Operation management device, operation management method, and operation management system

Abstract

An operation management device 1 for managing operation of a train comprises: a diagram drawing unit 7 that acquires a timetable of the train as an image; and a working arrangement model 6 that determines the presence or absence of a timetable disturbance on the basis of the image of the timetable of the train. Accordingly, it is possible to determine that a timetable disturbance has occurred from the operation state of the train, and to present the determination to a command person.

Description

Traffic management device, traffic management method, and traffic management system

[0001] The present invention relates to a traffic management device, a traffic management method, and a traffic management system, and more particularly to a traffic management device or the like that can determine whether or not a schedule (diagram) is disrupted.

[0002] A traffic management system has been proposed that proposes a train rescheduling plan to the dispatcher when a train schedule disruption occurs. In this case, the dispatcher checks the proposal, modifies it if necessary, and then approves it. This makes it easier to implement train rescheduling. One method for proposing train rescheduling plans is to register IF-THEN rules, which are pairs of operational status and operational rescheduling plans, in the system in advance, and when an operational status corresponding to the IF section occurs, the system proposes a train rescheduling plan written in the THEN section. The operational status is determined based on the actual train arrival and departure times and predicted arrival and departure times based on future operation forecasts.

[0003] Patent Literature 1 describes a train rescheduling support system. In this train rescheduling support system, a processing device generates input difference information that represents the difference between an input planned timetable, which is a planned timetable for a given situation, and an input predicted timetable, which is a predicted timetable predicted depending on the situation. The processing device then compares the input difference information with historical difference information. The historical difference information represents the difference between a historical planned timetable, which is a planned timetable set when a train rescheduling was implemented in the past, and a historical predicted timetable, which is a predicted timetable predicted at that time. Furthermore, the processing device determines a recommended train rescheduling operation for the situation based on the comparison result and historical train rescheduling operations, which are train rescheduling operations implemented when train rescheduling was implemented in the past.

[0004] Japanese Patent Application Laid-Open No. 2019-172142

[0005] In a method of registering a traffic rescheduling plan (IF-THEN rule) that focuses on the arrival and departure times of specific trains at specific stations in a system, the rules are localized, and the local operational status and traffic rescheduling plan are displayed, which can result in proposals that do not necessarily reflect the overall operational status. Regarding the overall operational status, for example, actual and predicted schedules are overlaid on the current day's schedule. In this case, as the amount of information in the schedule image increases due to high-density operation, quadruple tracks, and inter-rail access, it becomes difficult for a dispatcher to visually determine the operational status in a short amount of time, especially when a schedule disruption occurs at multiple locations. The present invention aims to provide a traffic management device, a traffic management method, and a traffic management system that can determine the occurrence of a schedule disruption from train operational status and present this information to a dispatcher.

[0006] In order to solve the above problems, the present invention provides a traffic management device that manages train operations, and includes an image acquisition unit that acquires train schedules as images, and a machine learning model that determines whether or not a schedule disruption has occurred based on the train schedule images. In this case, it is possible to provide a traffic management device that can determine the occurrence of a schedule disruption from the train operation status and notify the dispatcher.

[0007] Here, for example, the images include train schedules and operational records, and the machine learning model learns images previously acquired by the image acquisition unit, the operational status corresponding to the images, and the corresponding train schedule rescheduling details. Images acquired by the image acquisition unit during train traffic management are input, and the model outputs a determination result of whether a train schedule disruption will occur. In this case, the machine learning model can learn train operation statuses and train schedule rescheduling corresponding to a schedule disruption, and determine whether a schedule disruption will occur. If the machine learning model determines that a schedule disruption will occur, it outputs a train schedule rescheduling proposal. In this case, a train schedule rescheduling proposal based on the overall operational status can be presented to the dispatcher. Furthermore, for example, the machine learning model compares the current operational status with past operational statuses and outputs a train schedule rescheduling proposal for the current operational status based on a similar past operational status. In this case, a train schedule rescheduling proposal similar to a train schedule rescheduling proposal made in response to a similar schedule disruption in the past can be presented to the dispatcher. Furthermore, for example, the machine learning model further outputs operational status data indicating changes in train operation status over time, and creates a train schedule rescheduling proposal by adding the operational status data. In this case, a train schedule rescheduling proposal taking into account the overall operational status data can be output. For example, the machine learning model is created by learning based on the feature quantities of an image of a train schedule. In this case, learning can be performed according to the features of the schedule disruption. Furthermore, for example, the feature quantities include at least one of the position and pattern of the train schedule. In this case, more suitable feature quantities can be extracted. Furthermore, for example, the image acquisition unit draws train schedule images including the schedule set for the day the train is to run, the schedule determined based on actual train operation results, and the schedule predicted for the current and future train operations, and the machine learning model determines whether or not there is a schedule disruption based on the schedule images drawn by the image acquisition unit. In this case, feature quantities can be more easily identified. Furthermore, for example, the machine learning model is created by learning when an image of a schedule is used as input and train operation status and train rescheduling are used as output.In this case, it is possible to learn train operation status and train rescheduling in response to a schedule disruption. Furthermore, for example, the machine learning model is created by learning on cases where a schedule disruption occurs. In this case, it is possible to determine that a schedule disruption has occurred and to output a schedule rescheduling plan in response to the schedule disruption. Furthermore, for example, the machine learning model is created by learning on cases where a schedule disruption occurs as well as on normal cases where there is no schedule disruption. In this case, it is possible to learn the difference between cases where there is a schedule disruption and normal cases where there is no schedule disruption. Furthermore, for example, an image of a train schedule is represented on a time axis and a distance axis. In this case, the train operation status can be represented by the relationship between time and distance. Furthermore, for example, the image of a train schedule is represented on a time axis and stop station information as axes. In this case, the train operation status can be represented by the relationship between time and stop station information.

