JP7414474B2 - Data classification device, control method, and program - Google Patents

Description

The present invention relates to a technology for classifying data obtained from train travel routes.

Techniques have been developed to perform estimation (failure prediction, etc.) based on measured data. For example, Patent Document 1 discloses a device for predicting a failure state of a motor brake. This device achieves failure detection by learning the normal state using data obtained from the motor's brakes during normal times.

JP2019-146421A

The present inventor has studied a method of estimating equipment such as railroad tracks using data obtained from train travel routes (for example, voltage of a control line attached to the railroad tracks, etc.). Through this study, they discovered that data obtained from driving routes can vary due to the influence of things other than the state of the measurement target (control lines, etc.).

The present invention has been made in view of the above problems, and one of its purposes is to provide a technology that realizes estimation using data obtained on a train travel route with higher accuracy.

The data classification device of the present invention includes: 1) an acquisition unit that acquires a plurality of pieces of monitoring data regarding a monitored section on a train running route; and a classification unit that divides the monitoring data into a plurality of groups according to the state of the monitoring target section at the time of measurement of the monitoring data.
The status of the monitored section is represented by the on-track status, which indicates a combination of the presence or absence of trains in each blocked section included in the monitored section, or the status of the traffic lights included in the monitored section.

The control method of the present invention is executed by a computer. The control method includes 1) an acquisition step of acquiring a plurality of pieces of monitoring data for a monitored section on a train travel route, and 2) a step of acquiring a plurality of pieces of monitoring data for use in estimation regarding equipment installed in the monitored section. and a classification step of dividing the information into a plurality of groups according to the state of the monitored section at the time of measurement of the monitoring data.
The status of the monitored section is represented by the on-track status, which indicates a combination of the presence or absence of trains in each blocked section included in the monitored section, or the status of the traffic lights included in the monitored section.

A technology is provided that realizes estimation using data obtained from train travel routes with higher accuracy.

FIG. 1 is a diagram for explaining an overview of a data classification device according to the present embodiment. 1 is a diagram illustrating a functional configuration of a data classification device according to a first embodiment; FIG. FIG. 2 is a diagram illustrating a computer for realizing a data classification device. 3 is a flowchart illustrating the flow of processing executed by the data classification device of the first embodiment. FIG. 3 is a diagram illustrating the relationship between the track status and the traffic signal status in a monitored section. FIG. 3 is a diagram illustrating the relationship between the track status and the traffic signal status in a monitored section. 3 is a block diagram illustrating a functional configuration of a data classification device according to a second embodiment. FIG. 7 is a flowchart illustrating the flow of processing executed by the data classification device of the second embodiment.

Embodiments of the present invention will be described below with reference to the drawings. Note that in all the drawings, similar components are denoted by the same reference numerals, and descriptions thereof will be omitted as appropriate. Furthermore, in each block diagram, unless otherwise specified, each block represents a configuration of a functional unit rather than a configuration of a hardware unit.

[Embodiment 1]
<Summary>
FIG. 1 is a diagram for explaining an overview of a data classification device 2000 of this embodiment. Note that FIG. 1 is an example to facilitate understanding of the data classification device 2000, and the functions of the data classification device 2000 are not limited to those shown in FIG.

The data classification device 2000 handles classification of monitoring data 40 used for estimation regarding equipment installed within the monitored section 10 (for example, equipment failure prediction). The monitored section 10 is a part or all of the running section of the train 30, and is, for example, a section whose end point is a place (for example, a station) where two control devices that exchange control signals are installed. For example, when control signals are exchanged between a control device installed at A station and a control device installed at B station, the area between A station and B station can be treated as the monitoring target section 10. .

Signals sent and received between control devices include, for example, 1) control signals to maintain the current running direction of the train, and 2) in a single track section (a section where the track is shared by up and down trains). There are control signals to display the driving direction, etc. Note that the line on which the control signal in 1) flows is called an FC line, and the line on which the control signal in 2) flows is called an FK line. Hereinafter, lines through which control signals flow, such as FC lines and FK lines, are collectively referred to as control lines. The end points of the monitored section 10 are connected by a control line.

Note that the monitoring target section 10 is not limited to a single-track section, but may be a double-track section. Control signals exchanged between control devices in a double-track section include, for example, various control signals used in control of a double-track automatic closing system.

Various equipment set within the monitoring target section 10 can be used as the equipment to be estimated. For example, the equipment includes railroad tracks, blocking devices, signaling devices, switch devices, interlocking devices, train central control devices, automatic train stopping devices, or level crossing safety devices.

