JP2018112824A - Information processing device, information processing system, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an information processing device for improving the prediction accuracy of a required time for a target section of a road. SOLUTION: An acquisition unit for acquiring information indicating the occurrence of an event that restricts the passage of a vehicle in a target section of a road, a traffic density corresponding to a traffic volume in the target section, and a time required for passing through the target section. When the determination unit 150, which determines the parameters included in the model that defines the relationship with the predicted required time corresponding to the predicted value in association with the event by learning, and when the event occurs in the target section, A prediction unit 140 for calculating the prediction required time using the model to which the parameter corresponding to the event is applied is provided. [Selection diagram] Fig. 4

Description

  The present invention relates to an information processing apparatus, an information processing system, and a program.

  Conventionally, a required time, which is a time taken for passing through a target section of a road such as an expressway, has been predicted. The predicted value of the required time obtained by such prediction is used as useful information regarding traffic by notifying the driver, for example. In recent years, various techniques for predicting the required time have been proposed.

  For example, Patent Document 1 discloses a technique for predicting a required time by a method such as using a rate of change of a required travel time using an actual value of the required time up to the present time.

  Further, Patent Document 2 predicts future traffic volume transition on the current day by pattern matching the traffic volume transition performance in a certain period of time in the past with the traffic volume transition performance of the day. A technique for predicting the required time by doing so is disclosed.

  Further, Non-Patent Document 1 discloses a required time using a BPR function developed by the United States Road Authority as a model for defining a relationship between a traffic density that is a traffic volume in a target section and a predicted required time that is a predicted value of the required time. A technique for predicting the above is disclosed.

JP 2004-118700 A JP 2005-135282 A

Yoshioka Shinya et al., “Examination of Travel Time Data Collection and Processing Methods on General Roads”, Civil Engineering Planning Research and Lectures Vol. 40 CD-ROM, published in 2009

  By the way, in the conventional prediction of required time, it may be difficult to ensure sufficient prediction accuracy. For example, in the technologies disclosed in Patent Document 1 and Patent Document 2, the required time is predicted according to the transition of the actual value related to traffic such as travel time or traffic volume up to the present time. The prediction result of the real-time prediction that repeats the prediction with the progress is likely to depend on the past transition of the actual measurement values related to traffic. Here, in a traffic jam caused by an event that restricts the traffic of vehicles such as lane restrictions, the required time changes sharply compared to a natural traffic jam that occurs naturally regardless of the event occurrence. Cheap. Therefore, in a case where there is a traffic jam due to the occurrence of an event, the responsiveness to a change in the actual required time can be reduced in real time prediction. Further, in the technique disclosed in Patent Document 3, the required time can be predicted without using the past transition of the actual measurement value related to traffic. However, when traffic congestion occurs, sufficient prediction accuracy can be ensured. Can be difficult.

  Therefore, the present invention has been made in view of the above problems, and an object of the present invention is new and improved, which is capable of improving the prediction accuracy of the required time for the target section of the road. To provide an information processing apparatus, an information processing system, and a program.

  In order to solve the above problem, according to an aspect of the present invention, an acquisition unit that acquires information indicating an occurrence of an event that restricts vehicle traffic in a target section of a road, and corresponds to a traffic volume in the target section A determination unit that determines, by learning, a parameter included in a model that defines a relationship between a traffic density to be predicted and a predicted required time corresponding to a predicted value of the time required to pass through the target section, and the target There is provided an information processing apparatus comprising: a prediction unit that calculates the estimated required time using the model to which the parameter corresponding to the event is applied when the event occurs in a section.

  The model includes a free flow model that defines a relationship between the traffic density and the estimated required time when the traffic density is equal to or less than a threshold, and the traffic density and the estimated required time when the traffic density exceeds the threshold. A traffic flow model that defines a relationship between the traffic density and the prediction unit, when the traffic density is equal to or less than the threshold, the predicted time is calculated using the free flow model as the model, and the traffic density is When the threshold value is exceeded, the estimated required time may be calculated using the traffic flow model as the model.

  In the free flow model, the predicted travel time is specified to increase exponentially with the increase in traffic density, and in the congestion flow model, the predicted travel time is increased in proportion to the traffic density. It may be specified.

  The determining unit determines, for each of the plurality of target sections, the parameter in association with the event by the learning, and the predicting unit, when the event occurs in one target section, When the learning has not been completed for the one target section, the estimated required time for the one target section using the model to which the parameter corresponding to the event for the other target section is applied May be calculated.

  The determination unit determines the parameter in association with a plurality of the events by the learning, and the prediction unit determines the plurality of the events when the plurality of events occur in the target section. The estimated required time may be calculated using the model to which the corresponding parameter is applied.

  The event may include a change in the number of traffic lanes in the target section.

  The event may include a change in speed limit in the target section.

  The prediction unit may include a traffic density calculation unit that calculates the traffic density based on an actual required time corresponding to an actual value of the time taken to pass through the target section.

  The determination unit may determine the parameter in association with the event by the learning based on the actual required time and the traffic density in a period in which the event has occurred in the target section.

  In order to solve the above-described problem, according to another aspect of the present invention, an information processing device and a plurality of vehicle detections that respectively detect the passage of a vehicle at each of an upstream end and a downstream end of a target section of a road And an event detection device that detects the occurrence of an event that restricts the passage of the vehicle in the target section. The information processing device detects information indicating the occurrence of the event as the event detection device. A parameter included in a model that defines a relationship between an acquisition unit acquired from the apparatus, a traffic density corresponding to the traffic volume in the target section and a predicted required time corresponding to a predicted value of the time required to pass through the target section; A determination unit configured to determine the event corresponding to the event through learning based on detection results from the plurality of vehicle detection devices; and A prediction unit that calculates the estimated required time using the model to which the parameter corresponding to the event is applied and the detection results of the plurality of vehicle detection devices when the event occurs. A system is provided.

  In order to solve the above problem, according to another aspect of the present invention, an acquisition unit that acquires information indicating an occurrence of an event that restricts vehicle traffic in a target section of a road, and the target A parameter included in a model that defines a relationship between a traffic density corresponding to traffic volume in a section and a predicted required time corresponding to a predicted value of the time required to pass through the target section is determined by learning in association with the event. A program for functioning as a determination unit and a prediction unit that calculates the estimated required time using the model to which the parameter corresponding to the event is applied when the event occurs in the target section Is provided.

  As described above, according to the present invention, it is possible to improve the prediction accuracy of the required time for the target section of the road.

It is a block diagram which shows an example of schematic structure of the required time prediction system which concerns on embodiment of this invention. It is a schematic diagram which shows an example of arrangement | positioning of the vehicle detection apparatus which concerns on the same embodiment. It is a flowchart which shows an example of the flow of the vehicle detection process which the vehicle detection apparatus which concerns on the embodiment performs. It is a block diagram which shows an example of a function structure of the required time estimation apparatus which concerns on the same embodiment. It is explanatory drawing which shows an example of the relationship between the traffic density and estimated required time in a free flow model. It is explanatory drawing which shows an example of the relationship between the traffic density and estimated required time in a traffic jam model. It is explanatory drawing which shows an example of the data format of the information memorize | stored in the parameter memory | storage part which concerns on the embodiment. It is explanatory drawing for demonstrating an example of calculation of an integration parameter. It is a flowchart which shows an example of the flow of the production | generation of the vehicle information and the memory | storage process which the vehicle information generation part of the required time prediction apparatus which concerns on the same embodiment performs. It is a flowchart which shows an example of the flow of a memory | storage process of the event information which the communication part of the required time estimation apparatus which concerns on the same embodiment performs. It is a flowchart which shows an example of the flow of a required time prediction process which the prediction part of the required time prediction apparatus which concerns on the same embodiment performs. It is a flowchart which shows an example of the flow of the calculation process of the performance required time which the performance required time calculation part of the estimation part which concerns on the same embodiment performs. It is a flowchart which shows an example of the flow of a traffic density calculation process which the traffic density calculation part of the prediction part which concerns on the same embodiment performs. It is a flowchart which shows an example of the flow of a calculation process of the prediction required time which the prediction required time calculation part of the prediction part which concerns on the same embodiment performs. It is a flowchart which shows an example of the flow of the parameter determination process which the determination part of the required time prediction apparatus concerning the embodiment performs. It is a flowchart which shows an example of the flow of the calculation process of the free travel time per kilometer which the determination part of the required time prediction apparatus which concerns on the same embodiment performs. It is a flowchart which shows an example of the flow of a calculation process of the free flow maximum traffic density which the determination part of the required time prediction apparatus which concerns on the same embodiment performs. It is a flowchart which shows an example of the flow of the calculation process of the other parameter which the determination part of the required time prediction apparatus which concerns on the same embodiment performs.

