JP2001006090A - Traffic flow managing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a traffic flow managing method capable of highly accurately managing a traffic flow by finding the proper value of a coefficient in a car group diffusion model expression. SOLUTION: In the traffic flow managing method for managing the traffic flow using the car group diffusion model expression of TRANSYT, plural measuring instruments det0-det12 for measuring the volume of traffic are installed in the downstream direction from a crossing having a signal, the values of λ for correcting average traveling time and a car group diffusion coefficient αare found by using the measured data of these measuring instruments, and the traffic flow is estimated by the car group diffusion model expression while using the obtained values of λ and α. By using the values of λ and α reflecting a road state or traffic conditions, the traffic flow can be highly accurately managed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a traffic flow management method for predicting traffic flow at a downstream point from traffic flow data at an upstream point, and more particularly, to increasing traffic flow arriving at the next intersection from traffic flow data at an intersection. It is intended to be able to predict with high accuracy.

[0002]

2. Description of the Related Art In recent years, in order to realize smooth running by reducing the number of stops at intersections of vehicles running on a road, system signal control for controlling signal indication at intersections in a predetermined area has been performed. I have. The control parameters for signal indication include:
A cycle indicating the time from the green start time to the next start time, a split indicating the ratio of the green signal time of the intersecting road,
There is an offset indicating the shift of the green time start time from the adjacent intersection, and in the system signal control, at each intersection, the sum of the waiting time (delay time) and the number of stops at the vehicle signal is minimized. The control parameters for signal indication are set.

[0003] For the main purpose of determining the signal control parameters of the road network, TRA which is a simulator of street traffic flow is used.
NSYT (A Traffic Network Study Tool) 1967
It was developed at the TRRL Research Institute in the UK using a road outside London as a model.

In this TRANSYT, one cycle is divided into equally spaced steps (up to 60 steps), and the traffic flow at each step arriving at the downstream intersection is calculated based on the traffic flow flowing from the upstream intersection at each step. The prediction is made by the following vehicle group diffusion model equation (1). q ′ (k + t) = F · qk + (1−F) · q ′ (k + t−1) (1) where q′k: estimated flow arriving at the downstream intersection at step k qk: step k , The inflow from the upstream intersection at t: λ × average travel time F: smoothing coefficient F = 1 / (1 + αt) α: vehicle group diffusion coefficient

FIG. 5 schematically shows the transition of the traffic flow at each step according to the vehicle group diffusion model. The horizontal axis of the graph represents time (transition of steps), and the flow rate of the traffic flow that changes for each step is represented on the vertical axis. The upper shaded graph shows the traffic flow rate for each step flowing in from intersection 1, and the lower shaded graph shows the traffic flow rate for each step arriving at intersection 2. . The vehicle that has left the intersection 1 will arrive at the intersection 2 after the travel time has elapsed.

In addition, the outflow waveform of the vehicle flowing out of the intersection 2 when the traffic light at the intersection 2 turns green is shown on the opposite side of the lower hatched graph in FIG.

When the traffic light at the intersection 1 turns green, the group of vehicles stopped at the intersection 1 starts to run, but the vehicle group spreads and arrives at the intersection 2 because the speed of each vehicle varies. . Equation (1) is obtained by expanding (smoothing)
The traffic volume for each step of the arriving vehicle group is represented by a mathematical expression.

In the equation (1), the first term on the right-hand side represents the traffic volume when the vehicle group that has departed from the upstream intersection before the average travel time arrives at the downstream intersection after smoothing, and the second term on the right-hand side is Represents the traffic volume of the vehicle group that arrived at the step of k + t without being able to arrive at the downstream intersection at the step of k + t-1 for smoothing.

[0009] In consideration of the variation of the individual vehicle speed,
The average travel time is multiplied by λ for correction,
As this λ, 0.8 is used from an empirical rule. When the value of the vehicle group diffusion coefficient α is large, the model of the vehicle group is easily diffused. Conversely, when α is small, the vehicle group that has left the upstream intersection without spreading the vehicle group. The model arrives at the downstream intersection while maintaining that state. As α, a value of 0.2 to 0.6 is used from an empirical rule, and in the TRRL research institute that developed TRANSYT,
0.35 is used for α.

