WO2021149107A1

WO2021149107A1 - Elevator control system

Info

Publication number: WO2021149107A1
Application number: PCT/JP2020/001706
Authority: WO
Inventors: 祐貴梅田; 釜坂　等
Original assignee: 三菱電機株式会社
Priority date: 2020-01-20
Filing date: 2020-01-20
Publication date: 2021-07-29
Also published as: JP7276517B2; JPWO2021149107A1

Abstract

Provided is an elevator control system that uses image information obtained from an imaging device to predict the floors at which an unspecified number of user groups will get off, thereby making it possible to improve user convenience.　The elevator control system comprises: a processing device that, using image information in which users present at an elevator landing are recorded, identifies a group to which the elevator users belong, and that creates, as group attribute information, the overall characteristics of the group, including characteristics of the users belonging to the group; and an inference device that uses the group attribute information to create information on the floor at which the users are predicted to get off.

Description

Elevator control system

This disclosure relates to an elevator control system.

Patent Document 1 discloses an elevator control system. The control system forecasts the traffic demand for elevator car dispatch. The control system controls vehicle allocation based on the prediction.

Japanese Patent No. 6417292

However, the elevator control system described in Patent Document 1 predicts the traffic demand for car dispatch by detecting a registered individual. Therefore, when a user takes an unusual behavior by forming a group of a plurality of people, the control system cannot predict the behavior. That is, the control system cannot accurately predict traffic demand. As a result, the convenience of elevator users is not improved.

This disclosure was made to solve the above-mentioned problems. An object of the present disclosure is to provide an elevator control system capable of improving user convenience.

The elevator control system according to the present disclosure determines the group to which the elevator user belongs by using the imaging information recorded of the users existing at the elevator landing, and characterizes the users belonging to the group. It is provided with a processing device that creates the entire characteristics of the group including the group as group attribute information, and an inference device that creates information on the predicted disembarkation floor of the user using the group attribute information.

According to the present disclosure, the elevator control system predicts the disembarkation floor of an unspecified number of user groups by using the image pickup information from the image pickup device. Therefore, the convenience of the user can be improved.

It is a schematic diagram of the elevator in Embodiment 1. It is a block diagram of the control system of the elevator in Embodiment 1. FIG. This is an example of group information stored in the elevator usage status database according to the first embodiment. It is a flowchart of group information creation processing performed by the processing apparatus of an elevator in Embodiment 1. It is a flowchart of group information creation processing performed by the processing apparatus of an elevator in Embodiment 1. It is a block diagram of the learning apparatus of the elevator in Embodiment 1. FIG. This is an example of a machine learning method performed by the elevator learning device according to the first embodiment. This is an example of group information used for learning in the learning device of the elevator according to the first embodiment. FIG. 5 is a block diagram of an elevator inference device according to the first embodiment. This is an example of the inference processing result performed by the inference device of the elevator in the first embodiment. It is a flowchart of the learning operation performed by the learning apparatus of the elevator in Embodiment 1. It is a flowchart of the inference operation performed by the inference device of the elevator in Embodiment 1. FIG. 5 is a hardware configuration diagram of an elevator inference device according to the first embodiment. It is a schematic diagram of the elevator in Embodiment 2. This is a display example of the elevator car dispatch schedule display panel according to the second embodiment. It is a block diagram of the control panel of the elevator in Embodiment 3.

The embodiment will be described according to the attached drawings. In each figure, the same or corresponding parts are designated by the same reference numerals. The duplicate description of the relevant part will be simplified or omitted as appropriate.

Embodiment 1.
FIG. 1 is a schematic view of an elevator according to the first embodiment.

In the elevator system of FIG. 1, building 1 has two or more floors. The hoistway runs through each floor of Building 1. The machine room (not shown) is provided directly above the hoistway.

Each of the plurality of landings 2 is provided on each floor of the building 1. Each of the plurality of landings 2 faces the hoistway.

A plurality of image pickup devices 3 are provided on each floor of the building 1. The image pickup device 3 is provided at a position where a plurality of landings 2 on each floor can be imaged. For example, the image pickup device 3 is provided at a position where the behavior of a user who comes and goes between a plurality of landings 2 can be imaged. For example, the plurality of image pickup devices 3 photograph a plurality of landings 2 on each floor. The plurality of imaging devices 3 output imaging information.

The hoisting machine 4 is provided in the machine room. The main rope 5 is wound around the hoisting machine 4.

The car 6 is provided inside the hoistway. The car 6 is hung on one side of the main rope 5. Balanced weights (not shown) are provided inside the hoistway. The balance weight is hung on the other side of the main rope 5.

Each of the plurality of entrances and exits 7 is provided between the hoistway and each of the plurality of landings 2 on each floor. The user passes through the doorway 7 and moves between the landing 2 and the car 6.

For example, the control panel 8 is provided in the machine room. For example, the control panel 8 is provided at the upper part of the hoistway. The control panel 8 is electrically connected to the image pickup apparatus 3 by wire or wirelessly. The control panel 8 is electrically connected to the hoisting machine 4 by wire or wirelessly.

The control panel 8 controls the operation of the elevator system as a control system. The control panel 8 controls the movement of the hoisting machine 4. The control panel 8 controls the allocation of one or more cars 6.

