WO2019176196A1 - Elevator system - Google Patents

Abstract

Useless operations are generated in a repetition of round trips between the lowest floor and the highest floor. Accordingly, in controls of a plurality of elevators, estimated operation data is used, which is used for controlling at a prescribed future time point from an estimation time point. The estimated operation data is based on a plurality of output values estimated at the prescribed future time point from the estimation time point. The plurality of estimated output values are output from a learnt model to which a plurality of input values belonging to the estimation time point are input. The plurality of input values belonging to the estimation time point include a plurality of time input values that respectively indicate a plurality of time elements belonging to the estimation time point. The plurality of output values include values about a plurality of types of departure and arrival combinations. Each departure and arrival combination is a combination of a departure floor and an arrival floor.

Description

Elevator system

The present invention generally relates to the management of multiple elevators.

There are, for example, Japanese Patent No. 4606681 and Japanese Patent No. 4139819 as technologies related to management of a plurality of elevators.

Japanese Patent No. 4,606,681 discloses the following group management control device. That is, the group management control device detects traffic demands of a plurality of elevators and predicts near-future traffic demands based on the detected traffic demands. The group management control device discriminates a near-future traffic pattern from the predicted traffic demand, and automatically generates a plurality of control rule group candidates to be applied to the near future based on the discriminated traffic pattern. The group management control device evaluates each of the plurality of candidates and selects one of the candidates based on the evaluation result.

Patent No. 4,606,681

By the way, the elevator operation is usually repeated in a round trip between the lowest floor and the top floor in order to service passengers on multiple floors. Also, in order to be able to deal with anytime and anywhere car calls, multiple elevators can cover all the processes from the bottom floor to the top floor so that the car positions are distributed so that they do not overlap. Is common.

The technology of Patent Document 1 is a technology related to control of such round trips.

繰 り 返 し Repeated round trips between the lowest floor and the top floor have the problem of wasteful operation.

In the control of a plurality of elevators, estimated operation data used for control for a predetermined time in the future from the estimated time is used. The estimated operation data is operation data based on a plurality of estimated output values for a future time point for a predetermined time from the estimated time point. The plurality of estimated output values are a plurality of output values output from the learned model to which the plurality of input values belonging to the estimation time point are input. The plurality of input values belonging to the estimated time point include a plurality of time input values respectively indicating a plurality of time elements belonging to the estimated time point. The plurality of output values include values relating to a plurality of departure / arrival combinations. Each departure / arrival combination is a combination of a departure floor and an arrival floor.

According to the present invention, it is not assumed that the round trip between the lowermost floor and the uppermost floor is repeated, and the departure floor and arrival floor estimated by inputting a plurality of input values belonging to the estimated time point into the learned model. That is, based on the combination of the departure floor and arrival floor at a certain future time, the operation for the future time is controlled. Thereby, since useless operation can be reduced, the operation efficiency of a plurality of elevators can be improved.

The example of the block diagram of the whole system containing the group management controller as an example of the elevator system which concerns on Example 1. FIG. 6 is an example of a flowchart of learning storage processing according to the first embodiment. A first example of a neural network as a model. A second example of a neural network as a model. A third example of a neural network as a model. The example of the flowchart of a series of processes including an estimation process and a control process. The example of the flowchart of control processing. The example of the flowchart of a relief process. The example of the block diagram of the whole system containing the group management controller as an example of the elevator system which concerns on Example 2. FIG. 10 is an example of a flowchart of learning storage processing according to the second embodiment. The example of the block diagram of the whole system containing the core controller which is an example of the elevator system which concerns on Example 3, and an edge controller. 10 is an example of a flowchart of a series of processes including an estimation process and a control process according to the third embodiment. FIG. 10 is an example of a configuration diagram of the entire system according to a fourth embodiment. The example of the flowchart of the process containing the right / wrong determination of an estimation result, and the relearning based on the determination result. FIG. 10 is an example of a configuration diagram of the entire system according to Embodiment 5. 10 is an example of a neural network according to a fifth embodiment. 10 is an example of a flowchart of car door opening / closing control according to the sixth embodiment.

In the following description, the “interface unit” may be one or more interfaces. The one or more interfaces may be one or more similar interface devices or two or more different interface devices.

Further, in the following description, the “memory unit” is one or more memories, and may typically be a main storage device. At least one memory in the memory unit may be a volatile memory or a non-volatile memory.

In the following description, the “PDEV unit” is one or more PDEVs, and may typically be an auxiliary storage device. “PDEV” means a physical storage device (Physical storage device) and is typically a non-volatile storage device such as an HDD (Hard disk drive) or an SSD (Solid state drive).

In the following description, the “storage unit” is at least one of the memory unit and the PDEV unit (typically at least the memory unit).

In the following description, the “processor unit” is one or more processors. The at least one processor is typically a microprocessor such as a CPU (Central Processing Unit), but may be another type of processor such as a GPU (Graphics Processing Unit). At least one processor may be a single core or a multi-core. The at least one processor may be a broad processor such as a hardware circuit (for example, an FPGA (Field-Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit)) that performs part or all of the processing.

In the following description, the function may be described by the expression “kkk unit” (excluding the interface unit, the storage unit, and the processor unit). However, one or more computer programs are executed by the processor unit. It may be realized by one or more hardware circuits (for example, FPGA or ASIC). When the function is realized by the program being executed by the processor unit, since the predetermined processing is appropriately performed using the storage unit and / or the interface unit, the function is at least a part of the processor unit. May be. The processing described with the function as the subject may be processing performed by the processor unit or a device having the processor unit. The program may be installed from a program source. The program source may be, for example, a program distribution computer or a computer-readable recording medium (for example, a non-transitory recording medium). The description of each function is an example, and a plurality of functions may be combined into one function, or one function may be divided into a plurality of functions.

In the following description, when a description is made without distinguishing the same kind of elements, a common part of the reference signs may be used, and when the same kind of elements are distinguished and explained, a reference sign may be used. For example, when the elevator is described without being particularly distinguished, “elevator 1” is described. When the individual elevator is distinguished and described, it may be described as “elevator 1A” or “elevator 1B”. is there.

FIG. 1 is an example of a configuration diagram of the entire system including a group management controller as an example of an elevator system according to the first embodiment.

The group management controller 2 that controls the elevator group 200 (plural elevators 1) is an example of an elevator system. The “elevator system” may be a system that includes at least one of the learning unit 40, the estimation unit 50, and the control unit 60, for example, a system that includes only the learning unit 40 and does not include any of the plurality of elevators 1. possible. Since at least one of the learning unit 40 and the estimation unit 50 can be viewed as a module that supports the management of the elevator group 200, for example, a system that includes only the learning unit 40 has another name such as a group management support system. May be called. In the present embodiment, the group management controller 2 includes all of the learning unit 40, the estimation unit 50, and the control unit 60.

The learning unit 40 uses the plurality of input values belonging to the target time point, and is operation data used for controlling the elevator group 200, and the operation data 34 used for control of a future time point from the target time point. A model 44 for estimating a plurality of output values based on the above is learned. The model being learned is denoted as “model 44T”. The plurality of input values belonging to the target time point include a plurality of time input values respectively indicating a plurality of time elements belonging to the target time point. The plurality of output values include values relating to a plurality of departure / arrival combinations. Each departure / arrival combination is a combination of a departure floor and an arrival floor. By inputting a plurality of input values belonging to a certain time point to the learned model 44F (the model 44 learned by the learning unit 40), the departure floor and the arrival floor at a certain time in the future from the certain time point. A combination is estimated. Based on the estimation, operation for the future time point can be controlled. If there is no lowest floor or highest floor on the departure floor or arrival floor, there is no need to move the car 14 (hereinafter simply referred to as “car”) to the lowest floor or highest floor. Thereby, useless operation can be reduced. For example, for a certain elevator 1, it is estimated that a car 14 without passengers is descending, and that the lowest floor does not become the departure floor within a certain time, and that a certain middle floor becomes the departure floor. In this case, the direction of travel can be reversed without going to the lowest floor. As a result, useless strokes of round trips can be eliminated.

The estimation unit 50 inputs a plurality of input values belonging to the estimated time point to the learned model 44F, thereby providing a plurality of base data of operation data 34E used for control for a predetermined time point from the estimated time point. Estimate the output value. As a result, the operation at the future time point can be controlled based on the combination of the departure floor and the arrival floor at the future time point for the future time point from the estimated time point. Thereby, useless operation can be reduced.

