JP7031622B2 - Teacher data generator, gate setting learning system, teacher data generation method, and teacher data generation program - Google Patents

Description

The present invention is a teacher data generation device, a gate setting learning system, a teacher data generation method, and a teacher data generation that generate teacher data used for machine learning to make the passage regulation state of each of a plurality of gate devices appropriate. Regarding the program.

Conventionally, an automatic ticket gate that reads a toll ticket such as an admission ticket or a ticket and acquires information contained in the toll ticket has been known. Such automatic ticket gates are installed at traffic security gates in facilities and ticket gates in railways. The automatic ticket gate is equipped with a door (gate) for opening and closing the passage. The automatic ticket gate operates the door in the open position when the toll ticket is properly processed to open the passage through, and closes the door when the toll ticket is not read or defective. Hold it in place and prohibit passage.

Generally, a plurality of passages are divided at the ticket gate of a railway, and a plurality of automatic ticket gates are provided at the ticket gate corresponding to each passage. Conventionally, the station staff who manages each automatic ticket gate predicts the congestion situation near the ticket gate and changes the passage setting of each automatic ticket gate in advance in order to allow visitors and contestants to pass smoothly. , The passage setting of each automatic ticket gate is changed according to the current congestion situation near the ticket gate. Specifically, the station staff set the passage settings for each automatic ticket gate exclusively for entry, for entry only, and for entry and exit.

On the other hand, there are known devices that determine the passage setting by arithmetic processing by a computer regardless of human judgment. For example, a lane operation estimation device for determining the number of operating lanes at a plurality of gates of a tollhouse on a toll road is disclosed based on past operation results and current environmental information (see Patent Document 1). Further, there is disclosed a gate control device that determines a state to be set for each of a plurality of gates based on the position and traveling direction of a passerby estimated from the captured image around the gate (see Patent Document 2). ..

Japanese Unexamined Patent Publication No. 2001-143109 Japanese Unexamined Patent Publication No. 2005-70850

However, in the conventional method of determining the passage setting by arithmetic processing, an appropriate setting value according to the situation around the gate (ticket gate, etc.) is not always output, and a remarkably inappropriate setting value is output. It may be done. In this case, if the passage setting of each gate is automatically changed based on an inappropriate setting value, a passerby or a vehicle driver passing through the gate will be inconvenienced. In addition, the station staff who noticed the deterioration of the congestion situation after the change will have to change the passage setting in a hurry, and the work load of the station staff will increase.

Further, in recent years, it is conceivable to determine the passage setting of each of the plurality of gates by machine learning. However, in the machine learning, a large amount of teacher data is required in order to improve the determination accuracy, and it is difficult to generate a large amount of teacher data corresponding to a plurality of gates capable of setting a plurality of patterns of passages.

An object of the present invention is a teacher data generation device, a gate setting learning system, and a teacher data generation capable of efficiently generating teacher data used for machine learning for setting a passage regulation state of each of a plurality of gate devices. The method is to provide a teacher data generation program.

The teacher data generation device according to one aspect of the present invention is a teacher data generation device that generates teacher data used for machine learning, and has a passage state acquisition unit that acquires the passage regulation state of each of a plurality of gate devices. The congestion degree detection unit that detects the congestion degree of users around the plurality of gate devices in the passage regulation state acquired by the passage state acquisition unit and the congestion degree detected by the congestion degree detection unit are threshold values. Acquired by the congestion determination unit that determines whether or not the data exceeds the above, the congestion time calculation unit that calculates the total congestion time indicating the total time during which the congestion degree exceeds the threshold, and the passage state acquisition unit. It is provided with a teacher data generation unit that generates the teacher data in which the passage regulation state is associated with the total congestion time calculated by the congestion time calculation unit.

According to the above configuration, the teacher data generator does not calculate the total congestion time for all patterns of the passage regulation state composed of the plurality of gate devices, and the total congestion time for the predetermined number of patterns of the passage regulation state. Is calculated. Further, the total congestion time, which is the sum of the times when the degree of congestion exceeds the threshold value, is generated as the teacher data, and the teacher data is not generated for the time when the degree of congestion is equal to or less than the threshold value. Therefore, it is possible to generate specific teacher data that appropriately reflects the congestion situation. Therefore, it is possible to efficiently generate teacher data used for machine learning for setting the passage regulation state of each of the plurality of gate devices.

In the teacher data generation device, the congestion degree detection unit detects the number or density of users around the plurality of gate devices, and the congestion determination unit detects the number of users or the number of users detected by the congestion degree detection unit. It is determined whether or not the density exceeds the threshold value. As a result, it is possible to appropriately determine the congestion status around the plurality of gate devices.

In the teacher data generation device, the congestion time calculation unit is required until the congestion degree becomes equal to or less than the threshold value when the congestion degree is determined by the congestion determination unit to exceed the threshold value in the predetermined period. The total time is calculated as the total congestion time.

In the teacher data generation device, the predetermined period includes a first predetermined period and a second predetermined period, and the passage state acquisition unit has a first passage restricted state corresponding to the first predetermined period and the second predetermined period. The second passage regulation state corresponding to the period is acquired, and the congestion degree detection unit detects the first congestion degree corresponding to the first predetermined period and the second congestion degree corresponding to the second predetermined period. Then, the congestion determination unit determines whether or not the first congestion degree detected by the congestion degree detection unit exceeds the first threshold value, and the second congestion degree detected by the congestion degree detection unit is determined. The congestion time calculation unit determines whether or not the second threshold value is exceeded, and the congestion time calculation unit has a first congestion total time indicating the total time when the first congestion degree exceeds the first threshold value, and the second congestion degree is the said. The second total congestion time, which indicates the total time exceeding the second threshold value, is calculated, and the teacher data generation unit uses the first passage regulation state acquired by the passage state acquisition unit and the congestion time calculation unit. The first teacher data associated with the calculated first total congestion time, the second passage regulation state acquired by the passage state acquisition unit, and the second congestion total calculated by the congestion time calculation unit. Generate a second teacher data associated with time.

Further, the first predetermined period may be a period including the morning of the day, and the second predetermined period may be a period including the afternoon of the day. As a result, it is possible to generate teacher data according to the congestion situation in the morning and teacher data according to the congestion situation in the afternoon. The first threshold value and the second threshold value may be set to different values from each other. For example, the first threshold value and the second threshold value are set to different values depending on the congestion situation.

In the teacher data generation device, the passage state acquisition unit acquires the passage regulation state of a predetermined plurality of patterns, and the congestion degree detection unit obtains the congestion degree corresponding to the passage regulation state of each of the plurality of patterns. Upon detection, the congestion time calculation unit calculates the total congestion time corresponding to the passage regulation state of each of the plurality of patterns, and the teacher data generation unit corresponds to the passage regulation state of each of the plurality of patterns. Generate teacher data. As a result, teacher data can be generated by a plurality of predetermined patterns, so that it is not necessary to prepare a large number of patterns.

In the teacher data generation device, the passage restriction state is a first state that allows passage in only one direction to the passage of the gate device, and a first state that allows passage in only the other direction to the passage of the gate device. It may be one of the two states and the third state that allows passage in both directions to the passage of the gate device. Further, in the teacher data generation device, the gate device is an automatic ticket gate installed at the ticket gate of a railway station and set to any one of the first state, the second state, and the third state. There may be.

The gate setting learning system according to another aspect of the present invention is a learning device that generates a trained model by performing machine learning using the teacher data generation device and the teacher data generated by the teacher data generation device. And prepare. Further, in the gate setting learning system, the learning device generates the trained model for estimating the total congestion time corresponding to the arbitrary passage restriction state. Thereby, it is possible to generate a trained model capable of estimating the total congestion time corresponding to an arbitrary passage regulation state by using the teacher data corresponding to a predetermined number (predetermined pattern) of passage regulation states. ..

In the gate setting learning system, the aisle regulation state corresponding to the shortest total congestion time among the plurality of total congestion times estimated using the trained model generated by the learning device is the optimum passage regulation state. Further equipped with a passage setting device for determining. This makes it possible to set a plurality of gate devices to the optimum passage regulation.

In the gate setting learning system, the teacher data generation device further generates optimum teacher data corresponding to the optimum passage regulation state determined by the passage setting device, and the learning device is used by the teacher data generation device. A modified trained model is generated by modifying the trained model by performing machine learning using the generated teacher data and the optimum teacher data. Further, in the gate setting learning system, the passage setting device corresponds to the shortest total congestion time among the plurality of total congestion times estimated by using the modified trained model generated by the learning device. The passage regulation state is determined to be the new optimum passage regulation state. As a result, the passage regulation of each of the plurality of gate devices can be set to the optimum state according to the congestion situation around the ticket gate.

The teacher data generation method according to another aspect of the present invention is a teacher data generation method for generating teacher data used for machine learning, and includes a passage state acquisition step for acquiring the passage regulation state of each of a plurality of gate devices. The congestion degree detection step for detecting the congestion degree of users around the plurality of gate devices in the passage regulation state acquired by the passage state acquisition step and the congestion degree detected by the congestion degree detection step are threshold values. It is acquired by the congestion determination step for determining whether or not the data exceeds the above, the congestion time calculation step for calculating the total congestion time indicating the total time during which the congestion degree exceeds the threshold, and the passage state acquisition step. In a teacher data generation method in which one or more processors execute the teacher data generation step of generating the teacher data in which the passage regulation state is associated with the total congestion time calculated by the congestion time calculation step. be.

The teacher data generation program according to another aspect of the present invention is a teacher data generation program that generates teacher data used for machine learning, and includes a passage state acquisition step for acquiring the passage regulation state of each of a plurality of gate devices. The congestion degree detection step for detecting the congestion degree of users around the plurality of gate devices in the passage regulation state acquired by the passage state acquisition step and the congestion degree detected by the congestion degree detection step are threshold values. It is acquired by the congestion determination step for determining whether or not the data exceeds the above, the congestion time calculation step for calculating the total congestion time indicating the total time during which the congestion degree exceeds the threshold, and the passage state acquisition step. Teacher data generation for causing one or more processors to execute the teacher data generation step for generating the teacher data in which the passage regulation state is associated with the total congestion time calculated by the congestion time calculation step. It is a program.