The control panel 8 receives imaging information from the imaging device 3. The control panel 8 stores imaging information.

The control panel 8 uses the image pickup information to predict the floor on which the user at the landing 2 will get off. The control panel 8 includes a prediction model (not shown) for predicting the disembarkation floor of the user.

For example, the control panel 8 controls the operation of the elevator system by using the predicted exit floor information. The outline will be described below.

The imaging device 3 images a plurality of users who have arrived at the landing 2 and creates imaging information. The image pickup apparatus 3 outputs the image pickup information to the control panel 8.

The control panel 8 uses the imaging information to consider the plurality of users as one group. The control panel 8 creates personal attribute information for each user from the appearance characteristics of the user. The control panel 8 creates group attribute information from the human attribute information of a plurality of users belonging to the group.

The control panel 8 inputs the group attribute information into a prediction model (not shown). As a result, the control panel 8 predicts the floor on which the group will get off the elevator.

The control panel 8 registers the floor predicted to be disembarked by the group in the destination floor of the car 6. After that, the control panel 8 dispatches the elevator car.

The control panel 8 creates the prediction model using the stored imaging information. The outline will be described below.

The imaging device 3 gets off from the elevator, images a group of users existing at the landing 2, and creates imaging information. The image pickup device 3 outputs the image pickup information to the control panel 8.

The control panel 8 integrates the group attribute information of the user group and the getting-off floor and stores it as group information. For example, the control panel 8 integrates the group attribute information, the disembarkation floor, and a plurality of other information. Specifically, the information to be integrated is the boarding floor, boarding time, getting-off time, floor information of the getting-off floor, etc. of the user group.

The control panel 8 creates a prediction model using a plurality of stored group information. The prediction model is used to predict the drop-off floor of a user.

Next, the elevator control system according to the first embodiment will be described with reference to FIG.
FIG. 2 is a block diagram of the elevator control system according to the first embodiment.

As shown in FIG. 2, the control panel 8 includes a processing device 10, a learning device 20, an inference device 30, a vehicle allocation analysis device 40, and a vehicle allocation device 50.

The processing device 10 includes a conversion unit 11, an individual image extraction unit 12, a grouping unit 13, a person attribute determination unit 14, a boarding / alighting determination unit 15, an incidental information unit 16, an integration unit 17, and a group information storage unit 18.

The conversion unit 11 receives the imaging information from the imaging device 3. The conversion unit 11 converts the imaging information to create the first image information which is a still image. The conversion unit 11 identifies one or more users recorded in the first image information.

The conversion unit 11 creates behavioral characteristic information obtained from the imaging information for each of the users. For example, the behavioral characteristic information includes information on the position of the user in the imaging, information on the moving direction of the user, and information on the moving speed of the user.

The conversion unit 11 creates the second image information including the first image information and the behavior characteristic information.

The conversion unit 11 determines whether or not the user in the second image information is in the unprocessed state. For example, when the target user is not given the group ID information, the conversion unit 11 determines that the user is in the unprocessed state.

The individual image extraction unit 12 creates the third image information by extracting the image of only one user by using the second image information. For example, the third image information is an image in which the whole body of one user obtained by trimming the second image information is recorded. For example, the third image information is an image in which the face of one user obtained by trimming the second image information is recorded.

The grouping unit 13 classifies the users recorded in the second image information into one or more groups. For example, the grouping unit 13 determines a plurality of users having similar behavioral characteristic information as the same group.

The grouping unit 13 creates group ID information unique to the group to which the user belongs. If the group to which the user belongs does not yet have a unique group ID, the grouping unit 13 assigns new group ID information to the group.

The grouping unit 13 updates the second image information by adding the group ID information of the user to the second image information.

The grouping unit 13 outputs the group ID information to the integration unit 17.

The human attribute determination unit 14 includes a human attribute estimation model for estimating human attributes from image information. Human attributes mean personal characteristics that are inferred from a person's appearance. For example, the human attribute estimation model is a model created by machine learning outside the elevator system.

The human attribute determination unit 14 creates the personal attribute information of each user recorded in the second image information. For example, the human attribute determination unit 14 estimates the height, gender, age, clothes, etc. of the user recorded in the second image information. The human attribute determination unit 14 integrates the estimated classification items and creates human attribute information.

The boarding / alighting determination unit 15 determines the boarding state and the disembarking state of the user by using the behavior characteristic information of the imaging information. The boarding / alighting determination unit 15 determines whether the user is in a state after getting off the elevator or in a state of waiting for getting on the elevator. After that, the boarding / alighting determination unit 15 creates boarding floor information and getting-off floor information.

For example, if the user has behavioral characteristic information that the user has reached the landing 2 without passing through the entrance 7 on the third floor of the building 1, the boarding / alighting determination unit 15 puts the user in a state of waiting for the elevator to board. Judge that there is. After that, the boarding / alighting determination unit 15 creates boarding floor information "3" and getting-off floor information "0" for the user. "3" in the boarding floor information means that the boarding is waiting on the 3rd floor. "0" in the disembarkation floor information means that the user is not in the disembarked state.

