Control system, control method, and program

Abstract

To ensure that difference between the state of a control target resulting from control provided after switching in response to occurrence of abnormality in the control target and the state of the control target under normal conditions is kept relatively small.SOLUTION: A control system is provided, comprising switching means configured to switch between normal control and non-normal control of parts of a control target according to the state of at least one of the parts of the control target, and control means configured to control the parts of the control target according to the control switching done by the switching means.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a control system, a control method, and a program.

Background Art

[0002] In control, there are cases where a target state is set for the state of a control target, and control is performed to bring the state of the control target closer to the target state. For example, in the vehicle automatic control system described in Patent Document 1, a target diagram creation device creates and updates a target diagram, and an operation management system controls the operation of trains in a train operation network according to the target diagram.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] When an abnormality occurs in a control target, it is conceivable to switch the control for the control target. In that case, it is considered preferable that the difference between the state of the control target by the control after switching and the state of the control target in normal times is as small as possible.

[0005] An example of the object of the present invention is to provide a control system, a control method, and a program capable of solving the above problems.

Means for Solving the Problems

[0006] According to a first aspect of the present invention, the control system includes: a switching means that switches between normal control and abnormal control for each section of the train's route, which is the part of the railway to be controlled, based on the state of at least one part of the part to be controlled, and when a section is out of service, determines the control of each section between the section where turnaround operation is possible and the section including the out of service to be possible to be turned around to the abnormal control; and a control means that controls the part to be controlled in response to the switching of control by the switching means, and controls the part to be controlled in the abnormal control using a control method obtained by learning with an evaluation function that gives a higher evaluation the closer the state of the part to be controlled is to a state defined as the target state of the part to be controlled in normal conditions. 。

[0007] This invention 2 According to this embodiment, the control method includes a computer switching between normal control and abnormal control for each section of the train's route, which is the part of the railway being controlled, based on the state of at least one part of the part being controlled; if a section is out of service, the computer deciding to use the abnormal control for each section between the sections where turnaround operations are possible, including the out-of-service section; controlling the part being controlled in accordance with the control switching; and controlling the part being controlled in the abnormal control using a control method obtained through learning with an evaluation function that gives a higher evaluation the closer the state of the controlled object is to the state defined as the target state of the controlled object during normal operation.

[0008] This invention 3According to this embodiment, the program is a program that causes a computer to perform the following actions: switch between normal control and abnormal control for each section of the train's route, which is the part of the railway being controlled, based on the state of at least one part of the part being controlled; when a section is out of service, determine the control of each section between the section where turnaround operation is possible and the section including the out of service to be changed to the abnormal control; and, in response to the switch in control, control the part being controlled, and in the abnormal control, control the part being controlled using a control method obtained through learning with an evaluation function that gives a higher evaluation the closer the state of the controlled part is to the state defined as the target state of the controlled part in normal conditions. 。 [Effects of the Invention]

[0009] According to the present invention, it is expected that the difference between the state of the controlled object under the control switched in response to an abnormality in the controlled object and the state of the controlled object under normal conditions will be relatively small. [Brief explanation of the drawing]

[0010] [Figure 1] This figure shows an example of the configuration of a control system according to the embodiment. [Figure 2] This figure shows an example of the assignment of a control unit to a part of a controlled object during normal operation according to the embodiment. [Figure 3] This figure shows an example of assigning a control unit to a part of a controlled object when an abnormality occurs in the controlled object according to the embodiment. [Figure 4] This figure shows an example of assigning a control unit to a part of a controlled object when an abnormality in the controlled object is resolved according to the embodiment. [Figure 5] This figure shows an example of the processing procedure performed by the control system according to the embodiment. [Figure 6] This figure shows a second configuration example of the control system according to the embodiment. [Figure 7] This figure shows a third configuration example of the control system according to the embodiment. [Figure 8] It is a diagram showing a fourth configuration example of the control system according to the embodiment. [Figure 9] It is a diagram showing a fifth configuration example of the control system according to the embodiment. [Figure 10] It is a diagram showing a sixth configuration example of the control system according to the embodiment. [Figure 11] It is a diagram showing a seventh configuration example of the control system according to the embodiment. [Figure 12] It is a diagram showing an example of the processing procedure in the control method according to the embodiment. [Figure 13] It is a schematic block diagram showing the configuration of a computer according to at least one embodiment.

MODE FOR CARRYING OUT THE INVENTION

[0011] Hereinafter, embodiments of the present invention will be described. However, the following embodiments do not limit the invention according to the claims. Also, not all combinations of features described in the embodiments are essential for the solution means of the invention. FIG. 1 is a diagram showing an example of the configuration of a control system according to an embodiment. In the configuration shown in FIG. 1, the control system 1 includes a switching unit 10 and a control unit 20. The control unit 20 includes a plurality of normal control units 21, one or more abnormal-time control units 22, and one or more recovery control units 23. Further, a control target 30 is shown in FIG. 1.

[0012] The control system 1 is a system that controls the control target 30. The control target 30 operates according to the control by the control system 1. The target state of the control target 30 at normal times is set every time (that is, dynamically set), and the control target 30 can be controlled by controlling each part into which the control target 30 is divided into a plurality of parts, and various things in which local abnormalities can occur and is not limited to a specific one.

[0013] The controlled object 30 may be configured as a single device or as a system including a plurality of devices. Alternatively, the controlled object 30 may be configured as part of a device or a system.

[0014] When an abnormality occurs in the controlled object 30 here, it means that the state of at least one part of the parts of the controlled object 30 is different from the state preset as the normal state. The abnormality of the controlled object 30 may be an abnormality occurring due to an internal factor of the controlled object 30 or an abnormality occurring due to an external factor of the controlled object 30.

[0015] The normal time of the controlled object 30 is when the state of the entire controlled object 30 is in the state preset as the normal state. That the state of the controlled object 30 is in the normal state may mean that the error between the state of the controlled object 30 and the normal state is within a predetermined range. The state preset as the normal state may be a dynamic state (that is, a state that changes according to the passage of time). Therefore, the state of the controlled object 30 at normal time may be a dynamic state. When the state of the controlled object 30 is different from the normal state, the state of the controlled object 30 is also referred to as an abnormal state. The time when the state of the controlled object 30 is in the abnormal state is also referred to as the abnormal time.

[0016] The controlled object 30 may be configured as part of the control system 1 or may be an external configuration of the control system 1. Alternatively, when a traffic system includes a control system, for example, the control system 1 may be configured as part of the controlled object 30. Examples of the controlled object 30 include, but are not limited to, traffic systems such as railways, plants such as chemical plants, and traffic control systems. Hereinafter, the case where the controlled object 30 is a railway will be described as an example.

[0017] The control unit 20 controls the controlled object 30 by performing control for each part when the controlled object 30 is divided into a plurality of parts. The control unit 20 is an example of a control means.

[0018] The normal control unit 21 performs normal control on the portion of the controlled object 30. Normal control, in this context, means controlling the portion of the controlled object 30 so that its state becomes a state predetermined as the normal state of the controlled object 30. For example, if the controlled object 30 is a railway, the normal control unit 21 controls the part of the controlled object 30 so that the train of the controlled object 30 runs according to the normal operating schedule.

