Operation plan change support system, operation plan change support method, and operation system

Abstract

This invention provides a train operation plan change support system, a train operation plan change support method, and a train operation system that, when interacting with users, narrows down the areas for improvement to those where system-based change decisions are difficult, and allows for the modification of the train operation plan while interactively resolving these areas. [Solution] A train operation plan change support system 2 that assists in creating a revised train operation plan when a predetermined train operation plan is changed, comprising: a train operation prediction program 502 that predicts the future operating status of a mobile vehicle based on a predetermined train operation plan and the operating performance of the mobile vehicle; a change decision inference program 506 that estimates a change decision for improvement locations, which are locations where changes are recommended, based on the predicted operating status; and a change decision dialogue program 508 that determines questions to supplement the information lacking in order to apply the change decision to improvement locations based on the estimated change decision, and changes the train operation plan based on the answers to the questions.

Description

Technical Field

[0004] , , , , , , , ,

[0001] The present invention relates to an operation plan change support system, an operation plan change support method, and an operation system. In particular, the present invention relates to an operation plan change support system that can be suitably used when changing an operation plan based on the operation results of a moving body.

Background Art

[0002] For example, when disturbances occur in the operation of a moving body such as a train, it is necessary to change the operation plan in order to maintain passenger services. In particular, when large-scale operation disturbances occur, it is difficult to restore the operation only by adjusting the departure and arrival times, so it is necessary to make large-scale and extensive changes to the operation plan such as suspension of operation and reverse operation. Conventionally, for example, an operation plan is changed based on the knowledge and experience of users such as dispatchers by using the knowledge and rules possessed by companies that operate moving bodies. On the other hand, in recent years, operation plan change support systems that utilize machine learning have been proposed.

[0003] Patent Document 1 discloses an operation arrangement system that generates an operation arrangement plan. This operation arrangement system includes an arithmetic unit that executes predetermined processing and a storage device accessible by the arithmetic unit. The arithmetic unit has a diagram prediction unit that generates a predicted diagram, a state determination unit that determines whether the predicted diagram includes a pseudo-convergence state, and a congestion location extraction unit that determines whether a congestion state has occurred. The pseudo-convergence state is a state in which a congestion location has occurred in the predicted diagram output by the diagram prediction unit and a predicted diagram with high evaluation cannot be generated. The congestion location is a key point in the operation arrangement for resolving the pseudo-convergence state.

Prior Art Documents

Patent Documents

[0004] [[ID=Z7]]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, in complex railway lines, the system may require user verification and correction of the operational plan generated due to insufficient learning or other reasons. While an interactive approach is expected to leverage user insights in such situations, if the verification and correction process is cumbersome, it may become difficult to encourage users to utilize the system. The objective of the present invention is to provide a train operation plan change support system, a train operation plan change support method, and a train operation system that, when interacting with a user, can narrow down the areas for improvement to those where the system has difficulty making a change decision, and change the train operation plan while interactively resolving these areas. [Means for solving the problem]

[0006] To solve the above problems, the present invention provides a train operation plan change support system that assists in creating a revised train operation plan when a predetermined train operation plan is changed, comprising: a train operation prediction unit that predicts the future operating status of a mobile vehicle based on a predetermined train operation plan and the operating performance of the mobile vehicle; a change estimation unit that estimates a change judgment for improvement locations, which are locations where changes are recommended, based on the predicted operating status; and a change dialogue unit that determines questions to supplement the information lacking in order to apply the change judgment to the improvement locations based on the estimated change judgment, and changes the train operation plan based on the answers to the questions. In this case, when interacting with the user, the train operation plan change support system can be provided that narrows down the improvement locations to those where it is difficult for the system to make a change judgment, and changes the train operation plan while interactively resolving these locations.

[0007] Here, for example, the change dialogue unit uses change judgment to classify whether dialogue about the operation information is necessary, and decides to ask questions if dialogue is necessary. In this case, the questions can be narrowed down when interacting with the user. Furthermore, the system includes, for example, a change trial unit that tests estimated change judgments and creates trial results including the amount of change in the areas to be improved. The change dialogue unit uses the created trial results to classify whether or not dialogue about operational information is necessary. In this case, the process of narrowing down the questions can be performed using the trial results. Furthermore, for example, the change dialogue unit calculates feature quantities including the amount of change in the improvement area from the trial results, calculates the confidence level of the change judgment based on the feature quantities, and determines whether dialogue is necessary. In this case, the process of narrowing down the questions can be performed using the confidence level of the change judgment. Furthermore, for example, the change dialogue unit identifies whether or not there is insufficient information from the trial results and determines the content of the question. In this case, the content of the question can be determined based on the missing information. Furthermore, the change dialogue unit determines whether there is insufficient information from the trial results. If there is sufficient information to determine whether a change decision is necessary, the dialogue unit generates a question that includes options for selecting the parameters required for the change decision. If there is insufficient information to determine whether a change decision is necessary, the dialogue unit generates a question that includes options for accepting or rejecting the change decision. In this case, the dialogue unit can determine a more appropriate question depending on the missing information. Furthermore, the change dialogue unit determines whether a question is necessary using at least one of the following results: whether the change decision is confident, whether it matches the user's decision, whether the user's decision can be implemented, whether the proposed quality can be improved, and whether the processing time can be shortened. In this case, the determination of whether a question is necessary becomes more accurate. Furthermore, for example, regarding change decisions, the system includes a machine learning unit that calculates predetermined inference features for predetermined model types and learns pairs of inference features and correct inference labels. In this case, it can learn whether the change decision is successful or not. Furthermore, for example, the machine learning unit calculates predetermined dialogue features for predetermined model types regarding dialogue judgment, and then learns pairs of dialogue features and correct dialogue labels. In this case, it can learn about the success or failure of change judgments. Furthermore, the system may include a display unit that shows, for example, the revised flight schedule, along with questions and answers. In this case, it can learn about the success or failure of the dialogue decision.

[0008] Furthermore, the present invention provides a method for supporting the creation of a revised operation plan when a predetermined operation plan is to be changed. This method involves a processor executing a program stored in memory to predict the future operation status of a mobile vehicle based on the predetermined operation plan and the operation history of the mobile vehicle. Based on the predicted operation status, it estimates a decision to change improvement areas, which are areas where changes are recommended. Based on the estimated decision to change improvement areas, it determines questions to supplement the information lacking in order to apply the decision to improvement areas, and changes the operation plan based on the answers to the questions. In this case, when interacting with the user, the method provides a way to support the creation of a revised operation plan that allows for narrowing down the improvement areas to those where the system has difficulty making a decision on change, and resolving these areas interactively while changing the operation plan.

[0009] Here, for example, a change judgment is used to classify whether or not dialogue about operational information is necessary, and if dialogue is necessary, it is decided to ask questions. In this case, the questions can be narrowed down when interacting with the user. Furthermore, for example, the estimated change judgment can be tested, trial results including the amount of change in the areas to be improved can be created, and the necessity of dialogue regarding operational information can be classified using the created trial results. In this case, the process of narrowing down the questions can be performed using the trial results. Furthermore, for example, feature quantities including the amount of change in the areas of improvement can be calculated from the trial results, and the confidence level of the change decision can be calculated based on these feature quantities to determine whether dialogue is necessary. In this case, the process of narrowing down the questions can be carried out using the confidence level of the change decision. Furthermore, for example, the content of the questions can be determined by identifying whether or not there is a lack of information from the trial results. In this case, the content of the questions can be determined based on the missing information. Then, depending on whether the trial results indicate a lack of information, if there is sufficient information regarding the necessity of a change decision, the question will include options for selecting the parameters necessary for the change decision. If there is insufficient information regarding the necessity of a change decision, the question will include options for accepting or rejecting the change decision. In this case, the content of the question can be determined to be more appropriate depending on the missing information. Furthermore, the necessity of asking questions is determined using at least one of the following results: confidence in the change decision, agreement with the user's decision, feasibility of implementing the user's decision, feasibility of improving the proposed quality, and feasibility of reducing processing time. In this case, the determination of whether or not to ask questions becomes more accurate. For example, regarding change decisions, predetermined inference features are calculated for predetermined model types, and pairs of inference features and correct inference labels are learned. In this case, it is possible to learn whether a change decision is successful or not. Furthermore, for dialogue judgment, predetermined dialogue features are calculated for predetermined model types, and pairs of dialogue features and correct dialogue labels are further learned. In this case, it is possible to learn whether the dialogue judgment is successful or not.

[0010] Furthermore, the present invention provides an operation plan change support system that assists in creating a revised operation plan when a predetermined operation plan is changed, and an operation management system that controls multiple mobile units according to the operation plan. The operation plan change support system comprises an operation prediction unit that predicts the future operation status of mobile units based on a predetermined operation plan and the operation performance of the mobile units, a change estimation unit that estimates a change judgment for improvement locations, which are locations where changes are recommended, based on the predicted operation status, and a change dialogue unit that determines questions to supplement the information lacking in order to apply the change judgment to the improvement locations based on the estimated change judgment, and changes the operation plan based on the answers to the questions. In this case, when interacting with the user, the system can narrow down the improvement locations to those where it is difficult for the system to make a change judgment, and change the operation plan while interactively resolving these locations. [Effects of the Invention]

[0011] According to the present invention, when interacting with a user, it is possible to narrow down to the parts among the improvement points that are difficult for the system to judge for changes, and while interactively solving these parts, it is possible to provide an operation plan change support system, an operation plan change support method, and an operation system capable of changing an operation plan.