  Exemplary embodiments of the present invention will be described below in detail with reference to the accompanying drawings. In addition, in this specification and drawing, about the component which has the substantially same function structure, duplication description is abbreviate | omitted by attaching | subjecting the same code | symbol.

<1. Overview of time required prediction system>
First, with reference to FIGS. 1-3, schematic structure of the required time prediction system 1 which concerns on embodiment of this invention is demonstrated. FIG. 1 is a block diagram illustrating an example of a schematic configuration of a required time prediction system 1 according to the present embodiment. FIG. 2 is a schematic diagram illustrating an example of the arrangement of the vehicle detection devices 20a, 20b, 20c, and 20d according to the present embodiment.

  The required time prediction system 1 corresponds to an information processing system according to the present invention. The required time prediction system 1 includes, for example, a required time prediction device 10, vehicle detection devices 20a, 20b, 20c, and 20d, and an event detection device 30, as shown in FIG. The required time prediction device 10 can communicate with each of the vehicle detection devices 20a, 20b, 20c, 20d and the event detection device 30 via a wired or wireless communication network N1. Note that each of the vehicle detection devices 20a, 20b, 20c, and 20d is also simply referred to as a vehicle detection device 20 unless otherwise distinguished. Moreover, the number of the vehicle detection apparatuses 20 should just be two or more, and is not specifically limited to the number shown in FIG.

  The plurality of vehicle detection devices 20 respectively detect the passage of vehicles at points on the upstream end and the downstream end of the target section of the road. In this specification, the term “target section” refers to a section that is a target of prediction of required time. For example, one vehicle detection device 20 is installed at each of the start point that is the upstream end point of the target section and the end point that is the downstream end point of the target section.

  Here, the road for which the required time is predicted may be divided into a plurality of target sections. Specifically, as shown in FIG. 2, a plurality of continuous sections L001, L002, and L003 can be applied as the target sections. For example, the sections L001, L002, and L003 can be arranged along the vehicle flow in this order. L001, L002, and L003 can be assigned to the sections L001, L002, and L003, respectively, as section IDs that are identification information of the target section.

  Specifically, as shown in FIG. 2, the vehicle detection devices 20a, 20b, 20c, and 20d have a point P001 that is the start point of the section L001, a point P002 that is the end point of the section L001 and the start point of the section L002, and a section L002. And at the point P003 that is the start point of the section L003 and at the point P004 that is the end point of the section L003. P001, P002, P003, and P004 can be assigned to the points P001, P002, P003, and P004, respectively, as point IDs that are identification information of the points. The target sections adjacent to each other may be provided with an interval without being continuous. In this case, the end point of the upstream section does not coincide with the start point of the downstream section.

  The vehicle detection device 20 detects the passage of the vehicle at the installed location. For example, the vehicle detection device 20 detects a vehicle ID that is identification information of a vehicle that has passed through an installed point. The vehicle detection device 20 may detect the vehicle ID by communicating with the vehicle, or may detect the vehicle ID by performing image processing on an image obtained by imaging the license plate of the vehicle. In addition, the vehicle detection device 20 can store a device ID that is identification information of the vehicle detection device 20. The vehicle detection devices 20a, 20b, 20c, and 20d may be assigned T001, T002, T003, and T004 as device IDs, respectively.

  Further, the vehicle detection device 20 transmits a detection result of the passage of the vehicle to the required time prediction device 10 when the vehicle passes through the installed point. For example, the vehicle detection device 20 transmits, to the required time prediction device 10, information indicating the detection date and time when the passage of the vehicle is detected, the device ID to be stored, and the vehicle ID of the detected vehicle as a detection result.

  Specifically, when Y0123 is detected as the vehicle ID at 12:34:56 on May 5, 2016, the vehicle detection device 20a detects “detection date = 2016/05/05 12:34:56” as the detection result. Information such as “device ID = T001; vehicle ID = Y0123” may be transmitted to the required time prediction device 10.

  Here, with reference to FIG. 3, the flow of the vehicle detection process which the vehicle detection apparatus 20 which concerns on this embodiment performs is demonstrated. FIG. 3 is a flowchart illustrating an example of a flow of a vehicle detection process performed by the vehicle detection device 20 according to the present embodiment. As shown in FIG. 3, the vehicle detection device 20 determines whether or not the vehicle has passed through the installed point (step S <b> 201). When it is not determined that the vehicle has passed the installed point (step S201 / NO), the determination process of step S201 is repeated. On the other hand, when it is determined that the vehicle has passed through the installed point (step S201 / YES), the vehicle detection device 20 acquires the vehicle ID of the detected vehicle (step S203). And the vehicle detection apparatus 20 transmits the detection result which shows a detection date, apparatus ID, and vehicle ID to the required time prediction apparatus 10 (step S205), and the process shown in FIG. 3 is complete | finished.

  The event detection device 30 detects the occurrence of an event that restricts vehicle traffic in the target section. In addition, the event detection device 30 transmits event information, which is information related to the generated event, to the required time prediction device 10. In this specification, the term event means an event that restricts the passage of a vehicle in a target section. For example, the event includes a change in the number of lanes that can pass in the target section. The event may include a change in speed limit in the target section.

  For example, the event detection device 30 transmits information indicating the event type that is the type of the event that has occurred, the section in which the event has occurred, and the start date and time of occurrence of the event to the required time prediction device 10 as event information.

  Specifically, when the occurrence of one lane restriction as an event in the section L001 starts at 12:12:00 on May 5, 2016, the event detection device 30 displays “event type = 1 lane restriction” as event information. ; Section ID = L001; start date and time = 2016/05/05 12:12:00 ”can be transmitted to the required time prediction device 10.

  In addition, when the occurrence of the event ends, the event detection device 30 transmits information to which the end date / time of the occurrence of the event is added as event information to the required time prediction device 10.

  Specifically, when the occurrence of the one-lane restriction is completed in the section L001 at 13:33:00 on May 5, 2016, the event detection device 30 displays “event type = 1 lane restriction; section ID” as the event information. = L001; start date and time = 2016/05/05 12:12:00; end date and time = 2016/05/05 13:33:00 ”may be transmitted to the required time prediction apparatus 10.

  The required time prediction device 10 corresponds to an information processing device according to the present invention. The required time prediction device 10 temporarily stores a CPU (Central Processing Unit) that is an arithmetic processing device, a ROM (Read Only Memory) that stores programs used by the CPU, calculation parameters, and the like, parameters that change as appropriate during execution of the CPU, and the like. RAM (Random Access Memory) and the like.

  The required time predicting device 10 predicts a required time, which is a time required for passing through the target section. Specifically, the required time prediction device 10 calculates the predicted required time using a model that defines the relationship between the traffic density corresponding to the traffic volume in the target section and the predicted required time corresponding to the predicted value of the required time. . For example, the required time prediction device 10 calculates the predicted required time using information received from the vehicle detection device 20 and the event detection device 30. Further, the required time prediction device 10 transmits the calculated predicted required time to a public facility used by the driver or the driver's vehicle. Thereby, the estimated required time is notified to the driver, and is used as useful information regarding traffic.

  The required time predicting apparatus 10 according to the present embodiment uses a model in which parameters included in the model are determined in association with an event by learning, and parameters corresponding to the event are applied when the event occurs in the target section. To calculate the estimated required time. Thereby, the prediction accuracy of the required time about the target section of a road can be improved. The required time prediction apparatus 10 will be described in detail in the next section.

<2. Time required prediction device>
Next, the required time predicting apparatus 10 according to the present embodiment will be described with reference to FIGS.

[2-1. Functional configuration]
First, with reference to FIGS. 4-8, the function structure of the required time prediction apparatus 10 which concerns on this embodiment is demonstrated. FIG. 4 is a block diagram illustrating an example of a functional configuration of the required time prediction apparatus 10 according to the present embodiment.

  The required time prediction device 10 includes, for example, a communication unit 110, a storage unit 120, a vehicle information generation unit 130, a prediction unit 140, and a determination unit 150, as illustrated in FIG.

(Communication Department)
The communication unit 110 communicates with a device external to the required time prediction device 10. For example, the communication unit 110 receives a vehicle passage detection result transmitted from the vehicle detection device 20 and outputs the result to the vehicle information generation unit 130. In addition, the communication unit 110 receives event information transmitted from the event detection device 30 and stores the event information in the storage unit 120. Thus, the communication unit 110 corresponds to an acquisition unit according to the present invention that acquires information indicating the occurrence of an event. Further, the communication unit 110 may transmit information indicating the estimated required time calculated by the prediction unit 140 to a public facility used by the driver or the driver's vehicle.

(Memory part)
The storage unit 120 stores data referred to for various processes performed by the required time prediction device 10. For example, as illustrated in FIG. 4, the storage unit 120 includes a section information storage unit 121, a vehicle information storage unit 122, an event information storage unit 123, an actual required time storage unit 124, and a traffic density storage unit 125. Parameter storage unit 126.

