CN102169627A - Express way travel time prediction method based on virtual speed sensor - Google Patents

Abstract

The invention discloses an express way travel time prediction method based on a virtual speed sensor, which belongs to the technical field of traffic control and management. The method comprises the following steps: firstly arranging a virtual speed sensor between adjacent traffic flow sensors; then determining the speed of each virtual speed sensor by using a weight coefficient at a node of the virtual speed sensor and the speed data of the traffic flow sensor; finally obtaining the space distribution of the speed on the express way section, and then obtaining average travel time on the express way section through the estimated speed space distribution. The higher precision can be obtained in predicting the travel time by using the method; and the result can be directly sent to an ATIS (automatic terminal information system) and an ATMS (advanced terminal management system) to achieve the purpose of shortening the travel time and reducing the traffic jams.

Description

Through street predicting travel time method based on the pseudo-velocity sensor

Technical field

The invention belongs to the traffic control and management technical field, particularly a kind of through street predicting travel time method based on the pseudo-velocity sensor is specifically related to traffic flow sensing data Treatment Analysis and data fusion.

Background technology

Be the important indicator of reflection road section traffic volume state hourage, is the basis of carrying out traffic control and inducing, and has a wide range of applications in traffic control system.Its result can provide real-time and effective information to traveler, helps them to carry out routing better, and realizing route is induced, and with the reduction travel time, reduces congested in traffic.Predicting travel time is intelligent transportation system (Intelligent Transport System, abbreviation ITS) one of key problem of research, hourage, (journey time) data were to realize traveler infosystem (the Advanced Traveler Information system of the advanced person among the ITS, ATIS) and advanced control of traffic and road system (Advanced Transport Management System, the ATMS) important foundation of multiple function.

Be the transport information with time-space domain characteristic hourage.Traffic flow all has continuity in the time-space domain.Hourage is not only relevant with the current and historical traffic behavior in somewhere, and is simultaneously also relevant with the traffic behavior of upstream and downstream.To more reasonably determine hourage, need determine on the highway section continually varying traffic behavior on time and space.But owing to do not have measurement mechanism between the adjacent sensor node, the traffic data of sparse spatial sequence is difficult to directly obtain hourage accurately.

If the traffic behavior of each locus, sensor node place can be determined on the highway section, then when sensor abundant (trending towards infinite), can obtain traffic behavior accurately.Therefore we have defined virtual-sensor node (not actual traffic current detection functionality), if the traffic behavior of each locus, virtual-sensor node place can determine that the traffic behavior that then can obtain the highway section distributes.

Summary of the invention

The purpose of this invention is to provide a kind of through street predicting travel time method, it is characterized in that, may further comprise the steps based on the pseudo-velocity sensor:

The step that the pseudo-velocity sensor node is demarcated: adopt approach based on linear interpolation between adjacent traffic flow sensor, to set the pseudo-velocity sensor;

Utilize the pseudo-velocity sensor, the step of acquisition speed space distribution: utilize weight matrix and speed data to calculate the step of pseudo-velocity sensor node speed by multiple linear regression; According to the speed of the pseudo-velocity sensor node of trying to achieve, the step of acquisition speed space distribution.

The step that estimate hourage:, obtain the highway section step of hourage according to the space distribution of speed.

The described pseudo-velocity sensor node is demarcated is to suppose that the distance between traffic flow sensor node i and the i+1 is d _i, the distance setting between the adjacent node (node comprises traffic flow sensor and pseudo-velocity sensor) is a _i, then the number of the pseudo-velocity sensor node between traffic flow sensor node i and the i+1 is m _i=[d _i/ a _i].

Described weight matrix and pseudo-velocity sensor speed and traffic flow sensor velocity correlation are at t _NConstantly, (i, speed k) is defining virtual speed pickup node

Wherein, n _iNumber for the speed data of the traffic flow sensor i of needs; K=1,2......m _iw _KjBe weight coefficient,

t _NBe current time; Δ t is the time interval of traffic flow sensor acquisition data; v _i(t _N-q Δ t) (q=1,2...n _i) be traffic flow sensor i current time t _NThe q time preceding historical data.