[0019] Alternatively, if the controlled object is an air traffic control system, the normal control unit 21 controls each aircraft so that the aircraft corresponding to the controlled object 30 flies according to the flight plan. Alternatively, if the controlled object 30 is a chemical plant, the normal control unit 21 may control a portion of the controlled object 30 such that the measured values ​​from sensors installed on the controlled object 30 become values ​​that are pre-set as normal values.

[0020] The normal control unit 21 may perform normal control on a portion of the controlled object 30 based on the state information of the entire controlled object 30. Alternatively, the normal control unit 21 may perform normal control on a portion of the controlled object 30 based on the state information of the portion of the controlled object 30 that the normal control unit 21 controls itself.

[0021] The control unit 20 may have the same number of normal control units 21 as there are parts of the controlled object 30, and there may be a one-to-one correspondence between parts of the controlled object 30 and the normal control units 21 that control those parts. When distinguishing between individual normal control units 21, they are denoted as normal control unit 21-1, normal control unit 21-2, ..., normal control unit 21-Nn (where Nn is a positive integer representing the number of normal control units 21).

[0022] The abnormality control unit 22 performs abnormality control on the portion of the controlled object 30. Abnormality control, in this context, refers to controlling the portion of the controlled object 30 in response to an abnormality occurring in that portion of the controlled object 30, so that the controlled object 30 continues to operate.

[0023] For example, if the controlled object 30 is a railway, and an abnormality occurs that causes a section of track to become impassable, then under normal control conditions, if trains were to concentrate and stop just before the impassable section, it is conceivable that train service would not be available to passengers. Furthermore, if the controlled object 30 is a chemical plant, and the state of the chemical plant deviates significantly from its normal state due to an abnormality, the protection logic may shut down the chemical plant. This shutdown could lead to a significant decrease in production volume and a prolonged recovery time.

[0024] In contrast, if the operation of the controlled object 30 can be maintained by the control of the abnormality control unit 22, the functionality of the controlled object 30 can be maintained even if there is a decrease in functionality, and for example, the provision of transportation services to customers or the manufacturing of products can continue. Furthermore, if the operation of the controlled object 30 can be maintained by the control of the abnormality control unit 22, it is expected that the normal state will be restored relatively quickly once the abnormality of the controlled object 30 is resolved. When an abnormality occurs in the controlled object 30 (when an abnormality has occurred in the controlled object 30), the abnormality control unit 22 performs abnormality control on the portion of the controlled object 30 that has been set by the switching unit 10 as being near the location of the abnormality.

[0025] The abnormality control unit 22 controls parts of the controlled object 30 so that the overall state of the controlled object 30 is close to the state defined as the target state of the controlled object 30 during normal operation, in addition to preventing the entire controlled object 30 from becoming inoperable. The state defined as the target state of the controlled object 30 during normal operation is also called the target normal state.

[0026] For example, if the controlled object 30 is a railway, the abnormality control unit 22 controls parts of the controlled object 30 so that, in addition to being able to continue the operation of the train, it can operate the train as close as possible to the normal operating schedule. Because train operations will be close to the normal schedule, the burden on railway personnel is expected to be relatively light. Furthermore, because train operations will be close to the normal schedule, it is expected that normal operations will be restored relatively quickly once the abnormal situation is resolved.

[0027] The abnormality control unit 22 performs control to prevent the entire controlled object 30 from becoming inoperable and to bring the state of the entire controlled object 30 close to the target normal state. This is done, for example, by learning the control by the abnormality control unit 22 using an evaluation function that is highly evaluated when the operation of the controlled object 30 continues and when the state of the controlled object 30 is close to the target normal state.

[0028] The control unit 20 may have the same number of abnormality control units 22 as the number of parts of the controlled object 30. Alternatively, since only a portion of the controlled object 30 requires simultaneous control by abnormality control, the control unit 20 may have fewer abnormality control units 22 than the number of parts of the controlled object 30. When distinguishing individual abnormality control units 22, they are denoted as abnormality control unit 22-1, abnormality control unit 22-2, ..., abnormality control unit 22-Na (where Na is a positive integer representing the number of abnormality control units 22).

[0029] The recovery control unit 23 controls parts of the controlled object 30 so that the state of the controlled object 30 returns to its normal state (the state under normal circumstances). Specifically, the recovery control unit 23 is switched over from the abnormality control unit 22 when the switching unit 10 determines that the abnormality of the controlled object 30 has been resolved. The recovery control unit 23 then controls parts of the controlled object 30 so that the state of the controlled object 30 returns to its normal state from the state corresponding to the occurrence of the abnormality. When distinguishing between individual recovery control units 23, they are denoted as recovery control unit 23-1, recovery control unit 23-2, ..., recovery control unit 23-Nr (where Nr is a positive integer representing the number of recovery control units 23).

[0030] Furthermore, if the state of the controlled object 30 returns to a normal state without the need to switch control from the control unit 22, the control unit 22 may continue to control the portion of the controlled object 30 once the abnormality in the controlled object 30 is resolved. In this case, the control unit 20 does not need to have a recovery control unit 23.

[0031] The switching unit 10 assigns one of the normal control unit 21, the abnormal control unit 22, or the recovery control unit 23 as the control unit (part of the control unit 20) that controls the part of the controlled object 30. As a result, the switching unit 10 selects one of the normal control, abnormal control, or recovery control as the control for each part of the controlled object 30. The switching unit 10 is an example of a switching means.

[0032] Specifically, the switching unit 10 acquires status information of the controlled object 30 and determines the status of the controlled object 30 in real time. If it determines that the controlled object 30 is in a normal state, the switching unit 10 assigns the normal control unit 21 to control each part of the controlled object 30, and the normal control unit 21 performs normal control on the parts of the controlled object 30.

[0033] When the switching unit 10 detects an abnormality in the controlled object 30, it identifies the location of the abnormality. The switching unit 10 then sets a portion of the controlled object 30 that is near the location of the abnormality. The portion near the location of the abnormality may include the portion of the controlled object 30 that contains the location of the abnormality. The switching unit 10 switches the control unit assigned to the portion set near the location of the abnormality from the normal control unit 21 to the abnormality control unit 22.

[0034] The switching unit 10 may also set the portion of the controlled object 30 that includes the location where the abnormality occurred, as well as the portion that has a function to mitigate the spread of the abnormality's effects, as the portion near the location where the abnormality occurred. When the controlled object 30 is a railway, and the portion of the controlled object 30 is a section of the train's route, a section where train turnaround operations are possible is an example of a section that has the function of mitigating the spread of the effects of an abnormality. This point will be discussed later with reference to Figure 4.