Brief Description of the Drawings

[0012] [Figure 1] It is a diagram showing the functional configuration of the operation system of this embodiment. [Figure 2] It is a diagram showing the data structure of the operation information data used in this embodiment. [Figure 3] It is a diagram showing the data structure of the operation information data used in this embodiment. [Figure 4] It is a diagram showing the data structure of the improvement point data used in this embodiment. [Figure 5] It is a diagram showing the data structure of the trial result data used in this embodiment. [Figure 6] It is a diagram showing the data structure of the dialogue content data used in this embodiment. [Figure 7] It is a diagram showing the structure of the proposed information database used in this embodiment. [Figure 8] It is a diagram showing the structure of the inference information database used in this embodiment. [Figure 9] It is a diagram showing the structure of the dialogue information database used in this embodiment. [Figure 10] It is a flowchart explaining the operation of the operation plan change support system. [Figure 11] It is a flowchart explaining in detail the creation process of the operation plan change proposal of S1002 in FIG. 10. [Figure 12] It is a flowchart explaining in detail the process of trying to change the operation information of S1104 in FIG. 11. [Figure 13] This is a flowchart that details the process of changing the interactive operation information of S1105 in FIG. 11. [Figure 14] This is a flowchart that details the process of determining whether interactive communication is necessary for S1301 in FIG. 13. [Figure 15] This is a flowchart that details the process of determining the content of the question for S1303 in FIG. 13. [Figure 16] This is a flowchart that details the process of learning the change determination for S1005 in FIG. 10. [Figure 17] This is a flowchart that details the process of learning the interactive content for S1008 in FIG. 10. [Figure 18] This is a diagram showing the display content in the interactive operation plan change support on the display unit.

Embodiments for Carrying Out the Invention

[0013] Hereinafter, embodiments of the present invention will be described with reference to the drawings. The following description and drawings are examples for explaining the present invention, and for the sake of clarity of explanation, appropriate omissions and simplifications have been made. The present invention can be implemented in various other forms. Unless otherwise limited, each component may be singular or plural.

[0014] The positions, sizes, shapes, ranges, etc. of the respective components shown in the drawings may not represent the actual positions, sizes, shapes, ranges, etc. in order to facilitate understanding of the invention. For this reason, the present invention is not necessarily limited to the positions, sizes, shapes, ranges, etc. disclosed in the drawings.

[0015] While various types of information are sometimes described using terms such as "table" and "list," these types of information may also be represented using data structures other than these. To indicate independence from data structures, terms like "XX table" and "XX list" are sometimes referred to as "XX information" or "XX data." When describing identification information, if terms such as "identification information," "identifier," "name," "ID," and "number" are used, these terms are interchangeable.

[0016] When there are multiple components with the same or similar function, they may be described using the same symbol but with different subscripts. However, if it is not necessary to distinguish between these multiple components, the subscripts may be omitted in the description.

[0017] While the process performed by executing a program may be described, the processor (e.g., CPU (Central Processing Unit), GPU (Graphics Processing Unit)) executes the defined process, using memory resources (e.g., memory) and / or interface devices (e.g., communication ports) as appropriate; therefore, the processor may be the main entity performing the process. Similarly, the main entity performing the process by executing a program may be a controller, device, system, computer, or node that has a processor. The main entity performing the process by executing a program may be an arithmetic unit, and may include dedicated circuits that perform specific processing (e.g., FPGA (Field-Programmable Gate Array) or ASIC (Application Specific Integrated Circuit)).

[0018] A program may be installed from its program source into a device such as a computer. The program source may be, for example, a program distribution server or a computer-readable storage medium. If the program source is a program distribution server, the program distribution server includes a processor and storage resources for storing the program to be distributed, and the processor of the program distribution server may distribute the program to other computers. Furthermore, in the following description, two or more programs may be implemented as a single program, or one program may be implemented as two or more programs.

[0019] Furthermore, while terms such as inference model, dialogue model, constraints, and evaluation metrics are listed below, these terms are not limited to the meanings given herein and may include other meanings.

[0020] <Description of the functional configuration of the operating system 1> Figure 1 is a diagram showing the functional configuration of the operation system 1 of this embodiment. The illustrated operation system 1 comprises an operation plan change support system 2 and an operation management system 3. The Operation Plan Change Support System 2 is a device that assists in creating a revised operation plan when a predetermined operation plan for a moving object is changed. The following explanation will focus on the case where the moving object is a railway train.

[0021] The train operation management system 3 is connected to multiple trains via a communication network 4, enabling them to communicate with each other. The train operation management system 3 controls multiple trains within the train operation network under its management according to operation information, which is information that defines the operation of each train. The train operation management system 3 transmits various information related to operation management to the operation plan change support system 2 via the communication network 4. The train operation management system 3 receives the operation information changed by the operation plan change support system 2 via the communication network 4 and controls multiple trains within the train operation network according to the changed operation information. When the train operation management system 3 receives the changed operation information, it may request the user to confirm the changed operation information and reflect the approved changes in the control information, or it may automatically reflect the changes in the control information without requesting the user to confirm the changed operation information.

[0022] The operation plan change support system 2 is a computer device and includes a calculation processing unit 10, a main memory 20, an input unit 30, a display unit 40, a storage unit 50, and a communication unit 60. The arithmetic processing unit 10 is a processing unit that executes various software programs stored in the memory unit 50. The arithmetic processing unit 10 is, for example, a processor such as a CPU (Central Processing Unit). The main memory 20 is a memory device that serves as the working area for the arithmetic processing unit 10. The main memory 20 is, for example, RAM (Random Access Memory). The input unit 30 is a device operated by the operator of the operation plan change support system for inputting commands and data. The input unit 30 can be, for example, a mouse, keyboard, keypad, or buttons. The display unit 40 is a display device that outputs information from the operation plan change support system, such as operation information and dialogue content. The display unit 40 is, for example, a liquid crystal display or an organic EL display. The memory unit 50 is a storage device that stores various programs executed by the arithmetic processing unit 10 and various data used by the arithmetic processing unit 10 for processing. The memory unit 50 is an auxiliary storage device such as an HDD (Hard Disk Drive), SSD (Solid State Drive), or ROM (Read Only Memory). The communication unit 60 is connected to the communication network 4 and is a device that enables communication between the operation plan change support system 2 and the operation management system 3. The communication unit 60 is, for example, a communication interface (I / F).

[0023] The memory unit 50 stores software programs including a plan change proposal creation program 501, an operation prediction program 502, an operation control program 503, a machine learning program 504, a change decision proposal program 505, a change decision inference program 506, a change decision trial program 507, and a change decision dialogue program 508. The arithmetic processing unit 10 loads these software programs into the main memory 20 and executes them to realize each of these functions.

[0024] The plan change proposal creation program 501 is a software program that enables the creation of plan change proposals. The operation prediction program 502 is a software program that realizes operation predictions. As will be described in more detail later, the operation prediction program 502 functions as an operation prediction unit that predicts the future operation status of a mobile vehicle based on a predetermined operation plan and the operational history of the mobile vehicle. The operation control program 503 is a software program that implements operation control. The machine learning program 504 is a software program that implements machine learning. As will be explained in more detail later, the machine learning program 504 functions as a machine learning unit that learns how to make changes and learns how to interpret dialogue. The change decision proposal program 505 is a software program that implements the change decision proposal. The change decision inference program 506 is a software program that implements change decision inference. As will be described in more detail later, the change decision inference program 506 functions as a change estimation unit that estimates change decisions for improvement areas, which are areas where changes are recommended, based on predicted operating conditions. The change judgment trial program 507 is a software program that implements the trial of change judgment. As will be described in more detail later, the change judgment trial program 507 functions as a change trial unit that tests the estimated change judgment and creates trial results including the amount of change in the improvement area. The change decision dialogue program 508 is a software program that enables dialogue for change decisions. As will be explained in more detail later, the change decision dialogue program 508 functions as a change dialogue unit that determines questions to supplement the information missing when applying the change decision to the improvement area, based on the estimated change decision, and modifies the operation plan based on the answers to the questions.

[0025] Furthermore, the memory unit 50 stores operation plan data 511, operation performance data 512, proposed plan changes data 513, improvement point data 514, trial result data 515, dialogue content data 516, proposal information database 517, inference information database 518, and dialogue information database 519. The software program uses this data to perform processing.

[0026] Operation plan data 511 is data that shows predetermined operation information (and operational information). This can also be described as a predetermined operation plan. More specifically, it can be described as the operation plan for each train car of the railway, as defined in a timetable. Operational performance data 512 is data that shows actual operational information (and operational information). This can also be said to be the current operational performance of each railway vehicle. In other words, it is the operational performance of each train when a disruption occurs in train operations. Plan change data 513 is data showing the operational information (and operational information) of the revised plan. This can also be described as an operational plan that takes place when train cancellations or adjustments to departure intervals are made to improve train operations in the event of disruptions. Improvement data 514 indicates areas for improvement in operational information. For example, this refers to areas targeted for improvement when disruptions occur in train operations. Trial result data 515 shows the trial results for operational information. These trial results will be discussed in more detail later. Dialogue content data 516 is data showing the content of the dialogue regarding the operation information. The details of this dialogue content will be described later. The proposal information database 517 is a database that stores data used or calculated during the proposal process. Details about the proposal content will be described later. The inference information database 518 is a database that stores data used or calculated during inference. The details of the inference process will be described later. The dialogue information database 519 is a database that stores data used or calculated during dialogue.

[0027] <Explanation of data structure> Figure 2 shows the data structure of the operation information data used in this embodiment. The operational data shown includes train number, station name, platform number, arrival time, and departure time. The train number is the train number of the target train. For example, Figure 2 shows part of the description for trains with the train numbers "HR101" and "HR201". The station name refers to the station name related to the arrival, departure, and passing of the train in question. For example, the first row of Figure 2 describes stations where the "station name" is "St.A", and the second row of Figure 2 describes stations where the "station name" is "St.B". Track number refers to the name of the track used for the arrival, departure, and passing of the train in question. For example, the first and second rows of Figure 2 describe the track number being "Track 1". The arrival time is the time when the train arrives at the station. If there is no arrival time, such as at the starting station or a station the train passes through, a marker indicating that there is no data is entered, such as "-". For example, the first line of Figure 2 describes the case where there is no "arrival time", and the second line of Figure 2 describes the case where the "arrival time" is "08:03". The departure time is the time when the train departs from the station. If there is no departure time, such as at the terminal station, a marker indicating that there is no data is entered, such as "-". For example, the first row of Figure 2 describes the case where the departure time is "08:00", and the second row of Figure 2 describes the case where the arrival time is "08:05".