According to the present invention, it is possible to efficiently generate teacher data used for machine learning for setting a passage regulation state for each of a plurality of gate devices.

FIG. 1 is a diagram showing a network configuration of a gate system according to an embodiment of the present invention. FIG. 2 is a sketch showing a plurality of automatic ticket gates to which the gate system according to the embodiment of the present invention is applied and the situation around the ticket gate. FIG. 3 is a perspective view showing an automatic ticket gate to which the gate system according to the embodiment of the present invention is applied. FIG. 4 is a block diagram showing a configuration of an automatic ticket gate to which the gate system according to the embodiment of the present invention is applied. FIG. 5 is a block diagram showing a configuration of a station server to which the gate system according to the embodiment of the present invention is applied. FIG. 6 is a block diagram showing the configurations of a teacher data generation device, a learning device, and a passage setting device included in a station server. FIG. 7 is a graph showing an example of a change in the number of users representing a congestion situation around a ticket gate during a day at a station. FIG. 8 is a diagram showing an example of teacher data generated by the teacher data generation device according to the embodiment of the present invention. FIG. 9 is a block diagram showing a configuration of a station employee terminal device to which the gate system according to the embodiment of the present invention is applied. FIG. 10 is a flowchart showing an example of a teacher data generation process executed in the teacher data generation device according to the embodiment of the present invention. FIG. 11 is a flowchart showing an example of the passage setting process executed in the teacher data generation device, the learning device, and the passage setting device according to the embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described with reference to the drawings as appropriate. It should be noted that the embodiments described below are merely examples that embody the present invention, and do not limit the technical scope of the present invention.

[Gate system 100]
FIG. 1 is a network diagram showing a gate system 100 (an example of a gate setting learning system of the present invention) according to an embodiment of the present invention. The gate system 100 is applied to a plurality of automatic ticket gates 1 (an example of the gate device of the present invention) provided at the ticket gate 7 of a railway station 8 operated and managed by a railway operator, and is shown in FIG. As shown, it is configured to include an automatic ticket gate 1, a station employee terminal device 2, a station server 4 (an example of each of the teacher data generation device, the learning device, and the passage setting device of the present invention), a center device 5, and the like. There is. FIG. 1 illustrates a configuration in which a plurality of automatic ticket gates 1 are installed at each of two ticket gates 7 provided at a railway station 8, but the gate system 100 has at least two or more ticket gates at one ticket gate. This is applicable to a configuration in which a plurality of automatic ticket gates 1 are installed in the passage.

The gate system 100 is not limited to the one applied to the plurality of automatic ticket gates 1 of the station 8, and the gate system 100 is not limited to the one applied to the plurality of automatic ticket gates 1 of the station 8. It can also be applied to an apparatus (an example of a gate device of the present invention) and a plurality of automatic gate devices (an example of the gate device of the present invention) provided at a plurality of ETC gates defined in a tollhouse of a toll road.

A plurality of station affairs equipment belongs to the station 8, specifically, a plurality of automatic ticket gates 1, a station staff terminal device 2 used by station staff, a checkout terminal device 3 (see FIG. 2), a station server 4, and a ticket sales. Station equipment such as machine 6 (see FIG. 2) is installed around the ticket gate 7 of station 8. These station affairs devices are connected to each other so as to be able to communicate with each other via a network N1 such as a LAN. Further, the station server 4 of the station 8 is communicably connected to the center device 5 which is a server device managed by a railway operator (railway company) via a network N2 such as a dedicated line or a public line.

The automatic ticket gate 1 is installed at the ticket gate 7 of the station 8. A plurality of automatic ticket gates 1 are installed at each ticket gate 7. FIG. 2 is a sketch (layout diagram) showing the situation around the ticket gate 7A. As shown in FIG. 2, a plurality of ticket gate passages 9 are partitioned at the ticket gate 7A, and specifically, six ticket gate passages 9 are partitioned.

As shown in FIG. 3, the automatic ticket gate 1 is referred to as an entrance automatic ticket gate 1A (hereinafter referred to as an entrance ticket gate 1A) and an entry automatic ticket gate 1B (hereinafter referred to as an entry ticket gate 1B). .) And face to face with each other. The entrance ticket gate 1A and the entry ticket gate 1B are arranged so as to face each other. The ticket gate passage 9 is a space formed between the entrance ticket gate 1A and the exit ticket gate 1B, and connects the inside and the outside of the ticket gate 7A. One of the entrance ticket gates 1A is used for ticket gate processing (entrance ticket gate processing) for users (entrants) who enter from the outside to the inside of the ticket gate 7A along the traveling direction at the time of entry (direction of arrow D10). ). The other entry ticket gate 1B performs ticket gate processing (entry ticket gate processing) for users (contestants) who participate from the inside to the outside of the ticket gate 7A along the traveling direction at the time of entry (direction of arrow D11). Take on the role of doing. In the following, unless otherwise specified, the automatic ticket gate 1 includes the entrance ticket gate 1A and the entry ticket gate 1B, and will be described as having both entrance and exit ticket gate functions.

In the present embodiment, the gate system 100 can set a passage for each of the plurality of automatic ticket gates 1 so as to have a predetermined passage regulation. Three passage regulations are defined as the passage regulations that the automatic ticket gate 1 can take. That is, the gate system 100 can set the passage of the automatic ticket gate 1 so as to have a specific passage regulation selected from the three passage restrictions.

There are three types of passage restrictions that the automatic ticket gate 1 can take: "entrance only", "entry only", and "entrance / exit combined use" described below. Specifically, the entrance only is a passage regulation that stops the ticket gate function of the entry ticket gate 1B of the automatic ticket gate 1 and allows only the user (entrant) who is going to enter. .. The entry-only is a passage regulation that stops the ticket gate function of the entrance ticket gate 1A of the automatic ticket gate 1 and allows only the user (contestant) who intends to participate to pass. The entry / exit combined use is a passage regulation that enables both the entrance / exit ticket gates 1A and the entry / exit ticket gates 1B to enable both entry / exit functions. The passage setting of the plurality of automatic ticket gates 1 is set to one of the above-mentioned three types of traffic regulation.

In the automatic ticket gate 1 set exclusively for admission, when the admission ticket gate processing is performed, the ticket gate passage 9 is allowed to pass only in the traveling direction D10 at the time of admission, and the passage in the opposite direction is prohibited. Further, in the automatic ticket gate 1 set exclusively for entry, when the entry ticket gate processing is performed, the ticket gate passage 9 is allowed to pass only in the traveling direction D11 at the time of entry, and the passage in the opposite direction is prohibited. Ru. Further, in the automatic ticket gate 1 set for both entry and exit, it is possible to allow the ticket gate passage 9 to pass from both the traveling direction D10 at the time of entry and the traveling direction D11 at the time of entry. Allow the passage of users who have undergone ticket gate processing.

For the passage setting of each automatic ticket gate 1, the station staff himself sets the passage setting value corresponding to the passage regulation desired by the station staff, and the setting value information including the passage setting value is input to each automatic ticket gate through the station staff terminal device 2. It can be changed by transmitting to 1. Here, the passage setting value is a setting value indicating the content (traffic restriction type) of the passage setting for each of the plurality of automatic ticket gates 1. Further, the set value information is information for setting a passage for each of the plurality of automatic ticket gates 1.

Hereinafter, the configurations of the automatic ticket gate 1, the station server 4, the station staff terminal device 2, and the center device 5 to which the gate system 100 is applied will be described.

[Automatic ticket gate 1]
FIG. 3 is a perspective view showing the appearance of the automatic ticket gate 1, and FIG. 4 is a block diagram showing the configuration of the automatic ticket gate 1.

The automatic ticket gate 1 is installed at the ticket gate 7 of the station 8, and performs ticket gate processing for users who pass through the ticket gate 7. As shown in FIG. 3, the automatic ticket gate 1 includes an entrance ticket gate 1A and an entry ticket gate 1B installed so as to face each other. In the present embodiment, the automatic ticket gate 1 acquires the ticket information included in the ticket from a ticket such as an IC card or a ticket (magnetic ticket) owned by a user who intends to pass through the ticket gate passage 9. , Perform ticket gate processing based on the ticket information. Specifically, the automatic ticket gate 1 determines whether or not the acquired ticket information is valid, permits passage when it is determined to be valid, and when it is determined to be invalid or the ticket information is acquired. If not, the passage of users is prohibited. When passage is permitted, the automatic ticket gate 1 opens the flap door 15 and opens the ticket gate passage 9 so that the user can pass through. On the other hand, when the passage is prohibited, the automatic ticket gate 1 closes the flap door 15 and closes the ticket gate passage 9 so that the user cannot pass.

As the ticket, an information code such as a barcode or a QR code (registered trademark) in which the ticket information is recorded may be used. In this case, the paper medium on which the information code is printed or the display screen of a terminal device such as a smartphone on which the information code is displayed is substituted for the ticket.

As shown in FIGS. 3 and 4, the automatic ticket gate 1 includes a control unit 11, a storage unit 12, an upper display unit 13, a communication unit 14, flap doors 15 (15A, 15B), and an IC card reader. A unit 16, a side display unit 17, an image pickup unit 18, and a motion sensor 19 are provided, and these are provided in the housing 10 (see FIG. 3) of the automatic ticket gate 1.

The storage unit 12 is a non-volatile storage medium including an HDD or SSD that stores various types of information. The storage unit 12 stores a control program for causing the control unit 11 to execute various arithmetic processes executed by the automatic ticket gate 1, various data used for the various arithmetic processes, and the like. Further, the storage unit 12 stores various information such as passage setting values related to passage regulation transmitted from the station server 4 or the center device 5 to the automatic ticket gate 1.