[0035] When the switching unit 10 detects that the abnormality in the controlled object 30 has been resolved, it switches the assignment of the functional unit to the part of the controlled object 30 that was assigned to the abnormality control unit 22 from the abnormality control unit 22 to the recovery control unit 23. As a result, the control of these parts is switched from abnormality control to recovery control. However, as mentioned above, the recovery control unit 23 is not essential to the control system 1. The switching unit 10 may maintain the assignment of the abnormality control unit 22 to the portion of the controlled object 30 even after detecting that the abnormality of the controlled object 30 has been resolved, until the state of the controlled object 30 returns to a normal state.

[0036] Figure 2 shows an example of the allocation of control units to parts of the controlled object 30 during normal operation. Figure 2 shows an example where the controlled object 30 is a railway, and line L11 represents the train's route from left to right in Figure 2. Line L12 represents the train's route from right to left in Figure 2. The route is also called the track.

[0037] The rectangle "□" in Figure 2, exemplified by rectangle B1, represents a station platform. The line between line L11 and line L12, exemplified by line L21, represents the track connecting the track indicated by line L11 and the track indicated by line L12. The track connecting line L11 and line L12 allows for train turnaround operations.

[0038] The lines indicated by lines L11 and L12 are divided into multiple sections, with sections A1 through A9 shown in Figure 2. Each section from A1 to A9 contains one station. Train turnaround operations are possible within sections A3, A5, and A8. Each of the intervals A1 through A9 corresponds to an example of a part of the controlled object 30. However, the method of dividing the controlled object 30 into multiple parts is not limited to a specific method. For example, intervals A6 and A7 may be a single part of the controlled object 30. In other words, multiple intervals may be treated as a single part of the controlled object.

[0039] In the example shown in Figure 2, the switching unit 10 assigns a normal control unit 21 to each of the sections A1 through A9. As a result, the control unit 20 controls each section from A1 to A9 using normal control. When distinguishing between the normal control units 21 assigned to intervals A1 to A9 in Figure 2, the normal control unit 21 assigned to interval Ai (where i is an integer between 1 and 9) is denoted as normal control unit 21-i.

[0040] Figure 3 shows an example of the allocation of the control unit to a part of the controlled object 30 when an abnormality occurs in the controlled object 30. Figure 3 shows an example where an abnormality occurs in section A4 of the railway line shown in Figure 2, and section A4 becomes a closed section. A closed section is a section where trains cannot pass.

[0041] In the example shown in Figure 3, the switching unit 10 switches the control units assigned to each of the sections A3 to A5 from the normal control unit 21 to the abnormal control unit 22. The abnormal control unit 22 performs train turnaround operations in sections A3 and A5, which are sections where turnaround operations are possible before the closed section A4. A section where turnaround operations are possible is a section where train turnaround operations are possible.

[0042] Sections A3 and A5, which are sections where turnaround operation is possible, are examples of parts of the controlled object 30 that have a function to mitigate the spread of the effects of abnormalities. It is expected that by having trains entering section A3 from section A2 reverse direction towards section A2 in section A3, it will be possible to avoid trains becoming stuck at the location of the anomaly, and normal operation (operation based on the normal operating schedule) will be possible in each section from section A2 to section A1. Similarly, it is expected that trains entering section A5 from section A6 will reverse direction towards section A6 in section A5, thereby avoiding congestion at the point of the malfunction and enabling normal operation in each section from section A6 to section A7.

[0043] If the portion of the controlled object 30 that contains the location of the abnormality also has a function to mitigate the spread of the abnormality's effects, the switching unit 10 may set only one portion as the portion near the location of the abnormality. In the example in Figure 3, if the section from A3 to A5 constitutes one portion of the controlled object 30, when an abnormality occurs in section A4, the switching unit 10 may set the section from A3 to A5 as the portion near the location of the abnormality.

[0044] The setting method by which the switching unit 10 sets the portion of the controlled object 30 that is near the location where the abnormality occurred may be obtained through learning. Alternatively, the method for setting the switching unit 10 to the portion of the controlled object 30 near the location where the abnormality occurred may be predetermined by a person. For example, the designer of the control system 1 may pre-set the method for setting the switching unit 10 to the portion of the controlled object 30 near the location where the abnormality occurred using a rule-based approach. For example, in section A5 of Figure 3, the switching unit 10 may set section A5 as one of the sections near the location of the malfunction if any of sections A4 to A7 becomes a non-functional section.

[0045] Thus, the switching unit 10 may switch between normal control and abnormal control for one part of the controlled object 30 based on the state of at least one part of the controlled object 30. When distinguishing between the abnormality control units 22 assigned to intervals A3, A4, and A5, the abnormality control unit 22 assigned to interval Ai (where i is an integer between 3 and 5) is denoted as abnormality control unit 22-i.

[0046] The abnormal control performed by the abnormal control unit 22 may include, in addition to, turning the train around, or instead, sending the train to a depot, siding, or passing track, or moving it to a safe location. Furthermore, the abnormal control performed by the abnormal control unit 22 may also include operating the train that has been sent to a safe location or moved to a safe location.

[0047] The recovery control performed by the recovery control unit 23 may also include the decommissioning or evacuation of trains, and the operation of the decommissioned or evacuated trains. Through such control, the recovery control unit 23 can adjust the number of trains and their schedules. The abnormality control unit 22 may, under abnormality control, direct or evacuate the train, and the recovery control unit 23 may then operate the train under recovery control.

[0048] Figure 4 shows an example of the allocation of the control unit to a portion of the controlled object 30 when an abnormality in the controlled object 30 is resolved. Figure 4 shows an example from the example in Figure 3 where an abnormality that occurred in section A4 is resolved. In the example in Figure 4, section A4 is open due to the resolution of the abnormality. That is, trains can travel through section A4.

[0049] In the example shown in Figure 4, the switching unit 10 switches the control units assigned to each section from A3 to A5 from the abnormality control unit 22 to the recovery control unit 23. It is possible that the train operation status deviates from the normal operation status shown in the regular timetable due to the train turning back when an abnormality occurs. The recovery control unit 23 controls the part of the controlled object 30 so that the train operates according to the normal timetable. When distinguishing between the recovery control units 23 assigned to sections A3, A4, and A5, the recovery control unit 23 assigned to section Ai (where i is an integer between 3 and 5) is denoted as recovery control unit 23-i.

[0050] Figure 5 shows an example of the processing steps performed by the control system 1. At the start of the process shown in Figure 5, the state of the controlled object 30 is in the normal state, and the switching unit 10 assigns the normal control unit 21 to each part of the controlled object 30. As a result, the control unit 20 controls each part of the controlled object 30 using normal control (step S11).

[0051] Subsequently, the switching unit 10 detects the occurrence of an abnormality in the controlled object 30 (step S21). The switching unit 10 identifies the location of the abnormality in the controlled object 30 and sets the portion of the controlled object 30 that is near the location of the abnormality (step S22). Then, the switching unit 10 instructs the control unit 20 to switch the control when an abnormality is detected (step S23). In the instruction to switch the control when an abnormality is detected, the switching unit 10 instructs the control unit 20 to switch the control unit assigned to the part of the controlled object 30 that is set to be near the location where the abnormality occurred from the normal control unit 21 to the abnormality control unit 22. The control unit 20, in accordance with the instructions of the switching unit 10, switches the control unit assigned to the area near the location of the abnormality from the normal control unit 21 to the abnormality control unit 22. As a result, the control unit 20 controls the area near the location of the abnormality using abnormality control (step S24).

