JP5917280B2 - Traffic data analysis apparatus and computer program - Google Patents

Description

  The present invention relates to a traffic data analysis apparatus having a function for modeling traffic flow on a route (road) in an urban area, that is, a function for setting a traffic flow model for a route, and a computer for the apparatus Regarding the program.

  For example, in order to provide appropriate traffic control and information to each vehicle for routes such as urban roads where traffic conditions change from moment to moment, grasp traffic conditions such as traffic congestion on the routes in advance. As a method for grasping such a traffic situation, a simulation technique using a traffic flow model is known (for example, see Non-Patent Document 1).

  A traffic flow model consists of correlations (model formulas) such as traffic volume, traffic density, and vehicle speed, which are variables that express the traffic flow state. Using this traffic flow model, the traffic flow state of a route is analyzed. Is possible. Various studies have been made on the traffic flow model. For example, traffic flow models such as a stochastic statistical model, a fluid model, a follow-up model, a cellular automaton model, and an optimum speed model have been proposed.

  In a simulation using such a traffic flow model, for example, a predetermined traffic flow model is set for a certain route, and the traffic flow of the route is modeled and analyzed using the traffic flow model. As a result, it is possible to estimate traffic conditions on the route, that is, analyze traffic jam prediction on the route.

Masaaki Okushima and three others, “Analysis of traffic flow characteristics on urban expressways and application to traffic flow simulation” [online], September 2002, Japan Society of Civil Engineers, [April 10, 2012 search], Internet <URL: http://library.jsce.or.jp/jsce/open/00035/2002/57-4/57-4-0222.pdf>

  However, a route such as a road in an urban area includes a plurality of road links such as a VICS link (“VICS” is a registered trademark), for example, and road characteristics (road structure) are different for each road link. For this reason, it is difficult to approximate the traffic flows of all road links included in a certain route to a common traffic flow model.

  In other words, the set traffic flow model is suitable for one road link included in the route and enables high-accuracy analysis, but the traffic flow model is used for other road links included in the route. Therefore, it is not always suitable for the traffic jam prediction of the other road links by using the set traffic flow model.

  SUMMARY OF THE INVENTION An object of the present invention is to provide a traffic data analysis apparatus capable of setting a traffic flow model suitable for a route to be analyzed, and a computer program for this apparatus.

(1) The present invention provides a traffic flow model construction unit that sets a traffic flow model for a section based on traffic data acquired for a section included in a route to be analyzed, and a plurality of traffic flow model candidates. A storage unit that stores basic expressions of types of candidate models , and the traffic flow model construction unit uses the basic expressions of the plurality of types of candidate models to obtain a group of traffic data acquired for the section. , Each of the candidate models can be approximated, and the traffic flow model of the section can be set based on the accuracy of the approximation . Further, the traffic flow model construction unit can include a plurality of candidate models for one section. The traffic flow model can be set from the basic formula, and a model formula that continues the model formulas indicating the traffic flow model set from the basic formulas of the plurality of candidate models is further set .

According to the present invention, a candidate model suitable as a model representing a group of traffic data acquired for a section among a plurality of candidate models can be set as a traffic flow model for that section, and is suitable for each section. A traffic flow model can be set. As a result, using this set traffic flow model, for example, if the traffic situation of the route to be analyzed is analyzed, a highly accurate result can be obtained.
(2) Further, the present invention is based on the traffic data acquired for the section included in the route to be analyzed, and the traffic volume, the traffic density, and the vehicle that are variables expressing the traffic flow state of the section. A traffic flow model construction unit that sets a traffic flow model configured by at least two of the traveling speeds, and a storage unit that stores basic expressions of a plurality of types of candidate models that are candidates for the traffic flow model; The traffic flow model construction unit approximates a group of traffic data acquired for the section to each of the candidate models using basic expressions of the plurality of types of candidate models, and based on the accuracy of the approximation The traffic flow model for the section can be set, and the traffic flow model construction unit can select a basic candidate model corresponding to the traffic flow model that has been used for the section. And using the basic formula of a plurality of types of candidate models, including the basic type of the other candidate model, it is possible to update the traffic flow model for that segment.

( 3 ) In addition, a priority order is set for the plurality of candidate models stored in the storage unit, and the traffic flow model construction unit assigns a group of traffic data acquired for the section to a priority order. It is preferable that approximation is performed in order from the higher-ranking candidate model, and the candidate model is set as the traffic flow model of the section when the approximation accuracy is a certain level or higher.
In this case, it is possible to determine the traffic flow model of the section at an early stage without performing a process of approximating all candidate models stored in the storage unit.