The monitoring data 40 indicates measured values obtained for the monitoring target section 10. The monitoring data 40 indicates, for example, measured values regarding the control line described above. Specific examples include the voltage applied to the control line and the magnitude of the current flowing through the control line. The measuring device (measuring device 60 described later) for obtaining the monitoring data 40 includes a sensor (for example, a current sensor, a voltage sensor, etc.) for obtaining the above-mentioned measured values. The measuring device may be provided within the control device described above, or may be provided outside the control device. In the latter case, the measuring device can be placed anywhere on the control line.

The monitoring data 40 is classified according to the state of the monitoring target section 10 at the time when the monitoring data 40 is measured (the time when the measurement is performed by the measuring device). Here, the group used to classify the monitoring data 40 is referred to as a status group 50. A state group 50 is provided for each possible state of the monitored section 10. For each piece of monitoring data 40, the data classification device 2000 identifies the state of the monitored section 10 at the time of measurement of the monitoring data 40, and assigns the monitoring data 40 to the state group 50 corresponding to the state (state group 50 (add monitoring data 40 to the above).

The state of the monitored section 10 is expressed by the on-track state of the monitored section 10 (where in the monitored section 10 the train 30 is located) or a combination of the states of each signal in the monitored section 10. In the former case, a status group 50 is provided for each on-line status of the monitored section 10. In the latter case, a state group 50 is provided for each combination of the states of each traffic light in the monitored section 10.

<Example of effects>
The measurement data (monitoring data 40) obtained from the control line and the like described above may vary depending on the state of the monitored section 10 (the state of the line and the state of the traffic lights), as will be described later. Therefore, in order to accurately perform estimation using the monitoring data 40, it is important to distinguish between data fluctuations due to the influence of the state of the monitored section 10 and data fluctuations due to equipment failure or the like.

In this regard, according to the data classification device 2000 of the present embodiment, each piece of monitored data 40 is assigned to a status group 50 according to the status of the monitored section 10 at the time of measurement of the monitored data 40. Classification is done. Then, if estimation is performed using the monitoring data 40 grouped in this way, it is possible to make an estimation that separates fluctuations in measured values due to the influence of the state of the monitored section 10 from fluctuations in measured values due to equipment failure, etc. becomes possible. As a result, estimation regarding the equipment using the monitoring data 40 can be performed with higher accuracy. Further, such classification has the advantage that the monitoring data 40 can be utilized in a large amount.

This embodiment will be described in further detail below.

<Example of functional configuration>
FIG. 2 is a diagram illustrating the functional configuration of the data classification device 2000 of the first embodiment. The data classification device 2000 includes an acquisition section 2020 and a classification section 2040. The acquisition unit 2020 acquires a plurality of pieces of monitoring data 40 for the monitoring target section 10. The classification unit 2040 classifies the plurality of acquired monitoring data 40. Specifically, the classification unit 2040 identifies, for each piece of monitoring data 40, the state of the monitoring target section 10 at the time of measurement of the monitoring data 40. Then, the classification unit 2040 assigns each piece of monitoring data 40 to a state group 50 corresponding to the state specified for that piece of monitoring data 40.

<Example of hardware configuration of data classification device 2000>
Each functional component of the data classification device 2000 may be realized by hardware that implements each functional component (e.g., a hardwired electronic circuit, etc.), or by a combination of hardware and software (e.g., (e.g., a combination of an electronic circuit and a program that controls it). A case in which each functional component of the data classification device 2000 is realized by a combination of hardware and software will be further described below.

FIG. 3 is a diagram illustrating a computer 1000 for realizing the data classification device 2000. Computer 1000 is any computer. For example, the computer 1000 is a stationary computer such as a PC (Personal Computer) or a server machine. In addition, for example, the computer 1000 may be a portable computer such as a smartphone or a tablet terminal.

Computer 1000 may be a dedicated computer designed to implement data classification device 2000, or may be a general-purpose computer. In the latter case, the functions of the data classification device 2000 are realized in the computer 1000 by installing a predetermined application on the computer 1000, for example. The above-mentioned application is an application composed of a program for realizing the functional component of the data classification device 2000.