[0010] The road network is represented by a node indicating an intersection and a link which is a connecting road between the two nodes. When determining the signal control parameter of the system signal control, the delay time on the link (the time waited by the signal indication of the node connected to the link) and the number of stopped units on the link (link) expressed by the following equation (2) With the use of an evaluation function PI which means the sum of the number of vehicles and the number of vehicles stopped by the signal indication of the node connected to the node, the offset and the split are optimized such that the PI is minimized under the given cycle length. . PI = Σ (di + k · ci) (2) (Σ is added for i) where di: average delay time on the i-th link (passenger car converted number / time / time) ci: per second on the i-th link Average number of stops k: weighting factor

For this purpose, first, a signal control parameter is set to initial data, a traffic flow is estimated using the vehicle group diffusion model equation (1), and PI is calculated. Next, the offset is increased or decreased by a predetermined step size for each link to obtain an offset at which the PI becomes smaller.
By repeating this process while changing the step size, the “optimal offset” is finally obtained.

The same applies to the split. The blue time length is adjusted at the specified step width, and the optimum value is finally determined by observing the change in the PI value as in the offset optimizing process.

As described above, the signal control parameters can be optimized by using the vehicle group diffusion model equation represented by the equation (1).

[0014]

SUMMARY OF THE INVENTION However, TRANSY
Compared to the UK where the T was developed, the road conditions in Japan have differences such as smaller lane width, shorter link length, and shorter cycle length. It is far from the traffic conditions on the outskirts of London at the time of the development.

Therefore, when setting the signal control parameters of the intersection in Japan, λ (=
When the traffic flow is estimated using the values of 0.8) and α (= 0.2 to 0.6) as they are, there is a problem that the actual traffic flow is greatly different from the actual traffic flow and an optimal signal control parameter cannot be obtained. Can occur.

The present invention solves such a conventional problem, and provides a traffic flow management method capable of managing traffic flow with high accuracy by obtaining an appropriate value of a coefficient of a vehicle group diffusion model formula. It is intended to be.

[0017]

Therefore, in the present invention, T
When managing traffic flow using the vehicle group diffusion model formula of RANSYT, in the downstream direction from the intersection having a traffic light,
A plurality of measuring devices for measuring the traffic volume are installed, and the traffic flow is managed using the values of λ and α obtained using the measurement data of the measuring devices.

The value of λ is set to 1 and the value of α is set to 0.04.
The traffic flow is managed by a vehicle group diffusion model formula.

As described above, by calculating the coefficient value using the measurement data, the λ reflecting the road condition and the traffic condition is obtained.
And α can be obtained. According to the results of the simulation experiment, in Japan, the correction of the average travel time is unnecessary, and the vehicle group diffusion coefficient α is 0.04, which is extremely smaller than the conventionally used 0.2 to 0.6. Is found to be more suitable.

[0020]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The invention according to claim 1 of the present invention relates to a traffic flow management method for managing traffic flow using a TRANSYT vehicle group diffusion model formula, wherein a traffic flow is provided in a downstream direction from an intersection having a traffic light. A plurality of measuring devices for measuring the amount are installed, and the values of λ and the vehicle group diffusion coefficient α for correcting the average travel time are obtained using the measurement data of the measuring devices, and the obtained values of λ and α are obtained. The traffic flow is managed by the vehicle group diffusion model formula, and the traffic flow can be managed with high accuracy by using the values of λ and α reflecting the road condition and the traffic condition.

According to a second aspect of the present invention, a plurality of measuring instruments for measuring traffic volume are installed downstream from an intersection having a traffic light, and the average travel time is corrected using the measurement data of the measuring instruments. Assuming that λ is 1, the value of the vehicle group diffusion coefficient α is determined, and the traffic flow is managed by the vehicle group diffusion model formula using the obtained value of α. In the traffic situation in Japan, the average travel The traffic flow can be managed with high accuracy without correcting the time.