The control unit 60 controls the elevator group 200 using estimated operation data 34E, which is operation data based on a plurality of estimated output values for a time point after a predetermined time has elapsed from the estimated time point. Thereby, for a time point after a predetermined time has elapsed from the estimated time point, based on the combination of the departure floor and the arrival floor after the time point, the operation for the time point after the time point can be controlled. Thereby, useless operation can be reduced.

The elevator group 200 has a plurality of elevators 1 (for example, 1A to 1C). Hereinafter, as one elevator 1, an elevator 1A is taken as an example. In the drawing, α is added to the end of the reference numerals of the constituent elements of the elevator 1α (α = A, B, or C).

The elevator 1A has an elevator controller 10A, and the elevator controller 10A controls the hoisting machine 12A of the elevator 1A to control the movement of the car 14A. The elevator controller 10 </ b> A is connected to a network 20 such as a communication path, and can exchange information with other elevator controllers 10. It should be noted that the network 20 includes information on car call buttons installed on each floor (for example, a button for designating that a button for specifying going up or a button for specifying going down is pressed). Information indicating that the vehicle has been pressed), information on the traveling direction lamp of the elevator 1, information on the position lamp of the car 14, and information on the car door (for example, the car door is in a closed state or opened) State, open state, information indicating which state is in the closed state, etc.), and all other information related to the elevator flows. Such information is called “elevator information”. The car 14A is provided with a sensor such as a weight sensor 16A.

The group management controller 2 is connected to the network 20. The elevator group 200 and the group management controller 2 may be provided in the same building, or the group management controller 2 may exist in a place away from the building. For example, the group management controller 2 may be a software-defined controller realized on a computer system (one or more computers) such as a cloud system.

The group management controller 2 may have physical calculation resources such as an interface unit, a storage unit, and a processor unit connected thereto (or may be realized based on these physical calculation resources). A network 20 is connected to the interface unit. The group management controller 2 includes an input region 32, a data storage unit 30, a learning unit 40, a learning result region 42, an estimation unit 50, an estimation region 55, and a control unit 60. Each of the input area 32, the learning result area 42, and the estimation area 55 is a physical or logical storage area based on the storage unit. The data storage unit 30, the learning unit 40, the estimation unit 50, and the control unit 60 are realized by a processor unit that executes one or more computer programs.

In the input area 32, a number of values that can be input to at least one of the learning unit 40 and the estimation unit 50 are stored by the data storage unit 30. Each value stored in the input area 32 is, for example, a value based on any of the following:
・ Elevator information (for example, time-series elevator information) acquired from the present to the past target period,
・ Time information indicating the time,
・ Information on the structure of the building where the elevator group 200 is installed and each floor, and
-Disturbance factors of the operation of the elevator group 200 (for example, the repair work period of a building (an example of a building) in which the elevator group 200 is installed, the traffic volume around the building, the delay or suspension of the railway in the vicinity of the building, Or information about the destination floor registration system (DFRS: Destination Floor Reservation System)
It is. It is possible to obtain actual operation data 34R described later from the elevator information. There is at least one of the following (01) to (04) as a value stored in the input area 32, in other words, a value that can be an input value of the learning unit 40 or the estimation unit 50. Hereinafter, one time point is taken as an example (“target time point” in the description of (01) to (04) below).
(01) A plurality of time input values respectively indicating a plurality of time elements belonging to the target time point. For example, when the target time point is expressed in year / month / day / hour / minute / second, there are six "multiple time elements": "year", "month", "day", "hour", "minute", and "second". The “multiple time input values” are six values corresponding to the six time elements, respectively. Note that other time elements may be employed as the time elements employed instead of or in addition to the six time elements. For example, “day of the week” may be adopted as the time element. In this case, the time input value is a value indicating the day of the week to which the target time point belongs. Further, “festival holiday” may be adopted as the time element, and in this case, the time input value is a value indicating whether or not the target time point belongs to the festival holiday. Further, “week of the current month” may be adopted as the time element, and in this case, the time input value is a value indicating the week of the current month.
(02) Any one of a plurality of actual values serving as the basis of the actual operation data 34R. “Actual operation data” refers to operation data based on a plurality of actual values (for example, a plurality of specific values among various values that can be acquired from elevator information) specified based on sensors or the like at a target time. is there.
(03) A floor feature value that is a value indicating the feature of the floor.
(04) A disturbance factor value which is a value indicating a disturbance factor.

Regarding (02), the plurality of actual values that are the basis of the actual operation data 34R correspond to a plurality of actual values respectively corresponding to the plurality of output values (estimated values) from the learned model 44F. As described above, the plurality of output values (estimated values) include values relating to a plurality of departure / arrival combinations (combinations of departure floor and arrival floor) (details will be described later).

In the learning result area 42, a learned model 44F (for example, an image (for example, a file) of the learned model 44F) as a learning result by the learning unit 40 is stored. The learned model 44F is developed from the learning result area 42, and estimation using the developed learned model 44F is executed by the estimation unit 50.

The estimated operation data 34E based on a plurality of output values output from the estimation unit 50 (learned model 44F) is stored in the estimation area 55.

The learning unit 40, the estimation unit 50, and the control unit 60 are as described above.

Hereinafter, a specific example of FIG. 1 will be described.

Information related to the departure floor, arrival floor, and number of passengers is acquired from the elevator information, and values based on the acquired information are stored in the input area 32 by the data storage unit 30 as actual operation data 34R. Of course, the input area 32 may store values based on all elevator information. The information relating to the departure floor includes the position of the car 14 and the car up / down movement start event. Information relating to the arrival floor includes the position of the car 14 and a car door opening event. As information relating to the number of passengers, there is a measurement value of the weight sensor 16 installed in the car 14. The number of passengers can be calculated based on the measured value of the weight sensor 16 and the standard weight per person. The information related to the number of passengers output from the elevator controller 10 may be the number of passengers as the calculation result or the measured value of the weight sensor 16. In the former case, the number of passengers is calculated by the elevator controller 10, and in the latter case, the number of passengers is calculated by the data storage unit 30 (or the control unit 60). In an elevator system that can measure the number of passengers using an image sensor or the like, the number of passengers may be output to the network 20. The present invention does not limit the number of passengers to a mounting form such as an image sensor or a weight sensor 16.

The learning unit 40 acquires and learns the departure floor, the arrival floor, and the number of passengers from the input area 32, and outputs a learned model 44F as a learning result. The learning unit 40 performs, for example, deep learning as the learning process. In the case of deep learning, the learning result-completed model 44F is a learned deep network (neural network). The estimation unit 50 estimates the operation data 34 including at least the departure floor, the arrival floor, and the number of passengers, which will occur in the future, for example, after the time related to one stroke of the elevator, from the learned model 44F. As a result, estimated operation data 34E is obtained. The estimated operation data 34E is stored in the estimated area 55.

An example of the operation data 34 of the elevator 1 on the fifth floor is shown in FIG. The operation data 34 includes a matrix of arrival floor (vertical) and departure floor (horizontal). The numerical value in the cell in the matrix indicates the number of passengers in the departure / arrival combination (combination of departure floor and arrival floor) corresponding to the cell. Specifically, for example, according to the leftmost column, the number of passengers moving from the first floor to the third floor is 2, and the number of passengers moving from the first floor to the fifth floor is 4. According to the rightmost column, the number of passengers moving from the fifth floor to the first floor is two, and the number of passengers moving from the fifth floor to the second floor is seven. The same applies to the other floors. A departure / arrival combination corresponding to a blank cell means that the number of passengers is zero. A departure / arrival combination having one or more passengers can be referred to as an “effective departure / arrival combination”. The departure floor and arrival floor belonging to the effective departure / arrival combination can be referred to as “effective departure floor” and “effective arrival floor”. Hereinafter, the effective departure floor and the effective arrival floor related to the estimated operation data 34 may be referred to as “effective estimated departure floor” and “effective estimated arrival floor”, respectively. It should be noted that not all elevators 1 are 5 floors. For example, in an N-floor building (including an underground floor if there is an underground floor), for a certain elevator 1, the lowest floor is the lowest floor of the building and the highest floor is the p-th floor in the middle Well, for another elevator 1, the lowest floor may be the qth floor (between the lowest floor and the pth floor of the building), and the highest floor may be the highest floor of the building.