Computer 1000 has a bus 1020, a processor 1040, a memory 1060, a storage device 1080, an input/output interface 1100, and a network interface 1120. The bus 1020 is a data transmission path through which the processor 1040, memory 1060, storage device 1080, input/output interface 1100, and network interface 1120 exchange data with each other. However, the method for connecting the processors 1040 and the like to each other is not limited to bus connection.

The processor 1040 is a variety of processors such as a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or an FPGA (Field-Programmable Gate Array). Memory 1060 is a main storage device implemented using RAM (Random Access Memory) or the like. The storage device 1080 is an auxiliary storage device implemented using a hard disk, an SSD (Solid State Drive), a memory card, a ROM (Read Only Memory), or the like.

The input/output interface 1100 is an interface for connecting the computer 1000 and an input/output device. For example, an input device such as a keyboard, and an output device such as a display device are connected to the input/output interface 1100.

Network interface 1120 is an interface for connecting computer 1000 to a communication network. This communication network is, for example, a LAN (Local Area Network) or a WAN (Wide Area Network). In FIG. 3, computer 1000 is communicably connected to measurement device 60 via network interface 1120. The measuring device 60 is a device that performs measurements and generates the monitoring data 40. The measuring device 60 repeatedly performs measurements to generate monitoring data 40 in which measurement points and measured values are associated with each other in each measurement. The timing at which the measuring device 60 performs measurements may be regular timing or irregular timing. Note that the measurement device 60 and the data classification device 2000 (computer 1000) do not need to be communicably connected.

The storage device 1080 stores programs that implement each functional component of the data classification device 2000 (programs that implement the aforementioned applications). The processor 1040 reads each of these programs into the memory 1060 and executes them, thereby realizing the functions corresponding to each program.

<Processing flow>
FIG. 4 is a flowchart illustrating the flow of processing executed by the data classification device 2000 of the first embodiment. The acquisition unit 2020 acquires a plurality of pieces of monitoring data 40 for the monitoring target section 10 (S102). S104 to S110 are loop processing A that is repeatedly performed while changing the target monitoring data 40. In S104, the classification unit 2040 determines whether loop processing A has been executed for all of the monitoring data 40. If the loop process A has already been executed for all the monitoring data 40, the process in FIG. 4 proceeds to S112.

On the other hand, if there is monitoring data 40 that has not yet been subjected to loop processing A, the classification unit 2040 selects one of them. The monitoring data 40 selected here is called monitoring data i.

The classification unit 2040 identifies the state of the monitoring target section 10 at the time of measurement of the monitoring data i (S106). The classification unit 2040 assigns the monitoring data i to a group corresponding to the identified state (S108). Since S110 is the end of loop processing A, the processing in FIG. 4 returns to S104.

After loop processing A, the data classification device 2000 outputs the monitoring data 40 classified into groups (S112).

The flow of processing performed by the data classification device 2000 is not limited to the flow shown in FIG. 4. For example, the data classification device 2000 acquires the monitoring data 40 one by one, and each time it acquires one monitoring data 40, the data classification device 2000 identifies the state of the monitoring target section 10 corresponding to the monitoring data 40 (S106), Allocating the state group 50 (S108) and outputting the monitoring data 40 associated with the state group 50 (S110) may be executed.

<About the status of monitoring target section 10>
In order to classify the monitoring data 40, a state group 50 is provided that corresponds to each of a plurality of states that the monitoring target section 10 can take. As described above, the state of the monitored section 10 is determined by the track presence state of the monitored section 10 or a combination of the states of each traffic light included in the monitored section 10. Each case will be explained below.

<<About the track status of monitoring target section 10>>
The operation of the trains 30 is controlled so that only one train 30 exists in one blocked section at a time. Therefore, the track status of the monitored section 10 can be grasped for each blocked section. Specifically, the on-track status of the monitored section 10 can be expressed by a combination of the presence or absence of the train 30 in each blocked section included in the monitored section 10. Note that the monitored section 10 may include one or more blocked sections.

For example, assume that the monitored section 10 includes three blocked sections A to C. In this case, the track status of the monitored section 10 is represented by a combination of "the presence or absence of the train 30 in the blocked section A, the presence or absence of the train 30 in the blocked section B, and the presence or absence of the train 30 in the blocked section C." Therefore, there are a total of 8 (2^3) line statuses. Therefore, a maximum of eight state groups 50 are prepared.

The current and voltage of the control line can be affected not only by failures of equipment provided in the monitored section 10, but also by the state of the line in the monitored section 10. For example, when a train is between stations, the FK line voltage is set to 0 to prevent directional line switching.