( 4 ) The candidate model includes a variation parameter indicating the characteristics of the section, and the traffic flow model construction unit changes the variation parameter within a range that the variation parameter can take, It is preferable to repeatedly perform a process of approximating the group of traffic data acquired for the section to the candidate model.
In this case, even if each section has different characteristics (road characteristics), it is possible to set a traffic flow model according to the characteristics.

( 5 ) Further, when a traffic data group is newly acquired for the section, the traffic flow model construction unit approximates the newly acquired traffic data group to each of the plurality of candidate models, When the accuracy of the approximation becomes higher, it is preferable to update the candidate model as a traffic flow model of the section.
In this case, even if the section characteristics (road characteristics) change due to secular change, the traffic flow model can be updated according to the change.

( 6 ) Moreover, it is preferable that the said traffic flow model construction part sets the same traffic flow model about the area where the characteristic of a road is the same or similar.
In this case, it is only necessary to set one traffic flow model for sections with the same or similar road characteristics, and it is not necessary to perform the approximation processing for all sections, reducing processing time and processing load. Can be reduced.
For this reason, sections with the same or similar road characteristics are grouped, and it is preferable that the same group identification information (group ID) is given to these grouped sections. This facilitates the traffic flow model setting process.

( 7 ) In addition, the traffic data analysis device further includes a prediction unit that predicts the traffic situation of the section based on the traffic flow model of the section set by the traffic flow model construction unit. preferable.
In this case, when a traffic flow model of a section set by the traffic flow model construction unit is set, the traffic situation of the section can be predicted based on the traffic flow model.

Further, the traffic flow model construction unit of the traffic data analysis apparatus of (1) can set a traffic flow model from a plurality of basic formulas of the candidate models for one section .
For example, even if a traffic flow model is set from one candidate model for one section, the accuracy of approximation may not increase as a whole, but by setting a traffic flow model from a plurality of candidate models, This can be solved.

Further, the traffic flow model construction unit of the traffic data analysis device of (1) further adds and sets a model formula that continues the model formulas indicating the traffic flow model set from the basic formulas of the plurality of candidate models. To do .
If a traffic flow model is set from a plurality of candidate models for one section, the model equations indicating the traffic flow model set from these candidate models may be discontinuous. However, since these model formulas are additionally set to make the model formulas continuous, the traffic flow model in this section can be given continuity as a whole.

( 8 ) Further, the present invention is a computer program for causing a computer to execute a process of setting a traffic flow model of a section based on traffic data acquired regarding a section included in a route to be analyzed. Te, a computer, a group of traffic data obtained with respect to the section, using the basic formula of a plurality of types of candidate models that are candidates for traffic flow model, means to approximate to each the candidate model, the accuracy of the approximate A computer program for functioning as a means for setting a traffic flow model of the section based on the information, and the means for setting the traffic flow model based on a plurality of basic equations of the candidate models for a section. A flow model can be set, and model formulas indicating the traffic flow model set from a plurality of basic formulas of the candidate models Further set by adding a model formula for continuous.
According to this invention, there exists an effect similar to the traffic data analysis apparatus of said (1).
(9) Moreover, the present invention is based on the traffic data acquired regarding the section included in the route to be analyzed, based on the traffic volume, the traffic density, and the vehicle that are variables expressing the traffic flow state of the section. A computer program for causing a computer to execute a process for setting a traffic flow model constituted by at least two relationships of travel speeds, the computer, a group of traffic data acquired for the section, A function for approximating each of the candidate models using basic equations of a plurality of types of candidate models that are candidates for a traffic flow model, and for functioning as a means for setting the traffic flow model of the section based on the accuracy of the approximation A computer program, and the means for setting the traffic flow model has been used so far for the section Using the basic formula for a plurality of types of candidate models, including the basic type of the basic type and the other candidate model candidate model corresponding to the passing flow model, it is possible to update the traffic flow model for that segment.

  According to the present invention, it becomes possible to set a traffic flow model suitable for each section included in a route to be analyzed. By using this traffic flow model, for example, by analyzing the traffic situation of the route, the accuracy can be improved. Good results can be obtained.

It is a block diagram which shows the structure of the traffic data analyzer of this invention. It is explanatory drawing of the route used as analysis object. It is a graph which shows the relationship between traffic volume and traffic density. It is a graph which shows the relationship between driving speed and traffic density. It is explanatory drawing of the model type | formula of a traffic flow model. It is explanatory drawing of the information passed to a simulation part from a traffic flow model construction part. It is explanatory drawing of the process performed by the simulation part. It is a flowchart explaining the process until the setting of a traffic flow model. It is explanatory drawing which shows the relationship between the traffic volume and traffic density for demonstrating the function of a traffic flow model. It is explanatory drawing which shows the relationship between the traffic volume and traffic density for demonstrating the function of a traffic flow model. It is explanatory drawing which shows the relationship between the traffic volume and traffic density for demonstrating the function of a traffic flow model.