The control unit 60 is built in a single building or adjacent to the estimated operation data 34E and the current state of the elevator 1 (for example, the car position, the car traveling direction, and the number of passengers for each elevator 1). The allocation of the car 14 to the stop floor is determined for each elevator 1 in the elevator group 200 installed in a plurality of buildings. The controller 60 gives an instruction to the elevator controller 10 via the network 20 based on the determined information (car assignment), and efficiently controls the plurality of elevators 1. There is a group management controller 2 for controlling such a plurality of elevators 1.

There are various types of elevator controller 10, and elevator information is not standardized in the industry. A process of converting various elevator information into operation data 34 is required, and the process can be executed in the data storage unit 30. Thereby, various elevators 1 can be controlled by the group management controller 2.

FIG. 2 is a flowchart illustrating the learning storage process according to the first embodiment. “Learning storage processing” is processing from elevator information acquisition to learning result storage.

The data storage unit 30 acquires elevator information from the elevator controller 10 (S10).

The data storage unit 30 acquires at least information on the departure floor, arrival floor, number of passengers (number of passengers for each departure / arrival combination) and time information of the elevator information, and the actual operation data 34R based on the acquired information. A plurality of values based on it are stored in the input area 32 (S20).

The learning unit 40 performs a learning process. That is, the learning unit 40 acquires a plurality of input values based on the actual operation data 34R from the input area 32 (S30), and performs learning for inputting the acquired plurality of input values to the model 44T and obtaining a plurality of output values. (S40).

The learning unit 40 stores the learned model 44F as a learning result in the learning result area 42 (S50).

Note that the learning storage process may be executed sequentially or after the actual operation data 34R is accumulated to some extent. For example, it is possible to learn at once based on the actual operation data 34R accumulated for one year.

Further, S10 to S40 may be repeated as indicated by a broken line until the learning result is output as the learned model 44F.

Further, the learned model 44F may be updated by performing S10 to S50 again as indicated by the alternate long and short dash line (relearning). That is, the learning unit 40 may update the learned model 44F, and the estimation unit 50 may perform estimation according to the updated learned model 44F. The plurality of input values input in the update of the learned model 44F include the following so that the learned model 44F is updated (typically, a plurality of weighting coefficients respectively corresponding to a plurality of nodes in the learned model 44F) Is updated (adjusted). Thereby, the improvement of the accuracy of the learned model 44F is expected, and as a result, the accuracy of the estimation result by the estimation unit 50 is expected to be improved.
A plurality of time input values respectively indicating a plurality of time elements belonging to an elapsed time that is a time when a predetermined time has elapsed from an arbitrary estimated time.
A plurality of actual values that are the basis of the actual operation data 34R at the time of the passage and correspond to a plurality of output values (estimated values), respectively.

FIG. 3 is a diagram showing a first example of the neural network as the model 44.

The model 44 is composed of at least one neural network. In the present embodiment, for simplicity of explanation, the model 44 is composed of one neural network, but may be composed of a multi-stage neural network.

The model 44 includes an input layer 301, an output layer 303, and an intermediate layer 302. The input layer 301 includes a plurality of input nodes 70 to which a plurality of input values are respectively input. The output layer 303 includes a plurality of output nodes 400 that respectively output a plurality of output values based on the operation data 34 used for controlling the plurality of elevators 1. The intermediate layer 302 includes a plurality of intermediate nodes 352 between the input layer 301 and the output layer 303. The learned model 44F performs an operation based on a learned weighting factor in at least one neural network for a plurality of input values belonging to the target time point input to the input layer 301, and from the output layer 303, from the target time point. The computer is caused to function so as to output a plurality of output values based on the operation data 34 used for controlling the plurality of elevators 1 for a predetermined time in the future.

The plurality of input values respectively input to the plurality of input nodes 70 include a plurality of time input values respectively indicating a plurality of time elements belonging to the target time point. Accordingly, the plurality of input nodes 70 include a plurality of time input nodes 70-1 that respectively input a plurality of time input values. The time information may be the absolute value of year / month / day / hour / minute / second, but since the person does not act accurately according to the time, the accuracy of estimation cannot be expected even if the absolute value is input. In consideration of social activities, it is desirable to consider, for example, what day and what day of the week of any month including holiday information as time information. Therefore, in this embodiment, as the time input node 70-1, the month value input node 70-1a indicating the month, the week value input node 70-1b indicating the week, and the day value input node 70-1c indicating the day of the week are shown. , A time value input node 70-1d indicating time, a minute value input node 70-1e indicating minutes, and a festive holiday value input node 70-1f indicating whether the target time point belongs to a festive holiday. Is done.

Furthermore, as the input node 70, one or more disturbance factors that may disturb regular social activities such as building renovation work, traffic congestion on the roads around the building, and delays or suspension of trains near the building, etc. There may be a disturbance factor value input node 70-2. By inputting such one or more disturbance factor values, it is possible to improve the accuracy of estimation. The disturbance factor value may be, for example, a numerical value representing the degree of influence on human behavior. The disturbance factor value is not necessarily adopted as the input value.

Each of the plurality of output values output from the plurality of output nodes 400 is an output value that is the basis of the operation data 34 that is used for the control of a future time point for a predetermined time from the target time point. As described above, the plurality of output values include values relating to a plurality of departure / arrival combinations (combinations of departure floor and arrival floor). The value for the plurality of departure arrival combinations includes one or more departure arrival output values for each of the plurality of departure arrival combinations. For each of the plurality of departure / arrival combinations, one or more departure / arrival output values are values related to the object amount in the departure / arrival combination. Although the value related to the object amount in the departure / arrival combination may be whether or not there is a passenger (an example of an object) for the departure / arrival combination, in this embodiment, the value indicates the number of passengers corresponding to the departure / arrival combination. Since the number of passengers is estimated for each departure / arrival combination, the operation of the elevator group 200 can be expected to be more efficient. In FIG. 3, the notation “NOP (xF−yF)” in the vicinity of the output node 400 indicates the number of passengers from the x floor (departure floor) to the y floor (arrival floor) (NOP (the （NumberｙOf People). ).

The operation data 34 may vary depending on the operation mode of the elevator 1. In the case of two types of operation modes, the up and down buttons installed on each floor and the floor buttons installed in the car 14, the arrival floor is determined after the passenger presses the up and down buttons. Arrives and the passenger gets into the car 14 and presses the floor button. There is a lapse of time after the passenger presses the up / down button until the destination floor is determined. That is, there is a difference between the actual generation of passengers and the button press recognized by the elevator controller 10. Therefore, when inputting the number of passengers corresponding to the departure / arrival combination as the teacher data of the learning unit 40, the up and down buttons are pressed based on the operation data 34R obtained from the elevator information output from the elevator controller 10. It is desirable to set the passenger generation time before the time.

Also, a destination floor registration system (DFRS: Destination Floor Reservation System) is known, but the presence or absence of DFRS can be adopted as a disturbance factor value.

Moreover, since the model 44 is the model 44 for the elevator group 200, it is common to the plurality of elevators 1. For this reason, for each departure / arrival combination, the number of passengers as an output value may be a total value for a plurality of elevators 1.

As the structure of the model 44, there is a multistage regression structure as an example of a network structure, but the present invention is not limited to this structure.

By the way, if the input value is only the time input value, it is considered that there are many cases where the estimation is deviated when the person is not always acting on time. It is considered that the actual operation of the elevator reflects the change in human behavior with time.

Therefore, in this embodiment, each of a plurality of actual values corresponding to a plurality of output values (estimated values), which is the basis of the actual operation data 34R, is adopted as an input value.

FIG. 4 is a diagram showing a second example of the neural network as the model 44.

In order to improve the estimation accuracy, the input node 70 is further based on the actual operation data 34R (for example, actual operation data 34R at the target time (for example, the estimated time)), and each of the plurality of output values (estimated values). Each input node 70-3 of the corresponding plurality of actual values is employed. In order to realize such a network structure, it is possible to adopt Stacked Denoising Autoencoders, LSTM (Long Short-Term Memory) considering time factors, etc., but it is not limited to this structure.

FIG. 5 is a diagram showing a third example of the neural network as the model 44.

