Traffic jam prediction device
The traffic congestion prediction device uses a machine learning model to predict long-term traffic conditions by integrating historical and real-time data, addressing the limitations of short-term prediction methods and enhancing user experience with detailed forecasts.
Patent Information
- Application Number
- JP2022002614
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2022-01-11
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2042-01-11
AI Technical Summary
Existing traffic congestion prediction methods can only predict dynamically changing traffic conditions up to several hours into the future and fail to provide long-term forecasts, such as several days ahead.
A traffic congestion prediction device that utilizes a machine learning model to predict long-term road conditions by acquiring the number of people information and vehicle information, determining traffic congestion situations, and using a machine learning method to estimate future traffic congestion based on historical data and real-time vehicle and weather information.
Enables accurate long-term prediction of traffic congestion, reducing user load and providing detailed forecasts through a user-friendly presentation interface.
Smart Images

Figure 0007586102000001 
Figure 0007586102000002 
Figure 0007586102000003
Abstract
Description
[Technical field]
[0001] The present invention predicts road congestion Astringent This invention relates to a congestion prediction device. [Background technology]
[0002] Patent Document 1 discloses a road traffic server that collects vehicle position information and individual information from vehicles and predicts current and future traffic volume for the vehicle's travel route based on the collected position information and individual information. The road traffic server predicts current and future traffic volume for the next few hours based on information obtained from multiple vehicles and terminals owned by users who are in the vehicles, and presents the predictions to users. [Prior art documents] [Patent documents]
[0003] [Patent Document 1] JP 2014-191578 A Summary of the Invention [Problem to be solved by the invention]
[0004] Traffic congestion is a major social problem, and there is a demand for an excellent traffic congestion prediction method. Conventional technology predicts dynamically changing traffic congestion conditions based on information acquired from multiple vehicles currently traveling on the road. However, this technology is only capable of predicting dynamically changing traffic congestion conditions up to the immediate future (e.g., several hours into the future), and it is not always possible to predict long-term traffic congestion conditions, such as several days into the future.
[0005] This invention can predict long-term road conditions. Astringent The object of the present invention is to provide a congestion prediction device. [Means for solving the problem]
[0006] The method according to claim 1 Traffic jam prediction devicea machine learning method in which a computer executes a process of acquiring number-of-users information indicating the number of users, including users in vehicles, who depart from a facility for each predetermined period, and vehicle information regarding vehicles that have traveled around the facility, determining a traffic congestion state indicating the presence or absence of traffic congestion on roads located around the facility during the predetermined period using the vehicle information, and executing machine learning for predicting road congestion on a machine learning model using the number-of-users information and the traffic congestion state. and a processor. The machine learning model predicts the traffic congestion situation for each predetermined period on roads located in the vicinity of the facility using estimated number of users information indicating the number of users estimated to depart from the facility for each predetermined period, including users riding in vehicles, generated by a process similar to that described above, wherein the processor estimates the estimated number of users information for a target period for which the traffic congestion situation is predicted, inputs the estimated number of users information into the machine learning model to predict the traffic congestion situation for the target period, and presents, on a single screen, on which the traffic congestion situation is predicted using a map, the prediction result of the traffic congestion situation for each predetermined period in the target period, and traffic congestion information for each predetermined period on the same day of the week in the past during the target period.
[0007] The method according to claim 1 Traffic jam prediction device acquires number-of-people information indicating the number of users departing from a facility, and vehicle information of vehicles traveling on roads located in the vicinity of the facility. Traffic jam prediction device determines congestion information for roads located around a facility from vehicle information, and trains a machine learning model to learn the relationship between the acquired number of people information and the determined congestion information. Here, the facility refers to a company, a commercial facility, etc., and the user departing from the facility refers to a user leaving work from the company, a user leaving the facility and returning home, etc. This makes it possible to predict road conditions over a long period of time. Furthermore, according to the traffic congestion prediction device of claim 1, the burden on the user can be reduced in the process of predicting a traffic congestion situation. The traffic congestion prediction device of claim 2 is the traffic congestion prediction device of claim 1, wherein the processor presents on the prediction presentation screen a character string indicating the time for each of the predetermined periods and a rectangular mark indicating the presence or absence of congestion for each time indicated in the character string with a difference in color, and on an extension line of each time indicated in the character string, a rectangular mark indicating the presence or absence of congestion in the target period as a prediction result of the traffic congestion situation, and a rectangular mark indicating the presence or absence of congestion on the same day of the week in the past as the congestion information are placed.