Embodiments of the present invention will be described below.
[Overall configuration of traffic data analyzer]
FIG. 1 is a block diagram showing a configuration of a traffic data analysis apparatus 1 according to the present invention. The traffic data analysis device 1 (hereinafter also simply referred to as the analysis device 1) uses a traffic flow model to model a traffic flow in a section included in a route such as a road in an urban area, and the traffic volume on the route. It is an apparatus for estimating (predicting) information indicating a traffic situation on a road such as.

  The analysis device 1 includes a computer device including a processing device 2 including a CPU, a storage device 3 including a hard disk, an input / output device 4 including a keyboard and a monitor, and the like for causing the computer device to function as the analysis device 1. The analysis apparatus 1 is configured by installing a computer program. The functions of each functional unit, which will be described later, included in the analysis device 1 are exhibited when the computer program is executed by the computer device (processing device 2).

The processing device 2 of the analysis apparatus 1 includes a traffic flow model construction unit 10 (hereinafter simply referred to as construction unit 10) and a simulation unit 20 as functional units.
The construction unit 10 constructs a traffic flow model that models the traffic flow of the route based on the traffic data acquired on the route, and provides the information on the constructed traffic flow model to the simulation unit 20.

  By using this traffic flow model, it is possible to analyze the state of traffic flow on the route and the characteristics of the route (the road link included in the route). Therefore, the simulation unit 20 performs a process of predicting the traffic volume and travel time of the route using the information acquired from the construction unit 10 and the actual measurement value of the traffic data for performing the simulation.

  In this embodiment (see FIG. 2), the route R to be analyzed (simulation target) is an urban road, and there is one section between the sensors 6 installed at each position on the route R. It is set as a road link, and FIG. 1 shows road links L11 to L15. That is, the route R includes a plurality of road links (sections).

  The sensor 6 is a vehicle detection sensor that detects a vehicle traveling on the route R, and can detect the number of passing vehicles, and the sensor 6 can detect the road links L11 to L15 based on the detection result. The traffic density K and the (average) traveling speed V of the vehicle can be calculated. Further, the traffic volume Q can be calculated from the traffic density K and the traveling speed V of the vehicle. The unit of traffic volume Q is “number / time”, the unit of traffic density K is “number / distance”, and the unit of travel speed V is “distance / time”.

[Description of traffic flow model]
The traffic flow model is composed of relationships such as traffic volume Q, traffic density K, vehicle travel speed V, etc., which are variables expressing the state of traffic flow, and the relationship is expressed as a model formula (function). .
An example of a traffic flow model will be described. FIG. 3 is a graph showing the relationship between the traffic volume Q and the traffic density K, and FIG. 4 is a graph showing the relationship between the travel speed V and the traffic density K. The curves (KQ curve and KV curve) shown in FIGS. 3 and 4 can be expressed as model equations of the traffic flow model, and the model equations are as shown in FIGS. 3 and 4.

  A plurality of model formulas are used, and an example thereof is shown in FIG. In FIG. 5, models A to E are set as model formulas. These model formulas are formulas of candidate models that are candidates for road flow traffic flow models, and model formulas (models A to E) of the plurality of candidate models are stored in the storage device 3 (see FIG. 1). . That is, the models A to E are basic expressions of candidate models that can be traffic flow models. Note that the candidate model may be other than that shown in FIG. 5, or may be a uniquely defined model formula.

In FIG. 3, with the point on the curve that takes the maximum value of the traffic volume Q as a boundary, the side where the traffic density K is small is the “free flow region”, and the side where the traffic density K is large is the “congestion region”. The point at which the traffic volume Q takes the maximum value is the “critical density Kc”.
In FIG. 4, the maximum value of the velocity V is “free flow velocity Vf”.
In FIG. 5, “jam density Kj” in the model formula is the maximum value of the traffic density K, and “C” in the model formula is the possible traffic capacity. “Α, L, c” are “constants” indicating the characteristics of the road link, and can take values within a predetermined range. These “constants” are also referred to as variation parameters below.

[About the construction unit 10]
The construction unit 10 has a function of setting, for each road link, a traffic flow model adapted to the characteristics of the road link based on the actual traffic data.
Here, the “actual traffic data” is data indicating the traveling state of the vehicle generated based on the information actually acquired by the sensor 6 (see FIG. 2). For example, the traffic volume at each road link Q, at least two of the traffic density K and the traveling speed V of the vehicle.
In addition, since many vehicles pass through each road link, a lot of actual traffic data is acquired for each road link. A large number of actual traffic data for each road link is referred to as a “group of actual traffic data”. The group of actual traffic data may be a representative value.