  The section information storage unit 121 stores section information that is information related to the target section of the road. The section information is used for vehicle information generation and storage processing performed by the vehicle information generation unit 130.

  For example, the section information storage unit 121 includes a point ID, a point name that is the name of the point, a kilometer post indicating the distance from the reference point of the road, and the installed vehicle detection device 20 for the start point or end point of each target section. Is stored as section information.

  Specifically, the section information storage unit 121 may store information such as “Spot ID = P001; Point name = A, Downward; Kilometer post = 0.00; Device ID = T001” as the section information for the point P001. In addition, the section information storage unit 121 may store information such as “Spot ID = P002; Point name = B, Downward; Kilometer post = 10.00; Device ID = T002” as the section information for the point P002.

  In addition, the section information storage unit 121 includes a section ID, a section name that is a section name, a start point ID that is a start point ID, an end point ID that is an end point ID, and information indicating the number of lanes for each target section. Store as information.

  Specifically, for the section L001, the section information storage unit 121 includes “section ID = L001; section name = A to B (down); start point ID = P001, end point ID = P002; number of lanes = 2” as section information. Such information can be stored.

  The vehicle information storage unit 122 stores vehicle information that is information indicating the passage date and time for each point of the vehicle. The vehicle information is used for a required time prediction process performed by the prediction unit 140 and a parameter determination process performed by the determination unit 150. The vehicle information is generated by the vehicle information generation unit 130. Details of the vehicle information will be described later in the description of the vehicle information generation unit 130.

  The event information storage unit 123 stores event information. The event information is used for a required time prediction process performed by the prediction unit 140 and a parameter determination process performed by the determination unit 150. The event information is acquired by the communication unit 110. For example, the communication unit 110 receives event information from the event detection device 30 at a set time (for example, 1 minute) interval and stores the event information in the event information storage unit 123.

  The actual required time storage unit 124 stores information related to the actual required time corresponding to the actual value of the time taken to pass through the target section. Information on the actual required time is used for the required time prediction process performed by the prediction unit 140 and the parameter determination process performed by the determination unit 150. Information related to the actual required time is generated by the actual required time calculation unit 141 of the prediction unit 140. Details of the information related to the actual required time will be described later in the description of the actual required time calculation unit 141.

  The traffic density storage unit 125 stores information indicating the traffic density corresponding to the traffic volume in the target section. Information indicating the traffic density is used for the required time prediction process performed by the prediction unit 140 and the parameter determination process performed by the determination unit 150. Information indicating the traffic density is generated by the traffic density calculation unit 142 of the prediction unit 140. Details of the information indicating the traffic density will be described later in the description of the traffic density calculation unit 142.

  The parameter storage unit 126 stores parameters included in a model that defines the relationship between the traffic density and the estimated required time. The model is used for a required time prediction process performed by the prediction unit 140. The parameters included in the model are determined by parameter determination processing performed by the determination unit 150.

  For example, the model includes a free flow model and a traffic jam model, and the parameter storage unit 126 stores parameters for each model.

  The free flow model defines the relationship between the traffic density and the estimated required time when the traffic density is equal to or lower than the free flow maximum traffic density as a threshold. The free flow maximum traffic density is an index for determining whether the relationship between the traffic density and the estimated required time is close to the relationship defined by the free flow model or the congestion flow model described later. The free flow model is, for example, a BPR function developed by the United States Road Authority, and is represented by the following equation (1).

  In equation (1), T1, Q, L, T0, C, α, and β are referred to as the estimated travel time, traffic density, distance of the target section, free travel time per kilometer, free flow maximum traffic density, and slope, respectively. And a parameter called curvature. The free travel time per km corresponds to the estimated required time when the traffic density is zero. FIG. 5 shows an example of the relationship between the traffic density and the estimated required time in the free flow model represented by Expression (1). In the free flow model, for example, the estimated required time is specified to increase exponentially with an increase in traffic density. The parameter storage unit 126 stores a free travel time T0 per kilometer, a free flow maximum traffic density C, a slope α, and a curvature β as parameters included in the free flow model.

  The traffic jam model defines the relationship between the traffic density and the estimated required time when the traffic density exceeds the free flow maximum traffic density as a threshold. The traffic jam model is, for example, a function that linearly approximates the relationship between traffic density and estimated required time, and is represented by the following equation (2).

  In the equation (2), T1, Q, L, a, and b respectively indicate a predicted time, a traffic density, a distance of a target section, a parameter called a slope, and a parameter called an intercept. FIG. 6 shows an example of the relationship between the traffic density and the estimated required time in the traffic jam model represented by Expression (2). In the traffic jam model, for example, the estimated required time is specified to increase in proportion to the traffic density. The parameter storage unit 126 stores the slope a and the intercept b as parameters included in the traffic jam model.

  In the required time prediction apparatus 10 according to the present embodiment, the determination unit 150 determines a parameter in association with an event through learning, and the parameter storage unit 126 stores the parameter in association with an event. The parameter storage unit 126 may store parameters in association with the target section or the number of lanes. The parameter storage unit 126 may store parameters in association with a plurality of events.

  FIG. 7 is an explanatory diagram illustrating an example of a data format of information stored in the parameter storage unit 126. The parameter storage unit 126 stores information in the form of a decision tree, for example, as shown in FIG. In FIG. 7, nodes that are branch conditions are indicated by rounded rectangles, and leaves are indicated by rectangles. In FIG. 7, a part of the branch is omitted.

  The arrangement of the nodes is configured such that, for example, a node having a larger influence on the required time is positioned on the upstream side of the branch. Specifically, the decision tree shown in FIG. 7 first branches for traffic regulation as an event, then branches for the number of lanes, then branches for events other than traffic regulation, and branches for the target section. The traffic regulation includes a change in the number of lanes allowed to pass in the target section such as a traffic stop or one lane restriction, and a change in the number of lanes allowed to pass in the target section such as a 50 km limit. Further, events other than traffic regulation have a smaller influence on the required time than traffic regulation, and include events related to the environment such as crosswind, fog, snow, or rain. Specifically, each leaf is the free travel time T0 per kilometer of the free flow model, the maximum free flow traffic density C, the slope α and the curvature β, and the slope a and the intercept b of the jam flow model.

  Note that the data format of the information stored in the parameter storage unit 126 is not particularly limited to the format of the decision tree, and may be another format. For example, as a data format, a table format in which an event, a target section, a lane, and a parameter are associated with each row may be applied.

(Vehicle information generator)
When the communication unit 110 receives the detection result from the vehicle detection device 20, the vehicle information generation unit 130 acquires the interval information from the interval information storage unit 121, and based on the detection result and the interval information from the vehicle detection device 20 Information is generated and stored in the vehicle information storage unit 122.

  For example, the vehicle information generation unit 130 includes information indicating the detection date and time when the passage of the vehicle is detected, the point ID of the point where the vehicle detection device 20 that detects the passage of the vehicle is installed, and the vehicle ID of the detected vehicle. Generated as vehicle information.

  Specifically, the vehicle information generation unit 130 receives information such as “detection date = 2016/05/05 12:34:56; device ID = T001; vehicle ID = Y0123” as a detection result by the vehicle detection device 20. In this case, information such as “detection date / time = 2016/05/05 12:34:56; spot ID = P001; vehicle ID = Y0123” may be generated and stored in the vehicle information storage unit 122. Further, the vehicle information generation unit 130 receives information such as “detection date = 2016/05/05 13:20:00; device ID = T002; vehicle ID = Y0123” as a detection result by the vehicle detection device 20. Information such as “detection date / time = 2016/05/05 13:20:00; point ID = P002; vehicle ID = Y0123” may be generated as vehicle information and stored in the vehicle information storage unit 122.

(Prediction unit)
The prediction unit 140 executes a required time prediction process. For example, the prediction unit 140 executes a required time prediction process for each of a plurality of target sections at a set time (for example, 1 minute) interval. In the following, in order to facilitate understanding, the required time prediction process for the section L001 as the target section will be mainly described. As shown in FIG. 4, for example, the prediction unit 140 includes an actual required time calculation unit 141, a traffic density calculation unit 142, and a predicted required time calculation unit 143.

  The actual required time calculation unit 141 executes a process for calculating the actual required time. The actual required time is used for traffic density calculation processing by the traffic density calculation unit 142.

  For example, the actual required time calculation unit 141 includes the point P002 that is the end point of the section L001 between the current time (for example, 13:20 on May 5, 2016) and the time before the set time (for example, 5 minutes). The actual required time is calculated for each vehicle that has passed through. Specifically, the actual required time calculation unit 141 obtains vehicle information about the point P002 and the point P001 that is the start point of the section L001 for each vehicle that has passed the point P002 between the current time and the time before the set time. The actual time required is calculated based on the extracted vehicle information extracted from the vehicle information storage unit 122.