[0008] The present invention also provides a traffic control method for managing train operations, in which a processor executes a program stored in a memory to acquire a train schedule as an image, and determines whether or not a schedule disruption has occurred based on the image of the train schedule using a machine learning model for determining whether or not a schedule disruption has occurred. In this case, a traffic control method can be provided that can determine whether a schedule disruption has occurred based on the train operation status and notify a dispatcher.

[0009] Furthermore, the present invention provides a traffic management system that includes a learning model creation device that creates a machine learning model for outputting a determination result of whether or not a schedule disruption has occurred, and a traffic management device that manages train operations, wherein the traffic management device includes an image acquisition unit that acquires train schedules as images, a machine learning model that determines whether or not a schedule disruption has occurred based on the images of the train schedule, and a determination unit that determines whether or not a schedule disruption has occurred using the machine learning model.In this case, a traffic management system can be provided that can determine that a schedule disruption has occurred based on the train operation status and notify a dispatcher.

[0010] According to the present invention, it is possible to provide an operation management device, an operation management method, and an operation management system that can determine that a schedule disruption has occurred based on the train operation status and notify the dispatcher.

[0011] It is a block diagram showing the entire operation control system. It is a block diagram showing the functional configuration of the train operation command device according to the present embodiment. It is a diagram showing a timetable rescheduling learning device that creates a timetable rescheduling model. It is a diagram showing an example of a line image.

[0012] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. Here, the present invention will be described by way of first and second embodiments. Here, the first embodiment will be described first.

[0013] [First embodiment] <Explanation of the entire traffic control system> Fig. 1 is a block diagram showing the entire traffic control system S. As shown in the figure, the traffic control system S includes a timetable replanning learning device 51 and a train traffic control device 1. The timetable replanning learning device 51 is an example of a learning model creation device, and creates a timetable replanning model 6 as a machine learning model for outputting a determination result of whether or not there is a timetable disruption. The train traffic control device 1 is an example of a traffic control device that manages train operations. As will be described in detail later, the train traffic control device 1 includes the timetable replanning model 6 created by the timetable replanning learning device 51, and detects a timetable disruption based on an image of the timetable using the timetable replanning model 6. When a timetable disruption occurs, the device outputs a notice that a timetable disruption has occurred and a timetable replanning proposal in response.

[0014] <Description of train traffic control device 1> Figure 2 is a block diagram showing the functional configuration of the train traffic control device 1 according to this embodiment. The train traffic control device 1 shown in the figure includes a traffic management unit 2, a data receiving unit 3, a data transmitting unit 4, a traffic prediction unit 5, a traffic rescheduling model 6, a route drawing unit 7, an input processing unit 8, a traffic prediction constant DB (Data Base) 20, and an internal storage DB 21. Also shown in Figure 2 are a display 9, an input device 10, an interlocking device 11, and an I / F (interface) device 22 for other companies' lines, although they do not constitute the train traffic control device 1.