[0028] According to this train operation data, train "HR101" departs from platform "Track 1" at station "St.A" at "08:00" (first line). Train "HR101" arrives at station "St.W" on track "Track 1" at "09:42" and departs at "09:45" (line 5).

[0029] Hereafter, the term "XX Operation Information" may be used to describe the characteristics of the operation information. When explaining operation information, if expressions such as "Operation Information Before Change" or "Operation Information After Change" are used, these data structures should be considered to include the above items. However, additional items may be added to the Operation Information After Change, such as information that can be calculated at the time of the change.

[0030] Figure 3 shows the data structure of the operational information data used in this embodiment. Operational information data includes the operation number, train number, preceding train, and following train. The operation number is the operation number of the target operation. Trains belonging to the same operation number are operated by the same vehicle (or train set). The train number is the train number of the target train. The preceding train is the train number of the train that runs immediately before the target train. A "preceding train" refers to a train that uses the same rolling stock (or train set) as the target train and runs immediately before it. If there is no preceding train, a marker indicating that there is no data will be entered, such as "-". The "following train" is the train number of the train that runs immediately after the target train. A "following train" is a train that uses the same rolling stock (or train set) as the target train and runs immediately after it. If there is no following train, a marker indicating that there is no data will be entered, such as "-".

[0031] According to this operational data, there is no preceding train for train number "1", "HR101". Furthermore, the following train, "HR102", uses the same rolling stock and runs immediately afterward (first line). Furthermore, for train "HR102," which has operation number "1," the preceding train "HR101" is the train that runs immediately before it using the same rolling stock, and the following train "HR103" is the train that runs immediately after it using the same rolling stock (second line).

[0032] Here, the operational information in Figure 2 and the operational information in Figure 3 are described separately, but operational information may be included within the operational information. In the following explanation, even if both operational and operational information are included, it may simply be referred to as operational information. Here, we will consider what is referred to as "XX operation information" to also include operational information. For example, even when expressions such as "operation information before change" or "operation information after change" are used, we may explain them as including operational information.

[0033] Figure 4 shows the data structure of the improvement location data 514 used in this embodiment. The improvement data 514 includes the target train, station name, arrival / departure times, train type, scale, and the train that caused the problem. The ID is the identification information for the area that needs improvement. The target train is the train number of the train that is affected by the improvement in question. The station name is the name of the station related to the area being improved. Arrivals and departures are events related to the improvement in question. In this case, in addition to events related to the arrival and departure of trains, events related to the splitting and merging of carriages may also be included. The "Type" refers to the type of improvement needed. Specifically, this includes delays, conflicts, and operational issues (improvements are needed when train operation information is not correctly determined). Conflicts may include timetable irregularities such as reversed train arrival and departure times, or incorrect train operations. The scale refers to the size of the area requiring improvement. Specifically, this includes the amount of delay, the number of areas where competition is involved, and the number of trains whose operation has not yet been determined. The offending train is the train number of the train that caused the issue related to the improvement in question. If the offending train cannot be identified, a marker indicating that there is no data will be entered, such as "-".

[0034] According to this improvement data 514, the "Arrival" event of train "HR301" with ID "1" at station "St.M" has an improvement of a "Delay" of "10" minutes, and the train causing this is "HR511" (first line). Additionally, the departure event at station "St.L" for train "HR401" with ID "2" has an improvement where the delay is set to "20" minutes, and the train causing this is "HR301" (second line). Furthermore, the "Arrival" event at station "St.O" for train "HR501" with ID "3" has an improvement: there are "2" "conflicts," and the train causing these is "HR601" (3rd line).

[0035] Figure 5 shows the data structure of the trial result data 515 used in this embodiment. The trial result data 515 includes areas of improvement, change decisions, before change, after change, conflicts, delays, and evaluation values. The ID is the identification information for the trial result. The improvement location is the improvement location ID related to the trial result. A change decision is a change decision related to the results of the trial. A change decision is the type of change decision made within the operational information when modifying the operational plan. The "before change" section shows the operational information prior to the change based on the trial results. The revised information reflects the results of the trial. Conflict is the change in the result of the trial. Conflicts include, for example, tracks and platform numbers. The items may be defined separately depending on which resource the conflict relates to. Furthermore, conflicts may include timetable irrationalities, such as reversals in train arrival and departure times, or incorrectly defined train operations. Delay is the change in delay in the trial result. For example, delay could be the change in the delay of an arrival or departure. Note that delay may be defined separately depending on which event the delay relates to. The evaluation value is the change in the evaluation value in the trial results. The evaluation value is calculated based on criteria such as the delay time and location of competition, as well as the number of canceled trains and departure intervals. The evaluation value may be defined as a weighted linear sum of the evaluation indicators, or it may be defined separately for each evaluation indicator.

[0036] According to this trial result data 515, the trial result with ID "1" relates to the change decision "change order" for improvement area "1". The operational information before the trial change was "A1", the operational information after the trial change was "A9", the amount of change in conflict was "4", the amount of delay changed was "20" minutes, and the amount of change in evaluation value was "-100" (first row). The trial result with ID "7" relates to the change decision "Track change" for improvement area "10". The operational information before the trial change is "P1", and the operational information after the trial change is "P2". The amount of conflict change is "1", the amount of delay change is "0" minutes, and the amount of evaluation value change is "-20" (5th line).

[0037] Here, competition, delay, and evaluation values ​​are defined as relative values ​​representing the changes in the trial results, but they may also be defined using absolute values.

[0038] Figure 6 shows the data structure of the dialogue content data 516 used in this embodiment. The dialogue content data 516 includes ID, improvement area, change decision, sender, recipient, and dialogue content. The ID is identification information for the content of the conversation. The improvement location is the improvement location ID related to the content of the dialogue. The decision to change is a decision related to the content of the dialogue in question. The sender is the sender of the content of the conversation. The recipient is the recipient of the content of the conversation. The content of the dialogue is the message information contained within that dialogue.

[0039] According to this dialogue data 516, in the dialogue with ID "1", regarding the change decision "Track number change" in improvement area "3", the message "The conflict between HR501 and HR601 at St.O is being resolved." is sent from the sender "System A" to the recipient "User A" (first line). Furthermore, in the conversation with ID "1", regarding the change decision "track change" in improvement area "3", the message "Should we change the track used by HR601 at St.O?" is sent from the sender "System A" to the recipient "User A" (second line). Furthermore, in the conversation with ID "3", regarding the change decision "change track number" in improvement area "3", the message "Yes" is sent from the sender "User A" to the recipient "System A" (3rd line).

[0040] Figure 7 shows the structure of the proposal information database 517 used in this embodiment. The proposal information database 517 includes proposal management data 701, proposal configuration data 702, proposal history data 703, proposal model 704, pre-change operation information 705, post-change operation information 706, improvement points 707, change points 708, constraint conditions 709, and evaluation indicators 710. Multiple copies of proposal model 704, pre-change operation information 705, post-change operation information 706, improvement points 707, change points 708, constraint conditions 709, and evaluation indicators 710 exist, and are distinguished by adding A, B, ... to the end of each. Proposal management data 701 is management information related to a proposal. In addition to information about the date and time when the proposal was implemented, it defines the relationships between various data included in the proposal information database 517. For example, proposal management data 701 contains information such as which proposal model 704 took which pre-change operation information 705 as input and which post-change operation information 706 as output. Proposal configuration data 702 contains configuration information related to the proposal. It defines information such as thresholds and parameters to be used when implementing the proposal. Proposal history data 703 contains historical information about the proposal. It includes log information recorded when the proposal was implemented. Proposed model 704 is a model for implementing the proposal. It proposes changes 708 to the improvements 707 to the operational information. Proposed model 704 may be defined for a single change, or it may be defined by combining multiple changes. The pre-change operation information 705 is the pre-change operation information that will be used as input for proposed model 704. The revised operation information 706 is the revised operation information that will be output from the proposed model 704.

[0041] Improvement points 707 are areas that are processed when the proposed improvements are implemented. In other words, when the pre-defined operation plan (pre-change operation information 705) is changed due to a disruption in the schedule, etc., and when the revised operation plan (revised operation information 706) is created, these are the areas that are recommended to be changed in the pre-defined operation plan. Improvement points 707 are defined for both the pre-change operation information 705 and the revised operation information 706. Improvement points 707 may be defined one for each of the pre-change operation information 705 and the revised operation information 706, or multiple points may be defined together. Change 708 is a change adopted when the proposal is implemented. Change 708 may be defined individually for each change 708 adopted during the transition from the pre-change operation information 705 to the post-change information, or multiple change 708 may be defined together.

[0042] Constraint 709 is a constraint that will be referenced when the proposal is implemented. Constraint 709 includes facility conditions related to the route and turnaround, and time conditions related to driving time and stopping time. Of these, the facility conditions related to the route are conditions related to the arrangement of stations for each line and for each direction of travel, and the tracks and platforms to be used. The facility conditions related to turnaround are conditions related to stations where trains can turn around. Furthermore, the time conditions related to driving time are conditions related to the minimum time required for a train to travel between stations, from the departure of the first station to the arrival of the second station. Furthermore, the time conditions related to stopping time are conditions related to the minimum time required for a train to stop at a station, from the arrival of the train to the departure of the station. However, constraint 709 is not limited to these. For example, as a facility condition related to turnaround, a condition may be included regarding the minimum time required between the arrival of the first train and the departure of the second train when two trains turn around at a station. Also, as a continuation interval condition, a condition may be included regarding the minimum time required between the departure of the first train and the arrival of the second train at the station (or before the second train departs the station) when two trains depart from or arrive at a station. Furthermore, as an obstruction condition, a condition may be included regarding how long a train will be unable to use a certain facility.