[0008] Other Aspects The machine learning method is before The number of users information includes the number of users relating to a period prior to the period for which the traffic congestion situation is to be determined.
[0009] Other Aspects According to the machine learning method, traffic congestion is largely determined by the time it takes to reach the road after leaving the facility, and it is possible to make predictions taking into account factors that affect traffic congestion.
[0010] Other Aspects The machine learning method is before The vehicle information includes position information related to a position of the vehicle and speed information related to a speed of the vehicle, and the congestion situation is determined using the position information and the speed information.
[0011] Other Aspects According to the machine learning method, the traffic congestion situation can be determined without being affected by bad weather or the time of day, such as nighttime.
[0012] Other Aspects The machine learning method is before Using the position information and the speed information, the time required to pass through a predetermined section is derived, and if there are a predetermined number of consecutive sections in which the time required is equal to or exceeds a threshold, the congestion situation is determined to be a congestion.
[0013] Other Aspects According to the machine learning method, the criteria for indicating congestion can be clarified, making it possible to judge the congestion situation more accurately.
[0014] Other Aspects The machine learning method is before The machine learning model is a model that uses random forest.
[0015] Other Aspects This machine learning method can predict traffic congestion conditions more accurately than other machine learning models.
[0020] Other Aspects The traffic congestion prediction device of the present invention is 1 In the traffic congestion prediction device described in the above, the estimated number of users information for the target period is estimated using the number of users who have visited the facility in the past.
[0021] Other Aspects According to the traffic congestion prediction device, the number of visitors who will visit a facility on a day related to a target period can be estimated from the number of visitors in the past, and information on the number of people departing from the facility during the target period can be estimated from the estimated number of visitors. In other words, the number of visitors in the future can be estimated to predict traffic congestion information.
[0022] Other Aspects The traffic congestion prediction device of the present invention is 1 In the traffic congestion prediction device described in the above, the estimated traffic congestion situation for each target period is presented.
[0023] Other Aspects According to the traffic congestion prediction device, it is possible to allow a user to recognize the predicted traffic congestion situation. Effect of the Invention
[0024] According to the present invention, road conditions can be predicted over a long period of time. [Brief description of the drawings]
[0025] [Figure 1] 1 is a diagram showing a schematic configuration of a traffic congestion prediction system according to an embodiment of the present invention; [Diagram 2] 1 is a block diagram showing a hardware configuration of a vehicle according to an embodiment of the present invention. [Diagram 3] FIG. 2 is a block diagram showing a functional configuration of the vehicle-mounted device according to the embodiment; [Figure 4] FIG. 2 is a block diagram showing a hardware configuration of a center server according to the present embodiment. [Diagram 5] FIG. 2 is a block diagram showing a functional configuration of a center server according to the present embodiment. [Figure 6] 1 is a diagram illustrating an average required time for a vehicle to pass through each section and each time period, which is provided for explaining estimation of a congestion state according to the present embodiment. FIG. [Figure 7] FIG. 4 is a diagram showing a configuration of learning data according to the embodiment; [Figure 8] FIG. 4 is a diagram showing a configuration of setting data according to the embodiment. [Figure 9] 11 is a graph showing an example of the ratio of the number of people who leave work for each date and each time period, which is used to explain the estimation of the number of people who leave work according to the embodiment; [Figure 10] FIG. 13 is a schematic diagram showing a screen for displaying a prediction result according to the present embodiment. [Figure 11] FIG. 2 is a data flow diagram showing the flow of data in a process for generating a machine learning model according to the present embodiment. [Figure 12] FIG. 2 is a data flow diagram showing the flow of data in a process for predicting a traffic congestion situation according to the present embodiment. [Figure 13]13 is a flowchart showing the flow of a machine learning model generation process executed in the center server of this embodiment. [Figure 14] 13 is a flowchart showing the flow of a traffic congestion situation prediction process executed in the center server of this embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0026] A traffic congestion prediction system including a vehicle equipped with the on-board device of the present invention, a center server as a traffic congestion prediction device, and an information providing server that presents weather information indicating the weather and attendance record information indicating attendance records will be described. The traffic congestion prediction system performs processing to predict the traffic congestion situation on roads located around a facility for each predetermined period (time period) using the number of employees who attend work at the facility, including employees who commute by vehicle, and the number of employees who leave the facility for each predetermined period (time period).