The construction unit 10 has a function (approximation function) for approximating a group of actual traffic data acquired for each road link to each of a plurality of candidate models (models A to E in FIG. 5) stored in the storage device 3. Have. Further, the construction unit 10 has a function (setting function) for setting the candidate model having high approximation accuracy as a traffic flow model of the road link.
For example, a group of actual traffic data acquired for the road link L12 in FIG. 2 is approximated to each of the candidate models A to E shown in FIG. Then, the candidate model with high approximation accuracy is set as the traffic flow model of the road link L12.

Although the procedure of the approximation function and the setting function will be described later, the approximation method of each group of actual traffic data to the candidate models A to E can be, for example, linear regression.
In this case, the construction unit 10 calculates a determination coefficient as the “degree of fit” to each of the candidate models A to E of the group of the actual traffic data, and sets a candidate model with the preferable determination coefficient as the traffic flow model. To do. The approximation method may be another method, and the degree of fit may be an index other than the determination coefficient.

  As described above, the construction unit 10 sets the traffic flow model (model formula) of the road link based on the actual traffic data acquired for the road link included in the route R such as a road in an urban area. Can do. Then, the construction unit 10 sets a traffic flow model for each road link included in the route R.

  When the traffic flow model is set by the construction unit 10, various information is passed to the simulation unit 20. Information to be passed to the simulation unit 20 includes allowable traffic flow, critical traffic density, jam density, free flow velocity, KQ modeling function type, application range, and variation parameter. An example of the information (file format) is shown in FIG. Shown in

The KQ modeling function type indicates the type of model expression approximated when modeling the traffic flow, and is the type of candidate model (any one of models A to E) set by the construction unit 10. is there. The application range indicates the range of traffic density to which the candidate model is applied. The other information is obtained by the construction unit 10 based on the set traffic flow model (model formula).
The variation parameter is a constant that depends on the model formula. For example, when the KQ modeling function type is model D (see FIG. 5), “parameter 1” is a value of “Vf”, and “parameter” “2” is the value of “Kc”, and “Parameter 3” is the value of “L”.

[About the simulation unit 20]
The simulation unit 20 has a function (prediction function) as a prediction unit that predicts the traffic situation of the road link based on the traffic flow model of the road link set by the construction unit 10.
As a simulation method executed by the simulation unit 20, various methods can be adopted, but in this embodiment, a block density method is used (see FIG. 7). In the block density method, each road link is divided into finer blocks, and the movement between the blocks of the traffic flow that approximates the fluid is represented by the relationship between the traffic density and the traffic volume (KQ) and the traffic volume conservation law. This is a dynamic simulation method obtained using

[About the processing of the construction unit 10 and the simulation unit 20]
An analysis method executed by the analysis apparatus 1 having the above configuration will be described. FIG. 8 is a flowchart for explaining processing up to setting of a traffic flow model. Each process of FIG. 8 is performed by the construction unit 10.

  The construction unit 10 acquires actual traffic data for each road link (step S1 in FIG. 8). In the present embodiment, the traffic volume Q and the traffic density K are acquired as actual traffic data for each of the road links L11 to L15 (see FIG. 2). Since a large number of vehicles pass through each of the road links L11 to L15, a large number of actual traffic data is acquired and a “group of actual traffic data” is acquired. Further, in the present embodiment, the construction unit 10 does not sequentially acquire individual actual traffic data, but, for example, a central device (not shown) that manages the sensor 6 collects and acquires individual actual traffic data. The construction unit 10 collectively acquires the actual traffic data as a group of actual traffic data from the central device.

  In the storage device 3, basic formulas (models A to E in FIG. 5) of a plurality of candidate models that can be traffic flow models are registered. Therefore, the construction unit 10 selects a candidate model that may be set for each of the road links L11 to L15 from among a plurality of candidate models (models A to E) registered in the storage device 3 ( Step S2). Some of the candidate models may be selected, or all candidate models may be selected. As a case where a part is selected, for example, a priority order is set for a registered candidate model, and several candidate models at the top are selected.

  In addition, although each step shown in FIG. 8 is performed in order about each of the road links L11-L15, in the following description, the road link L11 is demonstrated as a representative. In step S2, candidate models that may be set for the road link L11 are selected from a plurality of candidate models (models A to E) registered in the storage device 3.