The upper display unit 13 displays a message to the user passing through the ticket gate passage 9 according to the instruction from the control unit 11. The upper display unit 13 has, for example, a liquid crystal panel. As shown in FIG. 3, the upper display unit 13 is provided on the upper surface of the housing 10 of the automatic ticket gate 1. Specifically, the upper display unit 13 is arranged on the upper surface of the housing 10 in front of the user's traveling direction (direction of arrow D10 or arrow D11) in the ticket gate passage 9 (see FIG. 3). When the passage of the user is permitted, a message indicating that the passage is possible is displayed on the upper display unit 13. Further, when the passage of the user is not permitted, a message indicating that the passage is impassable (prohibited) is displayed on the upper display unit 13. When the automatic ticket gate 1 is used for participation, for example, a message indicating that the fare is insufficient or a payment terminal device 3 (see FIG. 2) is urged to settle the payment on the upper display unit 13. A message may be displayed.

The communication unit 14 connects the automatic ticket gate 1 to the network N1 by wire or wirelessly, and data according to a predetermined communication protocol between the station server 4 and the station staff terminal device 2 via the network N1. It is a communication interface for executing communication.

As shown in FIG. 3, the motion sensor 19 is provided in the housing 10, and more specifically, the human sensor 19 is provided on the side surface of the housing 10 on the ticket gate passage 9 side. The motion sensor 19 is a sensor that detects a user who has entered the ticket gate passage 9, and is, for example, an optical proximity sensor that detects a user in a non-contact manner. In the present embodiment, motion sensors 19 are provided on the front side (entrance side) of the traveling direction D10 at the time of entry and on the front side (exit side) of the traveling direction in the ticket gate passage 9. In other words, the motion sensor 19 is provided on the side surface of the housing 10 at a position on the inner side of the ticket gate 7 and a position on the outer side of the ticket gate 7. For example, if the user in the ticket gate passage 9 is detected by the motion sensor 19 on the front side in the traveling direction before the ticket is read, it is determined that the user has entered the ticket gate passage 9 without performing the ticket gate processing. Further, when the user is detected by the motion sensor 19 on the front side in the traveling direction, it is determined that the ticket gate has passed through the ticket gate passage 9 without being processed.

The flap door 15 is provided in the vicinity of each of the passage ports 9A and 9B on both sides in the ticket gate passage 9 of the automatic ticket gate 1. In other words, the flap door 15 is installed in the ticket gate passage 9 at the passage 9A inside the ticket gate 7 and at the passage 9B outside the ticket gate 7. The former passage 9A is an exit (passage) on the front side of the traveling direction D10 at the time of entry, and the latter passage 9B is an exit (passage) on the front side of the traveling direction D11 at the time of entry.

The flap doors 15 are a pair of doors that can be opened and closed, and each door is rotatably provided on the side surfaces of the entrance ticket gate 1A and the exit ticket gate 1B. The flap door 15 opens and closes by obtaining a driving force from a driving unit (not shown) such as a motor. When the drive unit is driven and controlled by the flap door drive control unit 116 of the control unit 11, the flap door 15 is displaced between the open position and the closed position. The flap door 15 is opened when the entrance ticket gate processing or the entrance ticket gate processing is performed and the passage of the user is permitted. As a result, the user can pass through the ticket gate passage 9 of the automatic ticket gate 1. On the other hand, if the user is not allowed to pass, the flap door 15 is closed or the closed position is maintained.

In the present embodiment, in the automatic ticket gate 1 in which the passage restriction in the ticket gate passage 9 is set for both entrance and exit, the flap doors 15A and 15B of the ticket gate passage 9 are both arranged in the open position. In this state, when the user enters the ticket gate passage 9 and the user is detected by the motion sensor 19 (see FIG. 3) before the entrance ticket gate processing or the entrance ticket gate processing is performed, the flap door 15 opens. Displace from the open position to the closed position.

For example, if a user enters the ticket gate passage 9 from the passage 9B when the entrance ticket gate is not processed at the time of entry, the flap door on the side of the passage 9A is detected by the motion sensor 19 of the passage 9B. 15A is displaced from the open position to the closed position. Then, when the entrance ticket gate processing is performed after that, the flap door 15A is displaced from the closed position to the open position, and the ticket gate passage 9 is opened. On the other hand, if the entrance ticket gate processing is not performed, the flap door 15A is maintained in the closed position.

Further, when a user enters the ticket gate passage 9 from the passage 9A without the entry ticket gate processing being performed at the time of entry, the flap door on the side of the passage 9B is detected by the motion sensor 19 of the passage 9A. 15B is displaced from the open position to the closed position. Then, when the entry ticket gate processing is performed after that, the flap door 15B is displaced from the closed position to the open position, and the ticket gate passage 9 is opened. On the other hand, if the entry ticket gate processing is not performed, the flap door 15B is maintained in the closed position.

As described above, the state of the automatic ticket gate 1 (passage regulation state) in which the passage regulation is set for both entry and exit so that the flap door 15 operates corresponds to the third state of the present invention.

Further, in the automatic ticket gate 1 in which the passage restriction in the ticket gate passage 9 is set exclusively for the entrance, the flap door 15A of the passage 9A of the ticket gate 9 is arranged at the closed position, and the flap door 15B of the passage 9B is arranged. Is placed in the open position. In this state, the flap door 15A is opened and the ticket gate passage 9 is opened when the entrance ticket gate processing is performed and the passage of the user is permitted. Further, when the entrance ticket gate processing is not performed, the flap door 15A keeps the closed position and the ticket gate passage 9 keeps the closed state.

As described above, the state of the automatic ticket gate 1 (passage regulation state) in which the passage regulation is set exclusively for the entrance so that the flap door 15 operates corresponds to the first state of the present invention. When the passage regulation is set exclusively for the entrance, the entry ticket gate function is stopped, so that the user is prohibited from entering the ticket gate passage 9.

Further, in the automatic ticket gate 1 in which the passage restriction in the ticket gate passage 9 is set exclusively for the entry, the flap door 15B of the passage 9B of the ticket gate 9 is arranged at a closed position, and the flap door 15A of the passage 9A is arranged. Is placed in the open position. In this state, the flap door 15B is opened and the ticket gate passage 9 is opened when the entry ticket gate processing is performed and the passage of the user is permitted. Further, when the entry ticket gate processing is not performed, the flap door 15B keeps the closed position and the ticket gate passage 9 keeps the closed state.

As described above, the state of the automatic ticket gate 1 (passage regulation state) in which the passage regulation is set exclusively for the entry so that the flap door 15 operates corresponds to the second state of the present invention. When the passage regulation is set exclusively for the entry, the entrance ticket gate function is stopped, so that the user is prohibited from entering from the ticket gate passage 9.

The flap door 15 is not limited to a physical door, and may be, for example, a door represented by a stereoscopic image using a hologram. Further, instead of the flap door 15, a voice gate that allows or prohibits the passage of the user by voice may be used.

The side display unit 17 displays a message to the user who intends to pass through the ticket gate passage 9 according to the instruction from the control unit 11. The side display unit 17 has, for example, a liquid crystal panel. As shown in FIG. 3, the side display unit 17 is provided on the front surface (front side surface in the traveling direction) of the housing 10 of the automatic ticket gate 1. The side display unit 17 displays information indicating the content of the passage setting set in the automatic ticket gate 1, that is, information indicating the content of the passage regulation. For example, in the case of a ticket gate dedicated to an IC card, the characters "IC" are displayed on the side display unit 17. If the ticket gate processing is possible, an “◯” mark is displayed on the side display unit 17, and if the ticket gate processing is not possible, an “x” mark is displayed on the side display unit 17.

The IC card reading unit 16 is provided on the upper surface of the housing 10 on the front side in the traveling direction. The IC card reading unit 16 reads information (ticket information) recorded in an IC card format ticket (hereinafter referred to as "IC card") in a non-contact manner. The IC card reading unit 16 is, for example, a reader / writer device that reads information in an IC card in a non-contact manner and writes it in a non-contact manner by short-range communication. The information read by the IC card reading unit 16 is transmitted to the control unit 11, and the control unit 11 executes the ticket gate processing based on the information. A user who uses an IC card holds an IC card in the IC card reading unit 16 and causes the IC card reading unit 16 to read the ticket information.

Even when a ticket is inserted into the ticket insertion slot (not shown), the control unit 11 executes ticket gate processing based on the information read from the ticket.

The image pickup unit 18 is provided on the upper surface of the housing 10. The image pickup unit 18 captures a user who intends to use the automatic ticket gate 1, and in detail, captures the face of the user. The image pickup unit 18 is specifically a camera. The image pickup unit 18 is installed so that the lens faces the upstream side in the traveling direction. The image pickup unit 18 takes an image of the user's face when the ticket gate processing at the time of entry / exit is performed. The captured face image is used for the face image collation process. If it is determined by the face image matching process that the captured face image matches the registered face image of the user himself / herself registered in advance, the ticket gate processing at the time of entry / exit is performed and the ticket gate passage is performed. 9 passages are allowed. The imaging unit 18 may be omitted.

The control unit 11 controls the operation of each unit of the automatic ticket gate 1. The control unit 11 has control devices such as a CPU, a ROM, and a RAM. The ROM is a non-volatile storage medium in which control programs such as a BIOS and an OS for causing the CPU to execute various arithmetic processes are stored in advance. The RAM is a volatile or non-volatile storage medium that stores various types of information, and is used as a temporary storage memory (working area) for various arithmetic processes executed by the CPU. Then, the control unit 11 controls the automatic ticket gate 1 by executing various control programs stored in advance in the ROM or the storage unit 12 on the CPU.

As shown in FIG. 4, the control unit 11 includes various processing units such as a reading processing unit 111, a recording processing unit 112, a passage setting unit 113, a traffic control unit 114, a flap door drive control unit 116, and a notification processing unit 117. including.

The control unit 11 functions as the various processing units by the CPU executing various arithmetic processes according to the control program. In other words, by executing the control program, the CPU includes a reading processing unit 111, a recording processing unit 112, a passage setting unit 113, a passage control unit 114, a flap door drive control unit 116, a notification processing unit 117, and the like. Functions as a processing unit. Further, a part or all the processing unit included in the control unit 11 may be composed of an electronic circuit. Further, the control program may be a program for causing a plurality of processors to function as the various processing units. The control unit 11 or the CPU is an example of a computer that executes the control program.