[0027] Here, the number of people who leave the facility according to the present embodiment is an example of "information on the number of people who depart from the facility". Furthermore, "coming to work" according to the present embodiment is an example of "visit". Note that the facility according to the present embodiment is described as a company building owned by a company. However, the present embodiment is not limited to this. The facility may be, for example, a commercial facility where multiple stores are gathered, or a leisure facility such as an amusement park. In this case, the traffic congestion prediction system according to the present embodiment predicts the traffic congestion situation of roads located around the commercial facility and the leisure facility for each predetermined period (time period) based on the number of visitors to the commercial facility and the leisure facility and the number of people leaving for each predetermined period (time period). Note that, hereinafter, the predetermined period for predicting the road situation is referred to as a "time period". Furthermore, the "roads located in the vicinity" according to the present embodiment are roads that are directly or indirectly connected to the facility and are located within a predetermined distance from the facility.
[0028] (Overall composition) 1, a traffic congestion prediction system 10 according to an embodiment of the present invention includes a vehicle 12, a center server 30, and an information providing server 40. The vehicle 12 is equipped with an on-board device 20 as an on-board device, and the on-board devices 20 are mutually connected to the center server 30 via a network N.
[0029] The vehicle-mounted device 20 is a device that collects vehicle information related to the vehicle, including position information indicating the position of the vehicle 12 and speed information indicating the speed at which the vehicle 12 is traveling, and transmits the information to the center server 30.
[0030] The center server 30 is installed, for example, at a manufacturer that manufactures the vehicle 12 or at a car dealership affiliated with the manufacturer. The center server 30 acquires vehicle information from the in-vehicle device 20, and uses the vehicle information of the vehicle 12 traveling on a road located in the vicinity of the facility to determine a congestion state indicating the presence or absence of congestion on the road.
[0031] The center server 30 also acquires weather information, including weather and precipitation, and attendance record information, including the number of employees who report to work at a facility such as a company and the number of employees who leave work for each time period, from the information providing server 40 described below. The center server 30 uses the acquired weather information and attendance information to predict the traffic congestion situation on roads located around the facility for each time period.
[0032] The information providing server 40 is a server that provides weather information and attendance record information to the center server 30. The weather information according to this embodiment includes past weather observation results and future weather forecasts such as days when traffic congestion is predicted. The attendance record information includes the number of people who came to work in the past and the number of people who left work in the past as attendance records.
[0033] (vehicle) As shown in FIG. 2, the vehicle 12 according to this embodiment includes an in-vehicle device 20 and a plurality of ECUs (Electronic Control Units) 22.
[0034] The vehicle-mounted device 20 includes a central processing unit (CPU) 20A, a read only memory (ROM) 20B, a random access memory (RAM) 20C, an in-vehicle communication interface (I / F) 20D, a wireless communication I / F 20E, and an input / output I / F 20F. The CPU 20A, the ROM 20B, the RAM 20C, the in-vehicle communication I / F 20D, the wireless communication I / F 20E, and the input / output I / F 20F are connected to each other so as to be able to communicate with each other via an internal bus 20G.
[0035] The CPU 20A is a central processing unit that executes various programs and controls each part. That is, the CPU 20A reads out the programs from the ROM 20B and executes the programs using the RAM 20C as a working area.
[0036] The ROM 20B stores various programs and various data. In the present embodiment, the ROM 20B stores a collection program 100. When the collection program 100 is executed, the vehicle-mounted device 20 collects data from the vehicle-mounted devices and sensors (not shown) and transmits the data to the center server 30. The RAM 20C temporarily stores programs or data as a working area.
[0037] The in-vehicle communication I / F 20D is an interface for connecting to each ECU 22. The interface uses a communication standard based on the CAN protocol. The in-vehicle communication I / F 20D is connected to the external bus 20H.
[0038] The wireless communication I / F 20E is a wireless communication module for communicating with the center server 30. The wireless communication module uses communication standards such as 5G, LTE, and Wi-Fi (registered trademark). The wireless communication I / F 20E is connected to a network N.
[0039] The input / output I / F 20F is an interface for communicating with the microphone 24, the speaker 25, the monitor 26, the camera 27, the GPS device 28, and the input unit 29 mounted on the vehicle 12. Note that the microphone 24, the speaker 25, the monitor 26, the camera 27, the GPS device 28, and the input unit 29 may be directly connected to the internal bus 20G.
[0040] The ECU 22 is an electronic control module for controlling on-board devices and sensors (not shown) mounted on the vehicle 12. The ECU 22 is, for example, an ADAS (Advanced Driver Assistance System)-ECU, a steering ECU, an engine ECU, etc. The ECU 22 according to this embodiment detects the speed of the vehicle 12 from a vehicle speed sensor (not shown).