  More specifically, when the vehicle assigned Y0100 as the vehicle ID passes the point P002 between the current time and the time before the set time, the actual required time calculation unit 141 uses “detection date = 2016/05/05 13:19:59; point ID = P002; vehicle ID = Y0100 ”and“ detection date = 2016/05/05 12:33:33; point ID = P001; vehicle ID = Y0100 ”. Can be extracted. Further, when the vehicle assigned Y0123 as the vehicle ID passes the point P002 between the current time and the time before the set time, the actual required time calculation unit 141 displays “detection date = 2016/05 / 05 13:20:00; point ID = P002; vehicle ID = Y0123 ”and“ detection date = 2016/05/05 12:34:56; point ID = P001; vehicle ID = Y0123 ”.

  The actual required time calculation unit 141 calculates the actual required time for each vehicle, for example, the section ID, the vehicle ID, the start point detection date and time that is the detection date and time for the start point, the end point detection date and time that is the detection date and time for the end point, and Information indicating the actual required time is stored in the actual required time storage unit 124.

  Specifically, the result required time calculation unit 141 includes, as such information, “section ID = L001; vehicle ID = Y0100; start point detection date / time = 2016/05/05 12:33:33; end point detection date / time = 2016 / 05/05 13:19:59; actual required time = 46 minutes 26 seconds ”and“ section ID = L001; vehicle ID = Y0123; start point detection date = 2016/05/05 12:34:56; end point detection date = 2016 / 05/05 13:20:00; actual required time = 45 minutes 04 seconds ”may be stored in the actual required time storage unit 124.

  Further, the actual required time calculation unit 141 calculates a representative value of the actual required time. The representative value of the actual required time is a relatively high value in the actual required time.

  For example, the actual required time calculation unit 141 creates a histogram by dividing the calculated actual required time into a plurality of classes, and calculates the median value of the class having the highest frequency as a representative value of the actual required time. The actual required time calculation unit 141 may determine the number of classes using the Sturges formula, and in this case, the actual required time is obtained by dividing the difference between the minimum value and the maximum value of the actual required time by the number of classes. The value may be determined as the time width of each class. Further, the actual required time calculation unit 141 creates a histogram without using an abnormal value such as an excessively large value in the calculated actual required time, or a part of the upper or lower part when arranged in ascending order. May be.

  Specifically, when the class having the highest frequency is a class having a time width from 46.0 minutes to 47.4 minutes, the actual required time calculation unit 141 calculates 46.7 minutes as a representative value of the actual required time. Can do.

  The actual required time calculation unit 141 also includes, for example, the section ID, the representative start point detection date and time that is the earliest time among the times when the target vehicle for the actual required time passes the point P001, and the target vehicle for the actual required time calculation. Information indicating the representative end point detection date and time and the representative value of the actual required time, which is the latest time among the times when the vehicle has passed the point P002, is stored in the actual required time storage unit 124.

  Specifically, the result required time calculation unit 141 includes “section ID = L001; representative start point detection date / time = 2016/05/05 12:33:18; representative end point detection date / time = 2016/05/05” as such information. Information such as “13:20:00; representative value of actual required time = 46.7 minutes” may be stored in the actual required time storage unit 124.

  The traffic density calculation unit 142 executes a traffic density calculation process. The traffic density is used for the calculation process of the predicted required time by the predicted required time calculation unit 143. The traffic density calculation unit 142 calculates the traffic density based on the actual required time.

  For example, the traffic density calculation unit 142 may determine the interval L001 between the current time (for example, 13:20 on May 5, 2016) and the time before the representative value (for example, 46.7 minutes) of the actual required time. A value obtained by dividing the number of vehicles that have passed through the point P001, which is the starting point, by the distance of the section L001 is calculated as the traffic density. Specifically, the traffic density calculation unit 142 extracts vehicle information about each vehicle that has passed the point P001 between the current time and the time before the representative value of the actual required time from the vehicle information storage unit 122, Traffic density can be calculated based on the extracted vehicle information. The traffic density calculation unit 142 can calculate 10.00 [km] as the distance of the section L001 based on the information indicating the kiloposts at the points P001 and P002 stored in the section information storage unit 121.

  More specifically, the traffic density calculation unit 142 has a date and time from May 5, 2016, 12:33:18 to May 5, 2016, 13:20:00 as the start point detection date and time. When 987 pieces of information are extracted, 98.7 [vehicles / km] can be calculated as the traffic density.

  In addition, the traffic density calculation unit 142 causes the traffic density storage unit 125 to store, for example, the section ID, the representative start point detection date and time that is the earliest time among the start point detection dates and times of the extracted vehicle information, and the traffic density information.

  Specifically, the traffic density calculation unit 142 uses information such as “section ID = L001; representative start point detection date / time = 2016/05/05 12:33:20; traffic density = 98.7” as such information. It can be stored in the density storage unit 125.

  The predicted required time calculation unit 143 executes a process for calculating the predicted required time. Further, the predicted required time calculation unit 143 outputs information indicating the calculated predicted required time to the communication unit 110.

  The predicted required time calculation unit 143 calculates the predicted required time using a model that defines the relationship between the traffic density and the predicted required time.

  For example, when the traffic density is equal to or lower than the free flow maximum traffic density, the predicted required time calculation unit 143 calculates the predicted required time using a free flow model represented by Expression (1) as a model. Further, when the traffic density exceeds the free flow maximum traffic density, the predicted required time calculation unit 143 calculates the predicted required time using a traffic jam flow model represented by Expression (2) as a model.

  When an event occurs in the target section, the predicted required time calculation unit 143 calculates the predicted required time using a model to which a parameter corresponding to the event is applied. The estimated required time calculation unit 143 can acquire information indicating an event occurring in the target section from the event information storage unit 123. Further, the estimated required time calculation unit 143 can acquire a parameter corresponding to the event from the parameter storage unit 126.

  The estimated required time calculation unit 143 uses, for example, the decision tree shown in FIG. 7 to branch the condition so that it matches the type of event currently occurring, the number of lanes in the target section, and the section ID of the target section. Traces and gets the parameter that is the reached leaf. Thereby, the estimated required time calculation unit 143 can acquire a parameter corresponding to the event that is currently occurring.

  Specifically, when the required travel time calculation unit 143 obtains event information indicating that one lane restriction is currently occurring as an event in the section L001 as the target section, a parameter corresponding to the one lane restriction is set as a parameter. It can be acquired from the storage unit 126. In this case, the estimated required time calculation unit 143 acquires, from the parameter storage unit 126, parameters that are leaves that meet the conditions that the event type, the number of lanes, and the section ID are 1 lane restriction, 2 and L001, for example. Specifically, the estimated required time calculation unit 143 uses “free travel time per kilometer T0 = 0.66 minutes; free flow maximum traffic density C = 34.5; slope α as parameters of the free flow model and the congestion flow model, respectively. = 0.5; curvature β = 3.0 ”and“ slope a = 0.050; intercept b = −0.73 ”. In that case, since the traffic density (= 98.7) exceeds the free flow maximum traffic density C (= 34.5), the predicted required time calculation unit 143 calculates the predicted required time using a traffic jam flow model. . Specifically, the predicted required time calculation unit 143 calculates 42 minutes 03 seconds as the predicted required time by substituting the acquired parameters of the traffic jam flow model, the distance of the section L001, and the traffic density into Equation (2). .

  As described above, the estimated required time calculation unit 143 specifically uses the model to which the parameter corresponding to the event is applied and the detection results by the plurality of vehicle detection devices 20 when the event occurs in the target section. To calculate the estimated required time. More specifically, the predicted required time calculation unit 143 calculates the predicted required time using an integration parameter described later as a parameter corresponding to the event when an event occurs in the target section.

  Moreover, when a plurality of events occur in the target section, the predicted required time calculation unit 143 may calculate the predicted required time using a model to which parameters corresponding to the plurality of events are applied. As shown in FIG. 7, the parameter storage unit 126 can store parameters in association with a plurality of events. Therefore, the estimated required time calculation unit 143 can acquire parameters corresponding to the plurality of events when the plurality of events occur.

  In addition, when an event occurs in one target section and learning has not been completed for the one target section, the predicted required time calculation unit 143 sets a parameter corresponding to the event for the other target section. The predicted required time for the one target section may be calculated using the applied model. If the learning is not completed for the target section and the parameters are not determined, the leaf cannot be reached from the lowest layer node. In such a case, a parameter to be applied to the calculation of the predicted required time may be acquired based on a parameter that is a leaf corresponding to another target section that branches from the lowest layer node. For example, an average value of a parameter that is a leaf corresponding to another target section that branches from the lowermost node may be acquired as a parameter that is applied to the calculation of the estimated required time. Therefore, when the required duration calculation unit 143 has not completed learning for the target section, the predicted required time calculation unit 143 can acquire parameters corresponding to events for other target sections. The predicted required time calculation unit 143 determines a parameter that is a leaf that can be reached when the branch of the condition is traced so that the most conditions are met when the leaf cannot be reached from the lowermost node. You may acquire as a parameter applied to calculation of.