The reading processing unit 111 acquires the ticket information of the IC card read by the IC card reading unit 16 or the ticket information of the ticket read by the magnetic reading unit (not shown).

When the ticket information is acquired from the ticket by the reading processing unit 111, the recording processing unit 112 causes the IC card reading unit 16 and the magnetic reading unit (not shown) to update the information in the ticket. For example, the recording processing unit 112 writes the station name of the entrance station on the ticket. Further, the recording processing unit 112 deducts a predetermined amount (fare) from the amount charged in the IC card at the time of participation, and writes the remaining reduced amount in the information holding unit of the IC card. Further, the recording processing unit 112 writes the station name of the participating station on the IC card.

When the set value information is input to the automatic ticket gate 1, the passage setting unit 113 sets the passage in the automatic ticket gate 1 (passage regulation) based on the passage set value included in the input set value information. Contents) is changed. In the present embodiment, the passage setting unit 113 changes or maintains the passage setting of the automatic ticket gate 1 so as to be the passage regulation indicated by the passage setting value based on the passage setting value. As described above, there are three types of passage restrictions that can be taken by the automatic ticket gate 1, the entrance only, the entry only, and the entrance / exit combined use, and the passage setting unit 113 sets the passage of the automatic ticket gate 1. Change or maintain to be one of the three aisle regulations.

The set value information includes the identification information of each of the plurality of automatic ticket gates 1 and the passage set value indicating the setting contents of each automatic ticket gate 1. Upon receiving the set value information, the passage setting unit 113 extracts the passage set value corresponding to its own automatic ticket gate 1, and the passage of the automatic ticket gate 1 so as to be the passage regulation indicated by the passage set value. Change or maintain the settings.

In the present embodiment, there are cases where the set value information is input to the automatic ticket gate 1 from the station staff terminal device 2 and cases where the set value information is input to the automatic ticket gate 1 from the station server 4. In any case, when the passage setting unit 113 receives the set value information, the passage setting of the automatic ticket gate 1 is changed or maintained.

The set value information input from the station staff terminal device 2 to each automatic ticket gate 1 is information indicating passage regulation for each automatic ticket gate 1 determined by the judgment of the station staff. Further, the set value information input from the station server 4 to each automatic ticket gate 1 includes information determined by the passage setting device 43 (an example of the passage setting device of the present invention) of the station server 4. The passage setting device 43 determines the optimum passage regulation state (optimum passage regulation state) using the learned model generated by the learning device 42, and automatically inputs the set value information corresponding to the optimum passage regulation state. Send to ticket gate 1. When the passage setting unit 113 receives the set value information (optimal passage restriction state) including the passage set value determined by the passage setting device 43, the automatic ticket gate is an automatic ticket gate based on the passage set value included in the set value information. Change or maintain the passage setting of 1. The passage setting device 43 will be described later.

The traffic control unit 114 permits or prohibits the user from passing through the ticket gate passage 9 when a predetermined condition is satisfied. For example, if the traffic control unit 114 determines that the ticket information read from the ticket such as an IC card or a ticket is valid, the traffic control unit 114 permits the user to pass through the ticket gate passage 9. On the other hand, when the traffic control unit 114 determines that the ticket information is invalid, the traffic control unit 114 prohibits the user from passing through the ticket gate passage 9. When the passage is permitted, the flap door 15 is displaced to the open position by the flap door drive control unit 116 described later, and the ticket gate passage 9 is opened. On the other hand, if the passage is not permitted, the flap door 15 is displaced to the closed position by the flap door drive control unit 116, and the ticket gate passage 9 is closed.

The flap door drive control unit 116 controls the operation (opening / closing operation) of the flap door 15. For example, when the passage of a user in the ticket gate passage 9 is permitted, the flap door drive control unit 116 controls a drive unit (not shown) such as a motor to displace the flap door 15 to the open position. As a result, the ticket gate passage 9 is opened. On the other hand, when the passage of the user in the ticket gate passage 9 is prohibited, the flap door drive control unit 116 controls the drive unit to displace the flap door 15 to the closed position or keep the flap door 15 closed. As a result, the ticket gate passage 9 is closed.

The notification processing unit 117 outputs and displays predetermined information on the upper display unit 13 and the side display unit 17. In the present embodiment, the notification processing unit 117 displays a message indicating that the vehicle is passable on the upper display unit 13. If the user is not allowed to pass, a message indicating that the user is not allowed (prohibited) is displayed on the upper display unit 13. Further, the notification processing unit 117 displays a “◯” mark or an “x” mark on the side display unit 17 for identifying whether or not the ticket gate processing is possible. The notification processing unit 117 may display information indicating the content of the passage setting set in the automatic ticket gate 1, that is, information indicating the passage regulation on the side display unit 17.

[Station server 4]
FIG. 5 is a block diagram showing the configuration of the station server 4. The station server 4 is a server device installed in a server room or the like in the station 8, and is provided in each station 8. The station server 4 manages a plurality of station affairs devices such as an automatic ticket gate 1 connected to a network N1 which is an in-station network of a station 8. The station server 4 is not limited to one computer, and may be a computer system in which a plurality of computers cooperate to operate. Further, various processes executed by the station server 4 may be distributed and executed by a plurality of processors.

The station server 4 includes a teacher data generation device 41, a learning device 42, a passage setting device 43, a storage unit 44, and a communication unit 45. The teacher data generation device 41 is an example of the teacher data generation device of the present invention, the learning device 42 is an example of the learning device of the present invention, and the passage setting device 43 is an example of the passage setting device. At least one of the teacher data generation device 41, the learning device 42, and the passage setting device 43 may be an information processing device installed outside the station server 4. In this case, the information processing device is connected to the network N1.

The communication unit 45 connects the station server 4 to the networks N1 and N2 by wire or wirelessly, and according to a predetermined communication protocol, communicates with the station affairs equipment such as the automatic ticket gate 1 and the station staff terminal device 2 via the network N1. It is a communication interface for executing data communication and executing data communication with the center device 5 via the network N2.

The storage unit 44 is a non-volatile storage medium including an HDD or SSD that stores various types of information. The storage unit 44 contains a control program for causing the control unit of the station server 4 to execute various arithmetic processes, a teacher data generation process (see FIG. 10) and an optimum path determination process (see FIG. 10), which will be described later, which are executed by the station server 4. A control program such as a teacher data generation program used in 11) and data used for each process are stored. Further, the storage unit 44 stores the number of installed automatic ticket gates 1, the identification number of each automatic ticket gate 1, the position information, and the like.

In this embodiment, the configuration in which the storage unit 44 is provided in the station server 4 is exemplified. For example, a part or all of various information in the storage unit 44 is connected to the station server 4 through networks N1, N2 and the like. It may be stored in an external device such as another server device capable of data communication or a storage device capable of connecting to a network. Further, the passage setting history storage unit 441 and the teacher data storage unit 442, which will be described later, may be provided in the external device.

Further, the storage unit 44 is provided with a storage area such as a passage setting history storage unit 441 and a teacher data storage unit 442.

The passage setting history data is stored in the passage setting history storage unit 441. The passage setting history data includes the set value information sent from the station staff terminal device 2 to the automatic ticket gate 1. The passage setting history data is used for the teacher data generation processing and the optimum passage determination processing described later. The passage setting history data is log data that records the history of changes in the passage settings of a plurality of automatic ticket gates 1 when the passage settings of a plurality of automatic ticket gates 1 are changed at the discretion of the station staff. The passage setting history data includes the set value information used for setting change, the date and time when the setting is changed, and the like.

The teacher data storage unit 442 stores the teacher data (see FIG. 8) generated by the teacher data generation device 41. As shown in FIG. 8, the teacher data is information in which the set value information defining the passage setting of each of the plurality of automatic ticket gates 1 and the total congestion time described later are associated with each other. The teacher data may be composed of one set value information and the total congestion time, or may be composed of a plurality of set value information and the total congestion time. In the example shown in FIG. 8, the teacher data corresponding to each morning from the first day to the tenth day (an example of the first teacher data of the present invention) and each of the first day to the tenth day. The teacher data corresponding to the afternoon (an example of the second teacher data of the present invention) is shown.

The control unit controls the operation of each unit of the station server 4. The control unit includes control devices such as a CPU, ROM, and RAM. The ROM is a non-volatile storage medium in which control programs such as a BIOS and an OS for causing the CPU to execute various arithmetic processes are stored in advance. The RAM is a volatile or non-volatile storage medium that stores various types of information, and is used as a temporary storage memory (working area) for various arithmetic processes executed by the CPU. Then, the control unit controls the station server 4 by the CPU executing various control programs stored in advance in the ROM or the storage unit 44.

[Teacher data generator 41]
As shown in FIG. 6, the teacher data generation device 41 includes various processing units such as a passage state acquisition unit 411, a congestion degree detection unit 412, a congestion determination unit 413, a congestion time calculation unit 414, and a teacher data generation unit 415. .. The control unit of the teacher data generation device 41 functions as the various processing units by the CPU executing various arithmetic processes according to the control program (including the teacher data generation program). Further, a part or all the processing unit included in the control unit may be composed of an electronic circuit. The control program may be a program for allowing a plurality of processors to function as the various processing units.

The passage state acquisition unit 411 acquires the passage regulation state of each of the plurality of automatic ticket gates 1. Specifically, the passage state acquisition unit 411 acquires the state (passage regulation state) of the passage regulation (entrance only, entry only, and entry / exit combined use) set for each of the plurality of automatic ticket gates 1. For example, the passage state acquisition unit 411 acquires the passage regulation state based on the passage setting history data stored in the passage setting history storage unit 441. Further, the passage state acquisition unit 411 acquires a plurality of the set value information corresponding to each of the plurality of passage regulation states.

Further, the passage state acquisition unit 411 acquires a plurality of the passage regulation states based on predetermined conditions.