[0015] The traffic control unit 2 manages train operations. Specifically, for example, the traffic control unit 2 issues commands to control the route of trains based on the train's location on the line. Furthermore, for example, the traffic control unit 2 issues commands to change the timetable for operating trains and to prevent trains from departing. The data receiving unit 3 receives location status data of trains on the line from the interlocking device 11 and the other company's line I / F device 22, and sends it to the traffic control unit 2. The location status data is information indicating the location of each train.

[0016] The data transmission unit 4 sends commands from the traffic control unit 2 and the dispatcher to the interlocking device 11. The operation prediction unit 5 predicts the future operation of trains on the line based on the latest arrival and departure time information of trains on the line. The traffic rescheduling model 6 is a machine learning model that determines whether or not a timetable disruption has occurred. This allows the train operation status to be determined as a disruption and the dispatcher to be notified. Furthermore, if the train operation status is determined to be a disruption, the traffic rescheduling model 6 outputs a train operation rescheduling notification and a proposed train operation rescheduling. This allows the dispatcher to be presented with a traffic rescheduling plan that is tailored to the overall operational status. In other words, since the traffic rescheduling plan is created based on the entire image of the timetable, it can be tailored to the overall operational status. In this case, the traffic rescheduling model 6 compares the current operational status with past operational status and outputs a traffic rescheduling plan for the current operational status based on a traffic rescheduling plan implemented for a similar past operational status. This allows the dispatcher to be presented with a traffic rescheduling plan similar to a traffic rescheduling plan implemented in response to a similar timetable disruption in the past. The train line drawing unit 7 creates an image of the train timetable. The line drawing unit 7 functions as an image acquisition unit that acquires train timetables as images. The input processing unit 8 processes commands input by a train operator via the input device 10 and sends the commands to the data transmission unit 4.

[0017] The operation prediction constant DB 20 stores operation prediction constants 20a, which are constants required when the operation prediction unit 5 predicts the future operation of trains on the line from the present. The operation prediction constants 20a include, for example, running time between stations, stopping time at stations, and track conflict time intervals.

[0018] The internal memory DB 21 stores data required when the operation prediction unit 5 predicts the future operation of trains on the line and when the train rescheduling model 6 determines whether a train schedule disruption will occur. Specifically, the internal memory DB 21 stores a current day timetable 12, train tracking data 13, operation history data 14, other companies' line operation status data 15, operation status data 16, train rescheduling history 17, operation prediction data 18, and train route images 19. The current day timetable 12 is information about the timetable set for the day on which the train is to operate. The train tracking data 13 is created based on the train location status data and is data indicating changes in the train's location over time. The operation history data 14 is timetable data obtained from the results of actual train operations. In this case, the operation history data 14 indicates, for example, the actual times and stations at which trains departed or arrived for each train.

[0019] The other company's line operation status data 15 is operation status data on other company's lines, and the operation status data 16 is operation status data on the company's own lines. These operation status data indicate the train operation status that changes over time. The operation status data is data that indicates, for example, whether each train is currently in operation or has an operation disruption, its current position on the line if in operation, which station it is heading to, the delay time from the timetable 12 for that day, etc.

[0020] The train schedule history 17 is a history of train schedule changes made by dispatchers in the past, and is data showing what kind of schedule changes were made when a schedule disruption occurred. The operation prediction data 18 is data on a train schedule that predicts the future operation of trains from the present. The line image 19 is image data of a train schedule that includes lines that show the state of train operation.

[0021] The display 9 is a display device such as a liquid crystal display or an organic EL display. The display 9 displays information about train operations and presents it to the train dispatcher. In this embodiment, the display 9 displays the results of the train rescheduling model 6, and if there is a train schedule disruption, it displays a message that there is a schedule disruption. It can also display a train schedule rescheduling plan. The input device 10 is a device that allows the train dispatcher to input commands, and corresponds to, for example, a keyboard, a mouse, or buttons and levers on a control panel. The interlocking device 11 is a device that controls field equipment such as signals and points in accordance with predetermined rules. The other company's line I / F device 22 communicates with other company's train operation command devices and acquires the operation status of other company's lines.

[0022] <Operation of train traffic control device 1> The train traffic control device 1 operates, for example, as described below. The data receiving unit 3 receives train location data from the interlocking device 11 and other company's line I / F device 22, and inputs it to the traffic control unit 2. The traffic control unit 2 creates train tracking data 13 and operation performance data 14 based on the location data, and stores them in an internal storage DB 21. The traffic control unit 2 also creates route control data based on the location data and the current day's timetable 12, and transmits it to the interlocking device 11 via the data transmitting unit 4.