[0043] Evaluation indicator 710 is an evaluation indicator that will be referenced when implementing the proposal. Examples of evaluation indicator 710 include the number of canceled trains, the number of delayed trains, the delay time for departures, the delay time for arrivals, and the number of changes 708. Evaluation indicator 710 may also be calculated as a representative statistic for each indicator (sum, mean, variance, standard deviation, median, maximum value, minimum value, ratio, etc.). However, evaluation indicator 710 is not limited to these.

[0044] Figure 8 shows the structure of the inference information database 518 used in this embodiment. The inference information database 518 includes inference management data 801, inference configuration data 802, inference history data 803, inference model 804, inference features 805, inference labels 806, change judgment 807, judgment probability 808, judgment acceptance / rejection 809, and teacher judgment 810. Multiple copies of the inference model 804, inference features 805, inference labels 806, change judgment 807, judgment probability 808, judgment acceptance / rejection 809, and teacher judgment 810 exist, and they are distinguished by adding A, B, ... to the end of each.

[0045] Inference management data 801 is management information related to inference. In addition to information about the date and time when the inference is performed, inference management data 801 defines the relationships between various data contained in the inference information database 518. Inference configuration data 802 is configuration information related to inference. Inference configuration data 802 defines information such as thresholds and parameters used when performing inference. Inference history data 803 is historical information related to inference. Inference history data 803 contains log information recorded when inference is performed. The inference model 804 is a model that performs inference. The inference model 804 infers a change decision 807 for the improvement points 707 of the operational information. The inference model 804 may be defined for a single change, or it may be defined by combining multiple changes. Inference feature 805 is the feature that serves as input to inference model 804. Inference label 806 is the label that will be output by inference model 804. Change judgment 807 is a change judgment that is applied when the inference is performed. The decision probability of 808 is the decision probability calculated when the inference is performed. The decision of acceptance or rejection (809) is determined after the inference has been made. The teacher's decision 810 is a change decision that serves as the teacher during the training of the inference model 804.

[0046] Figure 9 shows the structure of the dialogue information database 519 used in this embodiment. The dialogue information database 519 includes dialogue management data 901, dialogue configuration data 902, dialogue history data 903, dialogue model 904, dialogue features 905, dialogue labels 906, user information 907, question content 908, answer content 909, user judgment 910, system judgment 911, proposal quality contribution 912, and processing time contribution 913. Multiple instances of dialogue model 904, dialogue features 905, dialogue labels 906, user information 907, question content 908, answer content 909, user judgment 910, system judgment 911, proposal quality contribution 912, and processing time contribution 913 exist, and are distinguished by adding A, B, ... to the end of each.

[0047] Dialogue management data 901 is management information related to the dialogue. In addition to information about the date and time when the dialogue took place, dialogue management data 901 defines the relationships between various data contained in the dialogue information database 519. Dialogue configuration data 902 is configuration information related to the dialogue. Dialogue configuration data 902 defines information such as thresholds and parameters used when conducting the dialogue. Dialogue history data 903 is historical information related to the dialogue. Dialogue history data 903 contains log information recorded when the dialogue takes place. Dialogue model 904 is a model for conducting dialogue. Dialogue model 904 determines whether dialogue is necessary regarding improvement points 707 in the operational information. Dialogue model 904 may be defined for a single change, or it may be defined by combining multiple changes. Dialogue feature 905 is a feature that serves as input to dialogue model 904. Dialogue label 906 is the label that will be output by dialogue model 904.

[0048] User information 907 is the user information of the user operating the system at the time of the interaction. Question 908 is the question that will be sent during the dialogue. Response 909 is the response received during the dialogue. User Decision 910 is a decision made by the user that is recorded during the interaction. User Decision 910 is used to create a correct label when learning whether or not there is a match with User Decision 910 during a change. It is also used to create a correct label when learning whether or not User Decision 910 can be implemented. System judgment 911 is a judgment made by the system during the interaction. System judgment 911 is used to create a correct label when learning whether or not it matches user judgment 910. The Proposal Quality Contribution 912 is the contribution to proposal quality recorded during the dialogue. The Proposal Quality Contribution 912 is used to create the correct label when learning whether proposal quality can be improved. The processing time contribution 913 is the processing time contribution recorded when the dialogue is performed. The processing time contribution 913 is used to create the correct label when learning whether or not the processing time can be reduced.

[0049] Question content 908, answer content 909, user judgment 910, system judgment 911, proposal quality contribution 912, and processing time contribution 913 may be created and stored in advance, not just during system operation.

[0050] <Explanation of Operation of the Operation Plan Change Support System 2> Figure 10 is a flowchart illustrating the operation of the train schedule change support system 2. First, the operation plan change support system 2 determines whether or not a change to the operation plan has been requested (S1001). As a result, if a change in the operation plan is requested (Yes in S1001), proceed to S1002; if no change in the operation plan is requested (No in S1001), proceed to S1004. In S1002, the operation plan change support system 2 creates a proposed operation plan change (S1002). The proposed operation plan change is a draft of the revised operation plan (revised operation information 706) when the pre-determined operation plan (original operation information 705) is changed due to a disruption in the schedule or other reasons. The specific details of this process will be described later in Figure 11. Furthermore, the operation plan change support system 2 displays a proposed change to the operation plan to the user (S1003).

[0051] Next, the operation plan change support system 2 determines whether or not the user has requested learning of the change judgment 807 (S1004). As a result, if learning the change decision 807 is required (Yes in S1004), proceed to S1005; otherwise, proceed to S1007. In S1005, the operation plan change support system 2 learns the change judgment 807 (S1005). The specific details of this process will be described later in Figure 16. Then, the operation plan change support system 2 displays the learning result of the change judgment 807 (S1006).

[0052] Next, the operation plan change support system 2 determines whether or not the user has requested learning of the dialogue content (S1007). As a result, if learning the dialogue content is required (Yes in S1007), the process proceeds to S1008; if learning the dialogue content is not required (No in S1007), the process proceeds to S1010. In S1008, the operation plan change support system 2 learns the content of the dialogue (S1008). The specific details of this process will be described later in Figure 17. Then, the operation plan change support system 2 displays the learning results of the dialogue (S1009).

[0053] Next, the operation plan change support system 2 determines whether or not a processing termination request exists (S1010). As a result, if a processing termination request exists (Yes in S1010), the series of processes is terminated; if no processing termination request exists (No in S1010), the process returns to S1001.

[0054] Figure 11 is a flowchart that provides a detailed explanation of the process for creating the proposed change to the operation plan in S1002 of Figure 10. First, the operation plan change support system 2 predicts and evaluates the operation based on the plan (S1101). This is a process that predicts and evaluates the operation of future trains based on a predetermined operation plan. For example, the operation plan change support system 2 uses the operation plan data 511, as well as the constraint conditions 709 and evaluation indicators 710 stored in the proposal information database 517, to predict and evaluate the operation of future trains. In this case, if an obstruction condition is defined in the constraint condition 709, the operation plan change support system 2 may predict the operation assuming that the train cannot run in the obstruction section and time indicated by the obstruction condition. Furthermore, if there is an arrival time or departure time in the operation performance data 512 (i.e., the train has already arrived or departed at the current time), the operation plan change support system 2 may predict the operation assuming that the train will run at the arrival time or departure time in the operation performance data 512 up to the current time. In addition, the operation plan change support system 2 may calculate evaluation values ​​for various indicators defined in the evaluation indicators 710 based on the predicted train operation. The evaluation indicator 710 includes, for example, the number of train cancellations and the departure intervals. Next, the operation plan change support system 2 identifies areas for improvement 707 (S1102). This is a process of identifying areas for improvement 707 based on the results of predicting and evaluating the operation. For example, the operation plan change support system 2 identifies areas of conflict or delays as areas for improvement 707.

[0055] Furthermore, the operation plan change support system 2 repeatedly processes all improvement points 707 identified in S1102 for the following S1003 to S1108. First, the operation plan change support system 2 infers a change decision 807 for the operation information (S1103). This is the process of inferring a change decision 807 for the operation information based on the improvement areas 707 predicted in S1102. Here, the inference of the change decision 807 can be performed using the inference model 804 stored in the inference information database 518. Next, the operation plan change support system 2 performs a trial change to the operation information (S1104). This is a process to test the change to the operation information. The specific details of this process will be described later in Figure 12. Furthermore, the operation plan change support system 2 performs interactive changes to operation information (S1105). The specific details of this process will be described later in Figure 13.

[0056] The operation plan change support system 2 then predicts and evaluates the operation based on the proposed change in the operation plan (S1106). This is a process of predicting and evaluating the operation of trains based on the proposed change. For example, the operation plan change support system 2 uses the proposed change in plan data 513, as well as the constraint conditions 709 and evaluation indicators 710 stored in the proposal information database 517, to predict and evaluate the operation of future trains. In this case, if an obstruction condition is defined in the constraint condition 709, the operation may be predicted assuming that the train cannot run in the obstruction section and time indicated by the obstruction condition. Also, if there is an arrival time or departure time in the operation performance data 512 (i.e., the train has already arrived or departed at the current time), the operation may be predicted assuming that the train will run at the arrival time or departure time in the operation performance data 512 up to the current time.

[0057] Next, the operation plan change support system 2 updates and displays the proposed operation plan change to the user (S1107). Then, the operation plan change support system 2 registers the proposed operation plan change in the proposal information database 517 (S1108). This is the process of registering the data used or calculated when the proposal was implemented in the proposal information database 517. Specifically, the operation plan change support system 2 registers data such as the proposed model 704, operation information before change 705, operation information after change 706, improvement points 707, change points 708, constraint conditions 709, and evaluation indicators 710 that were used or calculated when creating the proposed operation plan change in the proposal information database 517.