Here, the passage regulation state changes according to an element (variation element) that causes a change in the passage regulation content of each of the plurality of automatic ticket gates 1.

The variable element includes, for example, the number of automatic ticket gates 1 installed at the ticket gate 7A of the station 8 and the position of each automatic ticket gate 1. These variable elements can be obtained from the storage unit 44 in which these information are stored in advance.

The variable factors include the concourse inside or outside the ticket gate 7A of the station 8 (see FIG. 2), the vicinity of the ticket gate 7A, the vicinity of the checkout terminal device 3 and the ticket vending machine 6 (see FIG. 2), and the ticket gate. A flow index that shows the mobility and congestion of users at various places such as stairs and escalators connecting Exit 7A and the station platform (see Fig. 2), and elevators connecting the concourse floor and the platform floor (see Fig. 2). The progress speed and direction of the user, the age group of the user, the presence or absence of group users, and the like. In addition, if the user includes an unhealthy person, the variable factors include the traveling speed, traveling direction, position, age group, whether or not a wheelchair is used, whether or not a wheelchair is used, and the white cane. Whether or not it is used. These variable elements can be obtained by analyzing moving images or still images taken by a plurality of cameras (see FIG. 2) provided inside and outside the station 8.

Further, the variable factors include temperature, humidity, atmospheric pressure, altitude, and weather at the location of the station 8. In addition, the variable element includes the current season, calendar (year / month / day), and time. In addition, the variable factors include events held around the station 8, the presence or absence of large facilities around the station 8, the operation status (operation schedule) at the station 8, train delays, and the like. When the variable elements are temperature, humidity, atmospheric pressure, and altitude at the location of the station 8, these variable elements include a temperature sensor and a humidity sensor provided in the teacher data generator 41 indoors or outdoors of the station 8. , The barometric pressure sensor, receives the sensor output value output from the altitude sensor, and acquires the real-time temperature, humidity, barometric pressure, and altitude indicated by the sensor output value. When the variable element is the weather around the station 8, the control unit downloads the weather information from the weather information database connected to the station server 4 via the Internet, and the teacher data generation device 41 receives the weather information from the control unit. Receive the weather information. Further, the teacher data generation device 41 can acquire information such as the operation status and the delay information from the center device 5.

Further, the variable element may include statistical information (user history information) indicating the number of users by past month or day of the week.

Since such a variable element affects the passage status of the user passing through the ticket gate 7A, it can be said that it is information that causes fluctuations in the passage regulation contents of each of the plurality of automatic ticket gates 1. For example, if the space outside the ticket gate 7 or the concourse becomes more crowded, it can be estimated that the number of users entering from the ticket gate 7 will increase rapidly. In this case, the automatic ticket gate with passage restrictions set exclusively for admission. It will be necessary to increase the number of machines 1. In addition, during weekday work / commuting time, the number of users entering stations near business districts and schools will increase rapidly, and the number of users entering stations near residential areas will increase rapidly. Therefore, at the former station, it is necessary to increase the number of automatic ticket gates 1 whose passage restrictions are set exclusively for entry, and at the latter station, the number of automatic ticket gates 1 whose passage restrictions are set exclusively for entry is increased. Need arises. In addition, if the operation status (operation timetable) is disturbed or delayed, it can be estimated that the number of participating users who will participate in station 8 will increase rapidly. It will be necessary to increase the number of machines 1.

Therefore, it is desirable that the passage state acquisition unit 411 acquires the passage regulation state of various patterns set in the past based on the fluctuation element.

For example, among the above-mentioned variable factors, the commuting / school element has a large influence on the passage regulation content, so that the passage state acquisition unit 411 has the passage regulation state during the commuting / school time zone and the time zone excluding the time zone. To get the passage regulation status of. In addition, in general, there are times when it is crowded due to commuting to work or school during the day, but the contents of passage regulations are different. For example, in the case of a station having a large number of automatic ticket gates 1 set exclusively for entry in the morning, the number of automatic ticket gates 1 set exclusively for entry increases in the afternoon. Similarly, in the case of a station having a large number of automatic ticket gates 1 set exclusively for entry in the morning, the number of automatic ticket gates 1 set exclusively for entry increases in the afternoon. As described above, the passage regulation state of each of the plurality of automatic ticket gates 1 tends to be different during the day.

Therefore, the passage state acquisition unit 411 is based on the passage setting history data stored in the passage setting history storage unit 441, in the morning (first predetermined of the present invention) in one day (an example of a predetermined period of the present invention). A plurality of the passage regulation states (an example of the first passage regulation state of the present invention) set in the period (an example of the period) and a plurality of the passages set in the afternoon of the day (an example of the second predetermined period of the present invention). The passage regulation state (an example of the second passage regulation state of the present invention) is acquired.

Here, as an example, the passage state acquisition unit 411 has 10 (10 patterns) first passage regulation states set in the morning and 10 (10 patterns) second passage regulation states set in the afternoon. Shall be acquired. Therefore, the passage state acquisition unit 411 corresponds to the 10 first setting value information corresponding to the 10 first passage regulation states set in the morning and the 10 second passage regulation states set in the afternoon. Acquires 10 second set value information to be used. Each of the plurality of passage regulation states acquired by the passage state acquisition unit 411 is set in the plurality of automatic ticket gates 1 in order, and the processing of each of the following processing units is executed. For example, the ten first set value information is changed to the morning passage setting by transmitting one first set value information to a plurality of automatic ticket gates 1 every day. Similarly, for the ten second set value information, one second set value information is transmitted to the plurality of automatic ticket gates 1 every day, and the aisle setting is changed to the afternoon.

The congestion degree detection unit 412 detects the degree of congestion of users around the plurality of automatic ticket gates 1 in the passage restricted state acquired by the passage state acquisition unit 411. For example, when a plurality of automatic ticket gates 1 are set to the passage regulation of the passage regulation state acquired by the passage state acquisition unit 411 and operated, the degree of congestion of users around the plurality of automatic ticket gates 1 can be determined. To detect. For example, the congestion detection unit 412 may be the number of users around the plurality of automatic ticket gates 1 or the number of users around the plurality of automatic ticket gates 1 based on images captured by a plurality of cameras installed around the ticket gate 7 or the concourse of the station 8. Detect density. The density is represented by the number of users per unit area around the plurality of automatic ticket gates 1.

Further, the congestion degree detection unit 412 detects the congestion degree for a predetermined period (predetermined time zone). For example, when the day (predetermined period) is divided into the morning (first predetermined period) time zone and the afternoon (second predetermined period) time zone as described above, the congestion degree detection unit 412 uses the morning period. The congestion degree (an example of the first congestion degree of the present invention) and the congestion degree (an example of the second congestion degree of the present invention) during the afternoon period are detected. For example, the congestion degree detection unit 412 has a first congestion degree in the morning from the start time when the use of the station 8 starts to noon, and a second congestion degree in the afternoon from noon to the end time when the use of the station 8 ends. Is detected. For example, the congestion degree detection unit 412 detects the change in the number of users in the morning as the first congestion degree and detects the change in the number of users in the afternoon as the second congestion degree.

The congestion determination unit 413 determines whether or not the congestion degree detected by the congestion degree detection unit 412 exceeds the threshold value. Specifically, the congestion determination unit 413 determines whether or not the number or density of users around the plurality of automatic ticket gates 1 detected by the congestion degree detection unit 412 exceeds the threshold value. The threshold value is set to a value corresponding to a state in which a plurality of users are staying (matrix) around the plurality of automatic ticket gates 1. For example, the threshold value is set to a value obtained by multiplying the number of ticket gate passages 9 partitioned at the ticket gate by 10 (persons). That is, the threshold value corresponds to a situation in which 10 users are waiting for use (waiting for entry or waiting for entry) per ticket gate passage 9.

In the example of FIG. 2, since the ticket gate 7A is divided into 6 aisles, for example, the congestion determination unit 413 has 60 users around the ticket gate 7A detected by the congestion degree detection unit 412. Determine if it exceeds. The congestion determination unit 413 executes the determination process every time the congestion degree is detected by the congestion degree detection unit 412.

Further, when the day is divided into a morning time zone and an afternoon time zone as described above, in the congestion determination unit 413, the first congestion degree detected by the congestion degree detection unit 412 is the threshold value (the present invention). It is determined whether or not it exceeds the first threshold value), and it is determined whether or not the second degree of congestion detected by the congestion degree detection unit 412 exceeds the threshold value (an example of the second threshold value of the present invention). do. The first threshold value and the second threshold value may be set to the same value or may be set to different values from each other.

The congestion time calculation unit 414 calculates the total congestion time, which indicates the total time during which the degree of congestion exceeds the threshold value in a predetermined period. Specifically, when the congestion determination unit 413 determines that the congestion degree exceeds the threshold value in the predetermined period, the congestion time calculation unit 414 determines the time required until the congestion degree becomes equal to or lower than the threshold value. The total is calculated as the total congestion time. The required time is the time from when the degree of congestion exceeds the threshold value (congested state) to when the degree of congestion is equal to or lower than the threshold value (non-congested state) (the congested state is eliminated). Is.

Here, a specific example of the total congestion time will be described. For convenience of explanation, the examples given below focus only on the relationship between the number of users and the passage setting of the automatic ticket gate 1 without considering the above-mentioned variable factors.

In the first example, in the case where a plurality of automatic ticket gates 1 having the ability to process 100 ticket gates per minute are installed at the ticket gates and the ticket gates of 10 passages are divided, the morning commute to work or school. When 1000 users try to participate in the time zone, the congestion time calculation unit 414 calculates the total congestion time as shown below. For example, if nine ticket gate passages are set exclusively for entry, the number of users waiting for entry will be 100 or less in one minute, and the congestion state will be eliminated, so that the total congestion time will be "1 minute". Further, for example, when eight ticket gate passages are set exclusively for entry, the congestion state is eliminated in 1.125 minutes, so that the total congestion time is "1.125 minutes". Further, for example, when one ticket gate passage is set exclusively for entry, the congestion state is eliminated in 9 minutes, so that the total congestion time is "9 minutes". Therefore, in this example, when the nine ticket gate passages are set exclusively for entry, the total congestion time is the shortest.