[0023] The operation prediction unit 5 receives as input the operation prediction constants 20a of the operation prediction constant DB 20, the current day's timetable 12, train tracking data 13, operation history data 14, and other companies' line operation status data 15. The operation prediction unit 5 predicts future operations from the latest arrival and departure time information of trains on the line based on the train tracking data 13, operation history data 14, and other companies' line operation status data 15, and stores the result as operation prediction data 18 in the internal storage DB 21. At this time, the operation prediction unit 5 predicts the operation prediction data 18 taking into account the operation prediction constants 20a (running time between stations, stop time at stations, track conflict time intervals, etc.). The current day's timetable 12, operation history data 14, and operation prediction data 18 are input to the line drawing unit 7, and the line drawing unit 7 creates a line image 19.

[0024] As will be explained in detail in Figure 3, the timetable rescheduling model 6 has previously learned the relationship between the route image 19, the operation status data 16, and the timetable rescheduling history 17. The route image 19 generated in real time during operation management is input into the timetable rescheduling model 6, and the presence or absence of a timetable disruption as the operation status, and if it is determined that a timetable disruption exists, a timetable rescheduling proposal is output to the display 9.

[0025] The dispatcher views the operation status and proposed timetable rescheduling displayed on the display 9. If the proposed timetable rescheduling is adopted, the dispatcher uses the input device 10 to input approval for the proposed timetable rescheduling. The approved timetable rescheduling plan is input to the operation control unit 2 via the input processing unit 8. The operation control unit 2 outputs changes to the current day's timetable 12 or departure suppression information to the interlocking device 11.

[0026] <Method of Creating the Timetable Replanning Model 6> Next, a method of creating the timetable replanning model 6 will be described. FIG. 3 is a diagram showing a timetable replanning learning device 51 that creates the timetable replanning model 6. As described above, the timetable replanning learning device 51 creates a machine learning model for outputting a determination result of whether or not there is a schedule disruption. The timetable replanning learning device 51 creates the timetable replanning model 6 using a line image 19, operation status data 16, and a line replanning history 17. In this case, the timetable replanning learning device 51 learns about a case in which the line image 19 is input and the operation status data 16 and line replanning history 17 are output. In other words, the device 51 learns about the relationship between the line image 19 and the operation status data 16 and line replanning history 17 corresponding to this line image 19.

[0027] The timetable rescheduling learning device 51 then outputs the results of its learning as a timetable rescheduling model 6. This can also be said to mean that the timetable rescheduling model 6, which is a machine learning model, is created by learning when an image of a timetable is input and train operation status and train timetable rescheduling are output. This image also includes train timetables and operation records, and the timetable rescheduling model 6, which is a machine learning model, can also be said to machine-learn images previously acquired by the train track drawing unit 7, operation status corresponding to the images, and timetable rescheduling details corresponding to the images. This makes it possible to learn train operation status and train timetable rescheduling in response to timetable disruptions.

[0028] In this case, the timetable replanning model 6 is created by learning from cases where a timetable disruption has occurred. This makes it possible to determine that a timetable disruption has occurred and to output a timetable replanning plan that corresponds to this timetable disruption. However, this is not limited to this, and the timetable replanning model 6 may be created by learning from cases where a timetable disruption has occurred, as well as from normal cases where there is no timetable disruption. This makes it possible to learn the difference between cases where a timetable disruption has occurred and normal cases where there is no timetable disruption.

[0029] The train rescheduling learning device 51 creates the train rescheduling model 6 by learning based on the feature quantities of images of train schedules. This allows learning to be performed according to the characteristics of schedule disruptions. The feature quantities on the train schedule image 19 used for machine learning are learned by observing deviations of the predicted schedule created using the operation prediction data 18 from the schedule of the day 12, and abnormal image patterns that differ from the train schedule image during on-time operation. This means that the feature quantities include at least one of the position and pattern of the train schedule. This allows more appropriate feature quantities to be extracted. To make the feature quantities easier to identify, the train schedule drawing unit 7 may perform a process of color-coding the actual operation schedule and the predicted operation schedule according to the degree of delay when creating the train schedule image 19. In this case, the line drawing unit 7 draws, as train timetables, the line drawing unit 7 draws an image of the line drawing unit 7, which includes the line drawing set for the day the train is to be operated (in this case, the current day line drawing 12), the line drawing determined based on the actual train operation results (in this case, the operation performance data 14), and the line drawing unit 7 draws an image of the line drawing unit 7, which represents a timetable image that predicts the train's future operation (in this case, the operation forecast data 18). The train rescheduling model 6 determines whether or not there is a timetable disruption based on the image of the line drawing unit 7. In this case, the feature amount is the deviation in line position between the current day line drawing 12 and the operation forecast data 18. The feature amount can also be a color change that is determined according to the degree of delay when a deviation occurs between the current day line drawing 12 and the operation forecast data 18.