Therefore, for example, the data classification device 2000 classifies the monitoring data 40 according to the on-track status of the trains 30 in the monitoring target section 10. Thereby, it is possible to distinguish between a change in the monitoring data 40 according to the on-track status of the train 30 in the monitored section 10 and a change in the monitoring data 40 due to equipment failure or the like. Therefore, it becomes possible to estimate failures of equipment provided in the monitored section 10 with higher accuracy.

<<About traffic light status>>
The state of the monitored section 10 may be expressed by a combination of the states of each traffic light included in the monitored section 10. For example, if the monitored section 10 includes three traffic lights A to C, the combination of the states of the traffic lights included in the monitored section 10 is called "the state of traffic light A, the state of traffic light B, and the state of traffic light C". means a combination. The state of the traffic light is, for example, one of red, yellow, and blue.

Here, the combination of traffic light states can represent the on-track state in the monitored section 10. FIGS. 5 and 6 are diagrams illustrating the relationship between the track status and the traffic signal status of the monitored section 10. In this example, the monitored section 10 is a single track section. Further, for each direction of travel, traffic lights are provided at the entrance to the blocked section and the entrance to the station. In addition, in FIGS. 5 and 6, for convenience of illustration, the green light, yellow light, and red light are respectively represented by white paint, a dot pattern, and a checkered pattern.

In the example of FIGS. 5 and 6, the train 30 moves from station B to station A. Therefore, all the traffic lights in the direction from Station A to Station B are red.

The signal at the entrance of the blocked section where the train 30 is present is set to red. Furthermore, in this example, the signal at the entrance of the blocked section where the train 30 does not exist is set to yellow if the train 30 exists in the blocked section one block ahead (if the signal one ahead is red). , otherwise it is set to blue.

Since the signal is controlled according to the above rules, the state of the signal changes every time the blocked section in which the train 30 is present changes. For example, when the train 30 moves to the blocked section 3 (middle row in FIG. 5), the traffic light at the entrance to the blocked section 3 turns red. Further, when the train 30 moves to the blocked section 2 (lower part of FIG. 5), the traffic light at the entrance to the blocked section 2 turns red, and the signal at the entrance to the blocked section 3 turns yellow. Furthermore, when the train 30 moves to the blocked section 1 (upper row of FIG. 6), the traffic light at the entrance to the blocked section 1 turns red, and the signal at the entrance to the blocked section 2 turns yellow. Also, at this time, the color of the traffic light at the entrance to the blocked section 3 becomes blue.

Note that although there is only one train 30 in the examples of FIGS. 5 and 6, there may be a plurality of trains 30 in the monitored section 10. For example, in the example shown in the lower part of FIG. 5, another train 30 may enter the blocked section 3. In this case, the traffic light at the entrance to the blocked section 3 is set to red.

As described above, the combination of traffic light states can represent the on-line state of the monitored section 10. However, the combination of traffic light states and the on-line state of the monitored section 10 do not necessarily have a one-to-one correspondence. In other words, even if the line status of the monitored section 10 does not change, the combination of traffic light conditions can change. This is because the settings of the signal installed at the entrance of the station can be changed under conditions other than whether or not the train 30 is present in the station area.

For example, suppose that a person on the platform drops their luggage on the tracks while the traffic light at the entrance of the station is set to green because there is no train 30 in the station. In this case, the station staff picks up the luggage after setting the traffic light at the station entrance to red to prevent the train 30 from entering the station. After the work is completed, the traffic light at the station entrance will be changed to green again. In this way, even if the line status does not change, the signal status at the station entrance can change.

Also, as mentioned above, when operating under the rule that "if the next traffic light is red, the traffic light color is set to yellow", the traffic light at the entrance of the station may be set to red due to circumstances within the station. When set, the color of the traffic light immediately before it is set to yellow. Therefore, the status of signals other than those installed at station entrances can be set based on other than the line presence status.

Even if the line status of the monitored section 10 is the same, a difference in the combination of traffic light conditions can affect the monitoring data 40. That is, the current and voltage of the control line can be affected not only by a failure of equipment provided in the monitored section 10, but also by a combination of states of traffic lights included in the monitored section 10.

Therefore, for example, the data classification device 2000 classifies the monitoring data 40 for each combination of states of traffic lights included in the monitoring target section 10. Thereby, it is possible to distinguish between a change in the monitoring data 40 according to a combination of the states of traffic lights included in the monitored section 10 and a change in the monitoring data 40 due to equipment failure or the like. Therefore, it becomes possible to estimate failures of equipment provided in the monitored section 10 with higher accuracy.