[0041] The microphone 24 is a device that is provided on the instrument panel, the center console, the front pillar, the dashboard, etc., and collects voices emitted by the occupants of the vehicle 12.
[0042] The speaker 25 is a device that is provided on an instrument panel, a center console, a front pillar, a dashboard, or the like, and outputs audio.
[0043] The monitor 26 is a liquid crystal monitor that is provided on an instrument panel, dashboard, or the like of the vehicle 12 and displays various information. In addition, the monitor 26 in this embodiment displays images and text indicating the items to be replaced that are received from the center server.
[0044] The camera 27 is an imaging device that is provided above the windshield or adjacent to the rearview mirror, and captures images of passengers in the vehicle 12.
[0045] The GPS device 28 is a device that measures the current location of the vehicle 12. The GPS device 28 includes an antenna (not shown) that receives signals from GPS satellites. The GPS device 28 may be connected to the vehicle-mounted device 20 via a car navigation system (not shown).
[0046] The input unit 29 is a touch panel, buttons, or the like that accepts input from a passenger of the vehicle 12.
[0047] As shown in FIG. 3, in the vehicle-mounted device 20 of this embodiment, a CPU 20A executes a collection program 100 to function as a collection unit 200 and an output unit 210.
[0048] The collection unit 200 has a function of collecting speed information detected by a sensor from the ECU 22 and position information acquired by the GPS device 28 as vehicle information.
[0049] The output unit 210 has a function of outputting the collected vehicle information to the center server 30.
[0050] (Center server) 4, the center server 30 includes a CPU 30A, a ROM 30B, a RAM 30C, a storage 30D, and a communication I / F 30E. The CPU 30A, the ROM 30B, the RAM 30C, the storage 30D, and the communication I / F 30E are connected to each other via an internal bus 30F so as to be able to communicate with each other. The functions of the CPU 30A, the ROM 30B, the RAM 30C, and the communication I / F 30E are the same as those of the CPU 20A, the ROM 20B, the RAM 20C, and the wireless communication I / F 20E of the vehicle-mounted device 20 described above. The communication I / F 30E may perform wired communication.
[0051] The storage 30D as a memory is configured by an HDD (Hard Disk Drive) or an SSD (Solid State Drive) and stores various programs and various data. In the present embodiment, the storage 30D stores a prediction program 110, a vehicle information DB (database) 120, a attendance record information DB 130, a weather information DB 140, and a trained model 150. The ROM 30B may store the prediction program 110, the vehicle information DB 120, the attendance record information DB 130, the weather information DB 140, and the trained model 150.
[0052] The prediction program 110 as a program is a program for controlling the center server 30. With the execution of the prediction program 110, the center server 30 acquires vehicle information from the vehicle-mounted device 20, and acquires attendance record information and weather information from the information providing server 40. The center server 30 uses the acquired vehicle information, attendance record information, and weather information to generate a trained model 150 (described later) and predict the traffic congestion situation.
[0053] The vehicle information DB 120 stores vehicle information acquired from a plurality of vehicles 12. The attendance record information DB 130 stores attendance record information acquired from the information providing server 40. The weather information DB 140 stores weather information including observation results and forecasts acquired from the information providing server 40.
[0054] The trained model 150 is a trained model that has machine-learned vehicle information, attendance record information, and weather information. The trained model 150 predicts the traffic congestion situation for each time period using the number of people who come to work on the day to be predicted, the number of people who leave work for each time period, and weather information. Here, the trained model 150 predicts the traffic congestion situation using a random forest. A random forest creates multiple decision tree groups with low correlation with each other, and derives a prediction result by integrating and averaging the prediction results. Control parameters of the random forest include the number of explanatory variables to be selected, the number of branches of the decision tree, and a threshold value set for a question in the decision tree. Note that the trained model 150 according to this embodiment has been described as using a random forest. However, the trained model 150 is not limited to this. The trained model 150 may be a neural network or a support vector machine.
[0055] Next, functions of the center server will be described with reference to Fig. 5. As an example, as shown in Fig. 5, in the center server 30 of this embodiment, the CPU 30A executes the prediction program 110 to function as an acquisition unit 300, a determination unit 310, a setting unit 320, a learning unit 330, a storage unit 340, an estimation unit 350, a prediction unit 360, and a presentation unit 370.
[0056] The acquisition unit 300 has a function of acquiring vehicle information transmitted from the vehicle-mounted device 20 of the vehicle 12, weather information transmitted from the information providing server 40, and attendance record information.