According to a third aspect of the present invention, an actual measurement value of a smoothing coefficient is calculated using measurement data of a measurement device closest to a traffic light and measurement data of a downstream measurement device, and the actual measurement value and α are used. The value of α that minimizes the sum of squares of the difference between the smoothing coefficient and the theoretical value represented by the following formula is obtained, and the optimum α can be calculated.

According to a fourth aspect of the present invention, as the measuring device,
A vehicle sensor is used, and necessary measurement data can be obtained by using a vehicle sensor currently installed at an intersection or the like and adding only an insufficient number of vehicle sensors. .

According to the present invention, the value of λ for correcting the average travel time is set to 1 and the value of the vehicle group diffusion coefficient α is set to 0.0.
The traffic flow is managed by the vehicle group diffusion model formula as No. 4. From the results of the simulation experiment, it is not necessary to correct the average travel time. It is recognized that 0.04, which is extremely smaller than 0.2 to 0.6, is more suitable.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

In the traffic flow management method of the present invention, a plurality of sensors are installed at equal intervals on the downstream link of the intersection, and the coefficients λ and α of the vehicle group diffusion model equation (1) are obtained by using data obtained by each of the sensors. Set the value of.

Hereinafter, a simulation experiment of traffic flow estimation performed by the inventors and the like will be described, and a traffic flow management method of the present invention will be described based on the results of the experiment.

In the simulation experiment, as shown in FIG. 1, 13 detectors (det0 to det12) were installed at a downstream link of an intersection after passing a traffic light on a three-lane road at intervals of 50 meters, and the generated traffic volume was reduced. We changed variously and grasped the situation of vehicle group diffusion at that time.

The system parameters of the simulation in this experiment are as follows. -Signal control parameters Cycle length 100 seconds Split 60%-Detector installation location Immediately after passing through the intersection and at a point 50m to 600m downstream from the intersection (see Fig. 1)-Generated traffic volume 400 to 1600 vehicles / hour at intervals of 200 vehicles The traffic volume is changed, and the vehicle group diffusion situation in each case is simulated.・ Vehicle characteristics of each car Average deceleration rate 0.21G Maximum deceleration rate 0.30G Average acceleration rate 0.19G Reaction speed 1.0sec Critical deceleration rate 0.30G Minimum deceleration rate 0.15G Maximum acceleration rate 0.21G ・ Each Target vehicle speed and its mixing ratio The target speed is randomly given to each vehicle at the ratio shown in Table 1.

[0030]

[Table 1]

In this simulation experiment, equation (1)
The variables λ and α are verified. Therefore, λ is changed by 0.1 in the range of 0.5 to 1.5. As for α, the minimum α for each λ is obtained by the least square method.

Therefore, the following analysis is performed for each λ. [1] A smoothing coefficient F is obtained for each simulation in which the generated traffic volume differs. [2] Using the smoothing coefficient F obtained in each simulation and t (= λ × average travel time) corresponding thereto, an optimum α is obtained for each generated traffic volume.

First, the calculation of the smoothing coefficient F in [1] is performed in the following procedure.

Step 1: At each step k, the data of the sensor det0 immediately after passing through the intersection is used as link upstream data and used as qk in the equation (1). Also, det1
１２12 are used as downstream data q′k of equation (1).

Step 2: The average travel time between the upstream sensor (det0) and the downstream sensor (deti (i = 1, ‥, 12)) is obtained, and the average travel time is multiplied by λ to obtain t.
Find (i).

Step 3: q, q '(i) (i = 1,
‥, 12), the smoothing coefficient F between det0 and deti
(I) is obtained by the least squares method. Here, equation (1) q ′ (k + t) = F · qk + (1−F) · q ′ (k + t−1) is transformed as in the following equation (3). q ′ (k + t) −q ′ (k + t−1) = F · (qk−q ′ (k + t−1)) (3)

For the sake of simplicity, let q and q ′ be time k
Equation (3) is taken as the following equation (4).

(Equation 4)

Then, the equation (3) becomes the following equation (5).

(Equation 5)

Then, to apply the least squares method, the following equation (6) is used.

(Equation 6) Is calculated to minimize. Therefore, F in equation (6)
F that makes the first-order differential equation of 0 zero is obtained. This F is obtained as the following equation (7).