[0052] Subsequently, the switching unit 10 detects that the abnormality in the controlled object 30 has been resolved (step S31). The switching unit 10 instructs the control unit 20 to switch the control when the abnormality is resolved (step S32). In the instruction to switch the control when the abnormality is resolved, the switching unit 10 instructs the control unit 20 to switch the control unit assigned to the part of the controlled object 30 that is set to be near the location where the abnormality occurred from the abnormality control unit 22 to the recovery control unit 23. The control unit 20, in accordance with the instructions of the switching unit 10, switches the control unit assigned to the area near the location of the malfunction from the malfunction control unit 22 to the recovery control unit 23. As a result, the control unit 20 controls the area near the location of the malfunction using recovery control (step S33).

[0053] Subsequently, the switching unit 10 detects that the state of the controlled object 30 has returned to a normal state (step S41). The switching unit 10 instructs the control unit 20 to switch the control based on the recovery detection (step S42). In the control switching instruction upon recovery detection, the switching unit 10 instructs the control unit 20 to switch the control unit assigned to the part of the controlled object 30 that is set to be near the location of the abnormality from the recovery control unit 23 to the normal control unit 21. The control unit 20, in accordance with the instructions of the switching unit 10, switches the control unit assigned to the area near the location of the abnormality from the recovery control unit 23 to the normal control unit 21. As a result, the control unit 20 controls each part of the controlled object 30 using normal control (step S43).

[0054] As described above, the switching unit 10 switches between normal control and abnormal control for a portion of the controlled object 30 based on the state of at least one portion of the controlled object. The control unit 20 controls the portion of the controlled object 30 in accordance with the control switching by the switching unit 10. In this way, in response to an abnormality in the controlled object 30, the control unit 20 partially switches the control of the controlled object 30. For the parts of the controlled object 30 that were not subject to the control switch, the control is performed under normal conditions to bring those parts back to the target normal state.

[0055] As a result, it is expected that the difference between the state of the controlled object 30 under the control switched over in response to an abnormality in the controlled object 30 and the state of the controlled object 30 under normal conditions will be relatively small in the control system 1. Because this difference is small, it is expected that the state of the controlled object 30 will return to normal relatively quickly after the abnormality is resolved. Furthermore, because this difference is small, the work performed by the personnel handling the controlled object 30 will be similar to their normal work, and in this respect, it is expected that the workload of the personnel handling the controlled object 30 will be reduced.

[0056] Here, let's consider the case where the entire controlled system is controlled by a single control unit obtained through reinforcement learning. If an anomaly occurs in the controlled system, a single control unit controlling the entire system may cause parts other than the anomaly to perform actions different from normal operation.

[0057] In contrast, in control system 1, the switching unit 10 partially switches the control when an abnormality occurs, so that the parts of the controlled object 30 other than the area where the abnormality occurred are expected to maintain normal operation or operate in a manner close to normal operation. In this way, the control system 1 makes it relatively easy to predict the control of the controlled object 30.

[0058] Furthermore, it is expected that the control system 1 can prevent a localized anomaly in the controlled object 30 from affecting the entire controlled object 30. For example, if the controlled object 30 is an air traffic control system, and an anomaly occurs in one aircraft that is part of the controlled object 30, it is expected that the control system 1 will control the flight of each aircraft in such a way that the anomaly in that aircraft does not affect the operation of other aircraft as much as possible.

[0059] In the case of a controlled object 30, such as a railway, where localized abnormalities are likely to occur, it is assumed that the control system 1 applies abnormal control to only a portion of the controlled object 30, and normal control to the remaining portion. In this case, the control system 1 can apply abnormal control to only a portion of the controlled object 30 and normal control to the remaining portion without the need to provide a mechanism to limit the application of abnormal control to only a portion of the multiple portions of the controlled object 30.

[0060] Furthermore, if an abnormality occurs in a part of the controlled object 30, the switching unit 10 determines the control of one or more parts of the controlled object 30, including parts that have a function to mitigate the spread of the effects of the abnormality, to be in the non-normal control state. As a result, the abnormality control unit 22 can control the portion of the controlled object 30 in a way that reduces the spread of the effects of the abnormality during non-normal control, and it is expected that the effects of the abnormality can be limited.

[0061] Furthermore, the switching unit 10 switches between abnormal control, which is performed to respond to an abnormality occurring in a part of the controlled object 30, and recovery control, which is performed to restore the state of the part of the controlled object 30 to its normal state, as a control function for abnormal situations. The control unit 20 further controls the part of the controlled object 30 in accordance with the switching of the abnormal control by the switching unit 10. According to control system 1, when an abnormality in the controlled object 30 is resolved, it is expected that the state of the controlled object 30 will return to a normal state relatively quickly.

[0062] Furthermore, in non-normal operation, the control unit 20 controls the portion of the controlled object using a control method obtained through learning that employs an evaluation function, where the closer the state of the controlled object is to the state defined as the target state of the controlled object during normal operation, the higher the evaluation. According to control system 1, the state of the part of the controlled object 30 that is controlled by the abnormal control is expected to be relatively close to the normal state. Therefore, the state of each part of the controlled object 30 (both the part controlled by the normal control and the part controlled by the abnormal control) is expected to be relatively close to the normal state.

[0063] Because the state of each part of the controlled object 30 is relatively close to its normal state, it is expected that the state of the controlled object 30 will return to normal relatively quickly after the abnormality is resolved. Furthermore, because the state of each part of the controlled object 30 is relatively close to its normal state, the work performed by the personnel handling the controlled object 30 will be similar to their normal work, and in this respect, it is expected that the workload of the personnel handling the controlled object 30 will be reduced.

[0064] Furthermore, in non-normal operation, the control unit 20 uses a control method obtained through learning that employs an evaluation function, where the closer the state of the controlled object 30 is to a state defined time-wise as the target state of the controlled object 30 during normal operation, the higher the evaluation. According to control system 1, in cases where the target state of the controlled object 30 during normal operation is dynamically set, it is known that the state of each part of the controlled object 30 is relatively close to the normal state.

[0065] In this embodiment, we will describe an example where an evaluation function is used to learn the control performed by the control unit 20 or each of its parts. However, not only an evaluation function, but various things that can evaluate provisional learning results can be used for learning. For example, instead of an evaluation function, information that shows evaluation values ​​for provisional learning results in a tabular format may be used. The evaluation function and the tabular information, etc., which show evaluations of provisional learning results, are also called an evaluation model.

[0066] Furthermore, the controlled object 30 is a traffic system, and the control unit 20, in non-normal operation, uses a control method obtained through learning using an evaluation function that gives a higher evaluation the closer the operating status of the moving objects in the traffic system is to the operating status shown in the normal operating schedule of the traffic system. According to control system 1, the operation of the traffic system is expected to be relatively close to the operating conditions shown in the normal operating schedule.