<Acquisition of monitoring data 40: S102>
The acquisition unit 2020 acquires a plurality of pieces of monitoring data 40 (S102). Here, various existing methods can be adopted as a method for acquiring data obtained by measurement by the measuring device. For example, the measurement device 60 stores the monitoring data 40 in a storage device that is also accessible from the data classification device 2000. In this case, the acquisition unit 2020 acquires the monitoring data 40 by accessing this storage device. Note that this storage device may be provided inside the measuring device 60 or may be provided outside the measuring device 60.

In addition, for example, the measurement device 60 may transmit the monitoring data 40 to the data classification device 2000. In this case, the acquisition unit 2020 acquires the monitoring data 40 by receiving the monitoring data 40 transmitted from the measuring device 60.

Note that the acquisition unit 2020 may acquire the monitoring data 40 from a device other than the measuring device 60. For example, the aforementioned control device acquires the monitoring data 40 from the measuring device 60 and transmits the monitoring data 40 to the data classification device 2000.

Note that the acquisition unit 2020 may not acquire a plurality of pieces of monitoring data 40 at once, but at different timings. For example, the acquisition unit 2020 acquires the monitoring data 40 every time the monitoring data 40 is newly generated. In addition, for example, the acquisition unit 2020 may acquire unobtained monitoring data 40 at regular or irregular timings.

<Identification of the state of the monitoring target section 10: S106>
The classification unit 2040 identifies, for each piece of monitoring data 40, the state of the monitoring target section 10 at the time of measurement of the monitoring data 40 (S106). As described above, the state of the monitored section 10 is represented by the presence of a line in the monitored section 10 or a combination of the states of the respective traffic lights included in the monitored section 10.

Assume that the state of the monitored section 10 is represented by the track presence state of the monitored section 10. In this case, the classification unit 2040 identifies the presence or absence of the train 30 in each blocked section included in the monitored section 10, thereby identifying the on-track status of the monitored section 10. Note that an existing method can be used to identify the presence or absence of the train 30 in each blocked section. For example, the presence or absence of a train 30 in a blocked section can be ascertained from the state of the track circuit configured in the blocked section. Therefore, the classification unit 2040 identifies the presence or absence of the train 30 in each blocked section by acquiring information representing the state of the track circuit in each blocked section.

It is assumed that the state of the monitored section 10 is represented by a combination of the states of the respective traffic lights included in the monitored section 10. In this case, the classification unit 2040 identifies the state of the monitored section 10 by identifying the state of each traffic light included in the monitored section 10. Here, an existing method can be used to grasp the state of each traffic light. For example, assume that the status of each traffic light is managed by a management device that is connected to each traffic light via a network. In this case, for example, the classification unit 2040 acquires information representing the state of each traffic light from this management device.

Here, in order to specify the state of the monitored section 10 at the time of measurement of the monitoring data 40, it is necessary to understand the transition of the state of the monitored section 10 (correspondence between time and the state of the monitored section 10). There is a need. For example, the classification unit 2040 periodically specifies the state of the monitored section 10 using the method described above, and assigns information representing the state of the monitored section 10 (for example, an identifier assigned to each possible state) at the specified time. and store it in a storage device. The classification unit 2040 uses the information stored in this storage device to identify, for each piece of monitoring data 40, the state of the monitored section 10 at the time of measurement of the monitoring data 40.

<Classification of monitoring data 40: S108>
The classification unit 2040 classifies the monitoring data 40 by assigning each of the plurality of monitoring data 40 to a state group 50 corresponding to the state of the monitored section 10 at the time of measurement of the monitoring data 40 (S108). Note that any method can be adopted as a specific method for allocating data to one of a plurality of groups.

For example, a storage area (for example, a file) for each status group 50 is prepared in a storage device that stores the monitoring data 40. The classification unit 2040 assigns the monitored data 40 to the state group 50 by adding the monitored data 40 to the storage area corresponding to the state group 50 to which the monitored data 40 belongs. In addition, for example, the classification unit 2040 may classify the monitoring data 40 by generating information that associates the identifier of the state group 50 with the monitoring data 40 belonging to that state group 50.