[0030] FIG. 4 is a diagram showing an example of a train schedule image 19. The illustrated train schedule image 19 is a train schedule represented in a coordinate space consisting of a time axis and a distance axis. In FIG. 4, the horizontal axis is the time axis, on which time lines and time characters are drawn. The vertical axis is the distance axis, on which station names, station lines, and track numbers are drawn. That is, the train schedule image 19, which is an image of a train schedule, is represented by the time axis and the distance axis. Furthermore, since the train schedule image 19 also displays information about train stops, it can also be said that the train schedule image 19 is represented with the time axis and stop station information as axes. Here, the current schedule line, which is the line of the current day's schedule 12, is drawn using thin and solid lines. Furthermore, the actual line, which is the line of the operation record data 14, is drawn using thick and solid lines. Furthermore, the predicted line, which is the line of the operation forecast data 18, is drawn using thin and dotted lines. Here, any line width and line type may be used. In this way, the line image 19 draws the current day's timetable line based on the current day's timetable 12. The line image 19 also draws the actual line based on the operation record data 14. Furthermore, the line image 19 draws the predicted line based on the operation prediction data 18.

[0031] In Figure 4, the current day timetable, drawn with thin and solid lines, indicates that Train A will pass through two stations to reach Station 1, the terminal station on the company's line, then turn around at Station 1 and operate as Train A'. Similarly, Train B, drawn with the same current day timetable, indicates that Train A will enter from another company's line, pass through Station 1, and head to Station 2. In this case, Train A' will depart from Station 1 after Train B. The actual timetable, drawn with thick and solid lines, indicates that Train A will operate according to the current day timetable 12 and arrive at Station 1, the terminal station, at the current time. However, the predicted timetable, drawn with thin and dotted lines, indicates that Train B is delayed. As a result, the departure of Train A', which is scheduled to depart Station 1 after Train B, will also be delayed. The traffic rescheduling model 6 can use the above features to determine timetable disruptions. In this case, timetable disruptions can be determined by examining the positional discrepancies between the current day timetable 12 and the predicted timetable. In this case, the train rescheduling model 6, which is a machine learning model, receives as input images acquired by the train track drawing unit 7 during train operation management, and outputs a determination result as to whether or not there is a disruption to the schedule.

[0032] Second Embodiment As described above, the timetable replanning model 6 is a machine learning model that determines whether or not a train schedule disruption occurs. In this case, the timetable replanning model 6 receives an image of a train schedule as input, and if a disruption occurs, outputs a notice of the disruption and a train schedule replanning proposal. The timetable replanning model 6 also learns about the operation status data 16, and can output the operation status data 16 along with the image. In the second embodiment, in a conventional timetable replanning model that uses local features (station, platform, time interval between the first delayed train and the second delayed train, predicted arrival / departure delay time, etc.) at the location where a disruption occurs, the overall operation status data 16 output by the timetable replanning model 6 is added to the features. This enables the output of a timetable replanning proposal that takes into account the overall operation status data. In this case, the timetable replanning model 6 further outputs operation status data 16 indicating the train operation status that changes over time, and the train schedule replanning proposal is created by adding the operation status data 16.

[0033] According to the embodiment described above, it is possible to provide a traffic management device and a traffic management system that can determine the occurrence of a timetable disruption from train operation status and present the disruption to the dispatcher. Furthermore, when outputting a train operation rescheduling proposal along with a notice of a timetable disruption, it is possible to present the dispatcher with a timetable rescheduling proposal that is in line with the overall operation status. Conventional function for proposing timetable rescheduling proposals is established based on the timetable rescheduling rules obtained through interviews with dispatchers. In this case, when there is a change in the track alignment, such as a line extension, a change in points, or the addition of a siding, it is necessary to redefine the rules. In this embodiment, this is not necessary, and therefore it is expected that the quality of the rules can be improved and costs can be reduced.

[0034] In the above-described embodiment, the train rescheduling learning device 51 and the train operation control device 1 are separate devices, but they may be integrated into the same device to create a train rescheduling model 6, and use this train rescheduling model 6 to determine whether or not there will be a disruption to the schedule.