Equation 7

When these x and y are returned to q and q ', F (i) is calculated by equation (8). F (i) = {{q '(k + t) (i) -q' (k + t-1) (i)}. {Qk (i) -q '(k + t-1) (i) } / {{Q '(k + t) (i) -q' (k + t-1) (i)} ² (Σ is added for k) (8)

Next, α in the above [2] is calculated. From the data of the smoothing coefficient F (i) obtained in each simulation and the corresponding t (i) (= λ × average travel time),
For each generated traffic volume (generated traffic volume Q, mixing rate pattern)
Then, the optimal α that satisfies F = 1 / (1 + α · t) (9) is obtained by the least squares method.

This corresponds to the residual f defined by the following equation (10).
(Α) f (α) = {Fn−1 / (1 + α · tn)} ² (10) (Σ is an addition for n). Therefore, f
(Α) is differentiated, and the differential equation of the following equation (11): f ′ (α) = {2tn {Fn · (1 + α · tn) -tn} / (1 + α · tn) ³ (11) Α that satisfies' (α) = 0 is obtained by Newton's method, and α that minimizes the residual f (α) is calculated.

FIG. 2 is a flowchart showing the procedure of the simulation.

Step 1: Set λ to 0.5; Step 2: Set the generated traffic volume Q to 400 vehicles / hour; Step 3: Set the data of the upstream sensor (det0) to q; Step 4: Det1 as downstream sensor (deti)
(I = 1) is specified. Step 5: Data of the downstream sensor det1 is set to q ′. Step 6: ti = λ × average travel time (i = 1, λ =
0.5) is calculated. Step 8: The smoothing coefficient F (1) is calculated by the equation (8).

Step 9: Next, the value of i is increased by 1 and the procedure from step 5 is repeated. After performing the procedure from Step 5 onward for i = 12, Step 10: If there is a pattern of the mixing rate that has not been simulated, the process returns to Step 3 and returns to Step 3 based on the pattern of the mixing rate. Repeat the following steps. Step 11: Calculate α in the equation (9) by the least squares method when there is no unsimulated mixing ratio pattern.

Step 12: Next, the generated traffic volume Q is set to 20
0 is added at a time, and the procedure after step 2 is repeated.
After performing the procedure from step 2 for 1600 units / hour, step 13: add 0.1 to the value of λ.

This procedure is repeated until the process for λ = 1.5 is completed.

FIG. 3 shows the relationship between the generated traffic volume Q (z axis) at each λ (y axis) and the vehicle group diffusion coefficient α (x axis) calculated by the least square method. From this analysis result, it can be seen that the vehicle group diffusion coefficient α does not depend on the generated traffic volume Q.

Therefore, irrespective of the generated traffic volume, the optimum α obtained by the least square method at each λ and the residual f
(Α = minimum) is obtained as shown in FIG.

It is noted from FIG. 4 that the value of α is extremely small. Many values currently used are λ = 0.8 and α
= 0.35, but in this simulation experiment, except for λ = 0.5, all vehicle group diffusion coefficients were α ≦
0.1, which is a value considerably smaller than the TRANSYT recommended value.

From this, it can be said that a group of vehicles does not tend to spread on a Japanese road even when the traffic volume is small. This is thought to be due to the short link length as a feature of the road structure in Japan. In addition, due to recent improvements in vehicles, the magnitudes of the acceleration rate and deceleration rate have been significantly greater than when the TRANSYT model was created. This is considered to be one of the causes.

As for λ, the smaller the residual value (error value), the better. Therefore, from FIG.
= 0.5 is the optimal value of λ. However, as is clear from FIG. 4, if λ is 1.0 or less, the residual hardly changes. On the other hand, α shows the smallest value when λ is around 1.1.

Therefore, from this simulation experiment, it is necessary to set λ = 1.0 and take α = 0.04, which is the value of α when λ = 1.0, as the value of α. In some cases, no correction is required, and it is concluded that this is appropriate.