In order to improve the estimation accuracy, a floor feature value input node 70-4, which is a value indicating the feature of the floor, is further adopted as the input node 70 for each floor. The feature of the floor is considered to be a determinant of human behavior. As the floor feature value input node 70-4, a type value input node 70-4a indicating the type of the store or company on the floor, an attribute value input node indicating the floor attribute such as the customer draw rate of the store or the number of employees of the company 70-4b and a facility value input node 70-4c indicating the presence / absence of a special facility such as a smoking room, cafeteria, or event hall can be employed. The type value and facility value for each floor may be constant or variable depending on the form of use of the building. In a permanent case, the type value and the facility value may be handled as input parameters such as a weighting coefficient used in the intermediate layer 302. If it is variable, it is considered appropriate to handle it as an input value of the input layer 301 as in the example of FIG.

There is CNN (Convolutional Neural Network) for realizing such a network structure, but the present invention is not limited to this structure.

FIG. 6 is an example of a flowchart of a series of processes including an estimation process and a control process.

The estimation unit 50 acquires an image of the learned model 44F and develops the learned model 44F (S110).

The estimation unit 50 performs an estimation process. That is, the estimation unit 50 estimates a plurality of output values based on the operation data 34 after a predetermined time has elapsed from the estimated time point (for example, the current time point) based on the learned model 44F (S120). The estimation unit 50 presents estimated operation data 34E based on a plurality of estimated output values (that is, operation data used for control after a predetermined time has elapsed from the estimated time point) to the control unit 60 (stored in the estimation region 55). (S130).

The control unit 60 obtains estimated operation data 34E whose current time (control time) is the time after a predetermined time from the estimated time from the estimation area 55, and controls the allocation of cars based on the estimated operation data 34E. Processing is performed (S140).

The control unit 60 transmits an operation instruction to each elevator controller 10 based on the result of the control process (S150).

The start of execution of the estimation process may be any of the timing when the car door corresponding to the departure is closed, the timing when the car corresponding to the arrival is stopped, and the like. Further, the estimation process may be executed at a constant cycle.

FIG. 7 is an example of a flowchart of the control process (S140 in FIG. 6).

The control unit 60 grasps the current state at the time of control (S210). Specifically, for example, the control unit 60 acquires the following for each elevator 1. As a result, the control unit 60 calculates (calculates) the total tolerance of the elevator group 200 (the total tolerance of all elevators 1).
・ The current position of the car 14.
・ The arrival floor of basket 14.
-The direction of travel of the car 14.
The margin of the car 14 (difference between the allowable number of passengers in the car 14 and the number of passengers at the time of control).

Based on the current status ascertained, the control unit 60, for each elevator 1, in the middle of the current position and the arrival floor, is one of the effective estimated departure floors (the departure floor indicated by the estimated operation data 34E, and the estimated It is determined whether there is a departure floor whose number of passengers indicated by the operation data 34E is 1 or more (S220).

When the determination result of S220 is true (S220: Yes), the control unit 60 extracts the elevator 1 that sandwiches one of the effective estimated departure floors with the grasped current position and the grasped arrival floor (S230). The control unit 60 assigns the effective estimated departure floor to the extracted elevator 1 (S240). In S240, the control unit 60 subtracts the number of passengers (the number of passengers corresponding to the effective estimated departure floor) indicated by the estimated operation data 34E from the margin of the elevator 1, and updates the estimated operation data 34E (effective estimated departure floor). To zero passengers). By assigning an effective estimated departure floor to the elevator 1, even if the effective estimated departure floor is neither the current arrival floor of the elevator 1 nor the floor where the car has been called, the car 4 of the elevator 1 is effective. The door will only be opened on the estimated departure floor.

After S240 or when the determination result of S220 is false (S220: No), the control unit 60 determines whether the margin total is equal to or less than the estimated number of passengers (the total number of passengers indicated by the estimated operation data 34E). Is determined (S250). That is, the control unit 60 determines whether or not the elevator group 200 at the control time point (current grasping time point) has a vacant space that can mount all the passengers indicated by the estimated total number of passengers with the control time point as the future time point. . Therefore, the number of passengers corresponding to the departure / arrival combination (the number of passengers shown in the operation data 34 illustrated in FIG. 1) indicates the number of passengers newly generated from a certain time to a certain time. When S240 is performed, regarding S250, “tolerance total” is the updated margin total, and “estimated number of passengers” is the total number of passengers indicated by the updated estimated operation data 34E. is there.

If the determination result in S250 is true (S250: Yes), the control process ends. As a result, for each elevator 1, the relationship between the car 14 and the stop floor is maintained as it is.

When the determination result of S250 is false (S250: No), that is, when the total margin is greater than the current estimated number of passengers, the control unit 60 subtracts the estimated total number of passengers from the total tolerance. The remaining number is obtained (S260). The control unit 60 generates an elevator list in which the elevators 1 are arranged in ascending order (distance from the shortest distance) between the current position and the effective estimated departure floor (S270). The control unit 60 selects an elevator whose traveling direction (direction from the current position to the arrival floor) is directed to the effective estimated departure floor from the elevator list generated in S270 (S280). The control unit 60 calculates the margin of the elevator 1 from the number of passengers of the elevator 1 selected in S280 (S290). The control unit 60 assigns an effective estimated departure floor to the elevator 1 (S300). The control unit 60 obtains a new remaining number by subtracting the margin of the elevator 1 from the current remaining number (S310).

The control unit 60 determines whether the remaining number is 0 or less (S320). If the determination result in S320 is true (S320: Yes), the control process ends.

If the remaining number is greater than 0 (S320: No), the control unit 60 generates an elevator list in which the elevators 1 are arranged in descending order of tolerance (in descending order of tolerance) (S330). The elevator list generated in S330 may or may not include the elevator 1 already assigned in S300, but is not included in the present embodiment. The control unit 60 performs the following (S340).
(S340-1) From the elevator list generated in S330, the elevator 1 having the largest margin is selected from the elevators 1 that have not been selected.
(S340-2) An effective estimated departure floor is assigned to the elevator 1 selected in S340-1.
(S340-3) The margin of the elevator 1 selected in S340-1 is subtracted from the current remaining number of persons to obtain a new remaining number of persons.
(S340-4) If the new number of remaining persons calculated in S340-3 is 0 or less, S340 ends. As a result, the control process ends.
(S340-5) If the new remaining number calculated in S340-3 is greater than 0, it is determined whether all elevators 1 have been selected from the elevator list generated in S330.
(S340-6) If the determination result in S340-5 is true, S340 ends. As a result, the control process ends.
(S340-7) If the determination result in S340-5 is false, the process returns to S340-1.

According to the above control process, even if the elevator 1 is assigned to one of the effective estimated departure floors in S240, if the determination result in S250 is false, the assignment is canceled (discarded), and after S260, Another effective estimated departure floor may be assigned to the elevator 1.

Further, according to the above control processing, when the determination result of S250 is false, the difference between the estimated number of passengers and the total tolerance is calculated as the remaining number of people, and the difference (remaining number of people) becomes 0 or less. Thus, the elevator 1 is assigned. The remaining number is subtracted from the margin of the assigned elevator 1. That is, the determination result of S250 is false that the total number of passengers estimated to be newly generated is estimated to exceed the current margin total of the elevator group 200, in other words, all estimated passengers. Is estimated to be unable to be placed on the elevator group 200. In this case, the elevator 1 is assigned to the estimated departure floor from the viewpoint of subtracting the current margin from the excess number of passengers (remaining number of passengers).

Further, according to the above control processing, when the determination result of S250 is false, in S260 to S300, the elevator 1 whose distance to the effective estimated departure floor is short and the traveling direction is the direction toward the estimated departure floor is preferential. Assigned to the effective estimated departure floor. As a result, if the new remaining number is not less than or equal to 0, it is necessary to reverse the direction of travel for the unassigned elevator 1, but since such elevator 1 is assigned in descending order of margin, the elevator group 200 As an efficient operation, it is possible to expect an operation that reduces the number of elevators 1 that require reversal of the traveling direction.

Based on the above description of the control process, an outline of the control process can be expressed as follows, for example.