[0067] Because the operating status of the transportation system is relatively close to that shown in the normal operating schedule, it is expected that after the abnormality is resolved, the operating status of the transportation system will return to that shown in the normal operating schedule relatively quickly. Furthermore, because the operating status of the transportation system is relatively close to that shown in the normal operating schedule, the work of the transportation system personnel will be similar to their normal work, and in this respect, it is expected that the workload of the personnel handling the controlled object 30 will be reduced.

[0068] Figure 6 is a diagram showing a second configuration example of the control system according to the embodiment. In the configuration shown in Figure 6, the control system 1b includes a control device 100. The control device 100 includes a communication unit 110, a display unit 120, an operation input unit 130, a storage unit 180, and a processing unit 190. The processing unit 190 includes a switching unit 10 and a control unit 20. The control unit 20 includes a plurality of normal control units 21, one or more abnormal control units 22, and one or more recovery control units 23. Figure 6 also shows the controlled object 30.

[0069] In Figure 6, parts that have the same function as those in Figure 1 are denoted by the same reference numerals (10, 20, 21, 22, 23, 30), and detailed explanations are omitted here. Control system 1b is an example of control system 1. In control system 1b, control of the controlled object 30 is performed by a single control device 100.

[0070] The control device 100 controls the controlled object 30 by executing the functions of the switching unit 10 and the control unit 20. The control device 100 is configured using a computer such as a workstation. Alternatively, the control device 100 may be configured using hardware dedicated to the control device 100, such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

[0071] The communication unit 110 communicates with other devices. For example, the communication unit 110 may transmit control commands to the controlled object 30. Alternatively, the communication unit 110 may receive status information, such as sensor measurement values, from the controlled object 30. The display unit 120 includes a display screen such as a liquid crystal panel or an LED (Light Emitting Diode) panel, and displays various images. For example, the display unit 120 may display information related to the controlled object 30, such as status information of the controlled object 30 and, if the controlled object 30 is a railway, the operating schedule.

[0072] The operation input unit 130 includes, for example, input devices such as a keyboard and a mouse, and accepts user input. For example, the operation input unit 130 may be configured to accept user input that instructs the display of information related to the controlled object 30. The storage unit 180 stores various types of data. The storage unit 180 is configured using the storage devices provided by the control device 100.

[0073] The processing unit 190 controls various parts of the control device 100 to perform various processes. The functions of the processing unit 190 may also be performed, for example, by the CPU (Central Processing Unit) of the control device 100 reading a program from the storage unit 180 and executing it. As with control system 1b, controlling the controlled object 30 is performed by a single device, and because there is only one device, the maintenance burden on the device is expected to be relatively small.

[0074] Figure 7 shows a third configuration example of the control system according to the embodiment. In the configuration shown in Figure 7, the control system 1c comprises a plurality of control devices 200. The control devices 200 include a switching unit 10c, a normal control unit 21, an abnormal control unit 22, and a recovery control unit 23. Figure 7 also shows the controlled object 30. In Figure 7, parts that have the same function as those in Figure 1 are denoted by the same reference numerals (21, 22, 23, 30), and detailed explanations are omitted here.

[0075] Control system 1c is an example of control system 1. In the control system 1b shown in Figure 6, one control device 100 centrally controls each part of the controlled object 30, whereas in the control system 1c, there is a one-to-one correspondence between the control device 200 and parts of the controlled object 30, and one control device 200 controls one part of the controlled object 30. In other respects, the control system 1c is the same as the control system 1b, and the control device 200 is the same as the control device 100.

[0076] The control device 200 controls the controlled object 30 by executing the functions of the switching unit 10 and the control unit 20. The control device 100 is configured using a computer such as a workstation. Alternatively, the control device 100 may be configured using hardware dedicated to the control device 100, such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

[0077] Each switching unit 10c selects either the normal control unit 21, the abnormal control unit 22, or the recovery control unit 23 depending on the state of the controlled object 30. The combination of switching units 10c of all control devices 200 provided in the control system 1c corresponds to the example of the switching unit 10 (Figure 1) and the example of the switching means. The combination of switching units 10c of all control devices 200 provided in the control system 1c switches the control unit assigned to the part of the control device 30 that has been determined to be near the location of the abnormality from the normal control unit 21 to the abnormality control unit 22 when an abnormality occurs in the controlled object 30.

[0078] The combination of the normal control unit 21, abnormal control unit 22, and recovery control unit 23 of all control devices 200 in the control system 1c corresponds to the example of the control unit 20 (Figure 2) and the example of the control means. The combination of the normal control unit 21, abnormal control unit 22, and recovery control unit 23 of all control devices 200 in the control system 1c controls the controlled object 30 according to the switching by the switching means. As in control system 1c, by performing the control of the controlled object 30 using distributed control by multiple devices, the distance between the controlled object 30 and the control device 200 that controls that part is relatively short, and it is expected that the effects of communication delays and the like will be relatively small.

[0079] Figure 8 shows a fourth configuration example of the control system according to the embodiment. In the configuration shown in Figure 8, the control system 1d includes a plurality of switching devices 310, a normal control device 321, an abnormal control device 322, and a recovery control device 323. Figure 8 also shows the controlled object 30.

[0080] The controlled object 30 in Figure 8 is the same as in Figure 1, so a detailed explanation is omitted here. In the configuration shown in Figure 8, each part of the control device in Figure 7 is configured as a separate device. Specifically, in the configuration shown in Figure 8, the switching unit 10c, normal control unit 21, abnormal control unit 22, and recovery control unit 23 of the control device 200 in Figure 7 are configured as the switching device 310, normal control device 321, abnormal control device 322, and recovery control device 323, respectively.

[0081] In Figure 8, the control device 200 is the same as in Figure 7, except that it is configured as a switching device 310, a normal control device 321, an abnormal control device 322, and a recovery control device 323. The normal control device 321, the abnormal control device 322, and the recovery control device 323 switch depending on the selection of the switching device 310, and distribute control of the controlled object 30. All combinations of the normal control devices 321, abnormal control devices 322, and recovery control devices 323 provided by the control system 1d correspond to the example of the control unit 20 in Figure 1. All combinations of the switching devices 310 provided by the control system 1d correspond to the example of the switching unit 10 in Figure 1.

[0082] As in control system 1d, by performing the control of the controlled object 30 using distributed control by multiple devices, the distance between the controlled object 30 and the devices controlling that part is relatively short, and it is expected that the effects of communication delays and the like will be relatively small. Furthermore, since the normal control device 321, the abnormal control device 322, and the recovery control device 323 are each configured as separate devices, partial maintenance, replacement, and version upgrades can be performed relatively easily.

[0083] Figure 9 shows a fifth configuration example of the control system according to the embodiment. In the configuration shown in Figure 9, the control system 1e comprises a switching unit 10 and a control unit 20e. The control unit 20e comprises a normal control unit 21e, an abnormal control unit 22e, a recovery control unit 23e, and a detailed control unit 24. Figure 9 also shows the controlled object 30. In Figure 9, parts that have the same function as those in Figure 1 are denoted by the same reference numerals (10, 30), and detailed explanations are omitted here.