For example, when the user has the behavioral characteristic information that the user has passed the entrance 7 and reached the landing 2 on the third floor of the building 1, the boarding / alighting determination unit 15 is in the state after getting off the elevator. Is determined. After that, the boarding / alighting determination unit 15 creates boarding floor information "0" and getting-off floor information "3" for the user. "0" in the boarding floor information means that the user is not in the boarding waiting state. "3" in the disembarkation floor information means that the user has disembarked on the third floor.

The incidental information unit 16 gives the user incidental information that is information other than that related to the individual user. For example, the incidental information unit 16 imparts information on the date and time when the second image information was captured to the user recorded in the second image information. For example, the incidental information unit 16 gives the information of the floor on which the second image information was taken to the user recorded in the second image information. For example, the incidental information unit 16 gives the floor information of the floor on which the second image information is captured to the user recorded in the second image information. Specifically, the floor information on the floor is that the 5th floor is a toy department and the 12th floor is a restaurant.

The integration unit 17 acquires the second image information from the conversion unit 11. The integration unit 17 acquires the third image information from the individual image extraction unit 12. The integration unit 17 acquires group ID information from the grouping unit 13. The integration unit 17 acquires human attribute information from the human attribute determination unit 14. The integration unit 17 acquires boarding floor information and getting-off floor information from the boarding / alighting determination unit 15. The integration unit 17 acquires incidental information from the incidental information unit 16.

The integration unit 17 integrates the user information recorded in the second image information and creates the user information. For example, the integration unit 17 integrates the user's third image information, group ID information, personal attribute information, disembarkation information, boarding information, and incidental information recorded in the second image information to create user information. ..

The group information storage unit 18 acquires user information from the integration unit 17.

The group information storage unit 18 adds the user information to the group information that stores the user information. The group information storage unit 18 stores the group information to which the user information is added.

For example, when the group ID of the user information of a certain user A is a group ID already stored in the group information storage unit 18, the group information storage unit 18 adds the user A to the group information having the group ID. Add user information.

For example, when the group ID of the user information of a certain user B is a group ID that is not yet stored in the group information storage unit 18, the group information storage unit 18 creates new group information having the group ID. , Add the user information of user B to the new group information.

The group information storage unit 18 creates the boarding floor information of the group by using the boarding floor information of the users belonging to the group. The group information storage unit 18 creates the disembarkation floor information of the group by using the disembarkation floor information of the users belonging to the group. The group information storage unit 18 adds boarding floor information and getting-off floor information to the group information.

The group information storage unit 18 integrates the human attribute information of the users belonging to the group, the individual image information, and the incidental information to create the group attribute information which is the attribute information of the entire group. When the user information is newly added to the group information, the group information storage unit 18 newly creates the group attribute information. The group information storage unit 18 adds the group attribute information to the group information.

The learning device 20 uses group information as teacher data to perform so-called supervised learning. The learning device 20 creates a learned inference model that infers the disembarkation floor of the user by using the group information of the user.

The inference device 30 includes a learned inference model created by the learning device 20. The inference device 30 infers the disembarkation floor of the user by using the group information of the user.

For example, the vehicle allocation analysis device 40 acquires the predicted disembarkation floor information from the inference device 30. For example, the vehicle allocation analysis device 40 calculates the optimum elevator vehicle allocation information from the call information registered from the call buttons on each floor, the destination registration information registered from the inside of the car 6, and the predicted disembarkation information. For example, the vehicle allocation analysis device 40 uses DOAS (Destination Oriented Allocation System: elevator destination forecast system) to calculate the optimum elevator vehicle allocation information.

The vehicle allocation device 50 acquires vehicle allocation information from the vehicle allocation analysis device 40. For example, the vehicle dispatching device 50 allocates the car 6 based on the vehicle allocation information. For example, the vehicle dispatching device 50 outputs a drive command to the hoisting machine 4 in order to dispatch the car 6.

Next, an example of the group information stored in the group information storage unit 18 will be described with reference to FIG.
FIG. 3 is an example of group information stored in the elevator usage status database according to the first embodiment.

As shown in FIG. 3, the group attribute information includes a group ID, a group boarding floor information, a group getting-off floor information, and a group attribute information.

The group attribute information includes personal attribute information, incidental information, and individual image information of each group.

For example, human attribute information includes gender, height, age, body shape, clothing, and other appearance information.

For example, group A1005 means a group whose group ID is A1005. Group A1005 is a group to which one woman C belongs. The height of female C is about 160 cm. Female C is about 30 years old. The body shape of Female C is thin. Woman C wears a red jacket, white trousers and glasses. The group information of group A1005 is the information that female C boarded on the entrance floor on the first floor on October 1, 2019. “Yes” in the column of group A1005 means that the individual image extraction unit 12 created the third image information of the whole body and the face of the woman C.

For example, group A1006 is a group to which one woman belongs. Group A1006 has the same group attributes as group A1005. Therefore, group A1006 is the same group as group A1005. The woman belonging to group A1006 is woman C. The group information of group A1006 is the information that woman C got off at the women's clothing section on the 3rd floor on October 1, 2019.

In this way, the information that the woman C got on is recorded in A1005, and the information that the woman C got off is recorded in A1006.