[0058] Figure 12 is a flowchart that provides a detailed explanation of the process of attempting to change the operation information in S1104 of Figure 11. First, the operation plan change support system 2 applies the inferred change decision 807 (S1201). For example, the operation plan change support system 2 applies the inferred change decision 807 to the target pre-change operation information 705 and creates the corresponding post-change operation information 706. Next, the train operation plan change support system 2 predicts and evaluates the application results (S1202). This is a process that predicts and evaluates train operations based on the application results of the change decision 807. For example, the train operation plan change support system 2 uses the modified operation information 706, as well as the constraint conditions 709 and evaluation indicators 710 stored in the proposed information database 517, to predict and evaluate future train operations. Furthermore, the operation plan change support system 2 determines whether or not the competition has increased (S1203). This is the process of determining whether or not the competition has increased. The operation plan change support system 2 determines whether or not the competition has increased by comparing the operation information before the change 705 and the operation information after the change 706. If the competition has increased, proceed to step S1204. If the competition has not increased, proceed to step S1208.

[0059] If the number of conflicts increases (Yes in S1203), the train schedule change support system 2 classifies the increased conflicts (S1204). For example, the train schedule change support system 2 classifies whether the conflicting resources are tracks or platform numbers. Next, the operation plan change support system 2 evaluates the increased complexity of the competition (S1205). Furthermore, the operation plan change support system 2 attempts to resolve the increased conflicts (S1206). The operation plan change support system 2 attempts to resolve the increased conflicts using the proposed model 704 and the inference model 804. At this time, the operation plan change support system 2 may update the modified operation information 706 based on the results of the attempt to resolve the conflicts, or it may define it as a different modified operation information 706.

[0060] On the other hand, if the competition has not increased in S1203 (No in S1203), the operation plan change support system 2 determines whether the delay has increased or not (S1207). This is the process of determining whether the delay has increased or not. The operation plan change support system 2 determines whether the delay has increased or not by comparing the operation information before the change 705 and the operation information after the change 706. If the delay has increased, proceed to step S1208. If the delay has not increased, proceed to step S1210.

[0061] If the delay is increasing (Yes in S1207), the operation plan change support system 2 classifies the increased delay (S1208). This is the process of classifying the increased delay. The operation plan change support system 2 classifies them, for example, as arrival delays or departure delays. Alternatively, the operation plan change support system 2 may classify them by constraint conditions 709 that cause delays, such as obstruction conditions or continuation interval conditions.

[0062] Next, the operation plan change support system 2 attempts to resolve the increased delay (S1209). This is a process that attempts to resolve the increased delay. The operation plan change support system 2 attempts to resolve the increased delay using the proposed model 704 and the inference model 804. At this time, the operation plan change support system 2 may update the modified operation information 706 based on the result of the attempt to resolve the delay, or it may define a different modified operation information 706.

[0063] On the other hand, if the delay has not increased in S1208 (No in S1207), the operation plan change support system 2 calculates the change in the evaluation value (S1210). This is the process of calculating the change in the evaluation value. For each evaluation index 710, the operation plan change support system 2 calculates the evaluation value of the operation information before the change 705 and the evaluation value of the operation information after the change 706, respectively, and calculates the change.

[0064] Next, the operation plan change support system 2 attempts to improve the evaluation value (S1211). This is a process to attempt to improve the evaluation value. The operation plan change support system 2 attempts to improve the evaluation value using the proposed model 704 and the inference model 804. At this time, the operation plan change support system 2 may update the modified operation information 706 based on the results of the attempt to improve the evaluation value, or it may define it as a different modified operation information 706. In S1206, S1209, and S1211, the operation plan change support system 2 also performs a trial of the estimated change decision 807 and creates a trial result that includes the amount of change in the improvement area 707.

[0065] After S1206, S1209, and S1211, the Operation Plan Change Support System 2 classifies the trial results (S1212). This is the process of classifying the trial results from S1206, S1209, and S1211. The Operation Plan Change Support System 2 classifies the results based on items such as whether the increased conflicts or delays due to the trial were resolved, and whether the evaluation values ​​improved.

[0066] Figure 13 is a flowchart that provides a detailed explanation of the process for changing the operation information interactively, as shown in S1105 of Figure 11. First, the operation plan change support system 2 determines whether dialogue is necessary (S1301). The specific details of this process will be described later in Figure 14. Next, the operation plan change support system 2 determines whether or not dialogue is necessary (S1302). This is the process of determining whether or not dialogue is required. If dialogue is required, the system proceeds to step S1303. Conversely, if dialogue is not required, the series of processes ends.

[0067] If dialogue is required (Yes in S1302), the operation plan change support system 2 determines the content of question 908 (S1303). This is the process of determining the content of question 908 related to the change decision 807. In other words, the operation plan change support system 2 decides to ask questions when dialogue is required. The specific details of this process will be described later in Figure 15. Next, the operation plan change support system 2 presents a question (S1304). This is the process of presenting the user with a question related to the change decision 807. The operation plan change support system 2 presents the question to the user, for example, by transmitting the question content 908 determined in S1303 to the display unit 40. Furthermore, the operation plan change support system 2 waits for a response (S1305). This is the process of waiting for a response regarding the change decision 807. In other words, the operation plan change support system 2 waits until it receives the user's response content 909 to the question content 908 presented in S1304 from the input unit 30. When the user's response content 909 is received, the system proceeds to the next step S1306. If the user's response content 909 has not been received after a predetermined time has elapsed, the operation plan change support system 2 may resend a confirmation message, or it may proceed with processing assuming that the predetermined response content 909 for when no response was received has been received.

[0068] When response 909 is received, the operation plan change support system 2 changes the operation information based on response 909 (S1306). This is a process of changing the operation information based on response 909 regarding the change decision 807. Furthermore, the operation plan change support system 2 registers the data in the dialogue information database 519 (S1307). This is the process of registering the data used or calculated during the dialogue in the dialogue information database 519. The operation plan change support system 2 also registers data such as the dialogue model 904, dialogue features 905, dialogue labels 906, user information 907, question content 908, answer content 909, user judgment 910, system judgment 911, proposal quality contribution 912, and processing time contribution 913, which were used or calculated when the system interacts with the user, in the proposal information database 517.

[0069] Figure 14 is a flowchart that provides a detailed explanation of the process for determining whether or not dialogue is necessary in S1301 of Figure 13. First, the train schedule change support system 2 calculates features from the trial results (S1401). That is, the train schedule change support system 2 calculates dialogue features 905 related to the dialogue model 904. The train schedule change support system 2 calculates dialogue features 905 for which items have been defined in advance for multiple model types used in the processing described later. In this case, the train schedule change support system 2 calculates dialogue features 905 common to the model types, such as the current time and the type of train. Alternatively, as described later, the train schedule change support system 2 may calculate different dialogue features 905 for each model type. Note that the dialogue features 905 may be calculated from the pre-change operation information 705 and post-change operation information 706 shown in the trial result data 515. Furthermore, the train schedule change support system 2 may calculate dialogue features 905 using data contained in the proposal information database 517, the inference information database 518, and the dialogue information database 519.

[0070] Next, the operation plan change support system 2 classifies whether the change decision 807 is certain or not (S1402). The operation plan change support system 2 uses the dialogue model 904 that classifies whether the change decision 807 is certain or not to classify whether the inference model 804 is certain of the change decision 807. For example, the operation plan change support system 2 classifies whether the inference model 804 is certain of the change decision 807 by taking the dialogue feature 905 as input to the dialogue model 904 that classifies whether the change decision 807 is certain or not, and outputting the dialogue label 906. Here, the dialogue feature 905 of the dialogue model 904 that classifies whether the change decision 807 is certain or not may include, for example, the decision probability 808 or the features that the inference model 804 has learned. Furthermore, for the dialogue label 906 of the dialogue model 904 that classifies the certainty of the change decision 807, labels may be defined as follows: (1) the inference model 804 is certain of the change decision 807, and (0) the inference model 804 is not certain of the change decision 807. By classifying the certainty of the change decision 807, it is expected that questions about the change decision 807 that the system is certain to implement can be avoided, and the content of the dialogue regarding the improvement area 707 can be narrowed down.

[0071] Furthermore, the operation plan change support system 2 classifies whether or not it matches the user judgment 910 (S1403). The operation plan change support system 2 uses a dialogue model 904 that classifies whether or not it matches the user judgment 910 to classify whether or not the user judgment 910 and the system judgment 911 match. For example, the operation plan change support system 2 classifies whether or not the user judgment 910 and the system judgment 911 match by taking dialogue features 905 as input to the dialogue model 904 that classifies whether or not it matches the user judgment 910 and outputting a dialogue label 906. Here, the dialogue features 905 of the dialogue model 904 that classifies whether or not it matches the user judgment 910 may include, for example, the departure delay time and the departure time interval between trains. Furthermore, for the dialogue label 906 of the dialogue model 904 that classifies whether or not there is a match with the user judgment 910, labels may be defined as (1) when the user judgment 910 and the system judgment 911 match, and (0) when the user judgment 910 and the system judgment 911 do not match. By classifying whether or not there is a match with the user judgment 910, it is expected that questions about change judgments 807 whose content matches between the user and the system can be avoided, and the content of the dialogue regarding the improvement area 707 can be narrowed down.