[0084] The control unit 20e performs the processing that the control unit 20 (Figure 1) does in a two-stage process using the detailed control unit 24. Specifically, the combination of the normal control unit 21e and the detailed control unit 24 of the control unit 20e corresponds to the normal control unit 21 of the control unit 20. The combination of the abnormal control unit 22e and the detailed control unit 24 of the control unit 20e corresponds to the abnormal control unit 22 of the control unit 20. The combination of the recovery control unit 23e and the detailed control unit 24 of the control unit 20e corresponds to the recovery control unit 23 of the control unit 20.

[0085] In all other respects, control system 1e is the same as control system 1. The normal control unit 21e, abnormal control unit 22e, and recovery control unit 23e are the same as the normal control unit 21, abnormal control unit 22, and recovery control unit 23, except that they control the portion of the controlled object 30 in a two-stage process using the detailed control unit 24.

[0086] The detailed control unit 24 generates specific control commands for parts of the controlled object 30 in response to the output of the normal control unit 21e, the abnormal control unit 22e, or the recovery control unit 23e, and outputs them to the parts of the controlled object 30. For example, if the controlled object 30 is a railway, the normal control unit 21e may output a command indicating whether or not the train will reverse direction as one of the control commands. Then, the detailed control unit 24 may output specific commands for each piece of equipment included in the controlled object 30, such as commands for points and signals, depending on whether or not the train will reverse direction.

[0087] In this way, by having the control unit 20e generate control commands for the controlled object 30 in two stages, the possible values ​​of the outputs of the normal control unit 21e, the abnormal control unit 22e, and the recovery control unit 23e are limited. As a result, it is expected that learning the processes performed by each of the normal control unit 21e, the abnormal control unit 22e, and the recovery control unit 23e can be performed relatively easily and with relatively high accuracy. The detailed control unit 24 may be excluded from learning, for example, by being built using a rule-based method by the designer of the control system 1e.

[0088] Similar to the normal control unit 21, the control unit 20 may have the same number of detailed control units 24 as there are parts of the controlled object 30, and there may be a one-to-one correspondence between parts of the controlled object 30 and the detailed control unit 24 that controls those parts. When distinguishing between individual detailed control units 24, they are denoted as detailed control unit 24-1, detailed control unit 24-2, ..., detailed control unit 24-Nn. Here, Nn is a positive integer representing the number of detailed control units 24, and the number of detailed control units 24 is the same as the number of normal control units 21e.

[0089] Figure 10 shows a sixth configuration example of the control system according to the embodiment. Figure 10 shows a configuration example of the control system 1 shown in Figure 1 during the learning phase. As shown in Figure 10, the control system 1f comprises a switching unit 10, a control unit 20, and a learning unit 40. The control unit 20 includes a normal control unit 21, an abnormal control unit 22, and a recovery control unit 23. Figure 10 also shows a simulator 50.

[0090] The simulator 50 simulates the operation of the controlled object 30. The control system 1f uses the simulator 50 to acquire training data. Alternatively, in the configuration shown in Figure 10, the controlled object 30 may be connected to the control system 1f in addition to, or instead of, the simulator 50, and the control system 1f may use the controlled object 30 to acquire training data.

[0091] The learning unit 40 performs learning on the switching unit 10, the normal control unit 21, the abnormal control unit 22, and the recovery control unit 23, or a part of these. Learning, in this context, means setting or updating the parameter values ​​of the learning model. The learning unit 40 updates the parameter values ​​of the learning models that constitute each of the switching unit 10, the normal control unit 21, the abnormal control unit 22, and the recovery control unit 23, or the parameter values ​​of a part of these learning models. The learning unit 40 is an example of a learning means. The learning unit 40 may perform the learning of the switching unit 10, the normal control unit 21, the abnormal control unit 22, and the recovery control unit 23, or a part thereof, using reinforcement learning.

[0092] In the control unit 20, the normal control unit 21 and the abnormal control unit 22 are configured separately, so the learning unit 40 separately learns the control during normal operation by the normal control unit 21 and the control during abnormal operation by the abnormal control unit 22. This is expected to prevent the learning process from stalling when the learning unit 40 performs reinforcement learning for both the normal control unit 21 and the abnormal control unit 22, due to significant changes in the state of the controlled object 30 between normal and abnormal conditions.

[0093] The learning unit 40 may acquire learning data based on the results of a simulation run in the simulator 50 in which the time period and location of the anomaly are randomly set, and then perform learning on the switching unit 10, the normal control unit 21, the anomaly control unit 22, and the recovery control unit 23, or a part of these. As a result, the learning unit 40 can learn about the switching unit 10, the normal control unit 21, the abnormal control unit 22, and the recovery control unit 23 for various abnormal occurrences. This is expected to enable the control system 1 to appropriately control the controlled object 30 in response to various abnormalities during operation.

[0094] As described above, the learning unit 40 separately performs reinforcement learning for normal control by the normal control unit 21 and for abnormal control by the abnormal control unit 22. According to the control system 1f, it is expected that this will prevent the state of the controlled object 30 from changing significantly between normal and abnormal conditions, thus avoiding a situation where learning does not progress.

[0095] Furthermore, the learning unit 40 uses the results of a simulation run in the simulator 50, in which the time period and location of the anomaly are randomly set, to learn the normal control unit 21's control during normal times and the abnormal control unit 22's control during abnormal times. According to the control system 1f, the learning unit 40 can learn about the switching unit 10, the normal control unit 21, the abnormal control unit 22, and the recovery control unit 23 for various cases of abnormality. As a result, it is expected that when the control system 1 is in operation, the control system 1 will be able to appropriately control the controlled object 30 in response to various abnormalities.

[0096] Figure 11 shows a seventh configuration example of the control system according to the embodiment. In the configuration shown in Figure 11, the control system 610 comprises a switching unit 611 and a control unit 612. In this configuration, the switching unit 611 switches between normal control and abnormal control for the controlled portion based on the state of at least one portion of the controlled portion. The control unit 612 controls the controlled portion in response to the control switching by the switching unit 611. The switching unit 611 is an example of a switching means. The control unit 612 is an example of a control means.

[0097] In this way, in response to an abnormality in the controlled object, the control unit 612 partially switches the control of the controlled object. For the parts of the controlled object that were not subject to the control switch, the control is performed under normal conditions to bring the state of those parts back to the target normal state. As a result, it is expected that the control system 610 will have a relatively small difference between the state of the controlled object under the control switched over in response to an abnormality in the controlled object and the state of the controlled object under normal conditions. Because this difference is small, it is expected that the controlled system will return to its normal state relatively quickly after the anomaly is resolved. Furthermore, because this difference is small, the work performed by the personnel handling the controlled system will be similar to their normal work, and in this respect, it is expected that the workload of the personnel handling the controlled system will be reduced.

[0098] Here, let's consider the case where the entire controlled system is controlled by a single control unit obtained through reinforcement learning. If an anomaly occurs in the controlled system, a single control unit controlling the entire system may cause parts other than the anomaly to perform actions different from normal operation.

