_m\left(t_r\right)}{U_m\left(t_p\right)-U_m\left(t_r\right)}& U_m\left(t_p\right)\≠U_m\left(t_r\right)\\ 1& U_m\left(t_p\right)=U_m\left(t_r\right)\end{array}\right.---\left(5\right)$

in the formula: u shape_m' (t) is a recalculated U_m(t) m outlet obtainedThe number of vehicles accumulated upstream of the steering; c is a correction variable; s is a scaling factor.

4. Method for estimating urban road travel time taking account of exit turns according to claim 3, characterized in that said scaling factor V_m,pIs the U of each fixed time interval of the previous day_m' (t) and U_A(t) integration for predicting road network traffic conditions at future times, V_m,pThe calculation formula of (a) is as follows:

$V_{m,p}=\frac{Y_{A,p}-Y_{m,p}}{Y_{A,p}}---\left(6\right)$

wherein:

in the formula: t is t_e,p、t_s,pRespectively p time interval start and end times; y is_A,p、Y_m,pAre respectively U_A(t) and U_m' (t) cumulative count over p time intervals.

5. The method for estimating urban road travel time taking account of exit turns as claimed in claim 1, wherein said utilization estimation is performed by using an estimation of the travel time of the urban roadThe calculated upstream accumulated vehicle number of different exit turning directions of the vehicle and the calculated downstream accumulated vehicle number of different exit turning directions of the vehicle are used for estimating the average travel time of the specific exit turning direction of the road section: will U_m' (t) and detector data D_m(t) substituting into the formula (8), calculating the average travel time of the road section specific exit turning

$\stackrel{\‾}{T_m}=\frac{S}{N}=\frac{\underset{i=1}{\overset{N}{\Σ}}\[{D_m}^{-1}\left(i\right)-{\left(U_m^{\′}\right)}^{-1}\left(i\right)\]}{N}=\frac{\underset{i=1}{\overset{N}{\Σ}}{D_m}^{-1}\left(i\right)-\underset{i=1}{\overset{N}{\Σ}}{\left(U_m^{\′}\right)}^{-1}\left(i\right)}{N}---\left(8\right)$

Wherein:

N＝U′_m(t₂)-U′_m(t₁)＝D_m(t₄)-D_m(t₃) (9)

in the formula: u shape_m' (t) is the cumulative number of vehicles with the upstream m exit turns recalculated; s is the total travel time; n is the time t₁-t₂(t₃-t₄) The cumulative traffic volume arriving at the upstream point (leaving the downstream point).