<Output of classification results: S112>
The data classification device 2000 outputs the monitoring data 40 classified into each state group 50 (S112). Any method may be used to output information regarding the classification results by the classification unit 2040 (that is, the monitoring data 40 included in each state group 50). For example, the monitoring data 40 classified into each state group 50 is stored in a storage device that is accessible from the data classification device 2000. In addition, for example, the monitoring data 40 classified into each state group 50 may be transmitted to another device connected to the data classification device 2000. The monitoring data 40 thus classified and output for each status group 50 is used, for example, for estimation regarding equipment provided in the monitoring target section 10, as described in the second embodiment.

[Embodiment 2]
FIG. 7 is a block diagram illustrating the functional configuration of the data classification device 2000 of the second embodiment. Except for the points described below, the data classification device 2000 of the second embodiment has the same functions as the data classification device 2000 of the first embodiment.

The data classification device 2000 of the second embodiment includes an estimator 2060. For each state group 50, the estimating unit 2060 uses the monitoring data 40 included in the state group 50 to perform estimation regarding the equipment provided in the monitored section 10. The equipment here is any equipment that has some relationship with the monitoring data 40. For example, when the monitoring data 40 indicates a measured value regarding a control line, the equipment to be estimated may be one that affects the signal flowing through the control line. As mentioned above, such equipment includes railroad tracks, blocking devices, and the like.

The estimation performed by the estimation unit 2060 is, for example, detection or prediction of a failure of equipment provided in the monitored section 10. Here, the estimation performed by the estimation unit 2060 may be identification or regression. In the former case, for example, the estimating unit 2060 may perform functions such as "identifying whether the equipment will fail after a predetermined period of time," "identifying whether the equipment is currently malfunctioning," and "identifying whether the predicted failure will occur in the future." identification of which period it belongs to among multiple periods. Examples of multiple future periods include four categories: "Within six months," "After six months but within one year," "After one year but within two years," and "After two years." Examples include the period during which

When the estimation unit 2060 performs regression, for example, the estimation unit 2060 estimates “probability that the equipment will fail after a predetermined period of time”, “probability that the equipment is currently failing”, “time until equipment failure”, etc. .

Note that the estimation performed by the estimation unit 2060 is not limited to estimation regarding equipment failure. For example, the estimating unit 2060 may be able to "estimate when equipment needs to be replaced," "identify whether equipment needs to be replaced after a predetermined period," or "identify whether equipment needs to be replaced at this point." You may also make an estimate regarding the replacement of equipment, such as ``.

<Example of effects>
The measurement data (monitoring data 40) obtained from the control line and the like described above may vary depending on the state of the monitored section 10 (the state of the line and the state of the traffic lights), as will be described later. Therefore, in order to accurately perform estimation using the monitoring data 40, it is important to distinguish between data fluctuations due to the influence of the state of the monitored section 10 and data fluctuations due to equipment failure or the like.

In this regard, according to the data classification device 2000 of the present embodiment, each piece of monitored data 40 is assigned to a status group 50 according to the status of the monitored section 10 at the time of measurement of the monitored data 40. Classification is done. Then, if estimation is performed using the monitoring data 40 grouped in this way, it is possible to make an estimation that separates fluctuations in measured values due to the influence of the state of the monitored section 10 from fluctuations in measured values due to equipment failure, etc. becomes possible. As a result, estimation regarding the equipment using the monitoring data 40 can be performed with higher accuracy.

The details of this embodiment will be further explained below.

<Example of hardware configuration>
The hardware configuration of the data classification device 2000 of the second embodiment is represented, for example, in FIG. 3, similar to the hardware configuration of the data classification device 2000 of the first embodiment. However, the storage device 1080 of the second embodiment includes a program for realizing the functions of the data classification device 2000 of the second embodiment.

<Processing flow>
FIG. 8 is a flowchart illustrating the flow of processing executed by the data classification device 2000 of the second embodiment. The flowchart in FIG. 8 corresponds to the flowchart in FIG. 4 with S202 added. In S<b>202 , the estimation unit 2060 performs estimation regarding the equipment provided in the monitored section 10 for each state group 50 using the monitoring data 40 included in that state group 50 .

<Estimation by estimation unit 2060: S202>
The estimating unit 2060 uses the monitoring data 40 divided into each state group to perform estimation regarding the equipment provided in the monitored section 10 (S202). For example, an estimator for estimating equipment is provided for each state group. Various types of estimator models can be used, such as neural networks, SVM, heterogeneous mixture learning, and RAPID machine learning.