[Embodiment (Part 1)]
A case where all candidate models A to E are selected in the order in step S2 will be described.
The construction unit 10 approximates the “group of actual traffic data” of the road link L11 acquired in step S1 to the model formula A of the candidate model (step S3). The approximation process is linear regression as described above. Further, a determination coefficient in the case of approximating the “group of actual traffic data” to the model formula A is calculated (step S4). The calculation result is stored in the storage device 3.

It is determined whether or not the variation parameter is included in the candidate model selected in step S2 (step S5).
When the determination result in step S5 is negative (in the case of “NO” in step S5), among the candidate models selected in step S2, steps 2 to 4 are repeated for the next candidate model to calculate the determination coefficient. Further, this is repeated to calculate the determination coefficients of the remaining candidate models (step S6). As will be described later, the model equation D includes the fluctuation parameter “L”. However, in step S3, an initial value is set for this “L”.

  If the determination result in step S5 is affirmative (in the case of “YES” in step S5), that is, in the present embodiment, the variation parameter “L” is included in the model formula D (see FIG. 5). For candidate model D, the process proceeds to step S7. That is, for the model formula D, the variation parameter “L” is changed, approximation processing is performed, and the determination coefficient is obtained.

  Note that the fluctuation parameter is set in advance to a value within a predetermined range. For example, it is assumed that the fluctuation parameter “L” is set in a range of “1 to 3”. Further, it is assumed that the step value of the variation parameter “L” is set to “0.1”. In this case, the variation parameter “L” is changed in steps of “0.1” within the range of “1-3” and approximated to the model formula D, the coefficient of determination is calculated for each step value, and the end condition is This is repeated until it is satisfied (steps S7 to S9). The end condition is that the variation parameter “L” is completely changed by a value “1 to 3” within a predetermined range. That is, “L” is changed from 1.1 to 1.1, 1.2, 1.3... 2.8, 2.9, 3 in order, and the determination coefficient in each case is calculated. A determination of “NO” in step S9 is made.

  When the end condition is satisfied (in the case of “NO” in step S9), step 2 to step 2 are performed for the next candidate model among the candidate models selected in step S2, similarly to the case where the determination result in step S5 is negative. Step 5 and, in some cases, steps S7 to S9 are repeated to calculate the determination coefficients of all candidate models (step S6).

  As described above, when the candidate model includes the variation parameter (L) indicating the characteristic of the road link L11 (in the case of “YES” in step S5), the variation parameter can be taken as in steps S7 to S9. Within the range, the variation parameter is sequentially changed, and the process of approximating the group of actual traffic data acquired for the road link L11 to the candidate model is repeatedly performed. For this reason, even if the characteristic (road characteristic) which road link L11 has is different, it becomes possible to set up a traffic flow model according to the characteristic.

When the determination coefficients of all candidate models are calculated (in the case of “NO” in step S6), the construction unit 10 sets the candidate model having the largest determination coefficient as the traffic flow model of the road link L11 ( Step S10).
Such processing is also executed for other road links.

[Embodiment (2)]
A case where priority is set for the candidate models (model formulas A to E) registered in the storage device 3 will be described. For example, it is assumed that the priority order is set in the order of candidate models C, D, E, A, and B.
In this case, in step S2, candidate models are selected in this priority order. The construction unit 10 then approximates the “group of actual traffic data” acquired with respect to the road link L11 in order from the candidate model having the higher priority (step S3). That is, the approximation process is performed from the candidate model C having the highest priority. Then, the determination coefficient is calculated (step S4).

Then, an approximation process is performed on the candidate model D having the next highest priority, and a determination coefficient is calculated.
When the accuracy of approximation for a certain candidate model is equal to or higher than a certain level, that is, when the calculated determination coefficient is equal to or higher than a threshold, the candidate model is set as a traffic flow model of the road link L11. For example, when the determination coefficient calculated for the candidate model D is greater than or equal to the threshold, the candidate model D is set as the traffic flow model of the road link L11.
In this case, it is possible to determine the traffic flow model of the road link L11 at an early stage without performing the process of approximating all of the registered candidate models. That is, in the present embodiment, the determination coefficient may be calculated by performing approximation processing only on the candidate models D and E.
Note that the processing in steps S7 to S9 in FIG. 8 is also effective in this embodiment.

  In addition, for example, when setting the traffic flow model from the downstream road link, the priority order is determined based on the traffic flow model (model formula) already set for the adjacent road link on the downstream side. It is preferable to use a traffic flow model (model formula). In addition, when the traffic flow model is set from the upstream road link, the traffic flow model (model formula) already set for the adjacent road link on the upstream side is replaced with the upper traffic flow model (model formula). ). This is because the form and characteristics of a certain road link are likely to be the same as or similar to the form and characteristics of the adjacent road link.