[0099] In contrast, in the control system 610, the switching unit 611 partially switches the control when an abnormality occurs, so that the parts of the controlled object other than the area where the abnormality occurred are expected to maintain normal operation or operate in a manner close to normal operation. In this way, the control system 610 makes it relatively easy to predict the control of the controlled object. Furthermore, the control system 610 is expected to prevent localized abnormalities in the controlled object from affecting the entire controlled object.

[0100] The switching unit 611 can be implemented using, for example, the functions of the switching unit 10 shown in Figure 1. The control unit 612 can be implemented using, for example, the functions of the control unit 20 shown in Figure 1.

[0101] Figure 12 shows an example of the processing steps in the control method according to the embodiment. The control method shown in Figure 12 includes switching the control (step S611) and performing the control (step S612). In switching the control (step S611), the computer switches between normal control and abnormal control for the controlled part based on the state of at least one of the controlled parts. In performing the control (step S612), the computer performs control of the controlled part in accordance with the control switch.

[0102] In this way, by partially switching the control of the controlled object in response to an abnormality in the controlled object, the parts of the controlled object that were not subject to the control switch are controlled under normal conditions to bring those parts back to the target normal state. As a result, in the control method shown in Figure 12, it is expected that the difference between the state of the controlled object under the switched-out control in response to an abnormality and the state of the controlled object under normal conditions will be relatively small. Because this difference is small, it is expected that the controlled system will return to its normal state relatively quickly after the anomaly is resolved. Furthermore, because this difference is small, the work performed by the personnel handling the controlled system will be similar to their normal work, and in this respect, it is expected that the workload of the personnel handling the controlled system will be reduced.

[0103] Here, let's consider the case where the entire controlled system is controlled by a single control unit obtained through reinforcement learning. If an anomaly occurs in the controlled system, a single control unit controlling the entire system may cause parts other than the anomaly to perform actions different from normal operation.

[0104] In contrast, the control method shown in Figure 12 partially switches the control when an abnormality occurs, so that the parts of the controlled object other than the area where the abnormality occurred are expected to maintain normal operation or operate in a manner close to normal operation. Thus, the control method shown in Figure 12 makes it relatively easy to predict the control of the controlled object. Furthermore, the control method shown in Figure 12 is expected to prevent localized abnormalities in the controlled object from affecting the entire controlled object.

[0105] Figure 13 is a schematic block diagram showing the configuration of a computer according to at least one embodiment. In the configuration shown in Figure 13, the computer 700 comprises a CPU 710, a main memory 720, an auxiliary memory 730, an interface 740, and a non-volatile recording medium 750.

[0106] Any one or more of the above-mentioned control system 1, control device 100, control device 200, switching device 310, normal control device 321, abnormal control device 322, recovery control device 323, control system 1e, and control system 1f, or a part thereof, may be implemented in the computer 700. In that case, the operation of each of the above-mentioned processing units is stored in the auxiliary storage device 730 in the form of a program. The CPU 710 reads the program from the auxiliary storage device 730, expands it into the main memory device 720, and executes the above-mentioned processing according to the program. The CPU 710 also allocates memory areas in the main memory device 720 corresponding to each of the above-mentioned memory units according to the program. Communication between each device and other devices is performed by the interface 740 having a communication function and communicating according to the control of the CPU 710.

[0107] When the control system 1 is implemented in the computer 700, the operation of the switching unit 10, the control unit 20, and each of these units is stored in the auxiliary storage device 730 in the form of a program. The CPU 710 reads the program from the auxiliary storage device 730, loads it into the main memory 720, and executes the above processing according to the program.

[0108] Furthermore, the CPU 710 reserves memory space in the main memory 720 for processing the control system 1 according to the program. Communication between the control system 1 and other devices is performed by the interface 740 having a communication function and operating according to the control of the CPU 710. Interaction between the control system 1 and the user is performed by the interface 740 being equipped with a display device and an input device, displaying various images according to the control of the CPU 710, and accepting user operations.

[0109] When the control device 100 is implemented in the computer 700, the operation of the processing unit 190 and each of its parts is stored in auxiliary storage device 730 in the form of a program. The CPU 710 reads the program from the auxiliary storage device 730, loads it into the main memory device 720, and executes the above processing according to the program.

[0110] Furthermore, the CPU 710 reserves a memory area for the memory unit 180 in the main memory 720 according to the program. Communication with other devices by the communication unit 110 is performed by the interface 740 having a communication function and operating under the control of the CPU 710. Display of images by the display unit 120 is performed by the interface 740 having a display device and displaying various images under the control of the CPU 710. Acceptance of user operations by the operation input unit 130 is performed by the interface 740 having an input device and accepting user operations under the control of the CPU 710.

[0111] When the control device 200 is implemented in the computer 700, the operations of the switching unit 10c, the normal control unit 21, the abnormal control unit 22, and the recovery control unit 23 are stored in the auxiliary storage device 730 in the form of a program. The CPU 710 reads the program from the auxiliary storage device 730, loads it into the main memory 720, and executes the above processes according to the program.

[0112] Furthermore, the CPU 710 reserves memory space in the main memory 720 for processing by the control device 200 according to the program. Communication between the control device 200 and other devices is performed by the interface 740 having a communication function and operating according to the control of the CPU 710. Interaction between the control device 200 and the user is performed by the interface 740 being equipped with a display device and an input device, displaying various images according to the control of the CPU 710, and accepting user operations.

[0113] When the switching device 310 is implemented in the computer 700, its operation is stored in auxiliary storage device 730 in the form of a program. The CPU 710 reads the program from auxiliary storage device 730, loads it into main memory 720, and executes the above process according to the program.

[0114] Furthermore, the CPU 710 reserves memory space in the main memory 720 for processing the switching device 310 according to the program. Communication between the switching device 310 and other devices is performed by the interface 740 having a communication function and operating under the control of the CPU 710. Interaction between the switching device 310 and the user is performed by the interface 740 equipped with a display device and an input device, displaying various images under the control of the CPU 710 and accepting user operations.

[0115] When the normal control device 321 is implemented in the computer 700, its operation is stored in the auxiliary storage device 730 in the form of a program. The CPU 710 reads the program from the auxiliary storage device 730, loads it into the main memory 720, and executes the above process according to the program.

[0116] Furthermore, the CPU 710 reserves memory in the main memory 720 for processing by the normal control unit 321 according to the program. Communication between the normal control unit 321 and other devices is performed by the interface 740 having a communication function and operating according to the control of the CPU 710. Interaction between the normal control unit 321 and the user is performed by the interface 740 equipped with a display device and an input device, displaying various images according to the control of the CPU 710 and accepting user operations.

[0117] When the abnormality control device 322 is implemented in the computer 700, its operation is stored in auxiliary storage device 730 in the form of a program. The CPU 710 reads the program from auxiliary storage device 730, loads it into main memory device 720, and executes the above process according to the program.