[0057] The determination unit 310 uses the acquired vehicle information to determine the traffic congestion status indicating the presence or absence of traffic congestion on roads located around the facility. Specifically, the determination unit 310 divides sections into sections at predetermined distances (e.g., 50 m) from an intersection located around the facility, and derives the time required for the vehicle 12 to pass through the section using the vehicle information. The determination unit 310 derives the average time required for the vehicle 12 to pass through each section for each time period (hereinafter referred to as the "average required time"), and detects the number of sections in which the average required time exceeds a threshold and which are continuous. If the number of detected continuous sections in each time period is equal to or greater than a predetermined threshold (the number of sections in which the average required time exceeds the threshold is equal to or greater than a predetermined distance), the determination unit 310 determines that traffic congestion is occurring as the traffic congestion status.
[0058] As an example, Fig. 6 shows the average required time for each time period and for a predetermined section. For example, as shown in Fig. 6, the determination unit 310 derives the average required time for each section for each time period. The determination unit 310 detects sections in which the average required time is 4.0 seconds or more. Here, in Fig. 6, the shaded areas indicate sections detected by the determination unit.
[0059] When the detected sections are eight or more consecutive sections (400 m) in each time period, the determination unit 310 determines that congestion is occurring in that time period. For example, as shown in Fig. 6, the determination unit 310 detects sections A to I in time period e (17:00 to 17:15) where the average required time is 4.0 seconds or more, and determines that congestion is occurring in time period e because the detected sections are nine consecutive sections.
[0060] As an example, the base point (0 m point) when setting the section shown in FIG. 6 is, for example, an intersection or the like. However, this is not limited to this. Any location may be set as the base point (0 m point) as long as it is a location where there is a cause of congestion. For example, it may be a location where the number of lanes and road width are reduced, a location where construction work is being carried out, a location where an accident has occurred, or a location where acceleration is reduced, such as when climbing a slope.
[0061] The setting unit 320 shown in FIG. 5 uses the traffic congestion state, weather information, and attendance record information determined by the determining unit 310 to set the learning data shown in FIG.
[0062] As shown in FIG. 7, the learning data is set to include weather information, attendance record information, and a judgment result. The learning data includes date, time period, calendar, weather, and precipitation as weather information, employee attendance rate, number of employees who clock out, number of employees who clock out one period ago, number of employees who clock out two periods ago, and number of employees who clock out three periods ago as attendance record information, and traffic congestion as a judgment result. Here, the date is the target date on which the data was acquired, the time period is the target time period on which the data was acquired, and the calendar is information indicating the characteristics of the target date, such as a season, a weekday, and a holiday. The weather is information on the weather on the target date, and the precipitation is the amount of rain that fell on the target date. The attendance rate is the ratio of the number of employees who attended work on the target date to the total number of employees at the facility, and the number of employees who clock out is the number of employees who clock out from the facility in each time period. In addition, in the case of time period d, the number of people who left work one period ago is, for example, the number of people who left work in time period c, the number of people who left work two periods ago is the number of people who left work in time period b two periods ago, and the number of people who left work three periods ago is the number of people who left work in time period a three periods ago. In other words, the setting unit 320 sets the number of people who left work in one period as the learning data to each of the "number of people who left work one period ago", the "number of people who left work two periods ago", and the "number of people who left work three periods ago". As a result, the learning data includes the number of people who left work in a time period before the time period for which congestion is predicted, and it becomes possible to predict the congestion situation taking into account the time it takes for employees to reach the road to be predicted after leaving work.
[0063] The setting unit 320 also sets setting data for predicting congestion using weather information and an estimation result estimated by the estimation unit 350 described later. As an example, as shown in FIG. 8, the setting data is set to include weather information and an estimation result. The setting data includes, as weather information, a date, a time period, a calendar, weather, and precipitation, and includes, as estimation results, an estimated employee attendance rate, an estimated number of people who leave work, the number of people who leave work one period ago, the number of people who leave work two periods ago, and the number of people who leave work three periods ago, all of which are estimated by the estimation unit 350 described later. Here, the estimated number of people who leave work in this embodiment is an example of "estimated number of people information".
[0064] The learning unit 330 shown in Fig. 5 uses the learning data shown in Fig. 7 to perform machine learning for predicting congestion to generate a trained model 150. Specifically, the learning unit 330 performs machine learning using weather information and attendance record information included in the learning data as input data and the congestion situation included in the learning data as teacher data.