In the second example, when the plurality of automatic ticket gates 1 are installed at the ticket gates and the ticket gates of 10 passages are divided, 1000 users will participate in the morning commuting time zone. And when 325 users try to enter, the congestion time calculation unit 414 calculates the total congestion time as shown below. The crowded state here is a state in which 100 or more users are waiting for participation or admission. For example, if nine ticket gate passages are set exclusively for entry and one ticket gate passage is set exclusively for entry, the number of users waiting for entry will be 100 or less on the entry side in one minute, and the congestion state will be eliminated. On the entrance side, the number of users waiting for admission will be 100 or less in 2.25 minutes and the congestion state will be resolved, so the total congestion time will be "2.25 minutes". Also, for example, if eight ticket gate passages are set exclusively for entry and two ticket gate passages are set exclusively for entry, the congestion state will be resolved in 1.125 minutes on the entry side and 1.125 minutes on the entrance side. Since the congestion state is eliminated, the total congestion time is "1.125 minutes". Also, for example, if 7 ticket gate passages are set exclusively for entry and 3 ticket gate passages are set exclusively for entry, the congestion state will be resolved in 1.286 minutes on the entry side and 0.75 minutes on the entrance side. Since the congestion state is eliminated, the total congestion time is "1.286 minutes". Also, for example, if one ticket gate passage is set exclusively for entry and nine ticket gate passages are set exclusively for entry, the congestion state will be resolved in 9 minutes on the entry side and the congestion state will be cleared in 0.25 minutes on the entrance side. The total congestion time is "9 minutes". Therefore, in this example, when the eight ticket gate passages are set exclusively for entry and the two ticket gate passages are set exclusively for entry, the total congestion time is the shortest.

The above example is a calculation result on the premise that a plurality of users are evenly distributed to the plurality of automatic ticket gates 1 and the ticket gate processing is executed at a predetermined speed. However, in reality, it is difficult to calculate the total congestion time by a predetermined formula because the variable factor is added. Therefore, in the present embodiment, the congestion time calculation unit 414 calculates the total congestion time while the plurality of automatic ticket gates 1 are set and operated in the passage restricted state.

Further, when the day is divided into the morning time zone and the afternoon time zone as described above, the congestion time calculation unit 414 indicates the total time when the first congestion degree exceeds the first threshold value. The total time (an example of the first total congestion time of the present invention) and the total congestion time (an example of the second total congestion time of the present invention) indicating the total of the times when the second degree of congestion exceeds the second threshold value. Is calculated.

FIG. 7 shows a change in the number of users showing the congestion situation around the ticket gate during the period from the start time of use of the station 8 in a certain day to the end time of use. For example, in the morning, when there are two time zones in which the number of users exceeds the threshold value, the congestion time calculation unit 414 calculates "t1 + t2" as the total congestion time (first total congestion time T1) in the morning. do. Further, when there are three time zones in which the number of users exceeds the threshold value in the morning, the congestion time calculation unit 414 calculates "t3 + t4 + t5" as the total congestion time (second total congestion time T2) in the afternoon. do. In this way, the congestion time calculation unit 414 measures the time each time the congestion degree exceeds the threshold value and calculates the total congestion time for a predetermined period.

The teacher data generation unit 415 generates teacher data in which the passage regulation state acquired by the passage state acquisition unit 411 and the total congestion time calculated by the congestion time calculation unit 414 are associated with each other.

Further, when the day (predetermined period) is divided into the morning (first predetermined period) time zone and the afternoon (second predetermined period) time zone as described above, the teacher data generation unit 415 acquires the passage state. The first teacher data associating the first passage regulation state acquired by the unit 411 with the first total congestion time calculated by the congestion time calculation unit 414, and the first teacher data acquired by the passage state acquisition unit 411. 2 The second teacher data in which the passage regulation state and the second total congestion time calculated by the congestion time calculation unit 414 are associated with each other is generated. That is, the teacher data generation unit 415 generates teacher data at predetermined intervals.

The teacher data generation device 41 of the present embodiment generates the teacher data corresponding to the passage restriction state of each of the plurality of patterns. For example, the passage state acquisition unit 411 acquires the passage restriction state of a plurality of predetermined patterns. In the above example, the passage state acquisition unit 411 acquires the first passage regulation state of 10 patterns in the morning and the second passage regulation state of 10 patterns in the afternoon. The congestion degree detection unit 412 detects the congestion degree corresponding to the passage regulation state of each of the plurality of patterns. Further, the congestion time calculation unit 414 calculates the total congestion time corresponding to the passage regulation state of each of the plurality of patterns. Then, the teacher data generation unit 415 generates the teacher data corresponding to the passage restriction state of each of the plurality of patterns. FIG. 8 is a diagram showing an example of the teacher data generated by the teacher data generation device 41. As shown in FIG. 8, the setting information "D101 to D110" and the total congestion time "T101 to T110" corresponding to each of the 10 patterns of the passage regulation in the morning are registered in association with each other, and the 10 patterns in the afternoon are registered. The setting information "D201 to D210" and the total congestion time "T201 to T210" corresponding to each of the passage regulation states are registered in association with each other. The teacher data generated by the teacher data generation unit 415 is stored in the teacher data storage unit 442. Further, the teacher data generation device 41 transmits the teacher data to the learning device 42.

As described above, the teacher data generation device 41 does not calculate the total congestion time for all the patterns of the passage regulation state composed of the plurality of automatic ticket gates 1, and the total congestion time for the passage regulation state of a predetermined number of patterns. Is calculated. Further, the total congestion time, which is the sum of the times when the degree of congestion exceeds the threshold value, is generated as the teacher data, and the teacher data is not generated for the time when the degree of congestion is equal to or less than the threshold value. Further, the total congestion time is a time actually measured in a situation (environment) including the variable element. Therefore, it is possible to generate specific teacher data that appropriately reflects the congestion situation without preparing a pattern of a large amount of passage regulation state. Therefore, according to the teacher data generation device 41 of the present embodiment, it is possible to efficiently generate teacher data used for machine learning for setting the passage regulation state of each of the plurality of automatic ticket gates 1.

[Learning device 42]
The learning device 42 generates a trained model by performing machine learning using the teacher data generated by the teacher data generation device 41. Specifically, the learning device 42 generates the trained model that estimates the total congestion time corresponding to the arbitrary passage restriction state by performing machine learning using the teacher data.

The learning device 42 may perform machine learning using not only the teacher data but also the information of the passage restricted state. The learning device 42 can use a general-purpose CPU, but in order to enable higher-speed arithmetic processing, for example, GPGPU (General-Purpose computing on Graphics Processing Units), a large-scale PC cluster, or the like may be used. Is desirable.

Machine learning includes algorithms such as supervised learning, unsupervised learning, and reinforcement learning, and in order to realize these methods, the feature quantity itself is extracted. A technique called "deep learning" is used for learning. In the present embodiment, the learning device 42 has a learning model based on the various algorithms described above.

Here, supervised learning is an algorithm that learns the "relationship between input and output" from data input in advance. The input data is given the correct answer of the data together with the input value, and by inputting such data into the learning device 42, the learning device 42 learns the features in those data. Estimate the output (result) from the input. In supervised learning, an error function (such as a cross-entry pee function) is used to quantify the error between the input and the output. In supervised learning, the relationship between given input / output data is learned by adjusting the parameters (weights and biases) of the learning model so that the error becomes small. If this learning is possible, the output can be predicted by applying the relationship to unknown data. The learning device 42 of the present embodiment performs supervised learning using the teacher data.

Unsupervised learning is an algorithm that learns features such as the structure, characteristics, and new knowledge of the input data from the input data without being given the correct output data. It is different from supervised learning in that the correct answer is not given to the data that is the source of learning. The learning device 42 of the present embodiment may perform unsupervised learning using the information of the passage restricted state.

Intensified learning is not learning based on fixed and clear data such as supervised learning and unsupervised learning, but the program itself observes a given environment (current state) and a series of continuous actions. It is an algorithm that evaluates the behavior and learns the appropriate behavior based on the interaction that the behavior gives to the environment, that is, the behavior to maximize the reward obtained in the future. TD learning and Q-learning are known as typical methods. The learning device 42 of the present embodiment performs pre-learning by supervised learning, sets the pre-learned input / output relationship as an initial state, and then applies reinforcement learning.

The learning device 42 of the present embodiment generates a trained model by performing supervised learning particularly using the teacher data.

Further, the learning device 42 of the present embodiment generates a first trained model corresponding to the first predetermined period by performing machine learning using, for example, the first teacher data corresponding to the first predetermined period (morning). Then, machine learning is performed using the second teacher data corresponding to the second predetermined period (afternoon) to generate a second trained model corresponding to the second predetermined period.

For example, when information (set value information) of an arbitrary passage regulation state is input, the trained model estimates the total congestion time corresponding to the passage regulation state. For example, in the first trained model, when information (set value information) of an arbitrary passage regulation state is input, the first total congestion time corresponding to the characteristic of the morning corresponding to the passage regulation state is estimated. Similarly, when the information (set value information) of an arbitrary passage regulation state is input, the second trained model estimates the second total congestion time according to the characteristics of the afternoon corresponding to the passage regulation state. .. This makes it possible to obtain the total congestion time corresponding to any passage regulation state.

In this way, the learning device 42 can estimate the total congestion time corresponding to an arbitrary passage regulation state by using the teacher data corresponding to a predetermined number (predetermined pattern) of passage regulation states. To generate. Further, the teacher data corresponds to the total congestion time, which is the sum of the times when the degree of congestion exceeds the threshold value. Therefore, the total congestion time corresponding to an arbitrary passage regulation state can be estimated with high accuracy.

The learning device 42 outputs the generated learned model to the passage setting device 43. As a result, the trained model is applied to the passage setting device 43.