[0072] The operation plan change support system 2 then classifies whether the user decision 910 can be implemented (S1404). The operation plan change support system 2 uses a dialogue model 904 that classifies whether the user decision 910 can be implemented to classify whether the user decision can be implemented. For example, the operation plan change support system 2 takes dialogue features 905 as input to the dialogue model 904 that classifies whether the user decision 910 can be implemented and outputs a dialogue label 906 to classify whether the user decision can be implemented. Here, the dialogue features 905 of the dialogue model 904 that classifies whether the user decision 910 can be implemented may include, for example, the type of change decision 807 or the time of the event indicated by the improvement location 707. Also, for the dialogue label 906 of the dialogue model 904 that classifies whether the user decision 910 can be implemented, labels such as "the user decision can be implemented" (1) and "the user decision cannot be implemented" (0) may be defined. By classifying whether user decisions 910 can be implemented, it is expected that questions about change decisions 807, which are difficult for users to implement, can be avoided, and the content of the dialogue regarding improvement areas 707 can be narrowed down.

[0073] Next, the operation plan change support system 2 uses a dialogue model 904 that classifies whether the proposed quality can be improved to classify whether the proposed quality will be improved by accepting or rejecting the change decision 807 (S1405). For example, the operation plan change support system 2 takes dialogue features 905 as input to the dialogue model 904 that classifies whether the proposed quality can be improved and outputs a dialogue label 906 to classify whether the proposed quality will be improved by accepting or rejecting the change decision 807. Here, the dialogue features 905 of the dialogue model 904 that classifies whether the proposed quality can be improved may include, for example, the type of competition and various evaluation values. Also, for the dialogue label 906 of the dialogue model 904 that classifies whether the proposed quality can be improved, labels such as "proposed quality is improveable" (1) and "proposed quality is not improveable" (0) may be defined. By classifying whether the proposed quality can be improved, it is expected that questions about the change decision 807, which is unlikely to affect the proposed quality, can be avoided and the content of the dialogue regarding the improvement area 707 can be narrowed down.

[0074] Furthermore, the operation plan change support system 2 classifies whether processing time can be shortened (S1406). The operation plan change support system 2 uses a dialogue model 904 that classifies whether processing time can be shortened to classify whether processing time can be shortened or not based on the acceptance or rejection of the change decision 807. For example, the operation plan change support system 2 takes dialogue features 905 as input to the dialogue model 904 that classifies whether processing time can be improved and outputs dialogue labels 906 to classify whether processing time can be shortened or not based on the acceptance or rejection of the change decision 807. Here, the dialogue features 905 of the dialogue model 904 that classifies whether processing time can be shortened may include, for example, the complexity of the competition or the number of delayed trains. Also, for the dialogue labels 906 of the dialogue model 904 that classifies whether processing time can be shortened, labels such as processing time can be shortened (1) and processing time cannot be shortened (0) may be defined. By classifying whether processing time can be shortened or not, it is expected that questions about change decisions 807, which are unlikely to affect processing time, can be avoided, and the content of the dialogue regarding improvement areas 707 can be narrowed down.

[0075] Then, the operation plan change support system 2 determines whether questions are necessary (S1407). This is the process of determining whether questions are necessary regarding the change decision 807. The operation plan change support system 2 determines whether questions are necessary using the classification results of whether the change decision 807 is certain or not, whether it matches the user decision 910, whether the user decision 910 can be implemented or not, whether the proposed quality can be improved or not, and whether the processing time can be shortened or not. In this case, it can also be said that the operation plan change support system 2 determines whether questions are necessary using at least one of the following results: whether the change decision 807 is certain or not, whether it matches the user decision 910, whether the user decision 910 can be implemented or not, whether the proposed quality can be improved or not, and whether the processing time can be shortened or not. This makes the determination of whether questions are necessary more accurate. The operation plan change support system 2 may, for example, determine that questions are necessary if the classification result for whether or not there is agreement with user judgment 910 is "none", the classification result for whether or not user judgment 910 can be implemented is "yes", and the classification result for whether or not the proposed quality can be improved is "yes". Alternatively, the operation plan change support system 2 may determine that questions are unnecessary if the classification result for whether or not there is agreement with user judgment 910 is "none", the classification result for whether or not user judgment 910 can be implemented is "no", and the classification result for whether or not processing time can be shortened is "yes". Here, when determining whether or not questions are necessary, the combination of the classification result for whether or not there is certainty in the change judgment 807, the classification result for whether or not there is agreement with user judgment 910, the classification result for whether or not user judgment 910 can be implemented, the classification result for whether or not the proposed quality can be improved, and the classification result for whether or not processing time can be shortened is not limited to the combinations described above, and the combinations may be changed according to predetermined conditions. Note that the dialogue features 905 and dialogue labels 906 are not limited to the forms described above. Here, as shown in S1401, the operation plan change support system 2 uses the generated trial results to classify whether dialogue about operation information is necessary (in this case, whether questions are necessary). Also, as shown in S1402, the operation plan change support system 2 uses the change judgment to classify whether dialogue about operation information is necessary. Furthermore, according to the processing in S1401, S1402, and S1407, it can also be said that the operation plan change support system 2 calculates a feature quantity (in this case, dialogue feature quantity) that includes the amount of change in the improvement area from the trial results, and calculates the confidence level of the change judgment (in this case, confidence level) based on the feature quantity to determine whether dialogue is necessary.

[0076] Figure 15 is a flowchart that provides a detailed explanation of the process for determining the content of question 908 in S1303 of Figure 13. First, the operation plan change support system 2 identifies any deficiencies from the trial results (S1501). The operation plan change support system 2 extracts deficiency information using, for example, the pre-change operation information 705 and post-change operation information 706 shown in the trial result data 515, as well as related improvement locations 707, change locations 708, change decisions 807, and decision probabilities 808. For all change decisions 807 related to the improvement locations 707 being processed, the operation plan change support system 2 extracts deficiency information through the following procedures: (a) analysis of inference results, (b) analysis of trial results, and (c) determination of deficiencies. In (a) analysis of inference results, the confidence level for each change decision 807 is calculated using the decision probability 808 calculated by the inference model 804 and the dialogue label 906 and probability information calculated by the dialogue model 904. (b) In the analysis of trial results, the degree of improvement for each change decision 807 is calculated using classification information of trial results, such as whether the increased competition or delay due to the trial was resolved, or whether the evaluation value improved due to the trial, as well as the amount of change in competition, delay, and evaluation value. (c) In determining whether there is a deficiency, if the priority of each change decision 807, which is calculated by multiplying the degree of confidence by the degree of improvement, is smaller than a predetermined threshold, it is identified that there is a deficiency in the change decision 807.

[0077] Next, the operation plan change support system 2 classifies the missing information (S1502). The operation plan change support system 2 classifies the information into three classes: information on whether a change is necessary is not missing (1), information on the type of change is not missing (2), and information on the improvement location 707 is not missing (3). First, the initial value is set to "1". Next, if the priority of two or more change decisions 807 is greater than a predetermined threshold, the system detects that information on whether a change is necessary is missing and sets the classification result of the missing information to "2". However, if the priority of one of the change decisions 807 that meet the above conditions is greater than the priority of the remaining change decisions 807 by a predetermined value, the classification result of the missing information is set to "1". Finally, if the priority of all change decisions 807 is less than a predetermined threshold, the system detects that information on the type of change is missing and sets the classification result of the missing information to "3". However, if the priority of one of the change decisions 807 that meet the above conditions is greater than the priority of the remaining change decisions 807 by a predetermined value, the classification result of the missing information is set to "2" (S1503, S1507). In practice, if there is no missing information regarding the necessity of changes (1), and the classification result of the missing information is "1", proceed to S1504. Also, if there is no missing information regarding the type of change (2), and the classification result of the missing information is "2", proceed to S1508. Furthermore, if there is no missing information regarding the improvement location 707 (3), and the classification result of the missing information is "3", proceed to S1510.

[0078] In S1504, the operation plan change support system 2 refers to the change type parameter. The operation plan change support system 2 obtains predetermined parameters according to the change type. In this case, the parameters for changing the order include, for example, the target train and the other train, as well as the starting station and ending station. The target train is the train that will be changed in order. The other train is the train that will serve as the basis for the order change. The starting station is the name of the station where the order change begins. The ending station is the name of the station where the order change ends. For track changes, parameters include, for example, the train in question, the station in question, and the new track number. The train in question is the train that will be subject to the track change. The station in question is the name of the station to which the track change will occur. The new track number is the name of the track that the train will use at the station as a result of the track change.

[0079] Next, the operation plan change support system 2 identifies any uncertainties regarding the parameters (S1505). The operation plan change support system 2 identifies any uncertainties regarding the parameters using the trial results. At this time, if the trial results show that the change decision 807 resolves the conflict or delay or improves the evaluation value, it is determined that the parameters of that change decision 807 should be used, and the system considers there to be no uncertainties. On the other hand, if the trial results show that the change decision does not resolve the conflict or delay or improve the evaluation value, it is determined that it is better not to use the parameters of that change decision 807, and the system considers there to be uncertainties. For order changes, for example, if the conflict cannot be resolved as a result of changing the target train to run immediately after a certain other train, the parameters of the other train are extracted as uncertainties. At this time, the starting station and ending station may also be extracted as uncertainties. For track changes, for example, if the delay cannot be resolved as a result of changing the target train to a certain track, the parameters of the destination track are extracted as uncertainties. Even if the conflicts and delays in improvement area 707 are resolved, the overall conflicts and delays may increase, so it may be advisable to extract parameters based on the overall change. Furthermore, creating multiple trial results using multiple proposed models 704 may reduce the uncertainty regarding the parameters.

[0080] Then, the train schedule change support system 2 extracts input candidates for the unknown points (S1506). For changes in order, the train schedule change support system 2 extracts candidates for the opposing train and candidates for the starting and ending stations. For changes in track number, it extracts candidates for the destination track number.

[0081] In S1508, the operation plan change support system 2 refers to the candidate change types. Then, the operation plan change support system 2 identifies whether there are multiple candidates for the change type (S1509).

[0082] In S1510, the operation plan change support system 2 refers to improvement point 707. Then, the operation plan change support system 2 searches for the predetermined conditions (S1511).