[0065] The storage unit 340 stores the trained model 150 generated by the learning unit 330. The storage unit 340 also stores the vehicle information, attendance record information, and weather information acquired by the acquisition unit 300 in the vehicle information DB 120, the attendance record information DB 130, and the weather information DB 140, respectively.
[0066] The estimation unit 350 performs statistics using past attendance record information stored in the attendance record information DB 130, and estimates the number of employees who will be at work at the facility on the day to be predicted. Specifically, the estimation unit 350 uses the past attendance record information to derive the number of employees who will be at work on the day to be predicted (hereinafter referred to as the "estimated number of employees") from the average moving average of the number of employees who have worked on the same day of the week in the past four weeks (for example, the number of employees who worked on Wednesdays). The estimation unit 350 uses the derived estimated number of employees to estimate the ratio of the estimated number of employees to all employees (hereinafter referred to as the "estimated attendance rate").
[0067] 9, the estimation unit 350 performs statistics using past attendance record information and derives the ratio of the number of employees who have left work to the number of employees who have come to work for each date and time period. Using the derived ratio of employees who have left work and the estimated number of employees, the estimation unit 350 estimates the number of people who will leave work during each time period on the day to be predicted (hereinafter referred to as the "estimated number of people who will leave work").
[0068] The prediction unit 360 predicts the traffic congestion state in each time period on the day to be predicted, using the setting data shown in Fig. 8. Specifically, the prediction unit 360 predicts the traffic congestion state in each time period on the day to be predicted, by inputting the setting data to the trained model 150.
[0069] The presentation unit 370 presents the prediction result predicted by the prediction unit 360. Specifically, the presentation unit 370 presents a prediction presentation screen 400 shown in FIG. 10. As an example, the prediction presentation screen 400 shown in FIG. 10 includes a target road presentation area 410, a prediction result presentation area 420, and an observation result presentation area 430. The target road presentation area 410 presents a road for which a traffic jam situation is predicted using a map or the like, and the prediction result presentation area 420 presents the prediction results for each time period predicted by the prediction unit 360. The observation result presentation area 430 presents the judgment results of the traffic jam situation for each time period on the same day of the week (e.g., Wednesday) one week ago, two weeks ago, and three weeks ago.
[0070] Before describing the operation of the traffic congestion prediction system 10, the flow of data in the learning phase for generating the trained model 150 and the prediction phase for predicting traffic congestion in the center server 30 as a traffic congestion prediction device will be described with reference to Figures 11 and 12. Figure 11 is a data flow diagram showing an example of the flow of data in the learning phase for generating the trained model 150 in the center server 30.
[0071] First, the flow of data in the learning phase will be described with reference to Fig. 11. As shown in Fig. 11, the acquisition unit 300 acquires vehicle information, attendance record information, and weather information from the vehicle information DB 120, attendance record information DB 130, and weather information DB 140. The acquisition unit 300 inputs the vehicle information to the determination unit 310, and inputs the attendance record information and weather information to the setting unit 320.
[0072] 6, the determination unit 310 detects an average required time for each predetermined section and for each time period using the vehicle information, and determines the traffic congestion state as a determination result 500. The determination unit 310 inputs the determined traffic congestion state to the setting unit 320.
[0073] The setting unit 320 sets learning data 510 using the input attendance record information, weather information, and traffic congestion status, and inputs the data to the learning unit 330. Here, the setting unit 320 uses the number of people who clocked out included in the attendance record information to set the number of people who clocked out one period ago, the number of people who clocked out two periods ago, and the number of people who clocked out three periods ago in the learning data 510. The setting unit 320 also divides the learning data 510 into training data and test data and sets them.
[0074] The learning unit 330 executes machine learning using the input learning data 510, generates the learned model 150, and inputs it to the storage unit 340. Here, the learning unit 330 generates the learned model 150 using training data out of the learning data 510, and evaluates the generated learned model 150 using test data.
[0075] The memory unit 340 stores the input trained model 150 in the storage 30D.
[0076] Next, the data flow in the prediction phase will be described with reference to Fig. 12. As shown in Fig. 12, the acquisition unit 300 acquires attendance record information and weather information from the attendance record information DB 130 and the weather information DB 140. The acquisition unit 300 inputs the attendance record information to the estimation unit 350 and inputs the weather information to the setting unit 320.
[0077] The estimation unit 350 performs statistics using the input attendance record information, and estimates the estimated attendance rate on the day to be predicted and the estimated number of people who will leave work in each time period. The estimation unit 350 inputs the estimated attendance rate and the estimated number of people who will leave work to the setting unit 320 as the estimation result 530.