[Aisle setting device 43]
The passage setting device 43 sets the passage regulation state corresponding to the shortest total congestion time among the plurality of total congestion times estimated using the trained model generated by the learning device 42 to the optimum passage regulation state. decide. The passage setting device 43 inputs an arbitrary passage regulation state (set value information) into the trained model, and estimates the total congestion time corresponding to the passage regulation state. As a result, the total congestion time corresponding to a large amount of passage regulation state (set value information) can be acquired in advance. The aisle setting device 43 stores the acquired set value information and the data of the total congestion time in its own storage unit or storage unit 44. Then, the aisle setting device 43 determines the aisle regulation state corresponding to the shortest total congestion time from the acquired plurality of total congestion times to the optimum aisle regulation state. It can be said that the passage regulation state in which the total congestion time is the shortest is a state in which a plurality of users around the ticket gate can smoothly pass through the ticket gate.

For example, the aisle setting device 43 sets the aisle regulation state corresponding to the shortest total congestion time among the plurality of total congestion times estimated using the first trained model in the first predetermined period (am). Determine the optimum passage regulation state corresponding to. Similarly, the aisle setting device 43 sets the aisle regulation state corresponding to the shortest total congestion time among the plurality of total congestion times estimated using the second trained model in the second predetermined period (afternoon). ) Is determined to be the optimum passage regulation state.

In this way, by determining the optimum passage regulation state for each predetermined period, it is possible to perform appropriate passage regulation even when the congestion situation at the ticket gate differs for each predetermined period.

The passage setting device 43 transmits the determined information on the optimum passage regulation state (set value information) to the automatic ticket gate 1. When the set value information is input, each automatic ticket gate 1 changes the passage setting (contents of the passage regulation) based on the passage set value included in the set value information, or maintains the current state.

The gate system 100 of the present embodiment changes the passage setting of each of the plurality of automatic ticket gates 1 at the predetermined period. When the predetermined period is divided into a time zone of the morning (first predetermined period) and a time zone of the afternoon (second predetermined period), the gate system 100 has a plurality of automatic ticket gates 1 at the start time of the morning, respectively. Change the passage setting of, and change the passage setting of each of the multiple automatic ticket gates 1 at the start time of the afternoon.

[Station staff terminal device 2]
FIG. 9 is a block diagram showing the configuration of the station staff terminal device 2. The station staff terminal device 2 is a terminal device used by station staff who perform station affairs on railways. For example, an information processing device installed in a station clerk room 90 (see FIG. 2) adjacent to a ticket gate 7A or a station staff. Is a mobile terminal such as a smartphone, a mobile phone, or a tablet terminal that can be carried and carried. The station staff terminal device 2 is connected to the network N1 by wire or wirelessly, and performs data communication according to a predetermined communication protocol with other station affairs devices such as a station server 4 and an automatic ticket gate 1 via the network N1. conduct. By operating the station staff terminal device 2, the station staff can view and confirm various information received by the station staff terminal device 2, and can change the passage settings of the plurality of automatic ticket gates 1.

As shown in FIG. 7, the station staff terminal device 2 includes a control unit 21, an input unit 22, a storage unit 23, a display unit 24, and a communication unit 25.

The input unit 22 is a user interface such as a keyboard, a mouse, and a touch panel. The station staff operates the station staff terminal device 2 using the input unit 22.

The storage unit 23 is a non-volatile storage medium including an HDD or SSD that stores various types of information. The storage unit 23 stores a control program for causing the control unit 21 to execute various arithmetic processes executed by the station staff terminal device 2, various data used for the various arithmetic processes, and the like. In addition, various information transmitted from the station server 4, the center device 5, and the like may be temporarily stored in the storage unit 23.

The display unit 24 is, for example, a liquid crystal panel. On the display unit 24, a GUI screen for changing the passage setting of the automatic ticket gate 1, a message to the station staff, and the like are displayed according to instructions from the control unit 21.

The communication unit 25 connects the station staff terminal device 2 to the network N1 by wire or wirelessly, and follows a predetermined communication protocol with other station affairs devices such as the station server 4 and the automatic ticket gate 1 via the network N1. It is a communication interface for executing data communication.

The control unit 21 controls the operation of each unit of the station staff terminal device 2. The control unit 21 has control devices such as a CPU, a ROM, and a RAM. The ROM is a non-volatile storage medium in which control programs such as a BIOS and an OS for causing the CPU to execute various arithmetic processes are stored in advance. The RAM is a volatile or non-volatile storage medium that stores various types of information, and is used as a temporary storage memory (working area) for various arithmetic processes executed by the CPU. Then, the control unit 21 controls the station staff terminal device 2 by executing various control programs stored in advance in the ROM or the storage unit 32 on the CPU.

The control unit 21 includes a set value generation unit 211 and a notification processing unit 212.

The control unit 21 functions as the various processing units by the CPU executing various arithmetic processes according to the control program. In other words, the CPU functions as the setting value generation unit 211 and the notification processing unit 212 by executing the control program. Further, the various processing units included in the control unit 21 may be configured by an electronic circuit. Further, the control program may be a program for causing a plurality of processors to function as the various processing units. The control unit 21 or the CPU is an example of a computer that executes the control program.

The setting value generation unit 211 includes the setting value corresponding to the instruction information when the instruction information for changing the passage setting of the plurality of automatic ticket gates 1 is input by the operation by the station staff. Generate information. Then, when the setting change instruction is input by the operation by the station staff, the control unit 21 transmits the set value information to the plurality of automatic ticket gates 1, and sets the passage of each automatic ticket gate 1 to the passage setting unit 113. Change to (see Fig. 4). Further, the control unit 21 also transmits the set value information to the station server 4 and stores it in the passage setting history storage unit 441 of the station server 4.

The notification processing unit 212 outputs various messages to be notified to the station staff and displays them on the display unit 24.

[Center device 5]
The center device 5 is a central monitoring device managed by a railway operator that operates a railway business including a station 8, and is a station affairs device such as a station server 4 and an automatic ticket gate 1 of all stations 8 operated by the railway operator. To manage. As shown in FIG. 1, the center device 5 wirelessly connects to the network N2 and executes data communication with the station server 4 and other station affairs equipment via the network N2 according to a predetermined communication protocol.

[Teacher data generation process]
Hereinafter, an example of a teacher data generation process for generating teacher data executed by the control unit of the teacher data generation device 41 of the gate system 100 will be described with reference to FIG. In each figure, S11, S12, ... Indicates a processing procedure number (step number). The present invention can be regarded as an invention of a teacher data generation method for executing one or a plurality of steps included in the teacher data generation process.

As shown in FIG. 10, in step S11, the control unit acquires the passage regulation state. For example, the control unit acquires the passage regulation state of 10 patterns in the morning for 10 days from the passage setting history data. Then, in the morning of the first day, the control unit sets the passage regulation corresponding to the passage regulation state of one of the ten patterns in the plurality of automatic ticket gates 1. Step S11 is an example of the passage state acquisition step of the present invention.

Next, in step S12, the control unit of the teacher data generation device 41 determines whether or not the predetermined period has started. For example, the control unit determines whether or not the current time is the start time at which the station 8 starts to be used, the start time in the morning, or the start time in the afternoon.

Next, in step S13, the control unit starts detecting the degree of congestion of users around the plurality of automatic ticket gates 1 in the passage restricted state. For example, the control unit performs the congestion degree detection process for a predetermined period (morning period) in the passage restricted state. Step S13 is an example of the congestion degree detection step of the present invention.

Next, in step S14, the control unit determines whether or not the degree of congestion exceeds the threshold value. The control unit executes a determination process each time the degree of congestion is detected. When the degree of congestion exceeds the threshold value (S14: YES), the process proceeds to step S15. When the degree of congestion is equal to or less than the threshold value (S14: NO), the process proceeds to step S18. Step S14 is an example of the congestion determination step of the present invention.

In step S15, the control unit starts measuring the time. In step S16, the control unit determines whether or not the degree of congestion is equal to or less than the threshold value. When the degree of congestion becomes equal to or less than the threshold value (S16: YES), the process proceeds to step S17. In step S17, the control unit ends the time measurement. That is, the control unit measures the time until the degree of congestion becomes equal to or less than the threshold value.

In step S18, the control unit determines whether or not the predetermined period has ended. For example, the control unit determines whether or not the morning has ended. When the predetermined period ends (S18: YES), the process proceeds to step S19, and when the predetermined period does not end (S18: NO), the process returns to step S14.

In step S19, the control unit ends the detection of the degree of congestion. Next, in step S20, the control unit calculates the total congestion time. For example, the control unit calculates the total time (see “T1” in FIG. 7) as the total congestion time when the degree of congestion exceeds the threshold value in the morning. Step S20 is an example of the congestion time calculation step of the present invention.

Finally, in step S21, the control unit generates the teacher data associating the one passage restriction state acquired in step S12 with the total congestion time. The teacher data generated here is, for example, data including the set value information “D101” shown in FIG. 8 and the total congestion time “T101”. Step S21 is an example of the teacher data generation step of the present invention.

The control unit sets each of the remaining nine patterns of the passage regulation state acquired in step S11 in each automatic ticket gate 1 in each morning from the second day to the tenth day, and describes the above. The process of steps S12 to S21 of the above is repeated. In addition, the same processing is performed for each afternoon from the first day to the tenth day. As a result, the teacher data shown in FIG. 8 is generated.

[Optimal passage determination process]
Next, the optimum passage setting process will be described with reference to FIG.

As shown in FIG. 11, in the optimum passage setting process, the processes of steps S11 to S21 described above are performed, and then the processes of steps S22 to S2 are performed. The teacher data generation device 41 outputs the teacher data generated by the teacher data generation process to the learning device 42.

The control unit of the learning device 42 performs machine learning using the teacher data generated by the teacher data generation process (step S22), and generates a trained model (step S23). For example, the control unit uses the teacher data shown in FIG. 8 to generate a first trained model corresponding to the first predetermined period (morning), and the second learning corresponding to the second predetermined period (afternoon). Generate a finished model. The control unit outputs the generated learned model to the passage setting device 43.
In step S24, the control unit of the aisle setting device 43 applies the trained model to estimate the total congestion time corresponding to an arbitrary aisle regulation state.