[0035] <Description of Traffic Management Method> The processing performed by the train traffic control device 1 is realized by the cooperation of software and hardware resources. That is, a processor such as a CPU provided in the train traffic control device 1 loads programs that realize each function of the train traffic control device 1 into main memory and executes them to realize each function. Therefore, the processing performed by the train traffic control device 1 described above can be considered as a traffic management method for managing train operations, in which the processor executes a program recorded in memory to acquire a train schedule as an image, and determines whether or not a schedule disruption has occurred based on the image of the train schedule using a machine learning model for determining whether or not a schedule disruption has occurred. This makes it possible to provide a traffic management method that can determine the occurrence of a schedule disruption from the train operation status and notify the dispatcher.

[0036] The program for realizing this embodiment can be provided not only by communication means but also by being stored on a recording medium such as a CD-ROM.

[0037] Although the present embodiment has been described above, the technical scope of the present invention is not limited to the scope described in the above embodiment. It is clear from the claims that various modifications and improvements to the above embodiment are also included in the technical scope of the present invention.

[0038] 1...Train operation control device, 2...Operation management unit, 3...Data receiving unit, 4...Data transmitting unit, 5...Operation prediction unit, 6...Operation rescheduling model, 7...Train line drawing unit, 8...Input processing unit, 12...Today's timetable, 13...Train tracking data, 14...Operation performance data, 15...Operation status data of other companies' lines, 16...Operation status data, 17...Operation rescheduling history, 18...Operation prediction data, 19...Train line image, 20...Operation prediction constant DB, 20a...Operation prediction constant, 21...Internal memory DB, 51...Operation rescheduling learning device, S...Operation management system

Claims

1. A traffic management device that manages train operations, comprising: an image acquisition unit that acquires train schedules as images; and a machine learning model that determines whether or not there is a schedule disruption based on the images of the train schedule.

2. The operation management system described in claim 1, wherein the images include train timetables and operation records, the machine learning model machine-learns the images previously acquired by the image acquisition unit, the operation status corresponding to those images, and the operation rescheduling details corresponding to those images, and inputs the images acquired by the image acquisition unit during train operation management, and outputs a determination result as to whether or not there is a disruption to the timetable.

3. An operation management device as described in claim 1 or 2, wherein the machine learning model outputs a train operation rescheduling plan if the result of the judgment indicates a schedule disruption.

4. The operation management device of claim 3, wherein the machine learning model compares the current operation status with past operation status and outputs a timetable rescheduling plan for the current operation status based on timetable rescheduling plans made for similar past operation status.

5. The traffic management device of claim 3, wherein the machine learning model further outputs operational status data indicating the state of train operations that changes over time, and the operational status data is added to create train timetable rescheduling plans.

6. An operation management device according to any one of claims 1 to 5, wherein the machine learning model is created by learning based on the features of an image of a train schedule.

7. The traffic control device according to claim 6, wherein the feature quantity includes at least one of the position and pattern of train timetables.

8. The operation management device of claim 7, wherein the image acquisition unit draws images of train schedules, including the schedule set for the day the train is to operate, the schedule determined based on the results of actual train operations, and the schedule predicted for future train operations, and the machine learning model determines whether or not there is a schedule disruption based on the images of the schedule drawn by the image acquisition unit.

9. An operation management device as claimed in any one of claims 1 to 8, wherein the machine learning model is created by learning when an image of a timetable is used as input and train operation status and train rescheduling are used as output.

10. The operation management device described in claim 9, wherein the machine learning model is created by learning from cases where schedule disruptions occur.

11. The operation management device described in claim 10, wherein the machine learning model is created by learning not only cases where a schedule disruption occurs but also normal cases where there is no schedule disruption.

12. An operation control device according to any one of claims 1 to 11, wherein the image of the train schedule is displayed on a time axis and a distance axis.

13. An operation control device according to any one of claims 1 to 11, wherein the image of the train schedule is displayed with a time axis and stop station information as axes.

14. A train operation management method for managing train operations, comprising: a processor executing a program recorded in memory to acquire a train schedule as an image; and a machine learning model for determining whether or not there is a train schedule disruption based on the image of the train schedule, to determine whether or not there is a train schedule disruption.

15. An operation management system comprising: a learning model creation device that creates a machine learning model for outputting a determination result on whether or not there is a train schedule disruption; and an operation management device that manages train operations, wherein the operation management device comprises: an image acquisition unit that acquires train schedules as images; and a machine learning model that determines whether or not there is a train schedule disruption based on the images of the train schedule.

Images