Next, a method in which the processing apparatus processes the imaging information will be described with reference to FIGS. 4 and 5.
4 and 5 are flowcharts of group information creation processing performed by the elevator processing device according to the first embodiment.

In step S001, the conversion unit 11 acquires the imaging information from the imaging device 3.

After that, the operation of step S002 is performed. In step S002, the conversion unit 11 creates continuous still image information using the image pickup information received from the image pickup apparatus 3 as the first image information.

After that, the operation of step S003 is performed. In step S003, the conversion unit 11 identifies one or more users recorded in the first image information. The conversion unit 11 creates behavioral characteristic information for each of the users based on the continuous still image information. The conversion unit 11 creates the second image information including the first image information and the created behavioral characteristic information.

After that, the operation of step S004 is performed. In step S004, the conversion unit 11 designates one user recorded in the second image information as the first user to be processed.

After that, the operation of step S005 is performed. In step S005, the conversion unit 11 determines whether or not the first user is in the unprocessed state. For example, the conversion unit 11 determines that the first user is in the unprocessed state when the group ID information is not given to the first user.

When the conversion unit 11 determines in step S005 that the first user is in an unprocessed state, the operation of step S006 is performed. In step S006, the individual image extraction unit 12 creates the third image information of the first user by using the second image information.

After that, the operation of step S007 is performed. In step S007, the grouping unit 13 determines whether or not a second user, who is a user having behavioral characteristic information similar to that of the first user, exists in the second image information.

In step S007, when a user having behavioral characteristic information similar to that of the first user exists in the second image information, the operation of step S008 is performed. In step S008, the grouping unit 13 determines whether or not the second user has the group ID information.

In step S008, when the second user has the group ID information, the operation of step S009 is performed. In step S009, the grouping unit 13 creates group ID information of the first user having the same group ID as the second user. The grouping unit 13 outputs the created group ID information to the integration unit 17. The grouping unit 13 assigns the created group ID information to the first user in the second image information.

After that, the operation of step S010 is performed. In step S010, the human attribute determination unit 14 creates the human attribute information of the first user.

After that, the operation of step S011 is performed. In step S011, the boarding / alighting determination unit 15 creates boarding floor information and getting-off floor information of the first user based on the behavior characteristic information of the first user.

After that, the operation of step S012 is performed. In step S012, the incidental information unit 16 creates the incidental information from the second image information.

After that, the operation of step S013 is performed. In step S013, the integration unit 17 includes the third image information of the first user, the group ID information of the first user, the personal attribute information of the first user, the boarding floor information of the first user, and the first user. Get off floor information and incidental information. The integration unit 17 includes the third image information of the first user, the group ID information of the first user, the personal attribute information of the first user, the boarding floor information of the first user, and the disembarking floor information of the first user. And incidental information are integrated to create user information of the first user.

After that, the operation of step S014 is performed. In step S014, the group information storage unit 18 acquires the first user information. The group information storage unit 18 determines whether or not, among the plurality of group information stored in the group information storage unit 18, there is one having the same group ID as the group ID of the first user.

If there is group information having the same group ID as the first user in step S014, the operation of step S015 is performed. In step S015, the group information storage unit 18 adds the user information of the first user to the group information.

After that, the operation of step S016 is performed. In step S016, the conversion unit 11 determines whether or not there is a user who does not have the group ID information in the second image information.

In step S016, if there is no user who does not have the group ID information in the second image information, the processing device 10 ends the processing of the imaging information.

If the conversion unit 11 does not determine the first user as an unprocessed state in step S005, or if there is a user who does not have the group ID information in the second image information in step S016, the operation of step S004 is performed. It is said.

In step S007, when a user having behavioral characteristic information similar to that of the first user does not exist in the second image information, or in step S008, when the second user does not have group ID information, step S017 The operation of is performed. In step S017, the grouping unit 13 creates group ID information of the first user having a new group ID. The grouping unit 13 outputs new group ID information to the integration unit 17. The grouping unit 13 assigns new group ID information to the first user in the second image information.

After that, the operations after step S010 are performed.

If there is no group information having the same group ID as the first user in step S014, the operation of step S018 is performed. In step S018, the group information storage unit 18 creates new group information including the group ID of the first user. The group information storage unit 18 adds the user information of the first user to the new group information.

After that, the operation of step S019 is performed. In step S019, the group information storage unit 18 creates group attribute information, incidental information, individual image information, boarding floor information, and getting-off floor information of the new group using the user information of the first user.

After that, the operations after step S016 are performed.

Next, the learning device 20 will be described with reference to FIG.
FIG. 6 is a block diagram of the learning device for the elevator according to the first embodiment.

As shown in FIG. 6, the learning device 20 includes a usage status database 21 (hereinafter, usage status DB 21), a learning data acquisition unit 22, and a model generation unit 23.

The usage status DB 21 acquires group information from the processing device 10. The usage status DB 21 stores group information. The usage status DB 21 rewrites the boarding floor information of the group information already stored to the getting-off floor information of the same acquired group information.