For each state group 50, the estimation unit 2060 inputs the monitoring data 40 included in that state group 50 to the estimator corresponding to that state group. As a result, an estimation result by the estimator is obtained for each state group 50. The estimation unit 2060 performs estimation regarding equipment using estimation results obtained from each estimator.

Suppose the estimator performs regression. In this case, for example, the estimation unit 2060 calculates the statistical value of the value output from each estimator, and uses the calculated statistical value as the estimation result. For example, statistical values include an average value, a maximum value, a minimum value, and a median value.

Here, a weight may be given to each estimator. In this case, the estimation unit 2060 calculates a statistical value (for example, a weighted average) for the value obtained by multiplying the output of each estimator by the weight assigned to that estimator.

There are various methods for determining the weight to be given to an estimator. For example, the weight given to an estimator is determined based on the number of pieces of monitoring data 40 input to the estimator (used for estimation). That is, a larger weight is given to an estimator with a larger number of input monitoring data 40. Alternatively, for example, the accuracy of each estimator may be calculated during learning, and an estimator with higher accuracy may be given a larger weight. Note that various existing methods can be used as a specific method for calculating the accuracy of the estimator.

Suppose that the estimator makes a discrimination. In this case, the estimation unit 2060 identifies the final identification result from the identification results of the plurality of estimators. For example, the estimation unit 2060 obtains the final identification result by making a majority vote based on the estimation results of a plurality of estimators. That is, the estimation unit 2060 sets the estimation result output by the largest number of estimators as the final estimation result.

For example, it is assumed that there are seven status groups 50, and an estimator that "estimates whether equipment will break down within one month" is provided in association with each status group. In this case, if the number of estimators identified as ``failure'' is greater than the number of estimators identified as ``not failure,'' the estimation unit 2060 estimates that ``the equipment will fail within one month.'' On the other hand, if the number of estimators identified as ``failure'' is smaller than the number of estimators identified as ``not failure,'' estimating unit 2060 estimates that ``the equipment will not fail within one month.''

In addition, for example, when a predetermined percentage or more of estimators indicate a specific identification result, the estimation unit 2060 may use that identification result as the final identification result. For example, if 40% or more of the estimators estimate that "the equipment will break down within one month," the final estimation result is set to "the equipment will break down within one month." In this case, the estimating unit 2060 calculates the ratio of the number of estimators that have identified "equipment failure within one month" to the total number, and if the ratio is 40% or more, the estimation unit 2060 It is estimated that the equipment will break down. On the other hand, if the ratio is less than 40%, the estimation unit 2060 estimates that "the equipment will not break down within one month."

Note that weights may be given to the estimator that performs identification, similarly to the estimator that performs regression. For example, when performing majority voting based on the identification results of estimators, instead of simply counting up the number of estimators and performing majority voting, majority voting is performed by counting up the weights given to each estimator. In addition, in the case of determining whether ``a predetermined proportion or more of the estimators showed a specific identification result'', each estimator that showed the specific identification result is It is determined whether the ratio of the total value of the given weights is greater than or equal to a predetermined ratio.

It is assumed that the estimator described above has been trained in advance. The learning data used for learning the estimator is also divided into state groups 50 based on the state of the monitoring target section 10 using a method similar to that described in the first embodiment, and then used for learning. .

Various existing methods can be used as specific methods for learning the estimator using training data. For example, learning of an estimator that performs regression can be performed using a plurality of learning data configured as a pair of "monitoring data 40 for learning and correct output value." Further, learning of an estimator that performs discrimination can be achieved by using a plurality of learning data configured as a pair of "monitoring data 40 for learning and correct identification result".

<Output of estimation results>
The method of outputting the estimation result by the estimation unit 2060 is arbitrary. For example, information representing the estimation result is output to a display device connected to data classification device 2000, stored in a storage device, or transmitted to another device.

Although the embodiments of the present invention have been described above with reference to the drawings, these are merely examples of the present invention, and combinations of the embodiments described above or various configurations other than those described above may also be adopted.