Embodiment (No. 3)
In the first embodiment and the second embodiment, the case where one traffic flow model is set from one candidate model registered in the storage device 3 for one road link L11 has been described. A traffic flow model may be set from a plurality of candidate models registered in the storage device 3 for one road link L11.

That is, for example, as shown in FIG. 9, when the relationship between the traffic volume Q and the traffic density K is different between the free flow area and the traffic jam area with respect to the road link L11, one candidate model for the road link L11 is used. Therefore, even if a traffic flow model is set, the accuracy of approximation may not increase as a whole.
However, as shown in FIG. 10, it is considered that this can be solved by setting a traffic flow model from a plurality (two in the example) of candidate models.

Therefore, as shown in FIG. 10, when the relationship between the traffic volume Q and the traffic density K is different between the free flow area and the congestion flow area, the candidate models are approximated separately in the free flow area and the congestion flow area. Thus, the accuracy of approximation as a whole can be improved. Even in this case, the approximation process is performed using a group of actual traffic data including both the free flow area and the congestion flow area of the road link L11.
For this purpose, the construction unit 10 calculates the “critical density Kc” using the group of the actual traffic data, and the “critical density Kc” is used as a boundary and is different in both the free flow area and the congestion flow area. Approximate the candidate model, calculate the coefficient of determination, and set the traffic flow model.

In other words, the traffic flow model in the free flow area is set based on the group of actual traffic data in the free flow area, and independently of this, the traffic flow in the congested flow area is determined based on the group of actual traffic data in the congested flow area. Perform model setting processing. The setting of each traffic flow model is realized by the method described in the flowchart of FIG.
In the case of FIG. 10, the candidate models are the same in the free flow area and the jam flow area, but traffic flow models (model equations X1 and X2) having different variation parameters are set.

As shown in FIG. 10, when traffic flow models (model equations X1, X2) are set from two candidate models, the model equations X1, X2 indicating these traffic flow models are discontinuous. However, even in this case, as shown in FIG. 11, the construction unit 10 further obtains a model formula X3 in which the model formulas X1 and X2 indicating the traffic flow model are continuous, and this model formula X3 is obtained as a traffic flow model. It has a function to add and set to the model.
That is, in FIG. 10, the right end of the traffic flow model (model equation X1) in the free flow region and the left end of the traffic flow model (model equation X2) in the traffic flow region are separated, but as shown in FIG. , A model formula X3 for connecting them is obtained.

For example, the model formula X3 calculates the maximum values F1 and F2 of the model formulas X1 and X2 of the traffic flow model set by approximating each of two regions (free flow region and congestion flow region). A straight line connecting F2 is obtained as a model formula X3.
As described above, the construction unit 10 can set a traffic flow model from a plurality (two) of candidate models for one road link L11, and further shows a traffic flow model set from the plurality of candidate models. When the model formulas (X1, X2) are discontinuous, the model formula (X3) that makes these model formulas (X1, X2) continuous is further added and set.
As a result, the accuracy of the traffic flow model in the road link L11 can be set to be even higher, and the traffic flow model can be made continuous as a whole.

[Embodiment (Part 4)]
As shown in FIG. 2, a route R such as an urban road includes a plurality of road links (L11 to L15). Generally, in adjacent road links (for example, L11 and L12), Road characteristics (road shapes) may be the same or similar.
Therefore, in this case, when the traffic flow model is set according to the flow shown in FIG. 8 with respect to the road link L12, the adjacent road link L11 has the same or similar road characteristics as the road link L12. Regarding the road link L11, the construction unit 10 executes processing different from the flow shown in FIG. 8 and sets the same traffic flow model as that of the road link L12.

As described above, the construction unit 10 has a function of setting one (same) traffic flow model for road links having the same or similar road characteristics (for example, L11 and L12). In this case, it is not necessary to perform the approximation processing shown in FIG. 8 for all the road links L11 to L15, and the processing time can be shortened and the processing load can be reduced.
In order to facilitate this function, road links having the same or similar road characteristics are grouped, and the same group identification information (group ID) is assigned to these grouped road links. ing. For example, in the case of FIG. 2, a common group ID (111) is assigned to the road link L11 and the road link L12. Thereby, the setting process of the traffic flow model by the construction unit 10 becomes easy.

Embodiment (No. 5)
It is also conceivable that the characteristics of the road link included in the route R change due to a short-term change in the surrounding environment such as the weather on the route R. For example, when it is raining or snowing, the vehicle speed is relatively lowered and the vehicle interval is longer than when it is sunny. This affects the critical density and allowable traffic.
Therefore, in the analysis apparatus 1 according to the present embodiment, the traffic flow model can be properly used for each environmental characteristic around the road link. For example, the traffic flow model is switched between a sunny day, a rainy day, and a snowy day.