(Decision part)
The determination unit 150 executes parameter determination processing. Specifically, the determination unit 150 executes parameter determination processing for each of a plurality of target sections at a set time (for example, one day or one week) interval. In the following, in order to facilitate understanding, parameter determination processing for the section L001 as the target section will be mainly described.

  The determination unit 150 determines a parameter included in the model that defines the relationship between the traffic density and the estimated required time in association with the event by learning. Specifically, the determination unit 150 determines a parameter corresponding to the event by learning for each of the plurality of target sections. The determination unit 150 may determine the parameter in association with the target section or the number of lanes by learning. Further, the determination unit 150 may determine a parameter in association with a plurality of events by learning. The determined parameters are stored in the parameter storage unit 126 in the data format shown in FIG. 7, for example.

  The determination unit 150 determines a parameter by executing, for example, an event information acquisition process, a free travel time calculation process per kilometer, a free flow maximum traffic density calculation process, and another parameter calculation process. Hereinafter, an example of such parameter determination processing will be described.

  In the event information acquisition process, the determination unit 150 acquires event information about the target period (for example, the previous day or the previous week) that is the latest period corresponding to the set time from the event information storage unit 123. For example, the determination unit 150 acquires, as event information, information indicating an event type of an event that has occurred within the target period in the target section, a section in which the event has occurred, a start date / time of event occurrence, and an end date / time of event occurrence.

  Specifically, when the target period is May 4, 2016 as the previous day, the determining unit 150 sets “event type = 1 lane restriction; section ID = L001; start date / time = 2016/04/04” as event information. 12:12:00; end date and time = 2016/05/04 13:33:00 ”.

  In the calculation process of the free travel time per kilometer, the determination unit 150 extracts information indicating the actual required time in a period in which no event occurs in the target period from the actual required time storage unit 124 and extracts the extracted information. Based on the above, the free travel time per kilometer is calculated.

  For example, the determination unit 150 extracts information indicating the actual required time associated with the end point detection date and time included in a period in which no event occurs in the target period in the target section. In addition, the determination unit 150 creates a histogram by dividing the actual time required for the extracted information into a plurality of classes, and divides the median value of the class with the highest frequency by the distance of the section L001 in kilograms. Calculated as free travel time per hit. The determination unit 150 may determine the number of classes using the Sturges formula, in which case each value obtained by dividing the difference between the minimum value and the maximum value of the actual time required by the class number is determined. It may be determined as a class time width. Further, the determination unit 150 may create a histogram without using an abnormal value such as an excessively large value in the extracted result required time, or a part of the upper or lower part when arranged in ascending order. .

  Specifically, classes ranging from 6.1 minutes to 6.3 minutes, 6.3 minutes to 6.5 minutes, 6.5 minutes to 6.7 minutes, and 6.7 minutes to 6.9 minutes Are 433, 973, 1265, and 1091, respectively, the class having the highest frequency is a class having a time width from 6.5 minutes to 6.7 minutes. In that case, the determination unit 150 calculates 0.66 minutes (= 40 seconds) per kilometer obtained by dividing 6.6 minutes as the median value of the class by 10.00 [km] as the distance of the section L001. It can be calculated as free travel time.

  Further, the determination unit 150 may store information indicating the calculated free travel time per km in the parameter storage unit 126 in association with the target section. For example, the determination unit 150 routes the information indicating the calculation date and time and the free travel time per kilometer, which is the date and time when the free travel time per kilometer is calculated, to the section L001 at the branch of the target section of the decision tree in the parameter storage unit 126 You may memorize it as a leaf.

  Specifically, the determination unit 150 stores information such as “calculation date = 2016/5/5 2: 00: 00: 00; free travel time per kilometer = 0.66 minutes” in the parameter storage unit 126 as such information. Can be.

  In the calculation process of the free flow maximum traffic density, the determination unit 150 calculates a traffic jam determination required time as an index in determining whether or not a traffic jam has occurred, and represents a representative value of the actual required time that is equal to or less than the traffic jam determination required time. Is extracted from the actual required time storage unit 124, and the free flow maximum traffic density is calculated based on the extracted information. Since the term “congestion” can have various definitions, the time required for determining the traffic jam can be appropriately calculated according to the definition. For example, when the required time is less than the required time for determining the traffic jam, it can be determined that no traffic jam has occurred.

  For example, the determination unit 150 calculates a value obtained by multiplying the value obtained by increasing the free travel time per kilometer by the congestion determination delay rate and the distance of the target section as the congestion determination required time. The congestion determination delay rate is appropriately set according to the definition of the term “congestion”, and can be set to 50%, for example.

  Specifically, the determination unit 150 obtains the value obtained by increasing the free travel time per kilometer of the section L001 by 0.66 by 50% and 10.00 [km] as the distance of the section L001. The calculated 9.9 minutes (= 9 minutes 54 seconds) can be calculated as the time required for determining the traffic jam.

  In addition, the determination unit 150 extracts, from the actual required time storage unit 124, information indicating a representative value of the actual required time that is equal to or less than the traffic congestion determination required time during the target event within the target period in the target section. The target event is a parameter calculation target event. The target event may be a combination of a plurality of events. In addition, the determination unit 150 may execute parameter calculation processing for a plurality of target events. In that case, the determination unit 150 executes a free flow maximum traffic density calculation process and other parameter calculation processes for each target event.

  For example, the determination unit 150 includes information indicating a representative value of the actual required time that is associated with the representative start point detection date and time included in the target period in the target period and that is equal to or less than the congestion determination required time. Extract.

  Specifically, when the target period is May 4, 2016 as the previous day, the determination unit 150 includes “section ID = L001; representative start point detection date / time = 2016/05/04 0:00” as such information. : 24; Representative value of actual required time = 6.6 minutes ”,“ Section ID = L001; Representative start point detection date / time = 2016/05/04 0:01:54; Representative value of actual required time = 7.1 minutes ” And “section ID = L001; representative start point detection date and time = 2016/05/04 23:59:24; representative value of actual required time = 6.6 minutes” can be extracted for the target event that occurred in the section L001. Note that the extracted information may include information indicating the representative end point detection date and time.

  Further, the determination unit 150 extracts information indicating the traffic density corresponding to the extracted information indicating the representative value of the actual required time from the traffic density storage unit 125. For example, the determination unit 150 may extract the representative value of the actual required time extracted from the information indicating the traffic density associated with the representative start point detection date and time included in the period in which the target event occurs in the target period in the target section. Information indicating the traffic density calculated using the information indicating the traffic density is extracted from the traffic density storage unit 125.

  Specifically, the determination unit 150 includes, as such information, “section ID = L001; representative start point detection date and time = 2016/05/04 0:00; traffic density = 9.2”, “section ID = L001. ; Representative start point detection date = 2016/05/04 0:01:00; traffic density = 9.0 ”,“ section ID = L001; representative start point detection date = 2016/05/04 23:59:00; traffic density = 9.4 ”and“ section ID = L001; representative start point detection date / time = 2016/05/04 9:30:30; traffic density = 47.0 ”can be extracted for the target event that occurred in the section L001.

  Further, the determination unit 150 calculates the free flow maximum traffic density based on the extracted information indicating the traffic density. For example, the determination unit 150 calculates a value obtained by multiplying the maximum value of the traffic density indicated by the extracted information by a coefficient (for example, 0.9) as the free flow maximum traffic density.

  Specifically, the determination unit 150 calculates 42.3 obtained by multiplying 47.0 as the maximum value of the traffic density by 0.9 as the free flow maximum traffic density corresponding to the target event for the section L001. obtain.

  Further, the determination unit 150 may store information indicating the calculated free flow maximum traffic density in the parameter storage unit 126 in association with the target section and the event. For example, the determination unit 150 branches information indicating the calculation date and time and the free flow maximum traffic density, which is the date and time when the free flow maximum traffic density is calculated, to the target event by branching the decision tree event in the parameter storage unit 126, You may memorize | store as a leaf of the path | route which branches to the area L001 at the branch of an area.

  Specifically, the determination unit 150 causes the parameter storage unit 126 to store information such as “calculation date = 2016/5/5 2: 00: 00: 00; free flow maximum traffic density = 42.3” as such information. obtain.

  In the other parameter calculation processing, the determination unit 150 calculates the inclination α and the curvature β of the free flow model and the inclination a and the intercept b of the traffic flow model corresponding to the target event for the target section.