Weight coefficient w _KjCan constitute a weight matrix β, Pseudo-velocity sensor node number between traffic flow sensor node i and the i+1 is m _i, then the pass of weight matrix and pseudo-velocity sensor speed and traffic flow sensor speed is:

$\left[\begin{array}{cccc}{w}_{11}& w_{12}& \·\·\·& w_{{1n}_i}\\ w_{21}& w_{22}& \·\·\·& w_{2n_i}\\ \·\·\·& \·\·\·& \·\·\·& \·\·\·\\ w_{m_{i}1}& w_{m_{i}2}& \·\·\·& w_{m_i{n}_i}\end{array}\right]\left[\begin{array}{c}{v}_i(t_N-\mathrm{\Δt})\\ v_i(t_N-2\mathrm{\Δt})\\ \·\·\·\\ v_i(t_N-n_i\mathrm{\Δt})\end{array}\right]=\left[\begin{array}{c}{v}_{i,1}\left(t_N\right)\\ v_{i,2}\left(t_N\right)\\ \·\·\·\\ v_{i,m_i}\left(t_N\right)\end{array}\right]$

The invention has the beneficial effects as follows the speed of utilizing the pseudo-velocity sensor node, obtain the space distribution of traffic behavior, thereby dope the average hourage in highway section exactly, can provide real-time and effective information for traveler, help them to carry out routing better, realizing route is induced.This algorithm is more reasonable aspect traffic flow theory, improves accuracy of predicting, and provides foundation for traffic administration and control.

Description of drawings

Fig. 1 pseudo-velocity sensor node is demarcated synoptic diagram.

The time series speed data synoptic diagram of Fig. 2 sensor node i.

Fig. 3 t _NThe space distribution of moment speed.

Street microwave detector and detecting device node distribution hourage outside Fig. 4 west.

Fig. 5 is by the comparison of the hourage of distinct methods acquisition.

Embodiment

The invention provides a kind of through street predicting travel time method based on the pseudo-velocity sensor.Illustrated below in conjunction with accompanying drawing.

Embodiment 1

A kind of through street predicting travel time method based on the pseudo-velocity sensor, contain following steps:

(1) step that the pseudo-velocity sensor node is demarcated: adopt approach based on linear interpolation between adjacent traffic flow sensor, to set the step (as shown in Figure 1) of pseudo-velocity sensor

Suppose that the distance between traffic flow sensor node i and the i+1 is d _i(unit, rice), the distance setting between the adjacent node is a _i, then the number of the pseudo-velocity sensor node between traffic flow sensor node i and the i+1 is m _i=[d _i/ a _i]; Node in the described adjacent node comprises pseudo-velocity sensor node and traffic flow sensor node; Be labeled as (i, 1), (i, 2) respectively ... (i, m _i).

(2) utilize the speed data of weight matrix (with pseudo-velocity sensor speed and traffic flow sensor velocity correlation) and traffic flow sensor, the step of acquisition speed space distribution

Step 1: utilize weight matrix and speed data to calculate the step of pseudo-velocity sensor node speed

Traffic flow all has the characteristic of stream on time and space, so the continuity of the traffic behavior of the spatial sequence middle and lower reaches historical traffic behavior that is upstreams is flowed.Therefore can utilize the speed data on the known traffic flow sensor node time series, the speed of the pseudo-velocity sensor node of diverse location on the spatial sequence is estimated.