[0078] The setting unit 320 sets the setting data 540 using the input estimated attendance rate, estimated number of people leaving work, and weather information, and inputs the setting data 540 to the prediction unit 360. Here, the setting unit 320 uses the estimated number of people leaving work to set the number of people leaving work one period ago, the number of people leaving work two periods ago, and the number of people leaving work three periods ago in the setting data 540, respectively.
[0079] The prediction unit 360 uses the trained model 150 stored in the storage 30D to predict the traffic congestion situation based on the input setting data 540, and inputs the prediction result 550 to the presentation unit 370.
[0080] The presentation unit 370 presents the prediction presentation screen 400 using the input prediction result 550. Here, the presentation unit 370 acquires vehicle information and presents traffic congestion information for each time period on the same day of the week in the past together with the prediction result 550. When the presentation unit 370 receives an instruction from the user to present the prediction result 550, the presentation unit 370 transmits the prediction presentation screen 400 as the prediction result 550 to a terminal operated by the user and presents it.
[0081] (Flow of Control) The flow of processing executed in the traffic congestion prediction system 10 of this embodiment will be described with reference to the flowcharts of Figures 13 and 14. Each process in the center server 30 is executed by the CPU 30A of the center server 30 functioning as an acquisition unit 300, a determination unit 310, a setting unit 320, a learning unit 330, a storage unit 340, an estimation unit 350, a prediction unit 360, and a presentation unit 370. The generation process shown in Figure 13 is executed, for example, when an instruction to execute a process of generating a trained model 150 is input.
[0082] In step S101, CPU 30A acquires vehicle information, attendance record information, and weather information.
[0083] In step S102, CPU 30A determines the traffic congestion state by using the vehicle information.
[0084] In step S103, CPU 30A sets learning data using attendance record information, weather information, and traffic congestion conditions.
[0085] In step S104, CPU 30A executes machine learning using training data among the learning data.
[0086] In step S105, CPU 30A generates trained model 150 as a result of executing machine learning.
[0087] In step S106, CPU 30A uses test data from the learning data to evaluate the generated trained model 150. Here, as the evaluation result, a matching rate between the prediction of the traffic congestion situation by trained model 150 and the traffic congestion situation included in the test data is derived.
[0088] In step S107, CPU 30A determines whether the relevance rate derived as the evaluation result exceeds a threshold value. If the relevance rate exceeds the threshold value (step S107: YES), CPU 30A proceeds to step S108. On the other hand, if the relevance rate does not exceed the threshold value (the relevance rate is equal to or less than the threshold value) (step S107: NO), CPU 30A proceeds to step S104 and performs machine learning again.
[0089] In step S108, CPU 30A stores the generated trained model 150 in storage 30D.
[0090] Next, the flow of a process for predicting traffic congestion executed in the center server 30 of this embodiment will be described with reference to FIG.
[0091] In step S201, CPU 30A acquires attendance record information and weather information.
[0092] In step S202, CPU 30A uses the acquired attendance record information to estimate an estimated attendance rate and an estimated clock-out number as estimation results.
[0093] In step S203, CPU 30A sets the setting data using the estimation result and weather information.
[0094] In step S204, CPU 30A predicts a traffic congestion state as a prediction result using the setting data.
[0095] In step S205, CPU 30A determines whether or not the traffic congestion conditions have been predicted for all time periods on the day being predicted. If the traffic congestion conditions have been predicted for all time periods (step S205: YES), CPU 30A proceeds to step S206. On the other hand, if the traffic congestion conditions have not been predicted for all time periods (there are time periods for which traffic congestion conditions have not been predicted) (step S205: NO), CPU 30A proceeds to step S204 and predicts the traffic congestion conditions in the unpredicted time periods.
[0096] In step S206, CPU 30A presents the predicted traffic congestion situation on prediction presentation screen 400 as a prediction result.
[0097] In step S207, CPU 30A judges whether or not to end the prediction process. If the process is to be ended (step S207: YES), CPU 30A ends the prediction process. On the other hand, if the process is not to be ended (step S207: NO), CPU 30A proceeds to step S201 and acquires attendance record information and weather information.
[0098] (Summary of the embodiment) The center server 30 of this embodiment determines the traffic congestion status of roads located around the facility from vehicle information acquired from the vehicle 12, and generates a trained model 150 that performs machine learning on weather information, attendance record information, and traffic congestion status. The center server 30 estimates an estimated attendance rate and an estimated number of people who will leave work for each time period on the prediction target day, which are estimated from the attendance record information, and predicts the traffic congestion status for each time period on the prediction target day using the trained model 150.