For example, the usage status DB 21 stores some group information in which the disembarkation floor information is "0". The usage status DB 21 acquires the group information D in which the disembarkation floor information is "3" and the boarding floor information is "0". After that, the usage status DB 21 selects the group information E whose group attribute information is similar to the group information D in the group information already stored. The usage status DB 21 determines that the group information D and the group information E are the same group information. The disembarkation floor information of the group information D is "0". The boarding floor information of group information E is "1". The usage status DB 21 rewrites the disembarkation floor information of the group information E to the disembarkation floor information of the group information D. That is, the disembarkation floor information of the group information E is "3". In this way, the usage status DB 21 stores the disembarkation floor information and the boarding floor information of the same group so as to be paired.

The usage status DB 21 determines whether or not a sufficient amount of group information for learning is stored. For example, when the usage status DB 21 stores 1000 groups or more of group information, it determines that a sufficient amount of group information for learning is stored. For example, when the usage status DB 21 stores group information for a period of one week or more, it determines that it stores a sufficient amount of group information for learning.

The learning data acquisition unit 22 acquires group attribute information, boarding floor information, and getting-off floor information from the usage status DB 21.

The model generation unit 23 acquires group attribute information, boarding floor information, and getting-off floor information from the learning data acquisition unit 22.

The model generation unit 23 learns the predicted disembarkation floor information based on the learning data created based on the combination of the group attribute information, the boarding floor information, and the disembarking floor information. That is, the model generation unit 23 generates a learned model that infers the optimum predicted exit floor information from the elevator group attribute information and the boarding floor information.

The learning device 20 is used to learn the predicted exit floor information of elevator users. For example, it may be connected to the control system of the elevator via a network and may be a device separate from the elevator. Further, the learning device 20 may be built in the elevator system. Further, the learning device 20 may exist on the cloud server.

As the learning algorithm used by the model generation unit 23, known algorithms such as supervised learning, unsupervised learning, and reinforcement learning can be used.

For example, the model generation unit 23 learns the predicted disembarkation floor information by so-called supervised learning according to the neural network model. Here, supervised learning refers to a method of learning a feature in the learning data by giving a set of input and result (label) data to the learning device 20 and inferring the result from the input.

A neural network is composed of an input layer composed of a plurality of neurons, an intermediate layer (hidden layer) composed of a plurality of neurons, and an output layer composed of a plurality of neurons. The intermediate layer may be one layer or two or more layers.

For example, in the present disclosure, the neural network predicts disembarkation by so-called supervised learning according to learning data created based on a combination of group attribute information, boarding floor information, and disembarking floor information acquired by the data acquisition unit. Learn floor information.

Next, with reference to FIG. 7, a case where a neural network is applied will be described as an example of a machine learning method.
FIG. 7 is an example of a machine learning method performed by the elevator learning device according to the first embodiment.

For example, as shown in FIG. 7, in the case of a three-layer neural network, when a plurality of inputs are input to the input layer (X1-X3), the value is multiplied by the weight W1 (w11-w16) to be intermediate. It is input to the layer (Y1-Y2), and the result is further multiplied by the weight W2 (w21-w26) to be output from the output layer (Z1-Z3). This output result depends on the values of the weights W1 and W2.

That is, the neural network learns by inputting the group attribute information and the boarding floor information into the input layer and adjusting the weights W1 and W2 so that the result output from the output layer approaches the getting-off floor information.

Next, an example of group information used for learning will be described with reference to FIG.
FIG. 8 is an example of group information used for learning in the learning device of the elevator according to the first embodiment.

As shown in FIG. 8, the group attribute information includes a group ID, a group boarding floor information, a group getting-off floor information, and a group attribute information.

For example, group A1005 is a group to which female C belongs. The group information of Group A1005 is the information that Woman C got on the entrance floor on the 1st floor on October 1, 2019, and then got off at the women's clothing section on the 3rd floor.

Other information examples are also shown in FIG.

Next, the inference device 30 will be described with reference to FIG.
FIG. 9 is a block diagram of the elevator inference device according to the first embodiment.

As shown in FIG. 9, the inference device 30 includes a learned model storage unit 31, a usage data acquisition unit 32, and an inference unit 33.

The trained model storage unit 31 acquires the trained model from the model generation unit 23. The trained model storage unit 31 stores the learning model.

The usage data acquisition unit 32 acquires group attribute information and boarding floor information from the processing device 10.

The reasoning unit 33 infers the predicted disembarkation floor information obtained by using the trained model. The inference unit 33 acquires the trained model from the trained model storage unit 31. The inference unit 33 acquires group attribute information and boarding floor information from the usage data acquisition unit 32. After that, the inference unit 33 inputs the group attribute information and the boarding floor information into the trained model. The reasoning unit 33 creates the predicted disembarkation floor information inferred from the group attribute information and the boarding floor information.

Next, an example of inference performed by the inference device 30 will be described with reference to FIG.
FIG. 10 is an example of the inference processing result performed by the elevator inference device according to the first embodiment.

As shown in FIG. 10, the inference device 30 (not shown) acquires group attribute information and boarding floor information as INPUT information. The inference device 30 (not shown) outputs predicted disembarkation floor information as OUTPUT information.

For example, group A1101 is a group to which female F belongs. The group information of group A1101 is the information that female F boarded from the entrance on the first floor at 9:35 on October 2, 2019.