[0083] After S1506, S1509, and S1511, proceed to S1512. In S1512, the operation plan change support system 2 generates question content 908. The operation plan change support system 2 generates question content 908 according to the classification result of the missing information. In this case, if there is no lack of information regarding whether a change is necessary (1), the system generates a question 908 that includes input candidate options for the unclear points extracted in S1506. In this case, the operation plan change support system 2 makes the question 908 an option that allows the user to select the parameters required in the change judgment 807 (in this case, the station and platform number required for each change judgment). If there is no missing information regarding the type of change (2), a question 908 is generated that includes options for accepting or rejecting one or more change decisions 807. In this case, if it is determined in S1509 that there are multiple candidate change types, a question 908 is generated that includes options for accepting or rejecting two or more change decisions 807. On the other hand, if it is determined in S1509 that there are not multiple candidate change types, a question 908 is generated that includes options for accepting or rejecting one change decision 807. If the information for improvement location 707 is sufficient (3), then a question 908 is generated that includes the information for improvement location 707 referenced in S1510 and the dialogue content indicated by the predetermined conditions searched in S1511. In S1512, the operation plan change support system 2 can determine a more appropriate question 908 depending on the missing information. Note that a large-scale language model may be used to generate question 908.

[0084] Although the above process was explained using examples of order changes and track changes, the change decisions 807 related to the determination of question 908 are not limited to these. For example, various change decisions 807 can be applied, such as service suspension or operational changes, without departing from the spirit of the present invention. In the process shown in Figure 15, the operation plan change support system 2 can also be described as determining whether there is insufficient information from the trial results, as shown in S1501, and then deciding on the content of the question.

[0085] Figure 16 is a flowchart that provides a detailed explanation of the learning process for the change decision 807 in S1005 of Figure 10. First, the operation plan change support system 2 references the data range for training (S1601). For example, the operation plan change support system 2 obtains the data range defined in the inference configuration data 802. Note that when referencing the data range, it may be divided into training data and evaluation data. Next, the operation plan change support system 2 refers to the model type for learning (S1602). The operation plan change support system 2 obtains, for example, the model type defined in the inference configuration data 802. The model type is defined in units of change decisions 807, such as order changes or track changes. In addition, multiple model types may be defined for the same change decision 807. For example, a model for determining whether or not to implement a track change, or a model for determining the destination track in a track change, may be defined. Note that the model types are not limited to these.

[0086] Furthermore, the operation plan change support system 2 references a learning algorithm (S1603). For example, the operation plan change support system 2 obtains an algorithm defined in the inference configuration data 802. Examples of algorithms that can be used include linear regression, logistic regression, random forest, gradient boosting decision tree, and neural network. However, the algorithms are not limited to these. The train operation plan change support system 2 then calculates inference features 805 (S1604). The train operation plan change support system 2 calculates inference features 805 for which items have been defined in advance for each model type, for example. The train operation plan change support system 2 calculates common inference features 805 for each model type, such as the current time and the type of train. It is also possible to calculate different inference features 805 for each model type. In this case, the inference features 805 for the inference model 804 that determines the order change may include, for example, the departure delay time and the departure time interval between trains. Similarly, the inference features 805 for the inference model 804 that determines the track change may include, for example, the arrival delay time and the track occupancy time. It is also important to note that the inference features 805 are not limited to these.

[0087] Next, the train schedule change support system 2 learns a set of inference features 805 and inference labels 806 (S1605). The train schedule change support system 2 constructs an inference model 804 using, for example, the set of inference features 805 and the correct inference labels 806. The method of defining the correct inference labels 806 may be changed for each model type. For the inference labels 806 of the model that determines whether or not to implement a track change, for example, labels such as implement a track change (1) and do not implement a track change (0) may be defined. For the inference labels 806 of the model that determines the destination track in a track change, for example, labels such as use the current track (0), change to track 1 (1), change to track 2 (2), and change to track 3 (3) may be defined.

[0088] Furthermore, the operation plan change support system 2 evaluates the inference model 804 (S1606). The operation plan change support system 2 evaluates the inference model 804, for example, using evaluation data. If evaluation data is not available, the inference model 804 is evaluated using training data. Then, the operation plan change support system 2 registers the data in the inference information database 518 (S1607). This is the process of registering the data used or calculated during the learning of the change judgment 807 in the inference information database 518.

[0089] Although the above process was explained by separating the model type for each change judgment 807, such as changing the order or track number, it is also possible to perform training using a model type that combines multiple change judgments 807. Furthermore, although the above process was explained using examples of sequence changes and track changes, the change judgments 807 to be learned are not limited to these. It can be applied to various change judgments 807, such as service suspension and operational changes, as long as it does not deviate from the spirit of the invention. In the process shown in Figure 16, the operation plan change support system 2 calculates predetermined inference features 805 for predetermined model types when making a change decision, and learns pairs of inference features 805 and correct inference labels 806.

[0090] Figure 17 is a flowchart that provides a detailed explanation of the learning process for the dialogue content of S1008 in Figure 10. First, the operation plan change support system 2 references the data range for training (S1701). For example, the operation plan change support system 2 obtains the data range defined in the dialogue configuration data 902. When referencing the data range, it may be divided into training data and evaluation data. Next, the operation plan change support system 2 refers to the model type for learning (S1702). The operation plan change support system 2 obtains, for example, the model type defined in the dialogue configuration data 902. The model type is defined in units of dialogue decisions, such as whether or not there is certainty in the change decision 807 or whether or not there is agreement with the user decision 910. A dialogue decision is the one in the operation information that is the subject of the decision on whether or not dialogue is necessary. The operation plan change support system 2 may also define multiple model types for the same dialogue decision. For example, it may define a model for determining whether or not there is certainty about a change in order, or a model for determining whether or not there is certainty about a change in track number. Similarly, it may define a model for determining whether or not there is agreement with the user decision 910 regarding a change in order, or whether or not there is agreement with the user decision 910 regarding a change in track number. Note that the model types are not limited to these.

[0091] Furthermore, the operation plan change support system 2 references a learning algorithm (S1703). For example, the operation plan change support system 2 obtains an algorithm defined in the dialogue configuration data 902. Examples of algorithms that can be used include linear regression, logistic regression, random forest, gradient boosting decision tree, and neural network. However, the algorithms are not limited to these.

[0092] Then, the operation plan change support system 2 calculates dialogue features 905 (S1704). The operation plan change support system 2 calculates dialogue features 905 for which items have been defined in advance for each model type, for example. The operation plan change support system 2 calculates dialogue features 905 that are common to all model types, for example, the current time and the type of train. It is also possible to calculate different dialogue features 905 for each model type. In other words, the dialogue features 905 of the dialogue model 904 that classifies whether or not the change judgment 807 is certain may include, for example, the judgment probability 808 and the features that the inference model 804 has learned. The dialogue features 905 of the dialogue model 904 that classifies whether or not it matches the user judgment 910 may include, for example, the departure delay time and the departure time interval between trains. The dialogue features 905 of the dialogue model 904 that classifies whether or not the user judgment 910 can be implemented may include, for example, the type of change judgment 807 and the time of the event indicated by the improvement location 707. The dialogue features 905 of the dialogue model 904 that classifies whether the quality of the proposal can be improved may include, for example, the type of competition and various evaluation values. The dialogue features 905 of the dialogue model 904 that classifies whether the processing time can be reduced may include, for example, the complexity of the competition and the number of delayed trains. However, the dialogue features 905 are not limited to these.

[0093] Next, the operation plan change support system 2 learns a set of dialogue features 905 and dialogue labels 906 (S1705). The operation plan change support system 2 constructs a dialogue model 904 using, for example, the set of dialogue features 905 and the correct dialogue labels 906. The operation plan change support system 2 may change the method of defining the correct dialogue labels 906 for each model type. That is, for the dialogue labels 906 of the dialogue model 904 that classifies whether or not the change decision 807 is certain, labels such as (1) when the inference model 804 is certain of the change decision 807 and (0) when the inference model 804 is not certain of the change decision 807 may be defined. For the dialogue labels 906 of the dialogue model 904 that classifies whether or not the user decision 910 matches the system decision 911, labels such as (1) when the user decision 910 and the system decision 911 match and (0) when the user decision 910 and the system decision 911 do not match may be defined. For the dialogue label 906 of the dialogue model 904 that classifies whether or not a user decision 910 can be made, labels such as "the user can make a decision" (1) and "the user cannot make a decision" (0) may be defined. For the dialogue label 906 of the dialogue model 904 that classifies whether or not the proposal quality can be improved, labels such as "the proposal quality can be improved" (1) and "the proposal quality cannot be improved" (0) may be defined. For the dialogue label 906 of the dialogue model 904 that classifies whether or not processing time can be reduced, labels such as "processing time can be reduced" (1) and "processing time cannot be reduced" (0) may be defined. Note that the dialogue model 904 and dialogue label 906 are not limited to these.

[0094] Furthermore, the operation plan change support system 2 evaluates the dialogue model 904 (S1706). The operation plan change support system 2 evaluates the dialogue model 904, for example, using evaluation data. On the other hand, if evaluation data is not available, the dialogue model 904 is evaluated using training data. Then, the operation plan change support system 2 registers the information in the dialogue information database 519 (S1707). This is the process of registering the data used or calculated during the learning of the dialogue content in the dialogue information database 519.

[0095] Although the above process was explained by separating the model type for each change judgment 807, such as changing the order or track number, it is also possible to perform training using a model type that combines multiple change judgments 807. Furthermore, although the above process was explained using examples of changing the order and changing the track number, it is not limited to these when defining multiple model types. It can be applied to various change judgments 807, such as suspension of operation or changes in operation, as long as it does not deviate from the spirit of the invention. Furthermore, although the above process was explained using examples of whether or not there is certainty in the change judgment 807 and whether or not it matches the user judgment 910, the dialogue content to be learned is not limited to these. In addition to whether or not the user judgment 910 can be implemented, it can be applied to various dialogue judgments, such as whether or not the proposed quality can be improved or whether or not the processing time can be shortened, as long as it does not deviate from the spirit of the invention. In the process shown in Figure 17, the operation plan change support system 2 calculates predetermined dialogue features 905 for predetermined model types for dialogue judgment, and then further learns pairs of dialogue features 905 and correct dialogue labels 906.