[0099] As described above, according to this embodiment, it is possible to predict long-term road conditions.
[0100] [remarks] In the above embodiment, the presentation unit 370 acquires vehicle information, judges the traffic congestion state, and presents the judgment result together with the prediction result of traffic congestion information for each time period on the same day of the week in the past. However, the present invention is not limited to this. For example, the presentation unit 370 may acquire the judgment result judged by the judgment unit 310 and present it together with the prediction result.
[0101] In addition, various processes executed by the CPU 20A and the CPU 30A by reading the software (programs) in the above embodiment may be executed by various processors other than the CPU. In this case, examples of the processor include a PLD (Programmable Logic Device) such as an FPGA (Field-Programmable Gate Array) whose circuit configuration can be changed after manufacture, and a dedicated electric circuit such as an ASIC (Application Specific Integrated Circuit) which is a processor having a circuit configuration designed exclusively for executing a specific process. In addition, each of the above-mentioned processes may be executed by one of these various processors, or may be executed by a combination of two or more processors of the same or different types (for example, a plurality of FPGAs, a combination of a CPU and an FPGA, etc.). In addition, the hardware structure of these various processors is, more specifically, an electric circuit that combines circuit elements such as semiconductor elements.
[0102] In the above embodiment, each program has been described as being pre-stored (installed) in a non-transitory recording medium that can be read by a computer. For example, the prediction program 110 in the center server 30 is pre-stored in the ROM 30B. However, the present invention is not limited to this, and each program may be provided in a form recorded in a non-transitory recording medium such as a CD-ROM (Compact Disc Read Only Memory), a DVD-ROM (Digital Versatile Disc Read Only Memory), or a USB (Universal Serial Bus) memory. Also, the program may be downloaded from an external device via a network.
[0103] The process flow described in the above embodiment is merely an example, and unnecessary steps may be deleted, new steps may be added, or the process order may be changed without departing from the spirit of the invention. [Explanation of symbols]
[0104] 12 Vehicles 30 Center Server 30A CPU (processor) 120 Vehicle Information DB (Vehicle Information) 130 Attendance record information DB (number of people information) 150 trained models (machine learning models)
Claims
1. Acquire number-of-users information indicating the number of users who depart from a facility for each predetermined period, including users who board a vehicle, and vehicle information regarding vehicles that have traveled around the facility; Using the vehicle information, a traffic congestion situation indicating whether or not there is congestion on a road located in the vicinity of the facility during the predetermined period is determined; a memory that stores a machine learning model that predicts the traffic congestion status on a road located near a facility for each predetermined period of time using estimated number of users information indicating the number of users estimated to depart from the facility for each predetermined period of time, including users riding in a vehicle, generated by a machine learning method in which a computer executes processing to execute machine learning for predicting traffic congestion on a road using the number of users information and the traffic congestion status; and a processor; The processor, Estimating the estimated number of people information during a target period for predicting the traffic congestion situation; The estimated number of people information is input into the machine learning model to predict the traffic congestion situation during the target period; On a prediction presentation screen showing the predicted traffic congestion situation in the target period, roads for which the traffic congestion situation is predicted using a map, the prediction result of the traffic congestion situation for each of the predetermined periods in the target period, and traffic congestion information for each of the predetermined periods on the same day of the week in the past during the target period are presented on a single screen. Traffic congestion prediction device.
2. The processor, presenting, on the prediction presentation screen, a character string indicating a time for each of the predetermined periods and a rectangular mark indicating, by a difference in color, the presence or absence of congestion for each of the times indicated by the character string; On an extension line of each time shown in the character string, a rectangular mark indicating the presence or absence of congestion in the target period as the predicted result of the congestion situation and a rectangular mark indicating the presence or absence of congestion on the same day of the week in the past as the congestion information are arranged. The traffic congestion prediction device according to claim 1.
Citation Information
Patent Citations
Congestion prediction apparatus and navigation apparatus
JP2009251969A
Autonomous distribution type traffic information service system
JP2014191578A
Information processing system, information processing server, information processing method, and information processing program
JP2014235487A
Traffic flow control apparatus, traffic flow control system, mobile recording device, traffic flow control method, and program
JP2016157372A
Traffic congestion prediction system, traffic congestion prediction method, learning device, prediction device, program and learned model
JP2019169028A