  For this purpose, it is necessary to set traffic flow models for sunny days, rainy days, and snowy days. Therefore, the construction unit 10 sets a traffic flow model according to the flow of FIG. 8 based on the actual traffic data acquired for each condition having different environmental characteristics. And the simulation part 20 switches and analyzes the traffic flow model used according to the weather of the day.

[Traffic data analysis device 1]
According to the traffic data analyzing apparatus 1 according to each of the embodiments described above, the route R includes a plurality of road links, but the construction unit 10 collects a group of actual traffic data acquired for each of the road links. Is approximated to each of a plurality of candidate models A to E, and a candidate model with high accuracy of the approximation is set as a traffic flow model of the road link.
Thus, a candidate model suitable as a model representing a group of actual traffic data acquired for a certain road link (L11) among a plurality of candidate models A to E is set as a traffic flow model of the road link (11). Therefore, a traffic flow model suitable for each of a plurality of road links included in the route R can be set. That is, the traffic flow model can be optimized for each road link.

  In addition, the route R includes a plurality of road links, and these road links may have different road characteristics. In this case, in order to perform an accurate simulation, it is conventionally considered to set the traffic flow model by individually investigating the characteristics of each road link by manpower, for example. Work, and the efficiency is poor. However, according to the analysis device 1 according to the present embodiment, this labor can be greatly reduced.

  When a traffic flow model is set for each road link by the construction unit 10, the simulation unit 20 can predict a traffic situation for each road link based on the traffic flow model. In addition, since an appropriate traffic flow model suitable for each road link is set by the construction unit 10, an analysis of the traffic situation of the entire route R including these road links is performed using this traffic flow model. If 20 is performed, an accurate analysis result can be obtained. That is, it is possible to predict traffic jams.

  On a route R (see FIG. 2) such as a road in an urban area, a road link included in the route R due to an expansion work of the route R itself or a change in the environment of other roads connected to the route R. It is also conceivable that the characteristics of (L11 to L15) change over time. In this case, the traffic flow model once set for the road link included in the route R may become unsuitable with the passage of time.

Therefore, the traffic data analysis apparatus 1 according to the present embodiment executes the processing shown in the flow of FIG. 8 again when time elapses, for example, 3 months, 6 months, etc., updates the traffic flow model, and has real-time characteristics. .
In other words, even if a traffic flow model is set for each road link on the route R (see FIG. 2) at a certain timing (first time) based on the flow of FIG. Traffic data is acquired.

Therefore, when a group of actual traffic data is newly acquired for each road link included in the route R, the construction unit 10 converts the newly acquired group of actual traffic data into the flow shown in FIG. Based on this, each of the candidate models A to E is approximated. Then, when the accuracy of the approximation becomes higher than the current accuracy, that is, the determination coefficient when a newly acquired group of actual traffic data is approximated to a predetermined candidate model is the current traffic flow model. If it is higher than the coefficient of determination calculated at the time of setting, it has a function of updating the candidate model as a traffic flow model of the road link.
Thereby, even if the characteristics of the road link of the route R change due to secular change, it is possible to cope with the change.

  In this updating process, in step S2 of FIG. 8, the priority order is set for the candidate model. In other words, the highest candidate model is a candidate model corresponding to the traffic flow model used so far. For example, in the road link L11, when the traffic flow model is set from the candidate model D, the candidate model D is set to the higher priority in the update process. Then, the construction unit 10 approximates the “group of actual traffic data” acquired regarding the road link L11 in order from the candidate model D having the higher priority (step S3 in FIG. 8). Then, the coefficient of determination is calculated (step S4), and other processing is as shown in FIG.

The embodiments disclosed herein are illustrative and non-restrictive in every respect. The scope of rights of the present invention is not limited to the above-described embodiments, but includes all modifications within the scope equivalent to the configurations described in the claims.
For example, in the above-described embodiment, the actual traffic data is the traffic volume Q, the traffic density K, and the vehicle traveling speed V. However, the actual traffic data may be other than these, and the traffic volume Q, the traffic density K, and the vehicle traveling speed V may be used. It is also possible to use data according to the above.

  Moreover, although the said embodiment demonstrated as the case where the area between the sensors 6 installed in each position of the route R was set as one road link, the link of a DRM link unit or a VICS link unit is demonstrated. A link may be a single road link, and a section in which a plurality of such road links (DRM links, VICS links) are collected may be a single road link.