As shown by the results of this simulation experiment, the vehicle group diffusion coefficient α does not depend on the traffic Q. Therefore, when calculating the vehicle group diffusion coefficient α required for predicting the traffic volume in a certain area, multiple sensors are installed at equal intervals on the downstream link of the intersection to be surveyed, and a certain point in time obtained by each sensor The values of the coefficients λ and α of the vehicle group diffusion model equation (1) can be set to appropriate values using the traffic volume data of (1).

In this case, the procedure shown in FIG. 2 is executed by skipping steps 2, 10, and 12, and while changing the value of λ by 0.1, the minimum α for each λ is determined by the least squares method. The relationship between λ, the vehicle group diffusion coefficient α, and the residual (error) shown in FIG. 4 is obtained. Then, considering the magnitude of the residual (error) and the magnitude of the vehicle cluster diffusion (α), by setting λ and α corresponding thereto, road conditions and traffic conditions in the area can be determined. Λ and α reflected
Can be obtained.

As shown by the results of this simulation experiment, under the road conditions in Japan, the average travel time is λ
It is considered that the need for correction is small. Therefore, in the procedure of FIG.
Is skipped, and the procedure is executed with λ = 1, so that the vehicle group diffusion coefficient α can be easily obtained.

On a road in an environment similar to the road targeted in the current simulation experiment, λ = 1.0, α =
The traffic volume can be predicted using the value of 0.04.

Here, the case where actual measurement data is obtained using a vehicle sensor has been described. However, it is also possible to obtain data by video analysis using a video camera instead of the vehicle sensor.

[0059]

As is apparent from the above description, in the traffic flow management method of the present invention, appropriate values reflecting road conditions and traffic conditions are obtained as λ and α in the vehicle group diffusion model equation of TRANSYT. Can be. Therefore, the traffic flow can be managed with high accuracy by using the vehicle group diffusion model formula, and as a result, it is possible to appropriately set the signal control parameters in the system signal control.

In order to collect necessary measurement data,
A vehicle detector or the like currently installed at an intersection or the like can be used.

[Brief description of the drawings]

FIG. 1 is a diagram showing an installation position of a vehicle sensor installed for implementing a traffic flow management method according to an embodiment;

FIG. 2 is a flowchart showing the procedure of a simulation experiment;

FIG. 3 is a diagram showing a relationship between generated traffic volume, λ, and α obtained as a result of a simulation experiment;

FIG. 4 shows λ obtained as a result of a simulation experiment.
FIG.

FIG. 5 is a diagram showing a vehicle group diffusion model.

[Explanation of symbols]

 det0-det12 Vehicle detector

Claims (5)

[Claims] 1. A traffic flow management method for managing traffic flow using a TRANSYT vehicle group diffusion model formula, comprising: installing a plurality of measuring devices for measuring traffic volume in a downstream direction from an intersection having a traffic light; Using the measurement data of the vehicle, the value of λ for correcting the average travel time and the vehicle group diffusion coefficient α are obtained, and the traffic flow is managed by the vehicle group diffusion model formula using the obtained values of λ and α. Traffic flow management method characterized by performing. 2. A traffic flow management method for managing traffic flow using a vehicle group diffusion model formula of TRANSYT, comprising: installing a plurality of measuring devices for measuring traffic volume in a downstream direction from an intersection having a traffic light; The vehicle group diffusion coefficient α is determined by setting λ for correcting the average travel time to 1 using the measurement data of the vehicle, and the traffic flow is managed by the vehicle group diffusion model formula using the obtained value of α. Traffic flow management method characterized by performing. 3. An actual measurement value of a smoothing coefficient is calculated using measurement data of a measurement device closest to the traffic signal and measurement data of a downstream measurement device, and smoothing expressed using the actual measurement value and α. 3. The traffic flow management method according to claim 1, wherein a value of α that minimizes a sum of squares of a difference between the coefficient and a theoretical value is obtained. 4. The traffic flow management method according to claim 1, wherein a vehicle sensor is used as the measuring device. 5. A traffic flow management method for managing traffic flow using a TRANSYT vehicle group diffusion model formula, wherein the value of λ for correcting the average travel time is 1, and the value of the vehicle group diffusion coefficient α is 0.04. A traffic flow management method, wherein the traffic flow is managed by the vehicle group diffusion model formula.