The inference device 30 (not shown) applies a prediction model to the information of the group whose group ID is A1101, and predicts that the group will get off at the women's clothing section on the 3rd floor. The inference device 30 (not shown) creates the predicted disembarkation floor information “3”.

Examples of other inference processing are also as shown in FIG.

Next, the process of learning by the learning device will be described with reference to FIG.
FIG. 11 is a flowchart of a learning operation performed by the learning device of the elevator according to the first embodiment.

In step S101, the usage status DB 21 acquires the first group information, which is one group information, from the processing device 10.

After that, the operation of step S102 is performed. In step S102, the usage status DB 21 determines whether or not the first group information has the disembarkation floor information. When the disembarkation floor information of the first group information is not "0", the usage status DB 21 determines that the first group information has the disembarkation floor information.

In step S102, when the first group information has the disembarkation floor information, the operation of step S103 is performed. In step S103, the usage status DB 21 determines whether or not the stored group information has the disembarkation floor information "0" and the same group information as the acquired group information exists.

In step S103, when the disembarkation floor information is "0" and the same group information as the acquired group information (hereinafter referred to as the second group information) exists, the operation of step S104 is performed. In step S104, the usage status DB 21 rewrites the disembarkation floor information of the second group information to the disembarkation floor information of the first group information.

After that, the operation of step S105 is performed. In step S105, the usage status DB 21 determines whether or not a sufficient amount of group information for learning is stored.

In step S105, when the usage status DB 21 stores a sufficient amount of group information for learning, the operation of step S106 is performed. In step S106, the learning data acquisition unit 22 acquires group attribute information, boarding floor information, and getting-off floor information from the usage status DB 21 as learning data.

After that, the operation of step S107 is performed. In step S107, the model generation unit 23 performs so-called supervised learning according to the learning data created based on the combination of the group attribute information, the boarding floor information, and the disembarking floor information acquired by the learning data acquisition unit 22. Learn the predicted drop-off floor information and create a trained model.

After that, the operation of step S108 is performed. In step S108, the trained model storage unit 31 stores the trained model generated by the model generation unit 23.

After that, the learning process ends.

If the first group information does not have the disembarkation floor information in step S102, or if the disembarkation floor information is "0" and the same group information does not exist in step S103, the operation of step S109 is performed. In step S109, the usage status DB 21 stores the first group information.

After that, the operations after step S105 are performed.

Next, with reference to FIG. 12, a process for obtaining predicted disembarkation floor information using the inference device 30 will be described.
FIG. 12 is a flowchart of an inference operation performed by the elevator inference device according to the first embodiment.

In step S201, the usage data acquisition unit 32 acquires the third group information from the processing device 10 as the group information for inference.

After that, the operation of step S202 is performed. In step S202, the usage data acquisition unit 32 determines whether or not the third group information has boarding floor information. When the boarding floor information of the third group information is not "0", the usage data acquisition unit 32 determines that the third group information has the boarding floor information.

In step S202, when the third group information has boarding floor information, the operation of step S203 is performed. In step S203, the inference unit 33 acquires the group attribute information and the boarding floor information possessed by the third group information.

After that, the operation of step S204 is performed. In step S204, the inference unit 33 acquires the trained model stored in the trained model storage unit 31. The inference unit 33 calculates the boarding / alighting floor of the third group using the trained model. The inference unit 33 creates the predicted disembarkation floor information.

After that, the operation of step S205 is performed. In step S205, the inference unit 33 outputs the predicted disembarkation floor information to the vehicle allocation analysis device 40.

After that, the operation of step S206 is performed. In step S206, the vehicle allocation analyzer 40 calculates the elevator car allocation based on the predicted disembarkation floor information. The vehicle allocation analyzer 40 creates vehicle allocation information.

After that, the operation of step S207 is performed. In step S207, the vehicle dispatching device 50 acquires the car dispatching information. The vehicle dispatching device 50 allocates the elevator car 6 based on the car dispatching information.

According to the first embodiment described above, the processing device 10 creates group attribute information representing the characteristics of the group of elevator users by using the imaging information. The inference device 30 creates predicted disembarkation floor information of an unspecified number of user groups by using the group attribute information. Therefore, the inference device 30 can predict that the users will behave differently from usual when they form a group of a plurality of people. This means that the prediction accuracy of the inference device 30 is improved. As a result, the user can use the elevator for more efficient vehicle allocation. That is, the elevator control system can improve the convenience of the user.

Further, the control panel 8 is provided with a learning device 20 that infers the predicted disembarkation floor information by using the group attribute information and the disembarkation floor information. Therefore, the elevator control system can create a prediction model of the user's getting-off floor.

Further, the control panel 8 is provided with a vehicle allocation analyzer 40. The vehicle allocation analysis device 40 creates elevator car allocation information using the predicted exit floor information. The vehicle dispatching device 50 dispatches the elevator car 6 in which the predicted disembarkation floor information is registered as the destination by using the car dispatching information. Therefore, the elevator user can get on the elevator car 6 without registering the destination floor by himself / herself.

Here, the conversion unit 11 does not have to be provided in the processing device 10. For example, the conversion unit 11 is provided in the image pickup apparatus 3. In this case, the conversion unit outputs the converted imaging information to the processing device 10.