[0118] Furthermore, the CPU 710 reserves memory in the main memory 720 for processing by the abnormality control device 322 according to the program. Communication between the abnormality control device 322 and other devices is performed by the interface 740 having a communication function and operating according to the control of the CPU 710. Interaction between the abnormality control device 322 and the user is performed by the interface 740 being equipped with a display device and an input device, displaying various images according to the control of the CPU 710, and accepting user operations.

[0119] When the recovery control device 323 is implemented in the computer 700, its operation is stored in auxiliary storage device 730 in the form of a program. The CPU 710 reads the program from auxiliary storage device 730, loads it into main memory device 720, and executes the above process according to the program.

[0120] Furthermore, the CPU 710 reserves memory space in the main memory 720 for processing by the recovery control device 323 according to the program. Communication between the recovery control device 323 and other devices is performed by the interface 740 having a communication function and operating according to the control of the CPU 710. Interaction between the recovery control device 323 and the user is performed by the interface 740 equipped with a display device and an input device, displaying various images according to the control of the CPU 710 and accepting user operations.

[0121] When the control system 1e is implemented in the computer 700, the operation of the switching unit 10, the control unit 20e, and each of its components is stored in the auxiliary storage device 730 in the form of a program. The CPU 710 reads the program from the auxiliary storage device 730, loads it into the main memory 720, and executes the above processing according to the program.

[0122] Furthermore, the CPU 710 reserves memory space in the main memory 720 for processing the control system 1e according to the program. Communication between the control system 1e and other devices is performed by the interface 740 having a communication function and operating according to the control of the CPU 710. Interaction between the control system 1e and the user is performed by the interface 740 being equipped with a display device and an input device, displaying various images according to the control of the CPU 710, and accepting user operations.

[0123] When the control system 1f is implemented in the computer 700, the operation of the switching unit 10, the control unit 20, the learning unit 40, and each of these units is stored in the auxiliary storage device 730 in the form of a program. The CPU 710 reads the program from the auxiliary storage device 730, loads it into the main memory 720, and executes the above processing according to the program.

[0124] Furthermore, the CPU 710 reserves memory in the main memory 720 for processing the control system 1f according to the program. Communication between the control system 1f and other devices is performed by the interface 740 having a communication function and operating according to the control of the CPU 710. Interaction between the control system 1f and the user is performed by the interface 740 being equipped with a display device and an input device, displaying various images according to the control of the CPU 710, and accepting user operations.

[0125] One or more of the above-mentioned programs may be recorded on the non-volatile recording medium 750. In this case, the interface 740 may read the program from the non-volatile recording medium 750. The CPU 710 may then either directly execute the program read by the interface 740, or temporarily save it in the main memory 720 or auxiliary memory 730 before executing it.

[0126] Alternatively, programs for executing all or part of the processing performed by control system 1, control device 100, control device 200, switching device 310, normal control device 321, abnormal control device 322, recovery control device 323, control system 1e, and control system 1f may be recorded on a computer-readable recording medium, and the programs recorded on this recording medium may be loaded into a computer system and executed to perform the processing of each part. The term "computer system" here includes hardware such as the operating system and peripheral devices. Furthermore, "computer-readable recording media" refers to portable media such as flexible disks, magneto-optical disks, ROMs (Read Only Memory), CD-ROMs (Compact Disc Read Only Memory), and storage devices such as hard disks built into computer systems. The above-mentioned program may be intended to implement only a part of the functions described above, and may also be able to implement the above-mentioned functions in combination with programs already recorded in the computer system.

[0127] Although embodiments of this invention have been described in detail above with reference to the drawings, the specific configuration is not limited to these embodiments and includes designs and the like that do not depart from the spirit of this invention. [Explanation of symbols]

[0128] 1, 1b, 1c, 1d, 1e, 1f, 610 control system 10, 10c, 611 switching section 20, 20e, 612 Control Unit 21, 21e Normal Control Unit 22, 22e Abnormal Operation Control Unit 23, 23e Recovery Control Unit 24 Detailed Control Unit 30 Controlled object 40 Learning Department 50 Simulators 100, 200 control devices 110 Communications Department 120 Display section 130 Operation Input Section 180 Storage section 190 Processing Unit 310 Switching device 321 Normal Control System 322 Abnormal Operation Control System 323 Recovery control device

Claims

1. A switching means that switches between normal control and abnormal control for each section of the train's route, which is the part of the railway being controlled, based on the state of at least one part of the controlled section, and when a section is out of service, determines the control of each section from one section where turnaround operation is possible to the section including the out-of-service section to the next section where turnaround operation is possible to the abnormal control, A control means that controls the part of the controlled object in accordance with the switching of control by the switching means, and controls the part of the controlled object in the non-normal control using a control method obtained by learning with an evaluation function that gives a higher evaluation the closer the state of the controlled object is to the state defined as the target state of the controlled object in normal conditions, A control system equipped with the following features.

2. The switching means switches between abnormal control, which performs control in response to an abnormality occurring in a part of the controlled object, and recovery control, which performs control to restore the state of the part of the controlled object to its normal state, as the control during the abnormal period. The control means further controls the portion to be controlled in accordance with the switching of the control during the non-normal state by the switching means. The control system according to claim 1.

3. The control means uses a control method obtained through learning using the evaluation function, in the control during non-normal times, where the evaluation is higher the closer the state of the controlled object is to the state defined as the target state of the controlled object at each time during normal times. The control system according to claim 1 or claim 2.

4. The control means uses a control method obtained through learning using the evaluation function, in the control during non-normal times, where the evaluation is higher the closer the train's operating status is to the operating status shown in the railway's normal operating timetable. The control system according to claim 3.

5. A learning means that performs reinforcement learning for learning the control under normal conditions using the learning data for learning the control under normal conditions, and for learning the control under non-normal conditions using the learning data for learning the control under non-normal conditions. A control system according to any one of claims 1 to 4, further comprising the above.

6. The learning means uses the results of a simulation in which the time period and location of the abnormality are randomly set for the simulator of the controlled object to perform the learning of the control during normal times and the learning of the control during abnormal times. The control system according to claim 5.

7. Computers For each section of the train's route, which is the part of the railway being controlled, the control system switches between normal and abnormal control based on the state of at least one part of the controlled section. If a section is out of service, the control system for each section between the section where turnaround operations are possible and the section including the out-of-service section is set to the abnormal control system. In response to the switching of the control, the part of the controlled object is controlled, and in the control during the non-normal state, the part of the controlled object is controlled using a control method obtained by learning with an evaluation function that gives a higher evaluation the closer the state of the controlled object is to the state defined as the target state of the controlled object during normal state. A control method that includes the following.

8. On the computer, For each section of the train's route, which is the part of the railway being controlled, the system switches between normal and abnormal control based on the state of at least one part of the controlled section. If a section is out of service, the system determines the control for each section between the section where turnaround operations are possible and the section including the out-of-service section to be changed to the abnormal control. In response to the switching of the control, the control of the part to be controlled is performed, and in the control during the non-normal period, the control of the part to be controlled is performed using a control method obtained by learning with an evaluation function that gives a higher evaluation the closer the state of the controlled object is to the state defined as the target state of the controlled object during normal periods. A program to execute.

Images