Part or all of the above embodiments may be described as in the following additional notes, but are not limited to the following.
1. an acquisition unit that acquires a plurality of pieces of monitoring data regarding a monitored section on a train travel route;
a classification unit that divides the plurality of monitoring data into a plurality of groups according to the state of the monitoring target section at the time of measurement of the monitoring data, in order to use it for estimation regarding equipment installed in the monitoring target section; and,
The data classification device is characterized in that the state of the monitored section is represented by the presence of trains in the monitored section or the state of a traffic signal included in the monitored section.
2. The train presence status in the monitored section is determined by a combination of the presence or absence of trains in each of a plurality of blocked sections included in the monitored section; 1. The data classification device described in .
3. 1. The monitoring data indicates a measured value of current or voltage of a control line provided in the monitoring target section. Or 2. The data classification device described in .
4. 1. An estimation unit is provided for each group, using the monitoring data assigned to the group to estimate equipment installed in the monitoring target section. The data classification device according to any one of 3 to 3.
5. 4. The estimation performed by the estimating unit is an estimation regarding failure or replacement of the equipment. The data classification device described in .
6. The estimation unit is
comprising an estimator corresponding to each of the plurality of groups,
inputting the monitoring data assigned to each said group into an estimator corresponding to that group;
4. performing the estimation using the output obtained from each of the estimators; 4. or 5. The data classification device described in .
7. A control method performed by a computer, comprising:
an acquisition step of acquiring a plurality of pieces of monitoring data for a monitored section on the train travel route;
A classification step of dividing the plurality of monitoring data into a plurality of groups according to the state of the monitoring target section at the time of measurement of the monitoring data, in order to use it for estimation regarding equipment installed in the monitoring target section. and,
In the control method, the state of the monitored section is represented by the presence of trains in the monitored section or the state of a signal included in the monitored section.
8. 7. The train presence status in the monitored section is determined by a combination of the presence or absence of a train in each of a plurality of blocked sections included in the monitored section. Control method described in .
9. 7. The monitoring data indicates a measured value of current or voltage of a control line provided in the monitoring target section. or 8. Control method described in .
10. 7. comprising an estimation step of estimating, for each group, equipment provided in the monitoring target section using the monitoring data assigned to the group; 9. The control method according to any one of 9.
11. 10. The estimation in the estimation step is an estimation regarding failure or replacement of the equipment. Control method described in .
12. The computer has an estimator corresponding to each of the plurality of groups,
In the estimation step,
inputting the monitoring data assigned to each said group into an estimator corresponding to that group;
10. performing the estimation using the output obtained from each of the estimators; or 11. Control method described in .
13. 7. A program for causing a computer to execute the control method according to any one of 12 to 12.

10 Monitored section 30 Train 40 Monitored data 50 Status group 60 Measuring device 1000 Computer 1020 Bus 1040 Processor 1060 Memory 1080 Storage device 1100 Input/output interface 1120 Network interface 2000 Data classification device 2020 Acquisition unit 2040 Classification unit 2060 Estimation unit

Claims (8)

an acquisition unit that acquires a plurality of pieces of monitoring data regarding a monitored section on a train travel route;
a classification unit that divides the plurality of monitoring data into a plurality of groups according to the state of the monitoring target section at the time of measurement of the monitoring data, in order to use it for estimation regarding equipment installed in the monitoring target section; and,
The state of the monitored section is represented by a track presence state indicating a combination of the presence or absence of a train in each blocked section included in the monitored section , or a state of a signal included in the monitored section, the data classification device . The data classification device according to claim 1, wherein the train presence status in the monitored section is determined by a combination of the presence or absence of trains in each of a plurality of blocked sections included in the monitored section. 3. The data classification device according to claim 1, wherein the monitoring data indicates a measured value of current or voltage of a control line provided in the monitoring target section. The data classification according to any one of claims 1 to 3, further comprising, for each group, an estimation unit that estimates regarding equipment provided in the monitoring target section using the monitoring data assigned to the group. Device. 5. The data classification device according to claim 4, wherein the estimation performed by the estimation unit is estimation regarding failure or replacement of the equipment. The estimation unit is
comprising an estimator corresponding to each of the plurality of groups,
inputting the monitoring data assigned to each said group into an estimator corresponding to that group;
The data classification device according to claim 4 or 5, wherein the estimation is performed using an output obtained from each of the estimators. A control method performed by a computer, comprising:
an acquisition step of acquiring a plurality of pieces of monitoring data for a monitored section on the train travel route;
A classification step of dividing the plurality of monitoring data into a plurality of groups according to the state of the monitoring target section at the time of measurement of the monitoring data, in order to use it for estimation regarding equipment installed in the monitoring target section. and,
A control method in which the state of the monitored section is represented by an on-track state indicating a combination of the presence or absence of a train in each blocked section included in the monitored section , or a state of a signal included in the monitored section. A program that causes a computer to execute the control method according to claim 7.

Images