That is, the control part 60 is the present condition including the following about each of the some elevator 1,
・ The margin of the car 14 of the elevator 1,
-The direction of travel of the elevator 1,
・ The current position of the car 14 of the elevator 1 and
・ The arrival floor of the car 14 of the elevator 1
To figure out. The traveling direction of the elevator 1 can be grasped from the current position of the car 14 of the elevator 1 and the arrival floor. However, since the arrival floor may not be specified for the elevator 1, it is effective to grasp the traveling direction. It is believed that there is. Based on the estimated operation data 34E and the current grasped status of each of the plurality of elevators 1, the control unit 60 adds the current margin, the traveling direction, and the current position to each of the effective estimated departure floors. The elevator 1 determined based on at least one is assigned.

FIG. 8 is an example of a flowchart of a relief process, which is a process when a car call is made for a floor other than the effective floor indicated by the estimated operation data 34E (generic name of the effective estimated departure floor and the effective estimated arrival floor). A car call for a floor different from the effective floor is referred to as a “singular event”. When the singular event is detected by the control unit 60, the relief process of FIG. 8 is started.

The control unit 60 performs control processing according to the flowchart of FIG. 7 (S400). In addition, S400 may be skipped and S410 and subsequent steps may be performed based on the result of recent control processing.

The control unit 60 calculates the tolerance of each of the plurality of elevators 1 (S410).

The control unit 60 determines whether or not there is at least one elevator 1 in which the floor of the specific event exists in the assigned route (route connecting the current position and the assigned floor in the control process) (S420).

When the determination result of S420 is true (S420: Yes), the control unit 60 determines whether or not there is an elevator 1 having a floor of a specific event in the traveling direction in the elevator 1 of S420: Yes (S430). ). When the determination result of S430 is true (S430: Yes), the control unit 60 assigns a stop to the floor of the singular event to the elevator 1 having the largest tolerance among the elevators 1 of S430: Yes (S440). That is, for the elevator 1, the floor assigned by the control process is canceled and the floor of the unique event is assigned.

When the determination result of S420 is false (S420: No), or when the determination result of S430 is false (S430: No), the control unit 60 increases the distance from the arrival floor to the singular event in ascending order (in order of short distance). ) To generate an elevator list of the elevators 1 lined up (S450). The control unit 60 selects the elevator 1 whose traveling direction matches the direction to the floor of the singular event from the elevator list generated in S450 (S460). The control unit 60 changes the arrival floor (stop floor) of the selected elevator 1 to the floor of the unique event (S470).

By such a relief process, even if a specific event for which a car is called for a different floor from the estimation occurs, the specific event can be flexibly dealt with. As a result, even if a peculiar event occurs, it is possible to shorten the waiting time of the passengers, and thus it can be expected to improve the operation efficiency.

Moreover, according to such a relief process, even if there is no elevator 1 where the floor of the singular event is on the route, or even if there is such an elevator 1, there is no floor of the singular event in the traveling direction, This can be dealt with by changing the stop floor of another elevator 1 where the direction in which the floor of the singular event is located is the traveling direction to the floor of the singular event.

In addition, an update process may be performed in which a plurality of values (for example, a plurality of values corresponding to a plurality of output values of the model 44) according to the result of the relief process are input to the learned model 44F as a plurality of input values. Further improvement in estimation accuracy can be expected.

Based on the above description of the relief process, an outline of the relief process can be expressed as follows, for example.

That is, when the car call is generated on any floor other than the effective floor (for example, the departure floor in which the object amount of the departure floor indicated by the estimated operation data 34E is greater than 0), For each of the above, the margin of the elevator 1 is specified, and at least one of the following for the floor where the car call occurred:
・ Wealth,
・ The distance between the floor and the current location, and
・ Direction of travel
The elevator 1 determined based on is assigned.

Example 2 will be described. At that time, differences from the first embodiment will be mainly described, and description of common points with the first embodiment will be omitted or simplified.

An environment called a cloud system 100 is an environment that provides a storage device that can handle large amounts of data and a high-performance computing environment. The cloud system 100 is an example of a first device such as a computer system. Instead of the cloud system 100, another computer system such as a data center may be employed.

FIG. 9 is a configuration diagram of the entire system including a group management controller as an example of an elevator system according to the second embodiment.

The elevator information including the operation data 34 of the plurality of elevators 1 includes all the information related to the control and the state of the elevator 1, and thus has a large capacity and is continuously accumulated every day, and thus has a huge capacity. That is, the elevator information can correspond to so-called big data.

As an implementation example of the estimation unit 50, a learned model 44F according to deep learning is adopted. In order to realize deep learning, an environment in which big data such as elevator information including operation data 34 can be flexibly used and a high-load learning unit 40 can be executed is desirable.

In general, a group management controller is provided for the purpose of controlling a plurality of elevators. However, according to one comparative example, the group management controller is presumed to be incorporated into a device, and therefore, handling of enormous data and It is not designed to perform high-load calculations. According to the above description, a storage device and processing device capable of handling a large amount of data and a high-performance arithmetic processing device capable of executing deep learning learning and the like are necessary. It is considered difficult to install a large-capacity storage device and a high-speed processing device in an environment such as a hoistway like an embedded system such as the elevator controller 10 in terms of operation and location.

Therefore, the group management controller 2 is realized on the cloud system 100 as in the present embodiment. The cloud system 100 may be a global cloud system (or, for example, a data center) when a plurality of elevators in a wide area remote area are targets, and if a plurality of elevators in a small area or a building are targets, for example. An on-premises cloud system (or a local server, for example) may be used.

The group management controller 2 in the cloud system 100 may be a software-defined controller or a physical controller. The constituent elements of the group management controller 2 according to the second embodiment are the same as those of the first embodiment.

The elevator group 200 including a plurality of elevators 1 includes a wired network and a wide area network (an example of a communication network) 22 that is a wireless closed circuit network provided by a carrier, a network 20 and a communication device (for example, a network switch such as a router). ) It is connected via 28. The network 24 in the cloud system 100 is connected to the wide area network 22 via a communication device (for example, a network switch) 26.

FIG. 10 is an example of a flowchart of learning storage processing according to the second embodiment.

The elevator controller 10 transmits elevator information to the cloud system 100 via the wide area network 22 (S600). In order to communicate the information, a form using the IoT protocol and a form using a unique protocol are conceivable, but other forms may be adopted.

The group management controller 2 in the cloud system 100 receives the elevator information transmitted by the elevator controller 10 (S610).

The data storage unit 30 stores operation data 34R as information combining the departure floor, the arrival floor, and the number of passengers in the input area 32 from the acquired elevator information (S620).

The learning unit 40 performs a learning process. That is, the learning unit 40 acquires a plurality of input values based on the operation data 34R and the like from the input area 32 (S630), and executes learning based on the plurality of input values (S640).

The learning unit 40 stores the learned model 44F as a learning result in the learning result area 42 (S650).

Example 3 will be described. At that time, differences from the second embodiment will be mainly described, and description of common points with the second embodiment will be omitted or simplified.

In the second embodiment, it is considered that a communication delay may occur due to passing through the wide area network 22 and the plurality of communication devices 26 and 28, which may affect the control.

Therefore, in the third embodiment, as illustrated in FIG. 11, the group management controller 2 is roughly divided into a core controller 300C that is a part that performs learning processing and an edge controller 300E that is a part that performs estimation processing and control processing. The edge controller 300E is mounted on the elevator group 200. Thereby, prevention of the above-mentioned communication delay can be expected. The core controller 300C (or cloud system 100) is an example of the first device, and the edge controller 300E is an example of the second device.

Example 3 will be described in detail below.

FIG. 11 is an example of a configuration diagram of the entire system including a core controller 300C and an edge controller 300E which are an example of an elevator system according to the third embodiment.

Both the core controller 300C and the edge controller 300E may be physical controllers or software-defined controllers.

The core controller 300C is realized on the cloud system 100. The core controller 300C includes a data storage unit 30, an input area 32, a learning unit 40, and a learning result area 42C.

In many cases, the estimation unit 50 has a smaller processing load than the learning unit 40, but the estimation processing load is heavy for the elevator controller 10 that controls the hoisting machine 12. For this reason, in this embodiment, the estimation unit 50 is not mounted on the elevator controller 10, and an edge controller 300 </ b> E common to the plurality of elevators 1 is separately provided in the elevator group 200. The edge controller 300E has a learning result area 42E in which the learned model 44F stored in the learning result area 42C of the core controller 300C is copied (downloaded) from the core controller 300C and stored. Furthermore, the edge controller 300E includes an estimation unit 50, an estimation region 55, and a control unit 60.