Further, for example, the learning device 20 periodically executes the learning process. The frequency of processing that the learning device 20 learns is set in advance. For example, the learning device 20 executes a process of learning once a week. The amount of information stored in the usage status DB 21 increases as the storage period becomes longer. The larger the amount of information stored in the usage status DB 21, the better the prediction accuracy of the model.

Further, the usage status DB 21 stores the correct disembarkation floor information of the user who has calculated the wrong estimation result by the inference device 30 as learning data. Therefore, the learning device 20 creates a trained model capable of making more accurate inferences.

Further, for example, the learning data acquisition unit 22 may acquire the data of the group attribute information, the boarding floor information, and the getting-off floor information at different timings.

Further, the learning data acquisition unit 22 can add an elevator system for collecting learning data to the target on the way or remove it from the symmetry. Further, the learning device that has learned the predicted exit floor information for one elevator system may be applied to another elevator system, and the predicted exit floor information for the other elevator system may be relearned and updated. ..

Further, in the present embodiment, the case where supervised learning is applied to the learning algorithm used by the model generation unit 23 has been described, but the present invention is not limited to this. As for the learning algorithm, it is also possible to apply reinforcement learning, unsupervised learning, semi-supervised learning, etc. in addition to supervised learning.

Further, the model generation unit 23 may learn the predicted disembarkation floor information according to the learning data created for the plurality of elevator systems. The model generation unit 23 may acquire learning data from a plurality of elevator systems used in the same area. The model generation unit 23 may learn the predicted disembarkation floor information by using the learning data collected from a plurality of elevator systems that operate independently in different areas.

Further, as a learning algorithm used in the model generation unit 23, deep learning that learns the extraction of the feature amount itself can also be used, and other known methods such as genetic programming and functional logic programming can be used. Machine learning may be performed according to a support vector machine or the like.

Further, although the reasoning device 30 has been described as outputting the predicted disembarkation floor information using the trained model learned by the model generation unit 23 of the elevator, the trained model is acquired from the outside of another elevator or the like, and this The predicted drop-off floor information may be output based on the trained model.

For example, if the disembarkation floor desired by the user and the predicted disembarkation floor inferred by the inference device 30 are different, the user himself / herself registers the desired disembarkation floor as the destination. Therefore, the user can use the elevator having the usual convenience.

Next, an example of the inference device 30 will be described with reference to FIG.
FIG. 13 is a hardware configuration diagram of the elevator inference device according to the first embodiment.

Each function of the inference device 30 can be realized by a processing circuit. For example, the processing circuit includes at least one processor 100a and at least one memory 100b. For example, the processing circuit comprises at least one dedicated hardware 200.

When the processing circuit includes at least one processor 100a and at least one memory 100b, each function of the inference device 30 is realized by software, firmware, or a combination of software and firmware. At least one of the software and firmware is written as a program. At least one of the software and firmware is stored in at least one memory 100b. At least one processor 100a realizes each function of the inference device 30 by reading and executing a program stored in at least one memory 100b. At least one processor 100a is also referred to as a central processing unit, a processing unit, an arithmetic unit, a microprocessor, a microcomputer, and a DSP. For example, at least one memory 100b is a non-volatile or volatile semiconductor memory such as RAM, ROM, flash memory, EPROM, EEPROM, magnetic disk, flexible disk, optical disk, compact disk, mini disk, DVD, or the like.

If the processing circuit comprises at least one dedicated hardware 200, the processing circuit may be implemented, for example, as a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC, an FPGA, or a combination thereof. NS. For example, each function of the monitoring device 9 is realized by a processing circuit. For example, each function of the inference device 30 is collectively realized by a processing circuit.

For each function of the inference device 30, a part may be realized by the dedicated hardware 200, and the other part may be realized by software or firmware. For example, the function of the inference unit 33 is realized by a processing circuit as dedicated hardware 200, and for functions other than the function of the inference unit 33, at least one processor 100a reads a program stored in at least one memory 100b. It may be realized by executing.

In this way, the processing circuit realizes each function of the inference device 30 by hardware 200, software, firmware, or a combination thereof.

Although not shown, each function of the imaging device 3 is also realized by a processing circuit equivalent to a processing circuit that realizes each function of the inference device 30. Each function of the processing device 10 is also realized by a processing circuit equivalent to a processing circuit that realizes each function of the inference device 30. Each function of the learning device 20 is also realized by a processing circuit equivalent to a processing circuit that realizes each function of the inference device 30. Each function of the vehicle allocation analyzer 40 is also realized by a processing circuit equivalent to a processing circuit that realizes each function of the inference device 30. Each function of the vehicle dispatching device 50 is also realized by a processing circuit equivalent to a processing circuit that realizes each function of the inference device 30. Each function of the external information device 70 is also realized by a processing circuit equivalent to a processing circuit that realizes each function of the inference device 30.

Embodiment 2.
FIG. 14 is a schematic view of the elevator according to the second embodiment. The same or corresponding parts as those in the first embodiment are designated by the same reference numerals. The explanation of the relevant part is omitted.

As shown in FIG. 14, the vehicle allocation analyzer 40 has a function of outputting vehicle allocation information to the outside.