Traffic flow sensor node i (i=1,2 ... n) speed can directly be obtained, thus the pseudo-velocity sensor node (i, speed k) can be by the seasonal effect in time series data of sensor node i by calculating.(i, speed k) is the pseudo-velocity sensor node

$v_{i,k}\left(t_N\right)=\underset{j=1}{\overset{n_i}{\mathrm{\Σ}}}{w}_{\mathrm{kj}}{\×v}_i(t_N-j\·\mathrm{\Δt})......\left(1\right)$

Wherein, n _iNumber for the speed data of the traffic flow sensor i of needs; w _KjBe weight coefficient, w _Kj≤ 1; w _Kj〉=0; t _NBe current time; Δ t is the time interval of traffic flow sensor acquisition data; v _i(t _N-q Δ t) (q=1,2 ... n _i) be traffic flow sensor i current time t _NThe q time preceding historical data.The signal of traffic flow sensor node i seasonal effect in time series speed data as shown in Figure 2.

Weight coefficient w _KjCan constitute a weight matrix β,

Weight coefficient obtains by " a kind of pseudo-velocity sensor design method based on multiple linear regression ".Following relationship is arranged:

$\left[\begin{array}{c}{w}_{11}\\ w_{12}\\ \·\·\·\\ w_{1n_i}\\ w_{21}\\ w_{22}\\ \·\·\·\\ w_{2n_i}\\ \·\·\·\\ w_{m_{i}1}\\ w_{m_{i}2}\\ \·\·\·\\ w_{m_i{n}_i}\end{array}\right]\text{=}{\left({\left[\begin{array}{cccc}{v}_i(t_1-\mathrm{\Δt})& v_i(t_2-\mathrm{\Δt})& & v_i(t_l-\mathrm{\Δt})\\ v_i(t_1-2\mathrm{\Δt})& v_i(t_2-2\mathrm{\Δt})& & v_i(t_l-2\mathrm{\Δt})\\ \·\·\·& \·\·\·& & \·\·\·\\ v_i(t_1-n_i\mathrm{\Δt})& v_i(t_2-n_i\mathrm{\Δt})& & v_i(t_l-n_i\mathrm{\Δt})\\ v_i(t_1-\mathrm{\Δt})& v_i(t_2-\mathrm{\Δt})& & v_i(t_l-\mathrm{\Δt})\\ v_i(t_1-2\mathrm{\Δt})& v_i(t_2-2\mathrm{\Δt})& & v_i(t_l-2\mathrm{\Δt})\\ \·\·\·& \·\·\·& \text{\·\·\·}& \text{\·\·\·}\\ v_i(t_1-n_i\mathrm{\Δt})& v_i(t_2-n_i\mathrm{\Δt})& & v_i(t_l-n_i\mathrm{\Δt})\\ \·\·\·& \·\·\·& & \·\·\·\\ v_i(t_1-\mathrm{\Δt})& v_i(t_2-\mathrm{\Δt})& & v_i(t_l-\mathrm{\Δt})\\ v_i(t_1-2\mathrm{\Δt})& v_i(t_2-2\mathrm{\Δt})& & v_i(t_l-2\mathrm{\Δt})\\ \·\·\·& \·\·\·& & \·\·\·\\ v_i(t_1-n_i\mathrm{\Δt})& v_i(t_2-n_i\mathrm{\Δt})& & v_i(t_l-n_i\mathrm{\Δt})\end{array}\right]}^T\right)}^{-1}\left[\begin{array}{c}\frac{d_i}{T_i\left(t_1\right)}\×(m_i+2)-v_i\left(t_1\right)-v_{(i+1)}\left(t_1\right)\\ \frac{d_i}{T_i\left(t_2\right)}\×(m_i+2)-v_i\left(t_2\right)-v_{(i+1)}\left(t_2\right)\\ \·\·\·\\ \frac{d_i}{T_i\left(t_l\right)}\×(m_i+2)-v_i\left(t_l\right)-v_{(i+1)}\left(t_l\right)\end{array}\right]$

Wherein, T _i(t _N) be current time t _NSensor node i and the real hourage between the sensor node i+1.

Finally determine weight matrix β.