Although not shown in FIG. 11, the edge controller 300E is connected to the cloud system 100 (core controller 300C) via the communication device 28, the wide area network 22 and the communication device 26 illustrated in FIG. The learned model 44F developed and executed in the edge controller 300E is the learned model 44F copied from the cloud system 100 via the wide area network 22 as described above.

With such a system configuration, it is expected that the operation efficiency can be improved by retrofitting the installed elevator group 200 with the edge controller 300E and the edge controller 300E in cooperation with the core controller 300C on the cloud system 100.

Although the communication device 28 and the edge controller 300E are separated, the role of the edge controller 300E may be integrated into the communication device 28. That is, the communication device 28 may function as an edge controller.

FIG. 12 is an example of a flowchart of a series of processes including an estimation process and a control process according to the third embodiment.

The core controller 300C in the cloud system 100 transmits the learned model 44F to the edge controller 300E (S700).

The edge controller 300E receives the learned model 44F from the cloud system 100 (S701), and stores the received learned model 44F in the learning result area 42C (S702).

In the edge controller 300E, the estimation unit 50 performs an estimation process based on the learned model 44F acquired and developed from the learning result area 42C (S703).

In the edge controller 300, the control unit 60 performs control processing based on the estimated operation data 34E obtained by the estimation processing in S703 (S704). The control unit 60 gives an instruction to each elevator controller 10 based on the result of the control process (S705).

Each elevator controller 10 executes control according to an instruction from the edge controller 300 (S706).

Example 4 will be described. At that time, differences from the third embodiment will be mainly described, and description of common points with the third embodiment will be omitted or simplified.

FIG. 13 is an example of a configuration diagram of the entire system according to the fourth embodiment.

It is inefficient to learn the model 44 from scratch for the new elevator group 200B (an example of another elevator group) each time a building is built and the elevator group 200 is newly increased.

Therefore, in this embodiment, the learned model 44F for the existing elevator group 200A is copied for the new elevator group 200B, and the copied learned model (hereinafter referred to as copy model) 44F is the new model 44F. It is applied as a learned model used for estimating a plurality of values that are the basis of the operation data 34 of the elevator group 200B. The copy model 44F is relearned by the learning unit 40 based on a value different from the estimated value for the new elevator group 200B among the actual values for the new elevator group 200B. “Relearning” is to update the copy model 44F by learning a new case with reference to the copy model (learned model) 44F obtained as a result of progress of learning to some extent. The re-learning is expected to improve the estimation accuracy. Such re-learning may be performed on the learned model 44F of the existing elevator group 200A. In addition, the relearning may be performed not only on the part having the error in the estimated operation data 34E but on the entire part, but the calculation amount can be expected to be reduced only on the part having the error. The learned model 44F applied to the new elevator group 200B is applied to the elevator group 200 having the same or similar characteristics as the new elevator group 200B, such as structure and environment (for example, building structure), for example. It is desirable that the model 44 be learned. Such a learned model 44 is, for example, based on management information including information representing characteristics such as the structure and environment of the elevator group 200 for each elevator group 200 by the control unit 60 (or the core controller 300C). The elevator group 200 having the same or similar characteristics as the characteristics of the new elevator group 200B may be specified, and it may be determined to apply the learned model of the specified elevator group 200 to the new elevator group 200B.

For example, according to the system configuration of FIG. In FIG. 13, the edge controller 300E1 is the edge controller 300E in the existing elevator group 200A, and the edge controller 300E2 is the edge controller 300E in the new elevator group 200B. In the present embodiment, the learning unit 40E is provided in the edge controller 300E, and the learning unit 40 in the core controller 300C is referred to as a learning unit 40C.

When a new building is constructed and a new elevator group 200B is laid, a new edge controller 300E2 is also installed. The learned model 44F is copied from the learning result area 42C to the learning result area 42E of the edge controller 300E2 by the core controller 300C on the cloud system 100. Similarly to FIG. 12, in the edge controller 300E2, the estimation unit 50 estimates the operation based on the copy model 44F, and uses the estimated operation data 34E (the estimated operation data 34E stored in the estimation area 55) as the estimation result. Based on this, the control unit 60 performs control processing. The control unit 60 issues an instruction to each elevator controller 10 in the elevator group 200B based on the result of the control process. The learning unit 40E has a relearning function. The learning unit 40E compares the estimated operation data 34E stored in the estimation area 55 with the actual operation data 34R at the time of control, so that the actual operation data 34R differs from the estimated operation data 34E (in other words, An erroneous portion) of the estimated operation data 34E can be specified. The learning unit 40E updates the copy model 44F using a value corresponding to a portion different from the estimated operation data 34E in the actual operation data 34R. This update is relearning. The relearning is, for example, learning in which a value corresponding to a portion different from the estimated operation data 34E in the actual operation data 34R is a correct output value. Since only the operation data whose estimation has been deviated is used for relearning, the relearning processing load is low, and processing can be performed by the edge controller 300E2. In the re-learning, the method of distilling or refining the copy model 44F is equivalent to the learning result, the network structure is simplified, and the learning processing load can be further reduced by such a method. is there.

With this configuration, by storing various learned models 44F in the cloud system 100, even when a new elevator group 200 is installed, the stored learned models 44F can be used. Can start operation efficiently. The learned model 44F depends on the elevator group characteristics such as the floor plan of the building, the number and arrangement of the elevators 1, and the behavior of the user. For example, a department store with a large number of floors and a suburb with a small number of floors. The driving situation of the elevator group 200 varies depending on a shopping mall, a regional supermarket with a large number of floors, a complex office building based on a concept, and the like. On the other hand, store arrangements and the like are similar in the same department store series, shopping mall series, supermarket series, and as a result, the operation status of the elevator group 200 may be similar. In a complex office building with the same concept, the office area and the commercial area are often similar in layouts such as restrooms and smoking rooms, rest areas, and dining rooms. If the companies that move in are similar, time plans such as shifts in working hours and meal times may be similar. Correspondingly, the arrangement of the plurality of elevators 1 is similar, and the actual driving situation of the elevator group 200 may be similar. When the learned model 44F in a certain store or a complex office building in a certain place is used in another store, the accuracy of operation estimation may be expected. In a facility such as a department store event hall, the number of visitors may vary greatly depending on the planning of the event, and the value reflecting the plan and the number of visitors is entered in the facility value input node 70-4c for each floor. By performing relearning, it is possible to expect generation of a learned model 44F that enables operation control according to the plan held at the event hall.

Even if there are similar planes, the estimation accuracy may be low, but this is individuality for each case, and such a difference is caused by the learning unit 40E (learning unit 40 having a relearning function) of the edge controller 300E. By repeating relearning according to the target, the accuracy can be increased.

FIG. 14 is an example of a flowchart of processing including correct / incorrect determination of an estimation result and relearning based on the determination result.

The learned model 44F is copied from the cloud system 100 to the edge controller 300E2 (S800 and S801). In the edge controller 300E2, the estimation unit 50 estimates operation data based on the copy model 44F (S802). The estimation unit 50 stores the estimated operation data 34E in the estimation area 55 (S803). The control unit 60 performs control processing based on the estimated operation data 34E (S804), and issues an instruction to each elevator controller 10 based on the result of the control processing (S805). Receiving the instruction, the elevator controller 10 executes control according to the instruction (S806).

Each elevator controller 10 transmits information representing the actual operation status (information including a value based on the actual operation data 34R) to the edge controller 300E2 (S807).

The edge controller 300E2 receives the information transmitted from the elevator controller 10 (S808), and the learning unit 40E obtains the estimated operation data 34E stored in S803 and the actual operation data 34R based on the information from each elevator controller 10. By comparing, it is determined (correction determination) whether or not the actual operation data 34R has a different part from the estimated operation data 34E (S809). When the determination result in S809 is false (S809: No), that is, when the estimated and actual operation data are different, the learning unit 40E executes re-learning based on the actual operation data 34R (S810), As a result of the relearning, the copy model 44F is updated (S811).

By such processing, the initial cost of learning can be reduced by utilizing the learned model 44F. In addition, when it is detected that an erroneous estimation has occurred, re-learning can improve estimation accuracy.