  For example, for the period in which the target event occurs within the target period, the determination unit 150 sets a traffic density that is equal to or lower than the free flow maximum traffic density and a representative value of the actual travel time corresponding to the traffic density, respectively Then, the inclination α and the curvature β of the free flow model are calculated based on the extracted information from the result required time storage unit 124.

  Specifically, the determination unit 150 traffics information indicating a traffic density that is associated with a representative start point detection date and time included in a period in which the target event is occurring in the target period in the target section and is equal to or lower than the free flow maximum traffic density. Extracted from the density storage unit 125. In addition, the determination unit 150 includes information indicating a representative value of the actual required time that is equal to or less than the traffic congestion determination required time associated with the representative start point detection date and time included in the target event in the target period in the target period. Among them, information having the closest representative start point detection date and time is extracted from the actual required time storage unit 124 for each piece of information indicating the extracted traffic density. In addition, the determination unit 150 determines that the relationship between the traffic density defined by the free flow model represented by the equation (1) and the estimated required time corresponds to the traffic density in the extracted information and the representative value of the actual required time. The inclination α and the curvature β are calculated so as to approach the relationship.

  Specifically, the determination unit 150 sets “traffic density = 3.0; representative value of actual travel time = 6.6 minutes”, “as a pair of traffic density and the actual travel time corresponding to the traffic density”, “ Traffic density = 38.4; Typical value of actual travel time = 9.0 minutes "," Transport density = 39.0; Typical value of actual travel time = 9.5 minutes "and" Transport density = 39.2; Actual Information such as “representative value of required time = 9.6 minutes” can be extracted, and 0.50 and 3.0 can be calculated as the slope α and the curvature β, respectively.

  The determination unit 150 may store information indicating the calculated inclination α and curvature β in the parameter storage unit 126 in association with the target section and the event. For example, the determination unit 150 branches information indicating the calculation date and time, the slope α, and the curvature β, which are the dates and times when each parameter was calculated, to the target event by branching the decision tree event in the parameter storage unit 126, and branching the target section May be stored as a leaf of a route branching to the section L001.

  Specifically, the determination unit 150 uses information such as “calculation date / time = 2016/5/5 2: 00: 00: 00; slope α = 0.50; curvature β = 3.0” as the parameter storage unit. 126 can be stored.

  When the number of pairs of the extracted traffic density and the representative value of the actual travel time corresponding to the traffic density is less than a reference number (for example, 10), the determination unit 150 determines the prediction accuracy based on the free flow model. In order to suppress the decrease, it is not necessary to calculate the inclination α and the curvature β.

  In addition, for the period in which the target event occurs within the target period, the determination unit 150 sets the traffic density exceeding the maximum free-flow traffic density and the representative value of the actual required time corresponding to the traffic density, respectively. And the slope a and intercept b of the traffic jam model are calculated based on the extracted information.

  Specifically, the determination unit 150 traffics information indicating a traffic density that is associated with a representative start point detection date and time included in a period in which the target event occurs in the target period and exceeds the free flow maximum traffic density. Extracted from the density storage unit 125. In addition, the determination unit 150 includes information indicating a representative value of the actual required time exceeding the traffic congestion determination required time associated with the representative start point detection date and time included in the target event in the target period in the target period. For each piece of information indicating the extracted traffic density, information having the closest representative start point detection date and time is extracted from the actual required time storage unit 124. In addition, the determination unit 150 determines that the relationship between the traffic density defined by the traffic jam model represented by the equation (2) and the estimated required time corresponds to the traffic density in the extracted information and the representative value of the actual required time. The inclination a and the intercept b are calculated so as to approach the relationship.

  Specifically, the determining unit 150 sets “traffic density = 47.7; representative value of actual travel time = 12.3 minutes” and “representative value of actual travel time corresponding to the traffic density and the traffic density” and “ Information such as “traffic density = 45.6; representative value of actual travel time = 11.1 minutes” can be extracted, and 0.050 and −0.73 can be calculated as the slope a and the intercept b, respectively.

  Further, the determination unit 150 may store information indicating the calculated inclination a and intercept b in the parameter storage unit 126 in association with the target section and the event. For example, the determination unit 150 branches information indicating the calculation date and time, the slope a and the intercept b, which is the date and time when each parameter is calculated, to the target event by branching the decision tree event in the parameter storage unit 126, and branching the target section May be stored as a leaf of a route branching to the section L001.

  Specifically, the determination unit 150 stores, as such information, information such as “calculation date = 2016/5/5 2: 00: 00: 00; slope a = 0.050; intercept b = −0.73”. It can be stored in the unit 126.

  When the number of pairs of the extracted traffic density and the representative value of the actual travel time corresponding to the traffic density is less than a reference number (for example, 10), the determining unit 150 determines the prediction accuracy based on the traffic jam model. In order to suppress the decrease, it is not necessary to calculate the slope a and the intercept b.

  The determination unit 150 determines a parameter corresponding to the target event for the target section, based on the parameter calculated by the parameter calculation process described above.

  The determination unit 150 can store parameters for each calculation date in the parameter storage unit 126 by repeating the above-described processing at set time intervals. The determination unit 150 determines, for example, an integrated parameter obtained by averaging parameters for the most recent predetermined number of calculation dates and times as a parameter used for a required time prediction process.

  Specifically, as illustrated in FIG. 8, the determination unit 150 averages the parameters regarding the calculation date and time stored in each leaf from May 5, 2016 to April 29, 2016. The obtained integrated parameter may be determined as a parameter used for a required time prediction process. More specifically, the determination unit 150 determines the free travel time T0, the free flow maximum traffic density C, the slope α, the curvature β, and the calculation date and time from May 5, 2016 to April 29, 2016. By averaging the slope a and the intercept b, respectively, the integrated parameters including the average free travel time T0 per kilometer, the free flow maximum traffic density C, the slope α, the curvature β, the slope a and the intercept b can be determined. .

  As described above, the determination unit 150 determines the parameter in association with the event by learning based on the actual required time and the traffic density in the period in which the event occurred in the target section.

[2-2. Operation]
Next, with reference to FIGS. 9 to 18, the flow of processing performed by the required time prediction device 10 according to the present embodiment will be described.

  FIG. 9 is a flowchart illustrating an example of a flow of vehicle information generation and storage processing performed by the vehicle information generation unit 130 of the required time prediction apparatus 10 according to the present embodiment.

  As illustrated in FIG. 9, the vehicle information generation unit 130 first determines whether the communication unit 110 has received a detection result from the vehicle detection device 20 (step S <b> 301). When it is not determined that the communication unit 110 has received the detection result from the vehicle detection device 20 (step S301 / NO), the determination process of step S301 is repeated. On the other hand, when it determines with the communication part 110 having received the detection result by the vehicle detection apparatus 20 (step S301 / YES), the vehicle information generation part 130 acquires area information from the area information storage part 121 (step S303). . Next, the vehicle information generation part 130 produces | generates the vehicle information which shows a detection date, point ID, and vehicle ID (step S305). Next, the vehicle information generation unit 130 stores the generated vehicle information in the vehicle information storage unit 122 (step S307), and the process illustrated in FIG. 9 ends.

  FIG. 10 is a flowchart illustrating an example of a flow of storage processing of event information performed by the communication unit 110 of the required time predicting apparatus 10 according to the present embodiment.

  As shown in FIG. 10, the communication unit 110 first determines whether or not a set time (for example, 1 minute) has elapsed (step S401). If it is not determined that the set time has elapsed (step S401 / NO), the determination process of step S401 is repeated. On the other hand, when it is determined that the set time has passed (step S401 / YES), the communication unit 110 acquires event information indicating the event type, the section in which the event has occurred, and the start date and time of occurrence of the event from the event detection device 30. (Step S403). Note that the event information may include information indicating the end date and time of occurrence of the event. Next, the communication unit 110 stores the acquired event information in the event information storage unit 123 (step S405), and the process illustrated in FIG.

  FIG. 11 is a flowchart illustrating an example of a flow of a required time prediction process performed by the prediction unit 140 of the required time prediction apparatus 10 according to the present embodiment.

  As illustrated in FIG. 11, the prediction unit 140 first determines whether a set time (for example, 1 minute) has elapsed (step S501). If it is not determined that the set time has elapsed (step S501 / NO), the determination process of step S501 is repeated. On the other hand, when it is determined that the set time has elapsed (step S501 / YES), the actual required time calculation unit 141 of the prediction unit 140 executes the actual required time calculation process (step S510). Next, the traffic density calculation unit 142 of the prediction unit 140 executes a traffic density calculation process (step S520). Next, the predicted required time calculation unit 143 of the prediction unit 140 executes a calculation process of the predicted required time (step S530). The processes of steps S510, S520, and S530 can be executed for one target section among a plurality of target sections, for example.