The car dispatch schedule display board 60 is provided on each floor of the building 1. For example, the car dispatch schedule display board 60 is provided around the doorway 7 on each floor. For example, the car dispatch schedule display board 60 is provided at a position where the user at the landing 2 can easily see it. The car dispatch schedule display panel 60 is electrically connected to the vehicle dispatch analyzer 40 by wire or wirelessly. The car dispatch schedule display board 60 acquires the car dispatch information of the elevator from the vehicle dispatch analysis device 40. The car dispatch schedule display board 60 displays the acquired car dispatch information.

FIG. 15 is a display example of the elevator car dispatch schedule display panel in the second embodiment.

As shown in FIG. 15, the car dispatch schedule display panel 60 displays information related to the car dispatch of the elevator by using the car dispatch information. For example, the car dispatch schedule display panel 60 displays the stop schedule floors of a plurality of elevator cars. For example, the car dispatch schedule display board 60 displays the estimated time until each elevator car arrives.

According to the second embodiment described above, the vehicle allocation analysis device 40 has a function of outputting vehicle allocation information to the vehicle allocation schedule display panel 60. The car dispatch schedule display panel 60 conveys car dispatch information to elevator users. Therefore, the user can know the number of the elevator car that requires less time to reach the target floor before the car arrives. The user can move to the vicinity of the door of the elevator car in which he / she rides before the car arrives. As a result, the convenience of the elevator user can be improved.

The notification device that notifies the user of the car dispatch information is not limited to the car dispatch schedule display panel 60. For example, the notification device may be a voice device capable of outputting car allocation information by voice. The voice device acquires car allocation information from the vehicle allocation analysis device 40. The voice device notifies the user of the car dispatch information by outputting a voice representing the car dispatch information.

Embodiment 3.
FIG. 16 is a block diagram of an elevator control panel according to the third embodiment. The same or corresponding parts as those in the first embodiment are designated by the same reference numerals. The explanation of the relevant part is omitted.

As shown in FIG. 16, the control panel 8 includes an external information device 70.

The external information device 70 receives arbitrary external information from an input terminal (not shown). For example, the external information device 70 receives weather information, temperature information, event information held in the building 1, sales information on the sales floor, and the like.

The incidental information unit 16 acquires external information from the external information device 70. The incidental information unit 16 creates incidental information including external information.

The integration unit 17 acquires incidental information including external information from the incidental information unit 16. The integration unit 17 creates user information including external information.

The group information storage unit 18 acquires user information including external information from the integration unit 17. The group information storage unit 18 creates group attribute information including external information.

The learning device 20 acquires group attribute information including external information from the group information storage unit 18. The learning device 20 uses group attribute information including external information at the time of learning. Therefore, the learning device 20 generates a learned model in which external information is reflected.

The inference device 30 acquires group attribute information including external information from the group information storage unit 18. The inference device 30 uses group attribute information including external information at the time of inference. Therefore, the inference device 30 creates the predicted disembarkation floor information that reflects the external information.

According to the third embodiment described above, the control panel 8 includes an external information device 70 that receives arbitrary external information from the outside. The learning device 20 generates a trained model in which external information is reflected. The inference device 30 outputs predicted disembarkation floor information that reflects external information. Therefore, even if the external information changes, the predicted disembarkation floor information will be the information that reflects the change. That is, the prediction accuracy of the predicted disembarkation floor information is improved. Elevators are dispatched more efficiently and faster. As a result, the convenience of the elevator user can be improved.

As described above, the elevator control system according to the present disclosure can be used for the elevator system.

1 building, 2 landing, 3 imaging device, 4 hoisting machine, 5 main rope, 6 basket, 7 doorway, 8 control panel, 10 processing device, 11 conversion unit, 12 individual image extraction unit, 13 grouping unit, 14 Human attribute judgment unit, 15 boarding / alighting judgment unit, 16 incidental information unit, 17 integration unit, 18 group information storage unit, 20 learning device, 21 usage status database, 22 learning data acquisition unit, 23 model generation unit, 30 inference device, 31 Learned model storage unit, 32 usage data acquisition unit, 33 inference unit, 40 vehicle allocation analyzer, 50 vehicle allocation device, 60 car allocation schedule display panel, 70 external information device

Claims

Using the imaging information recorded of the users existing at the elevator landing, the group to which the elevator user belongs is determined, and the overall characteristics of the group including the characteristics of the users belonging to the group are grouped. The processing device created as attribute information and
An inference device that creates information on the predicted exit floor of the user using the group attribute information, and
Elevator control system with.
The first aspect of claim 1 is provided with a learning device for generating a trained model for inferring predicted disembarkation floor information from group attribute information by using the disembarkation floor information representing the floor on which the group disembarked and the group attribute information. Described elevator control system.
The elevator control system according to claim 1 or 2, further comprising a vehicle allocation analyzer that creates vehicle allocation information that controls the elevator that reflects the information on the predicted exit floor.
The elevator control system according to claim 3, wherein the vehicle allocation analysis device outputs the car allocation information to a notification device that notifies the user of the elevator control information.
An external information device that outputs external information input from the outside,
The processing device that creates the group attribute information including the external information, and
The elevator control system according to any one of claims 1 to 4.