  Further, in step S2 of FIG. 8, there is no restriction on the acquired actual traffic data, but specific actual traffic data may be excluded from the actual traffic data acquired for setting the traffic flow model. In other words, the actual traffic data may include temporary event traffic data such as accidents and road accidents. Therefore, a traffic flow model that expresses the characteristics at normal time for each road link more clearly by detecting such a sudden event by the analysis device 1 and excluding the actual traffic data acquired at that time. It becomes possible to set.

  1: Traffic data analysis device 3: Storage device (storage unit) 10: Traffic flow model construction unit 20: Simulation unit (prediction unit) R: Route L11 to L15: Road link (section)

Claims (9)

A traffic flow model construction unit that sets a traffic flow model for the section based on traffic data acquired for the section included in the route to be analyzed;
A storage unit storing basic expressions of a plurality of types of candidate models that are candidates for a traffic flow model,
The traffic flow model construction unit approximates a group of traffic data acquired for the section to each of the candidate models using basic expressions of the plurality of types of candidate models , and the section based on the accuracy of the approximation Traffic flow model can be set ,
Further, the traffic flow model construction unit can set a traffic flow model from a plurality of basic equations of the candidate models for one section, and the traffic flow model set from a plurality of basic equations of the candidate models Add and set a model formula that continues the model formulas shown.
Traffic data analysis device. Based on the traffic data acquired for the section included in the route to be analyzed , at least two relationships among traffic volume, traffic density, and vehicle travel speed, which are variables expressing the traffic flow status of the section and traffic flow model construction unit for setting a traffic flow model which is constituted by,
A storage unit storing basic expressions of a plurality of types of candidate models that are candidates for a traffic flow model,
The traffic flow model construction unit approximates a group of traffic data acquired for the section to each of the candidate models using basic expressions of the plurality of types of candidate models , and the section based on the accuracy of the approximation Traffic flow model can be set ,
Further, the traffic flow model construction unit includes basic expressions of a plurality of types of candidate models including a basic expression of a candidate model corresponding to the traffic flow model that has been used for the section and a basic expression of other candidate models. The traffic flow model for the section can be updated using
Traffic data analysis device. Priorities are set for the plurality of candidate models stored in the storage unit,
The traffic flow model construction unit approximates a group of traffic data acquired for the section in order from a candidate model having a higher priority, and if the accuracy of the approximation is a certain level or more, the candidate model is The traffic data analysis device according to claim 1 or 2, which is set as a traffic flow model of the section. The candidate model includes a variation parameter indicating the characteristics of the section,
The said traffic flow model construction part changes the said fluctuation parameter within the range which the said fluctuation parameter can take, and repeats the process which approximates the group of the traffic data acquired regarding the said section to the said candidate model . 4. The traffic data analysis device according to any one of 3 above. When a new group of traffic data is acquired for the section,
The traffic flow model construction unit approximates the group of newly acquired traffic data to each of the plurality of candidate models, and when the accuracy of the approximation becomes higher, the candidate model is converted to the traffic flow model of the section. The traffic data analyzing apparatus according to any one of claims 1 to 4 , which is updated as follows. The traffic data analysis device according to any one of claims 1 to 5 , wherein the traffic flow model construction unit sets the same traffic flow model for sections having the same or similar road characteristics. The traffic according to any one of claims 1 to 6 , further comprising a prediction unit that predicts a traffic situation of the section based on the traffic flow model of the section set by the traffic flow model construction unit. Data analysis device. A computer program for causing a computer to execute a process of setting a traffic flow model of a section based on traffic data acquired for a section included in a route to be analyzed,
Computer
Means for approximating a group of traffic data acquired for the section to each of the candidate models using basic equations of a plurality of types of candidate models that are candidates for a traffic flow model;
Means for setting the traffic flow model of the section based on the accuracy of the approximation;
Is a computer program to function as
Further, the means for setting the traffic flow model can set a traffic flow model from a plurality of basic equations of the candidate models for one section, and the traffic flow model set from a plurality of basic equations of the candidate models. Add and set model formulas that make model formulas that represent models continuous.
Computer program. Based on the traffic data acquired for the section included in the route to be analyzed , at least two relationships among traffic volume, traffic density, and vehicle travel speed, which are variables expressing the traffic flow status of the section a process of setting a traffic flow model that is configured by a computer program to be executed by a computer,
Computer
Means for approximating a group of traffic data acquired for the section to each of the candidate models using basic equations of a plurality of types of candidate models that are candidates for a traffic flow model;
Means for setting the traffic flow model of the section based on the accuracy of the approximation;
Is a computer program to function as
Further, the means for setting the traffic flow model includes a plurality of types of candidate models including a basic expression of a candidate model corresponding to the traffic flow model that has been used for the section and a basic expression of other candidate models. The traffic flow model for the section can be updated using the basic formula.
Computer program.

Images