[0096] Figure 18 shows the display content for interactive operation plan change support on the display unit 40. The operational planning support information presented to the user consists of operational information 1801 and dialogue content 1802. The operation information displays, for example, operation information with the vertical axis representing each station and the horizontal axis representing time. The screen switches by selecting the Plan, Forecast, Proposed Changes, or Trial Details tabs. The illustrated example shows the case where the Proposed Changes tab is selected. Dialogue Content 1802 displays the message history between the user and the system. In this case, it can also be said that the display unit 40 displays the revised operation plan, along with the questions and answers.

[0097] <Variation> The above explanation described the case where the moving object is a railway train, but it can also be applied to other moving objects such as buses that have operational information.

[0098] According to the operation plan change support system 2 described in detail above, when interacting with the user, the system can narrow down the areas for improvement to those where it is difficult for the system to make a change decision, and change the operation plan by interactively resolving these areas.

[0099] <Explanation of methods for supporting changes to the operation plan> The processing performed by the Operation Plan Change Support System 2 is realized through the cooperation of software and hardware resources. Specifically, a processor such as the arithmetic processing unit 10 provided in the Operation Plan Change Support System 2 loads programs that realize each function of the Operation Plan Change Support System 2 into memory such as the main memory 20 and executes them to realize each of these functions. Therefore, the process performed by the aforementioned operation plan change support system 2 can be understood as an operation plan change support method that assists in creating a revised operation plan when a predetermined operation plan is changed. The processor executes a program stored in memory to predict the future operation status of the mobile vehicle based on the predetermined operation plan and the operation history of the mobile vehicle. Based on the predicted operation status, it estimates a change judgment for improvement areas where changes are recommended. Based on the estimated change judgment, it determines questions to supplement the information lacking in order to apply the change judgment to the improvement areas, and changes the operation plan based on the answers to the questions. This makes it possible to provide an operation plan change support method that, when interacting with the user, narrows down the improvement areas to those where the system has difficulty making a change judgment, and allows the operation plan to be changed while interactively resolving these areas.

[0100] It should be noted that the present invention is not limited to the embodiments described above, and various modifications are included. For example, the embodiments described above are explained in detail to make the present invention easier to understand, and are not necessarily limited to those having all the configurations described. Furthermore, it is possible to replace parts of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add configurations from other embodiments to the configuration of one embodiment. In addition, it is possible to add, delete, or replace parts of the configuration of each embodiment with other configurations. Furthermore, each of the above configurations, functions, processing units, and processing means may be implemented in hardware, either partially or entirely, by designing them as integrated circuits, for example. Alternatively, each of the above configurations and functions may be implemented in software by having the processor interpret and execute programs that implement each function. Information such as programs, tables, and files that implement each function can be stored in memory, a recording device such as a hard disk or SSD (Solid State Drive), or a recording medium such as an IC card, SD card, or DVD. Furthermore, the control lines and information lines shown are those deemed necessary for explanatory purposes, and not all control lines and information lines are necessarily shown in the actual product. In reality, it is safe to assume that almost all components are interconnected. [Explanation of Symbols]

[0101] 1...Operation system, 2...Operation plan change support system, 3...Operation management system, 10...Processing unit, 20...Main memory, 30...Input unit, 40...Display unit, 50...Storage unit, 60...Communication unit, 501...Program to create plan change proposal, 502...Operation forecast program, 503...Operation control program, 504...Machine learning program, 505...Program to propose change decision, 506...Program to infer change decision, 507...Program to try change decision, 508...Dialogue program for change decision, 511...Operation 512…Operational data, 513…Proposed plan changes, 514…Improvement data, 515…Trial results data, 516…Dialogue content data, 517…Proposal information database, 518…Inference information database, 519…Dialogue information database, 705…Pre-change operation information, 706…Post-change operation information, 707…Improvement points, 708…Changes, 804…Inference model, 805…Inference features, 806…Inference labels, 904…Dialogue model, 905…Dialogue features, 906…Dialogue labels

Claims

1. A system for supporting changes to a pre-determined operational plan, which assists in creating a revised operational plan when the operational plan is changed, Based on predetermined operating plans and the operating history of the mobile vehicles, the operation prediction unit predicts the future operating status of the mobile vehicles, Based on predicted operating conditions, a change estimation unit estimates the decision to make changes to areas that are recommended for improvement, and Based on the estimated change judgment, a change dialogue unit determines questions to supplement the information lacking in applying the change judgment to the improvement area, and modifies the operation plan based on the answers to the questions. A system for supporting changes to train schedules.

2. The operation plan change support system according to claim 1, wherein the change dialogue unit uses the change judgment to classify whether or not dialogue regarding operation information is necessary, and decides to ask the question if such dialogue is necessary.

3. The system further includes a change trial unit that tests the estimated change judgment and creates trial results including the amount of change in the improved area. The operation plan change support system according to claim 2, wherein the change dialogue unit classifies whether or not dialogue regarding the operation information is necessary using the created trial results.

4. The operation plan change support system according to claim 3, wherein the change dialogue unit calculates feature quantities including the amount of change in the improvement area from the trial results, calculates the confidence level of the change judgment, and determines whether dialogue is necessary.

5. The operation plan change support system according to claim 3, wherein the change dialogue unit identifies whether or not there is insufficient information from the trial results and determines the content of the question.

6. The aforementioned change dialogue unit, If the information from the aforementioned trial results is insufficient, and if there is no shortage of information regarding the necessity of the change decision, the question will include options to select the parameters necessary for the change decision. If the results of the trial indicate that there is insufficient information, and if there is insufficient information regarding the necessity of the change decision, the content of the question will be generated, including the option of accepting or rejecting the change decision. The operation plan change support system according to claim 5.

7. The operation plan change support system according to claim 2, wherein the change dialogue unit determines whether a question is necessary using at least one result of whether there is confidence in the change judgment, whether it matches the user judgment, whether the user judgment can be implemented, whether the proposed quality can be improved, and whether the processing time can be shortened.

8. The operation plan change support system according to claim 1, further comprising a machine learning unit that calculates predetermined inference features for predetermined model types and learns pairs of the predetermined inference features and correct inference labels for the aforementioned change judgment.

9. The operation plan change support system according to claim 8, wherein the machine learning unit calculates predetermined dialogue features for predetermined model types for dialogue judgment, and further learns pairs of dialogue features and correct dialogue labels.

10. The operation plan change support system according to claim 1, further comprising a display unit that displays the changed operation plan, the question and the answer.

11. A method for supporting changes to a pre-determined operational plan, which assists in creating a revised operational plan when changing the predetermined operational plan, The processor executes the program stored in memory, Based on predetermined operating plans and the operational history of the mobile vehicles, future operational status of the mobile vehicles is predicted. Based on the predicted operating conditions, we estimate the decision to make changes to areas where improvements are recommended. Based on the estimated change judgment, questions are determined to supplement the information lacking in order to apply the change judgment to the improvement area, and the operation plan is modified based on the answers to the questions. Methods for supporting changes to the train schedule.

12. A method for supporting changes to an operation plan according to claim 11, which uses the change judgment to classify whether or not dialogue regarding operation information is necessary, and decides to ask the questions if such dialogue is necessary.

13. The estimated change judgment is put to the test, and the test results, including the amount of change in the improved area, are created. A method for supporting changes to an operation plan according to claim 12, which uses the trial results created to classify whether or not dialogue regarding the operation information is necessary.

14. The method for supporting changes to an operation plan according to claim 13, which involves calculating a feature quantity including the amount of change in the improvement area from the trial results, calculating the confidence level of the change judgment, and determining whether dialogue is necessary.

15. A method for supporting changes to an operation plan according to claim 13, which identifies whether or not there is insufficient information from the results of the trial and determines the content of the question.

16. If the information from the aforementioned trial results is insufficient, and if there is no shortage of information regarding the necessity of the change decision, the question will include options to select the parameters necessary for the change decision. If the results of the trial indicate that there is insufficient information, and if there is insufficient information regarding the necessity of the change decision, the content of the question will be generated, including the option of accepting or rejecting the change decision. The method for supporting changes to a train schedule as described in claim 15.

17. A method for supporting changes to an operation plan according to claim 12, which determines whether a question is necessary using at least one result of whether there is confidence in the decision to change, whether it matches the user's decision, whether the user's decision can be implemented, whether the quality of the proposal can be improved, and whether the processing time can be shortened.

18. The method for supporting changes to an operation plan according to claim 11, wherein, with respect to the aforementioned change judgment, predetermined inference features are calculated for predetermined model types, and pairs of the aforementioned inference features and correct inference labels are learned.

19. The method for supporting changes to an operation plan according to claim 18, wherein, regarding dialogue judgment, predetermined dialogue features are calculated for predetermined model types, and pairs of dialogue features and correct dialogue labels are further learned.

20. When changing a predetermined operating plan, a system to assist in creating a revised operating plan is provided. A traffic management system that controls multiple mobile vehicles according to a traffic plan, Equipped with, The aforementioned operation plan change support system is: Based on predetermined operating plans and the operating history of the mobile vehicles, the operation prediction unit predicts the future operating status of the mobile vehicles, Based on predicted operating conditions, a change estimation unit estimates the decision to make changes to areas that are recommended for improvement, and Based on the estimated change judgment, a change dialogue unit determines questions to supplement the information lacking in applying the change judgment to the improvement area, and modifies the operation plan based on the answers to the questions. An operating system equipped with the following features.

Images