Example 5 will be described. At that time, differences from the first to fourth embodiments will be mainly described, and description of common points with the first to fourth embodiments will be omitted or simplified.

FIG. 15 is an example of a configuration diagram of the entire system according to the fifth embodiment.

The number of passengers calculated based only on the weight sensor 16 installed in the car 14 is based on a fixed standard weight, so that when a passenger with a weight difference rides, the value is accurate. Not exclusively.

Therefore, in the present embodiment, the image sensor 17 is arranged in the car 14, and the image sensor 18 is installed on the floor of each floor. The edge controller 300E includes a recognition unit 54 and a recognition area 46. The recognition area 46 is a storage area that stores a recognition model. The recognition model is a model for object recognition (for example, a learned neural network according to deep learning). Based on the image data from the image sensors 17 and 18, the recognition unit 54 performs object recognition using the recognition model.

Specifically, the recognition unit 54 in the edge controller 300E acquires the image data captured by the image sensor 17 installed in the car 14 via the network 20. The recognition unit 54 can recognize the object with high accuracy by using the recognition model in the recognition region 46. Specifically, the transported object and the person can be distinguished (that is, in the present embodiment, “object” is a general term for a person and a transported object (thing other than a person)). For this reason, the number of passengers in the car 14 can be estimated more accurately. The recognition model may be a result of separately learning using a lot of image data. For example, the imaging data of the image sensors 17 and 18 may be sent to the cloud system 100 via the communication device 28, the recognition model is learned by the core controller 300C, and the recognition model may be applied to the edge controller 300E. . Alternatively, the recognition model may be a model learned by a completely different system.

Although the image sensor 18 is installed on each floor, the recognition unit 54 in the edge controller 300E acquires image data captured by the image sensor 18 on each floor via the network 20. Since the recognition unit 54 can recognize the object with high accuracy, it is possible to more accurately estimate the number of people waiting on the floor of each floor.

By utilizing the image sensors 17 and 18, depending on the learning method of image recognition, not only the number of persons and transported objects, but also the identification of persons and transported objects, the age of persons, the sex of persons, the volume of persons or objects, It is possible to estimate. In addition to or in addition to the image sensors 17 and 18, by setting a sensor such as a security gate capable of personal authentication using an ID card, it is possible to accurately grasp the number of passengers and the behavior of individuals.

FIG. 16 is a diagram illustrating an example of the neural network according to the fifth embodiment. In FIG. 16, the meaning of each notation is as follows. The reason why the following values can be adopted is that the following values can be recognized by the recognition unit 54 based on the image data from the image sensors 17 and 18.
“NOT (xF−yF)” means the number of items (NOT (the Number Of Things)) from the x floor (departure floor) to the y floor (arrival floor).
“A_NOP (xF−yF)” means the number of passengers for each age group from the xth floor (departure floor) to the yth floor (arrival floor). The range of each age group may be arbitrarily wide.
“G_NOP (xF−yF)” means the number of passengers for each gender from the x floor (departure floor) to the y floor (arrival floor). The sum of A_NOP (xF-yF) or the sum of G_NOP (xF-yF) corresponds to the number of passengers from the x floor (departure floor) to the y floor (arrival floor).
“Volume (xF−yF)” means the volume of all objects from the xth floor (departure floor) to the yth floor (arrival floor).
“NOT_C” means the number of items in the car 14.
“A_NOP_C” means the number of passengers for each age group in the car 14. The same age group may be employed for input and output.
“G_NOP_C” means the number of passengers for each gender in the car 14. The sum of A_NOP_C or the sum of G_NOP_C corresponds to the number of passengers in the car 14.
“Volume_C” means the volume of all objects in the car 14.
・ "NOT (vF)" means the number of items transported on the v-th floor.
“A_NOP (vF)” means the number of passengers (number of persons) for each age group on the floor of the v-th floor. Each age group may be the same for A_NOP (xF-yF) and A_NOP_C.
“G_NOP (vF)” means the number of passengers (number of persons) for each gender on the floor of the v-th floor. The sum of A_NOP (vF) or the sum of G_NOP (vF) corresponds to the number of passengers (number of persons) on the floor.
“Volume (vF)” means the volume of all objects on the floor of the v-th floor.

According to FIG. 16, for each of a plurality of departure / arrival combinations, among the number of passengers for each gender, the number of passengers for each age group, and the volume of all objects, There is at least one of Specifically, for example, it is as follows.

The output node 400 includes the following. For example, NOP (xF-yF) in the first embodiment is output as A_NOP (xF-yF) and G_NOP (xF-yF) in this embodiment.
NOT (xF-yF) output node 400T-xy (eg, NOT (1F-2F) output node 400T-12).
A_NOP (xF-yF) output node 400A-xy (eg, A_NOP (1F-2F) output node 400A-12).
G_NOP (xF-yF) output node 400G-xy (eg, G_NOP (1F-2F) output node 400G-12).
Output node 400V-xy of Volume (xF-yF) (for example, output node 400V-12 of Volume (1F-2F)).

On the other hand, there are the following as input nodes.
NOT_C input node 70-6.
• A_NOP_C input node 70-7.
G_NOP_C input node 70-8.
Volume_C input node 70-9.
NOT (vF) input node 70-10Tv (eg, NOT (5F) input node 70-10T5).
A_NOP (vF) input node 70-10Av (eg, A_NOP (5F) input node 70-10A5).
G_NOP (vF) input node 70-10Gv (eg, G_NOP (5F) input node 70-10G5).
Volume (vF) input node 70-10Vv (eg, Volume (5F) input node 70-10V5).

The input values input to the input nodes 70-6 to 70-10 are actual values based on the actual operation data 34R at the time of control. Input values input to the input nodes 70-6 to 70-10 correspond to output values (estimated values) output from the output nodes 400T, 400A, 400G, and 400V.

Since there are the input node 70 and the output node 400 as described above, for example, at least one of the following can be expected.
-According to the ratio of the number of passengers for each age group and the ratio of the number of passengers for each gender, (1) the hoisting machine 12 is controlled to adjust acceleration / deceleration, and (2) the upper limit of the number of passengers is adjusted. At least one of them can be performed by the control unit 60. Thereby, according to the ratio of the number of passengers for every age group, and the ratio of the number of passengers for every sex, adjustment of the riding comfort of the elevator 1 and adjustment of the congestion degree in the cage | basket | car 14 are attained.
-For example, in the case where the car 14 is a car in which a large-capacity cart (an example of a transported item) is mounted, the weight of the car tends to be light, and therefore it is easy to determine that the number of passengers is small. It is in. For this reason, it is considered that there is a high possibility that more floors are assigned to the car 14 as stop floors than the car without a carriage as described above. Since there is a carriage as described above, there is a relatively large margin for the number of passengers (in this case, for example, the difference between the internal volume of the car and the occupied volume (Volume_C)) is relatively small. Even if a floor is assigned as a stop floor, there is a high possibility that people will not be able to get on. From the stop of the car, a waste time that is the sum of the opening time of the door, the time when the passenger recognizes that he / she cannot get on the car, and the closing time of the door may be generated for each assigned floor. As a result, operation efficiency can be reduced. According to the present embodiment, Volume (xF−yF) is estimated. Therefore, if the car internal space is largely occupied, the control unit 60 determines that the car is full, and removes the car from the allocation target. be able to. For this reason, generation | occurrence | production of the above-mentioned dead time can be avoided and operation efficiency can be improved.

Example 6 will be described. At that time, the differences from the first to fifth embodiments will be mainly described, and the description of the common points with the first to fifth embodiments will be omitted or simplified.

FIG. 17 is an example of a flowchart of car door opening / closing control according to the sixth embodiment.

In Example 6, when it is estimated that a car call is generated after a predetermined time on the floor where the door of the car 14 is opened, the control unit 60 delays the closing operation of the door of the car 14. Specifically, for example, it is as follows. Take one floor as an example.

That is, the control unit 60 determines whether or not the car door opening standby time has timed out (that is, whether or not a certain period of time has elapsed with the car door open) (S500). Further, the control unit 60 determines whether or not the door close button in the car 14 has been pressed (S510).

If any determination result is true (S500: Yes or S510: Yes), the control unit 60 causes the estimation unit 50 to estimate the operation data after the car opening extension allowable time (S520).

The controller 60 determines whether or not a car call is generated based on the estimated operation data (S530).