In step S25, the control unit of the passage setting device 43 sets the passage regulation state corresponding to the shortest total congestion time among the plurality of total congestion times estimated using the learned model to the optimum passage regulation state. decide.

The control unit of the passage setting device 43 transmits the determined information (set value information) of the optimum passage regulation state to the automatic ticket gate 1. When the set value information is input, each automatic ticket gate 1 changes the passage setting (contents of the passage regulation) based on the passage set value included in the set value information, or maintains the current state.

In the optimum passage determination process, when the process of step S25 is completed, the process may proceed to step S12. In this case, in the teacher data generation process, the control unit of the teacher data generation device 41 detects the congestion degree and calculates the total congestion time in the optimum passage regulation state set in the plurality of automatic ticket gates 1. That is, the control unit generates teacher data (an example of the optimum teacher data of the present invention) corresponding to the optimum passage regulation state (S21).

After that, in step S22, the learning device 42 modifies the trained model by performing machine learning (step S22) using the teacher data generated by the teacher data generation process and the optimum teacher data. Generate a finished model (step S23).

Next, in step S24, the control unit of the passage setting device 43 applies the modified and learned model to estimate the total congestion time corresponding to an arbitrary passage restriction state.

Then, in step S25, the control unit of the passage setting device 43 newly sets the passage regulation state corresponding to the shortest total congestion time among the plurality of total congestion times estimated by using the modified and learned model. Determine the optimum passage regulation state.

The control unit of the passage setting device 43 transmits the determined information (set value information) of the optimum passage regulation state to the automatic ticket gate 1. When the set value information is input, each automatic ticket gate 1 changes the passage setting based on the passage set value included in the set value information, or maintains the current state.

In this way, the teacher data generation process and the optimum passage determination process are repeated. Thereby, the passage regulation of each of the plurality of automatic ticket gates 1 can be set to the optimum state according to the congestion situation around the ticket gate.

In the above-described embodiment, an example in which the teacher data generation process and the optimum passage determination process shown in FIGS. 10 and 11 are executed by the station server 4 has been described, but the processes in each step are the station server 4 and the automatic ticket gate. 1. It may be executed by any of the control units of the station staff terminal device 2 and the center device 5, or may be distributed and executed by each control unit of each device.

Further, each of the teacher data generation device 41, the learning device 42, and the passage setting device 43 may be configured by individual information processing devices. Further, the teacher data generation device 41 and the learning device 42 may be configured by one information processing device installed outside the station server 4, and the passage setting device 43 may be configured by the station server 4. The gate setting learning system of the present invention may be composed of a teacher data generation device 41 and a learning device 42, and may further include a passage setting device 43. Further, the gate setting learning system of the present invention may be the gate system 100.

Further, in the above-described embodiment, the configuration in which the passage setting of the traffic regulation of each automatic ticket gate 1 can be set to any of entrance-only, entry-only, and entry / exit combined has been described, but the present invention has this configuration. Not limited. For example, the automatic ticket gate 1 can be set to either entry-only or entry-only, entry-only and entry-entry combined configuration, or entry-only and entry-entry combined use. The present invention can also be applied to a configuration that can be set to.

1: Automatic ticket gate 2: Station staff terminal device 3: Settlement terminal device 4: Station server 5: Center device 6: Ticket vending machine 7: Ticket gate 8: Station 9: Ticket gate passage 41: Teacher data generation device 42: Learning device 43: Passage setting device 44: Storage unit 45: Communication unit 90: Station office 100: Gate system 111: Read processing unit 112: Recording processing unit 113: Passage setting unit 114: Traffic control unit 116: Flap door drive control unit 117: Notification Processing unit 211: Setting value generation unit 212: Notification processing unit 411: Passage state acquisition unit 412: Congestion degree detection unit 413: Congestion determination unit 414: Congestion time calculation unit 415: Teacher data generation unit 441: Passage setting history storage unit 442 : Teacher data storage

Claims (16)

A teacher data generator that generates teacher data used for machine learning.
A passage state acquisition unit that acquires the passage regulation state of each of multiple gate devices,
A congestion degree detection unit that detects the congestion degree of users around the plurality of gate devices in the passage regulation state acquired by the passage state acquisition unit, and a congestion degree detection unit.
A congestion determination unit that determines whether or not the congestion degree detected by the congestion degree detection unit exceeds a threshold value, and a congestion determination unit.
A congestion time calculation unit that calculates the total congestion time, which indicates the total time during which the congestion degree exceeds the threshold value in a predetermined period.
A teacher data generation unit that generates the teacher data in which the passage regulation state acquired by the passage state acquisition unit and the total congestion time calculated by the congestion time calculation unit are associated with each other.
A teacher data generator equipped with. The congestion degree detection unit detects the number or density of users around the plurality of gate devices, and detects the number or density of users.
The congestion determination unit determines whether or not the number of people or the density detected by the congestion degree detection unit exceeds the threshold value.
The teacher data generation device according to claim 1. The congestion time calculation unit calculates the total time required until the congestion degree becomes equal to or less than the threshold value when the congestion degree is determined by the congestion determination unit to exceed the threshold value in the predetermined period. Calculated as total time,
The teacher data generator according to claim 1 or 2. The predetermined period includes a first predetermined period and a second predetermined period.
The passage state acquisition unit acquires the first passage regulation state corresponding to the first predetermined period and the second passage regulation state corresponding to the second predetermined period.
The congestion degree detection unit detects the first congestion degree corresponding to the first predetermined period and the second congestion degree corresponding to the second predetermined period.
The congestion determination unit determines whether or not the first congestion degree detected by the congestion degree detection unit exceeds the first threshold value, and the second congestion degree detected by the congestion degree detection unit is the second. Determine if the threshold is exceeded and
The congestion time calculation unit is a second unit that indicates the total of the first congestion total time indicating the total time when the first congestion degree exceeds the first threshold value and the total time when the second congestion degree exceeds the second threshold value. Calculate the total congestion time and
The teacher data generation unit has the first teacher data in which the first passage regulation state acquired by the passage state acquisition unit and the first total congestion time calculated by the congestion time calculation unit are associated with each other. The second teacher data in which the second passage regulation state acquired by the passage state acquisition unit and the second total congestion time calculated by the congestion time calculation unit are associated with each other is generated.
The teacher data generator according to any one of claims 1 to 3. The first predetermined period is a period including the morning of the day, and the second predetermined period is a period including the afternoon of the day.
The teacher data generation device according to claim 4. The first threshold value and the second threshold value are set to different values.
The teacher data generator according to claim 4 or 5. The passage state acquisition unit acquires the passage regulation state of a predetermined plurality of patterns, and obtains the passage regulation state.
The congestion degree detection unit detects the congestion degree corresponding to the passage regulation state of each of the plurality of patterns, and detects the congestion degree.
The congestion time calculation unit calculates the total congestion time corresponding to the passage regulation state of each of the plurality of patterns.
The teacher data generation unit generates the teacher data corresponding to the passage restriction state of each of the plurality of patterns.
The teacher data generator according to any one of claims 1 to 6. The passage restriction state is a first state that allows passage in only one direction to the passage of the gate device, a second state that allows passage in only the other direction to the passage of the gate device, and the gate. One of the third states that allows bidirectional passage through the passage of the device,
The teacher data generator according to any one of claims 1 to 7. The gate device is an automatic ticket gate installed at the ticket gate of a railway station and set in any one of the first state, the second state, and the third state.
The teacher data generation device according to claim 8. The teacher data generator according to any one of claims 1 to 9,
A learning device that generates a trained model by performing machine learning using the teacher data generated by the teacher data generation device.
A gate setting learning system equipped with. The learning device generates the trained model that estimates the total congestion time corresponding to any of the aisle restricted conditions.
The gate setting learning system according to claim 10. An aisle setting device that determines the aisle regulation state corresponding to the shortest total congestion time among a plurality of total congestion times estimated using the trained model generated by the learning device to the optimum aisle regulation state. Further prepare
The gate setting learning system according to claim 11. The teacher data generation device further generates optimum teacher data corresponding to the optimum passage regulation state determined by the passage setting device.
The learning device generates a modified trained model that is a modification of the trained model by performing machine learning using the teacher data generated by the teacher data generation device and the optimum teacher data.
The gate setting learning system according to claim 12. The aisle setting device newly optimizes the aisle regulation state corresponding to the shortest total aisle time among the plurality of total congestion times estimated using the modified trained model generated by the learning device. Determine the passage regulation status,
The gate setting learning system according to claim 13. It is a teacher data generation method that generates teacher data used for machine learning.
A passage state acquisition step for acquiring the passage regulation state of each of the plurality of gate devices,
A congestion degree detection step for detecting the congestion degree of users around the plurality of gate devices in the passage regulation state acquired by the passage state acquisition step, and a congestion degree detection step.
A congestion determination step for determining whether or not the congestion degree detected by the congestion degree detection step exceeds a threshold value, and a congestion determination step.
A congestion time calculation step for calculating the total congestion time indicating the total time during which the congestion degree exceeds the threshold value in a predetermined period.
A teacher data generation step that generates the teacher data in which the passage regulation state acquired by the passage state acquisition step and the total congestion time calculated by the congestion time calculation step are associated with each other.
A method of generating teacher data that is executed by one or more processors. A teacher data generation program that generates teacher data used for machine learning.
A passage state acquisition step for acquiring the passage regulation state of each of the plurality of gate devices,
A congestion degree detection step for detecting the congestion degree of users around the plurality of gate devices in the passage regulation state acquired by the passage state acquisition step, and a congestion degree detection step.
A congestion determination step for determining whether or not the congestion degree detected by the congestion degree detection step exceeds a threshold value, and a congestion determination step.
A congestion time calculation step for calculating the total congestion time indicating the total time during which the congestion degree exceeds the threshold value in a predetermined period.
A teacher data generation step that generates the teacher data in which the passage regulation state acquired by the passage state acquisition step and the total congestion time calculated by the congestion time calculation step are associated with each other.
A teacher data generation program for running one or more processors.

Images