  Then, the prediction unit 140 determines whether or not the calculation of the predicted required time has been completed for all target sections (step S503). If it is not determined that the calculation of the predicted required time has been completed for all target sections (step S503 / NO), the process returns to step S510, and steps S510, S520, and the target sections for which the calculation of the predicted required time has not been completed. The process of S530 is performed. On the other hand, when it is determined that the calculation of the predicted required time has been completed for all target sections (step S503 / YES), the processing illustrated in FIG. 11 ends.

  FIG. 12 is a flowchart illustrating an example of a flow of calculation processing of the required time required performed by the actual required time calculation unit 141 of the prediction unit 140 according to the present embodiment. 12 is equivalent to the process in step S510 in FIG.

  As shown in FIG. 12, the actual required time calculation unit 141 first sets the target section of each vehicle that has passed through the end point of the target section between the current time and the time before the set time (for example, 5 minutes). Vehicle information about the end point and the start point is extracted from the vehicle information storage unit 122 (step S511). Next, the actual required time calculation unit 141 calculates the actual required time for each vehicle based on the extracted vehicle information (step S513). Next, the actual required time calculation unit 141 stores information indicating the calculated actual required time for each vehicle in the actual required time storage unit 124 (step S515).

  Then, the actual required time calculating unit 141 calculates a representative value of the actual required time based on the calculated actual required time (step S517). Next, the actual required time calculation unit 141 stores information indicating the representative value of the calculated actual required time in the actual required time storage unit 124 (step S519), and the process illustrated in FIG. 12 ends.

  FIG. 13 is a flowchart illustrating an example of a flow of traffic density calculation processing performed by the traffic density calculation unit 142 of the prediction unit 140 according to the present embodiment. The traffic density calculation process shown in FIG. 13 corresponds to the process of step S520 in FIG.

  As shown in FIG. 13, the traffic density calculation unit 142 first obtains vehicle information about each vehicle that has passed the starting point of the target section between the current time and the time before the representative value of the actual required time by the vehicle information. Extracted from the storage unit 122 (step S521). Next, the traffic density calculation unit 142 calculates the traffic density based on the extracted vehicle information (step S523). Next, the traffic density calculation unit 142 stores information indicating the traffic density in the traffic density storage unit 125 (step S525), and the process illustrated in FIG. 13 ends.

  FIG. 14 is a flowchart illustrating an example of a flow of calculation processing for a predicted required time performed by the predicted required time calculation unit 143 of the prediction unit 140 according to the present embodiment. Note that the process for calculating the estimated required time shown in FIG. 14 corresponds to the process in step S530 in FIG.

  As illustrated in FIG. 14, the predicted required time calculation unit 143 first acquires information indicating an event occurring in the target section from the event information storage unit 123 (step S531). Next, the estimated required time calculation unit 143 acquires a parameter corresponding to the event from the parameter storage unit 126 (step S533). Next, the estimated required time calculation unit 143 calculates the estimated required time using a model to which the parameter corresponding to the event is applied (step S535), and the process shown in FIG. 14 ends.

  As described above, the predicted required time calculation unit 143 can calculate the predicted required time using a model to which a parameter corresponding to the event is applied when an event occurs in the target section. When no event occurs in the target section, the predicted required time calculation unit 143 can acquire parameters that can be applied to the model when no event occurs in step S533.

  FIG. 15 is a flowchart illustrating an example of a flow of parameter determination processing performed by the determination unit 150 of the required time prediction apparatus 10 according to the present embodiment.

  As illustrated in FIG. 15, the determination unit 150 first determines whether a set time (for example, one day or one week) has elapsed (step S601). If it is not determined that the set time has elapsed (step S601 / NO), the determination process of step S601 is repeated. On the other hand, when it is determined that the set time has elapsed (step S601 / YES), the determination unit 150 sets the event information about the target period (for example, the previous day or the previous week) that is the latest period corresponding to the set time as the event information. Obtained from the storage unit 123 (step S603). Next, the determination unit 150 executes a calculation process of free travel time per kilometer (step S610). Next, the determination unit 150 executes a free flow maximum traffic density calculation process (step S620). Next, the determination unit 150 executes another parameter calculation process (step S630). Next, the determination unit 150 determines a parameter corresponding to the target event for the target section based on the calculated parameter (step S605). The processes of steps S603, S610, S620, S630, and S605 can be executed for one target section among a plurality of target sections, for example.

  Then, the determination unit 150 determines whether the parameter determination has been completed for all target sections (step S607). If it is not determined that the parameter determination has been completed for all target sections (step S607 / NO), the processing returns to step S603, and steps S603, S610, S620, S630 and the target sections for which parameter determination has not been completed. The process of S605 is performed. On the other hand, when it is determined that the parameter determination has been completed for all target sections (step S607 / YES), the processing illustrated in FIG. 15 ends.

  FIG. 16 is a flowchart illustrating an example of a flow of a free travel time calculation process per kilometer performed by the determination unit 150 of the required time prediction apparatus 10 according to the present embodiment. The calculation process of the free travel time per kilometer shown in FIG. 16 corresponds to the process of step S610 in FIG.

  As illustrated in FIG. 16, the determination unit 150 first extracts information indicating the actual required time in a period in which no event occurs in the target period from the actual required time storage unit 124 (step S611). . Next, the determination unit 150 calculates a free travel time per kilometer based on the extracted information (step S613). Next, the determination unit 150 stores information indicating the calculated free travel time per kilometer in the parameter storage unit 126 (step S615), and the process illustrated in FIG. 16 ends.

  FIG. 17 is a flowchart showing an example of the flow of the free flow maximum traffic density calculation process performed by the determination unit 150 of the required time prediction apparatus 10 according to the present embodiment. The calculation process of the free flow maximum traffic density shown in FIG. 17 corresponds to the process of step S620 in FIG.

  As shown in FIG. 17, the determination unit 150 first calculates a traffic jam determination required time as an index in determining whether or not a traffic jam has occurred (step S621). Next, the determination unit 150 extracts, from the actual required time storage unit 124, information indicating a representative value of the actual required time that is equal to or less than the traffic congestion determination required time in the target period in the target period in the target period. (Step S623). Next, the determination unit 150 extracts information indicating the traffic density corresponding to the extracted information indicating the representative value of the actual required time from the traffic density storage unit 125 (step S625).

  And the determination part 150 calculates a free flow maximum traffic density based on the information which shows the extracted traffic density (step S627). Next, the determination unit 150 stores information indicating the calculated free flow maximum traffic density in the parameter storage unit 126 (step S629), and the process illustrated in FIG. 17 ends.

  FIG. 18 is a flowchart illustrating an example of the flow of another parameter calculation process performed by the determination unit 150 of the required time prediction apparatus 10 according to the present embodiment. The other parameter calculation process shown in FIG. 18 corresponds to the process of step S630 in FIG.

  As illustrated in FIG. 18, the determination unit 150 first represents a traffic density equal to or lower than the free flow maximum traffic density and the actual required time corresponding to the traffic density for the period in which the target event occurs within the target period. Values are extracted from the traffic density storage unit 125 and the actual required time storage unit 124, respectively (step S631). Next, the determination unit 150 calculates the inclination α and the curvature β, which are parameters of the free flow model, based on the extracted information (step S632). Next, the determination unit 150 causes the parameter storage unit 126 to store the calculated slope α and curvature β, which are parameters of the free flow model (step S633).

  Then, the determination unit 150 determines the traffic density exceeding the free flow maximum traffic density and the representative value of the actual travel time corresponding to the traffic density for the period in which the target event occurs within the target period, respectively. And it extracts from the performance required time memory | storage part 124 (step S634). Next, the determination unit 150 calculates the slope a and the intercept b, which are parameters of the traffic jam flow model, based on the extracted information (step S635). Next, the determination unit 150 stores the slope a and the intercept b, which are the parameters of the calculated congestion flow model, in the parameter storage unit 126 (step S636), and the process illustrated in FIG. 18 ends.

<3. Effect>
As described above, according to the present embodiment, the determination unit 150 determines the parameter included in the model that defines the relationship between the traffic density and the estimated required time in association with the event by learning. In addition, when an event occurs in the target section, the predicted required time calculation unit 143 of the prediction unit 140 calculates the predicted required time using a model to which a parameter corresponding to the event is applied. As a result, real-time prediction that repeats prediction with the passage of time in a day can be appropriately performed according to the occurrence of an event without using past transitions of actual measurement values related to traffic. Therefore, even when there is a traffic jam due to the occurrence of an event whose required time is likely to change relatively steeply, it is possible to improve the responsiveness to changes in the actual required time for real-time prediction. Therefore, the prediction accuracy of the required time about the object area of a road can be improved.

  In addition, for example, when the traffic density is equal to or lower than the free flow maximum traffic density, the predicted required time calculation unit 143 of the prediction unit 140 calculates the predicted required time using a free flow model, and the traffic density is the free flow maximum traffic density. If it exceeds, the estimated required time is calculated using the traffic jam model. Thereby, a required time can be appropriately predicted according to the traffic density. Therefore, the prediction accuracy of the required time for the target section of the road can be improved more effectively.