When the car call is generated (S530: Yes), the control unit 60 instructs the elevator controller 10 to extend the car door opening time by a predetermined time (S540).

When the car call is not generated (S530: No), the control unit 60 instructs the elevator controller 10 to immediately close the car door (S550).

This process can cope with a car call that will occur in the near future, so that the waiting time of the person who comes later can be eliminated. As a result, the overall operation efficiency is improved. In addition, when extending the door opening time, the control unit 60 informs the elevator controller 10 to make an announcement such as lamp lighting, voice or screen display to the car 14 so that the passenger can recognize the extension. It is desirable to give instructions.

Based on the description of the present embodiment, the outline of the car door opening / closing control can be expressed as follows, for example.

For at least one elevator 1 of the plurality of elevators 1, operation data 34 </ b> E used for the control of the estimation unit 50 at a second time point a predetermined time after the first time point when the door of the car 14 of the elevator 1 is closed. Estimate a plurality of output values based on When the operation data 34E based on the output value estimated for the second time point indicates that a car call is generated at the second time point on the floor of the car 14, the control unit 60 opens the door. Extend the time to stay on.

Although several embodiments of the present invention have been described above, these are merely examples for explaining the present invention, and the scope of the present invention is not limited to these embodiments. The present invention can be implemented in various other forms. For example, any two or more of the embodiments can be combined. For example, in the third embodiment, the group management controller 2 as in the first embodiment may be adopted instead of the cloud system 100 as in the second embodiment, and in the fourth embodiment, the edge controller as in the third embodiment is adopted. Instead of 300, a group management controller 2 as in the first embodiment may be employed.

1: elevator, 2: group management controller, 10: elevator controller, 40: learning unit, 50: estimation unit, 60: control unit

Claims (15)

1. An elevator system that manages an elevator group that is a plurality of elevators,
   An input area that is a storage area for storing input data including a plurality of input values for each of one or more time points;
   A plurality of input values belonging to a target time point among the one or more time points are acquired from the input area, and the operation data used for controlling the plurality of elevators using the acquired plurality of input values, A learning unit that learns a model that estimates a plurality of output values based on operation data used for control for a predetermined time in the future from the target time point,
   The plurality of input values belonging to the target time point include a plurality of time input values respectively indicating a plurality of time elements belonging to the target time point,
   The plurality of output values include values relating to a plurality of departure / arrival combinations;
   Each departure arrival combination is a combination of departure and arrival floors,
   An elevator system characterized by that.
2. In the elevator system according to claim 1,
   The value for the plurality of departure arrival combinations includes one or more departure arrival output values for each of the plurality of departure arrival combinations;
   For each of the plurality of departure arrival combinations, the one or more departure arrival output values are values related to the amount of objects in the departure arrival combination.
   An elevator system characterized by that.
3. In the elevator system according to claim 1,
   By inputting a plurality of input values belonging to the estimated time to a learned model that is a model learned by the learning unit, it becomes a basis of operation data used for control for the predetermined time in the future from the estimated time An estimation unit for estimating a plurality of output values;
   An elevator system characterized by that.
4. In the elevator system according to claim 3,
   A controller that controls the plurality of elevators using estimated operation data that is operation data based on a plurality of output values estimated for a time after the predetermined time has elapsed from the estimated time;
   An elevator system characterized by that.
5. In the elevator system according to claim 4,
   A control device having the input area, the learning unit, the estimation unit, and the control unit;
   An elevator system characterized by that.
6. In the elevator system according to claim 4,
   A device connected to a first device having the input area and the learning unit, the second device having the estimation unit and the control unit;
   The elevator system, wherein the learned model is transmitted from the first device to the second device.
7. In the elevator system according to claim 3,
   The learning unit updates the learned model,
   The estimation unit performs estimation according to the updated learned model,
   The plurality of input values input in the update of the learned model is
   A plurality of time input values respectively indicating a plurality of time elements belonging to the elapsed time point that is a time point when the predetermined time has elapsed from an arbitrary estimated time point;
   It is a basis of actual operation data at the time of the passage, and includes a plurality of actual values respectively corresponding to the plurality of output values.
   An elevator system characterized by that.
8. In the elevator system according to claim 1,
   In addition to a plurality of time input values respectively indicating a plurality of time elements belonging to the target time point, the plurality of input values belonging to the target time point are:
   One or more disturbance factor values that are one or more values respectively indicating one or more disturbance factors, and
   For each floor, a floor feature value that is a value indicating the feature of the floor,
   Including at least one of
   An elevator system characterized by that.
9. In the elevator system according to claim 2,
   For each of the plurality of departure / arrival combinations, there is at least one of the number of passengers for each gender, the number of passengers for each age group, and the volume of all objects as a value related to the object amount in the departure / arrival combination. ,
   An elevator system characterized by that.
10. In the elevator system according to claim 1,
    The learned model by the learning unit is copied for another elevator group that is an elevator group installed in a building different from the building in which the elevator group is installed,
    The copied learned model is applied as a learned model that is used to estimate multiple values that are the basis for the operation data of the other elevator group,
    The copied learned model is relearned based on a value different from the estimated value for the other elevator group among the actual values for the other elevator group,
    An elevator system characterized by that.
11. In the elevator system according to claim 4,
    About at least one elevator of the plurality of elevators,
    The estimation unit estimates a plurality of output values based on operation data used for control for a second time point in the future for a predetermined time from the first time point when the elevator car door is closed,
    When the operation data based on the output value estimated for the second time point indicates that a car call is generated at the second time point on the floor of the car, the control unit Extend the time to keep it open,
    An elevator system characterized by that.
12. In the elevator system according to claim 4,
    In the control using the estimated operation data, the control unit,
    For each of the plurality of elevators, grasp the current situation including the following,
    The current margin of the elevator car,
    The current position of the elevator car, and
    Arrival floor of the elevator car,
    Based on the estimated operation data and the grasped current status of each of the plurality of elevators, each of the departure floors estimated to generate an object among the departure floors indicated by the estimated operation data Assigning an elevator determined based on at least one of tolerance, direction of travel and current position;
    An elevator system characterized by that.
13. The elevator system according to claim 12,
    The control unit, when a car call occurs for any floor other than the departure floor where the object amount of the departure floor indicated by the estimated operation data exceeds 0,
    For each of the plurality of elevators, identify the margin of the elevator,
    Assign an elevator to the floor determined based on at least one of the following:
    Margin,
    The distance between the floor and the current location, and
    Direction of travel,
    An elevator system characterized by that.
14. An elevator system that manages an elevator group that is a plurality of elevators,
    An estimated area that is a storage area for storing estimated operation data;
    A controller that controls the plurality of elevators using estimated operation data in the estimation area;
    The estimated operation data is operation data used for controlling the plurality of elevators, and is operation data based on a plurality of output values estimated for a predetermined time in the future from an estimated time.
    The plurality of estimated output values are a plurality of output values output from a learned model to which a plurality of input values belonging to the estimation time point are input,
    The plurality of input values belonging to the estimated time point include a plurality of time input values respectively indicating a plurality of time elements belonging to the estimated time point,
    The plurality of output values include values relating to a plurality of departure / arrival combinations;
    Each departure arrival combination is a combination of departure and arrival floors,
    An elevator system characterized by that.
15. A trained model composed of at least one neural network,
    The at least one neural network comprises:
    An input layer including a plurality of input nodes to which a plurality of input values are respectively input;
    An output layer including a plurality of output nodes that respectively output a plurality of output values based on operation data used to control an elevator group that is a plurality of elevators;
    An intermediate layer including a plurality of intermediate nodes between the input layer and the output layer;
    Each of the plurality of input values is an input value belonging to a target time point,
    Each of the plurality of output values is an output value that is a basis of operation data used for control about a predetermined time in the future from the target time point,
    The plurality of input values include a plurality of time input values respectively indicating a plurality of time elements belonging to the target time point,
    The plurality of output values include values relating to a plurality of departure / arrival combinations;
    Each departure arrival combination is a combination of departure and arrival floors,
    An operation based on a weighted coefficient learned in the at least one neural network is performed on the plurality of input values belonging to the target time point input to the input layer, and from the output layer for a predetermined time from the target time point To output the plurality of output values that are the basis of the operation data used to control the plurality of elevators for a future time point,
    A trained model to make the computer work.
    

Images