Step 2: according to the speed of the pseudo-velocity sensor node of trying to achieve, the step of acquisition speed space distribution.

Obtain t _NThe space distribution (as shown in Figure 3) of moment traffic behavior parameter.

(3) step of estimating hourage:, obtain the highway section step of hourage according to the space distribution of speed.

t _NConstantly, the average stroke time between definition traffic flow sensor node i and the pseudo-velocity sensor node (i, 1) is Δ T _I1(t _N), average velocity is Δ v _I1(t _N); Pseudo-velocity sensor node (i, (k-1)) is with (i, k) the average stroke time between is Δ T _Ik(t _N), average velocity is Δ v _Ik(t _N) (k=2,3 ... m _i); Pseudo-velocity sensor node (i, m _i) and traffic flow sensor node i+1 between the average stroke time be

Average velocity is

Then

$\left\{\begin{array}{c}\mathrm{\Δ}{v}_{i1}\left(t_N\right)=\frac{v_i\left(t_N\right)+v_{i,1}\left(t_N\right)}{2}\\ \mathrm{\Δ}{v}_{\mathrm{ik}}\left(t_N\right)=\frac{v_{i,(k-1)}\left(t_N\right)+v_{i,k}\left(t_N\right)}{2}\\ \mathrm{\Δ}{v}_{i(m_i+1)}\left(t_N\right)=\frac{v_{i,m_i}\left(t_N\right)+v_{(i+1)}\left(t_N\right)}{2}\end{array}\right.\·\·\·\·\·\·\left(2\right)$

t _NThe average stroke time between traffic flow sensor node i and the i+1 is constantly

Distance is between the adjacent node (comprising real sensor and pseudo-velocity sensor)

Then

$\left\{\begin{array}{c}{\mathrm{\ΔT}}_{\mathrm{ij}}\left(t_N\right)=\frac{{\stackrel{\‾}{s}}_i}{\mathrm{\Δ}{v}_{\mathrm{ij}}\left(t_N\right)}\\ T_i\left(t_N\right)=\underset{j=1}{\overset{m_i+1}{\mathrm{\Σ}}}\frac{{\stackrel{\‾}{s}}_i}{\mathrm{\Δ}{v}_{\mathrm{ij}}\left(t_N\right)}\end{array}\right.(j=\mathrm{1,2}{\·\·\·m}_i+1)\·\·\·\·\·\·\left(3\right)$

If total p traffic flow sensor node on this section, numbering is respectively 1,2 ... p, this section length is T then _NThe average stroke time of this section is constantly

Embodiment 2

Adopt the Xizhimenwai Dajie to carry out case verification.The traffic flow sensor placement in this highway section and between distance as shown in Figure 4.Wherein, 41003,41004 is microwave detector, and speed data is provided; 67,68,69 is detecting device hourage, and the journey time of each car process adjacent hourage of detecting device is provided.

Extract four days (2009.08.02-2009.08.05,41003 and 41004 speed datas 09:00-12:00).Utilize the parameter matrix with pseudo-velocity sensor speed and traffic flow sensor velocity correlation, obtain the speed of each pseudo-velocity sensor node.

Based on the pseudo-velocity sensor node, estimate the average stroke time between these two sensors.In order to make the result who draws have comparability, by

Try to achieve rough average hourage, and compare with hourage of predicting based on the pseudo-velocity sensor node and real hourage, its result as shown in Figure 5.

Utilize the standard variance σ of absolute error e, percentage error PE and absolute error to come checking measurements precision, e=|T ^*-T|,

Wherein, T ^*It is the hourage of calculating; T is real hourage; N is an experiment numbers.