  In the free flow model, for example, the predicted travel time is specified to increase exponentially with an increase in traffic density, and in the congestion flow model, for example, the predicted travel time is increased in proportion to the traffic density. It is prescribed. Accordingly, the relationship between the traffic density and the estimated required time can be appropriately expressed according to the traffic density by the free flow model and the jam flow model.

  In addition, when an event has occurred in one target section and the learning has not been completed for the one target section, the predicted required time calculation unit 143 of the prediction unit 140 determines an event for another target section as The predicted required time for the one target section may be calculated using a model to which the corresponding parameter is applied. As a result, even if an event that has not occurred in the past occurs in one target section, the required time can be accurately predicted by using parameters corresponding to the event for the other target section. be able to. Here, the required time prediction device 10 may control the display of the calculation result of the predicted required time on a display device such as a public facility or a vehicle. As described above, when the required time predicting apparatus 10 calculates an estimated required time using parameters corresponding to the event for another target section that has not been learned yet, the calculation result is a target for which learning has not been completed. In order to make it possible for a driver or the like to determine that the result is a result of prediction of the required time for a section, the display of the calculation result may be different from the display of the calculation result for the target section for which learning has been completed. For example, in such a case, the required time predicting device 10 performs display control such as making the display color of the calculation result different from the display color of the calculation result for the target section for which learning has been completed, or adding some annotations. You may go.

  Further, as a data format of information stored in the parameter storage unit 126, for example, a decision tree format may be applied. In that case, the arrangement of the nodes is configured such that, for example, a node having a greater influence on the required time is positioned upstream of the branch. Accordingly, when an event occurs in one target section and the learning for the one target section is not completed when the predicted required time calculation unit 143 of the prediction unit 140 is incomplete, Calculating a predicted required time for the one target section using a model that preferentially applies a parameter corresponding to an event (for example, the number of lanes) common to the one target section among parameters corresponding to the event it can. Therefore, even when an event that has not occurred in the past occurs in one target section, for example, by using a parameter corresponding to the event for another target section having the same lane number, Prediction can be performed with higher accuracy.

  In addition, when a plurality of events occur in the target section, the predicted required time calculation unit 143 of the prediction unit 140 calculates the predicted required time using a model to which parameters corresponding to the plurality of events are applied. Also good. Thereby, even when a plurality of events occur in the target section, the required time can be accurately predicted according to the combination of the plurality of events.

<4. Conclusion>
As described above, the information processing apparatus according to the present invention includes a determination unit that determines a parameter included in a model that defines a relationship between a traffic density and a predicted required time in association with an event by learning, and is generated in a target section. A predicting unit that calculates a predicted required time using a model to which a parameter corresponding to the event is applied when the event being performed occurs. As a result, real-time prediction that repeats prediction with the passage of time in a day can be appropriately performed according to the occurrence of an event without using past transitions of actual measurement values related to traffic. Therefore, even when there is a traffic jam due to the occurrence of an event whose required time is likely to change relatively steeply, it is possible to improve the responsiveness to changes in the actual required time for real-time prediction. Therefore, the prediction accuracy of the required time about the object area of a road can be improved.

  Note that a series of control processing by each device described in this specification may be realized using any of software, hardware, and a combination of software and hardware. For example, a program constituting the software is stored in advance in a storage medium (non-transitory media) provided inside or outside each device. Each program is read into a RAM at the time of execution, for example, and executed by a processor such as a CPU. The number of processors that execute each program may be one or more.

  Specifically, it is possible to create a computer program for realizing each function of the required time prediction apparatus 10 according to the present embodiment as described above, and to mount the computer program on a PC or the like. The required time prediction apparatus 10 according to the present embodiment may correspond to a computer. In addition, a computer-readable recording medium storing such a computer program can be provided. The recording medium is, for example, a magnetic disk, an optical disk, a magneto-optical disk, a flash memory, or the like. Further, the above computer program may be distributed via a network, for example, without using a recording medium. In addition, each function of the required time prediction apparatus 10 according to the present embodiment may be divided by a plurality of computers. In this case, each function of the plurality of computers may be realized by the above computer program.

  Further, the processing described using the flowchart in the present specification may not necessarily be executed in the order shown in the flowchart. Some processing steps may be performed in parallel. Further, additional processing steps may be employed, and some processing steps may be omitted.

  The preferred embodiments of the present invention have been described in detail above with reference to the accompanying drawings, but the present invention is not limited to such examples. It is obvious that a person having ordinary knowledge in the technical field to which the present invention pertains can make various modifications or application examples within the scope of the technical idea described in the claims. Of course, it is understood that these also belong to the technical scope of the present invention.

DESCRIPTION OF SYMBOLS 1 Required time prediction system 10 Required time prediction apparatus 20 Vehicle detection apparatus 30 Event detection apparatus 110 Communication part 120 Storage part 121 Section information storage part 122 Vehicle information storage part 123 Event information storage part 124 Actual required time storage part 125 Traffic density storage part 126 Parameter storage unit 130 Vehicle information generation unit 140 Prediction unit 141 Actual required time calculation unit 142 Traffic density calculation unit 143 Predicted required time calculation unit 150 Determination unit

Claims (11)

An acquisition unit that acquires information indicating an occurrence of an event that restricts vehicle traffic in a target section of a road;
A parameter included in a model that defines a relationship between a traffic density corresponding to the traffic volume in the target section and a predicted required time corresponding to a predicted value of the time required to pass through the target section is associated with the event by learning. A decision part to decide;
When the event occurring in the target section occurs, a prediction unit that calculates the predicted time using the model to which the parameter corresponding to the event is applied,
Comprising
Information processing device. The model includes a free flow model that defines a relationship between the traffic density and the estimated required time when the traffic density is equal to or less than a threshold, and the traffic density and the estimated required time when the traffic density exceeds the threshold. Including a traffic jam model that regulates the relationship between
When the traffic density is equal to or less than the threshold, the prediction unit calculates the predicted time using the free flow model as the model, and when the traffic density exceeds the threshold, the congestion flow as the model Calculating the estimated required time using a model;
The information processing apparatus according to claim 1. In the free flow model, the estimated travel time is defined to increase exponentially with increasing traffic density,
In the traffic flow model, the estimated required time is specified to increase in proportion to the traffic density.
The information processing apparatus according to claim 2. The determination unit determines the parameter in association with the event by the learning for each of the plurality of target sections,
When the event is occurring in one target section and the learning is not completed for the one target section, the prediction unit corresponds to the event for the other target section. Calculating the estimated required time for the one target section using the model to which a parameter is applied;
The information processing apparatus according to any one of claims 1 to 3. The determination unit determines the parameter in association with a plurality of the events by the learning,
The predicting unit calculates the predicted required time using the model to which the parameters corresponding to the plurality of events are applied when the plurality of events occur in the target section.
The information processing apparatus according to any one of claims 1 to 4.   The information processing apparatus according to claim 1, wherein the event includes a change in the number of lanes that can pass in the target section.   The information processing apparatus according to claim 1, wherein the event includes a change in speed limit in the target section.   The said prediction part is provided with the traffic density calculation part which calculates the said traffic density based on the performance required time corresponded to the actual value of the time which passed for the said object area, The any one of Claims 1-7. The information processing apparatus described in 1.   The said determination part determines the said parameter corresponding to the said event by the said learning based on the said performance required time and the said traffic density in the period when the said event generate | occur | produced in the said object area. Information processing device. An information processing device;
A plurality of vehicle detection devices that respectively detect the passage of vehicles at each point of the upstream end and the downstream end of the target section of the road;
An event detection device that detects the occurrence of an event that restricts the passage of the vehicle in the target section;
In an information processing system including
The information processing apparatus includes:
An acquisition unit for acquiring information indicating the occurrence of the event from the event detection device;
Detecting parameters included in a model that defines a relationship between a traffic density corresponding to the traffic volume in the target section and a predicted required time corresponding to a predicted value of the time required to pass through the target section by the plurality of vehicle detection devices A determination unit configured to determine the event corresponding to the event by learning based on the result;
When the event occurs in the target section, a prediction unit that calculates the estimated required time using the model to which the parameter corresponding to the event is applied and detection results by the plurality of vehicle detection devices;
Comprising
Information processing system. Computer
An acquisition unit that acquires information indicating an occurrence of an event that restricts vehicle traffic in a target section of a road;
A parameter included in a model that defines a relationship between a traffic density corresponding to the traffic volume in the target section and a predicted required time corresponding to a predicted value of the time required to pass through the target section is associated with the event by learning. A decision part to decide;
When the event occurs in the target section, a prediction unit that calculates the estimated required time using the model to which the parameter corresponding to the event is applied;
Program to function as.

Images