The tabulation of table 1 experimental data

Experimental data with 2009.08.02 is an example, and its testing result is as shown in table 1.Wherein, T _EstBe based on value hourage of pseudo-velocity sensor prediction, e _Est, PE _EstBe its corresponding absolute error and percentage error, σ _EstIt is the standard variance of its absolute error; T _MeanBe value hourage of simple speed average computation, e _Mean, PE _MeanBe its corresponding absolute error and percentage error, σ _MeanIt is the standard variance of its absolute error.The mean absolute error of the hourage that prediction obtains based on the pseudo-velocity sensor is 2.54s, and the average percent error is 3.45%, and the standard variance of absolute error is 3.07; The mean absolute error of the hourage that obtains by simple speed average computation is 7.56s, and the average percent error is 10.33%, and the standard variance of absolute error is 8.13.Use the same method and handle the data of 2009.08.03-2009.08.05, the comparing result that obtains is as shown in table 2.

Table 2 experimental result contrast tabulation

From the experimental result of table 1 and table 2 as can be seen, value hourage that prediction obtains based on the pseudo-velocity sensor, mean absolute error is 3.38s, the average percent error is 4.64%, the mean value of the standard variance of absolute error is 3.99, every accuracy of detection all is better than the hourage of simple speed average computation, has higher precision.

Claims (3)

1. the through street predicting travel time method based on the pseudo-velocity sensor is characterized in that, may further comprise the steps:

The step that the pseudo-velocity sensor node is demarcated: adopt approach based on linear interpolation between adjacent traffic flow sensor, to set the pseudo-velocity sensor;

Utilize the pseudo-velocity sensor, the step of acquisition speed space distribution:

Utilize weight matrix and speed data to calculate the step of pseudo-velocity sensor node speed by multiple linear regression;

According to the speed of the pseudo-velocity sensor node of trying to achieve, the step of acquisition speed space distribution;

The step of estimating hourage:, obtain highway section hourage according to the space distribution of speed.

2. the through street predicting travel time method based on the pseudo-velocity sensor according to claim 1 is characterized in that, the described pseudo-velocity sensor node is demarcated is to suppose that the distance between traffic flow sensor node i and the i+1 is d _i, the distance setting between the adjacent node (node comprises traffic flow sensor and pseudo-velocity sensor) is a _i, then the number of the pseudo-velocity sensor node between traffic flow sensor node i and the i+1 is m _i=[d _i/ a _i].

3. the through street predicting travel time method based on the pseudo-velocity sensor according to claim 1 is characterized in that described weight matrix and pseudo-velocity sensor speed and traffic flow sensor velocity correlation are at t _NConstantly, (i, speed k) is defining virtual speed pickup node

Wherein, n _iNumber for the speed data of the traffic flow sensor i of needs; K=1,2......m _iw _KjBe weight coefficient,

t _NBe current time; Δ t is the time interval of traffic flow sensor acquisition data; v _i(t _N-q Δ t) (q=1,2...n _i) be traffic flow sensor i current time t _NThe q time preceding historical data.

Weight coefficient w _KjCan constitute a weight matrix β, Pseudo-velocity sensor node number between traffic flow sensor node i and the i+1 is m _i, then the pass of weight matrix and pseudo-velocity sensor speed and traffic flow sensor speed is:

$\left[\begin{array}{cccc}{w}_{11}& w_{12}& \·\·\·& w_{{1n}_i}\\ w_{21}& w_{22}& \·\·\·& w_{2n_i}\\ \·\·\·& \·\·\·& \·\·\·& \·\·\·\\ w_{m_{i}1}& w_{m_{i}2}& \·\·\·& w_{m_i{n}_i}\end{array}\right]\left[\begin{array}{c}{v}_i(t_N-\mathrm{\Δt})\\ v_i(t_N-2\mathrm{\Δt})\\ \·\·\·\\ v_i(t_N-n_i\mathrm{\Δt})\end{array}\right]=\left[\begin{array}{c}{v}_{i,1}\left(t_N\right)\\ v_{i,2}\left(t_N\right)\\ \·\·\·\\ v_{i,m_i}\left(t_N\right)\end{array}\right].$

Images