JP6915762B1 - Allocation support program, allocation support device, allocation learning program, allocation learning device, and arithmetic unit readable storage medium - Google Patents

Abstract

Obtained from an information acquisition unit that acquires the first input information or the second input information, and an inference unit (20, 20b) that makes an inference using a trained model by inputting the first input information or the second input information. , The power consumption of the automatic control system allocated by the allocation information is reduced. The allocation information output unit (12), which outputs the allocation information of the controlled device, is provided so that the power usage of the automatic control system is reduced. Provided are an allocation support program, an allocation support device (1, 1b, 1c), an allocation learning program, an allocation learning device (2, 2b), and an arithmetic terminal readable storage medium that support the review of the allocation of the control target device in the automatic control system. ..

Description

The present invention relates to an allocation support program, an allocation support device, an allocation learning program, an allocation learning device, and an arithmetic unit readable storage medium for supporting a review of the allocation of controlled devices in an automatic control system.

In recent years, in order to improve the operating rate of factories and plants, various devices (sensors, motors, actuators, safety devices, etc.) that compose a production line by a control device using a programmable logic controller (hereinafter, simply referred to as "PLC") The introduction of an automatic control system that automatically controls the devices to be controlled such as) is being promoted.

A PLC is a control device constructed using a desired unit. The unit for constructing the PLC includes, for example, a power supply unit that is a power supply source, a CPU unit that controls a PLC, an input / output that transmits a signal to a plurality of various devices, or receives a signal from a plurality of various devices. There are output units, communication units for connecting to communication networks, etc.

The input / output unit has a plurality of input / output terminals in one input / output unit, and a control target device is assigned to each input / output terminal. The allocation refers to associating the identification symbol of the input / output terminal with the device to be controlled. Then, the PLC controls the PLC by designating the input / output terminals of the input / output unit and receiving and transmitting signals based on the control program stored in the CPU unit, and controls the PLC of various devices connected to the PLC. The operation is automatically controlled.

However, when an automatic control system using PLC is introduced, the power consumption of the production line tends to increase, and the power consumption of the entire factory increases. Therefore, an automatic control system with a power-saving design is required. ..

A field network system, which is an example of a conventional automatic control system, is an I / O input / output that transmits signals to a plurality of field devices (controlled devices) or receives signals from a plurality of field devices (controlled devices). Sends a power operation signal to the unit (input / output unit). The I / O input / output unit controls the on / off of the supply power supply in the I / O input / output unit based on the power supply operation signal, and reduces the power consumption of the field network system (for example, Patent Document). 1).

Specifically, in a plurality of I / O input / output units included in the field network system, a plurality of field devices are assigned to one I / O input / output unit. Further, inside the I / O input / output unit, a switch for switching on / off of power supply to the internal circuit CPU of the I / O input / output unit and all of a plurality of field devices (referred to as "whole switch" for convenience of explanation). It is equipped with a plurality of switches (referred to as "individual switches" for convenience of explanation) for switching the power supply on / off for each of the plurality of field devices. The field network system turns off the entire switch when it is not necessary to supply power to all the plurality of field devices assigned to one I / O input / output unit. In the field network system, when any one of the plurality of field devices needs to be supplied with power, the whole switch is turned on and the individual switches corresponding to the field devices that do not need to be supplied with power are turned off. The amount used was reduced.

Japanese Unexamined Patent Publication No. 2012-1988841

However, among the plurality of controlled devices, there are devices such as sensors and safety devices that require constant power supply during the operation of the production line. In the field network system disclosed in Patent Document 1 described above, when any one of the plurality of field devices assigned to one I / O input / output unit is operated, the above-mentioned overall switch is used. It cannot be turned off. That is, since the allocation state of the field equipment in the field network system is not taken into consideration, there is no period during which the above-mentioned overall switch can be turned off depending on the allocation state of the field equipment to the I / O input / output unit, and the field network There is a problem that the amount of power used in the system can be reduced only in a limited manner.

The present invention has been made to solve the above-mentioned problems, and is an allocation support program that supports a review of the allocation of controlled devices in an automatic control system so that the power consumption of the automatic control system is reduced. An object of the present invention is to provide an allocation support device, an allocation learning program, an allocation learning device, and an arithmetic unit readable storage medium.

The allocation support program according to the present invention is an allocation support program that supports a review of allocation in an automatic control system in which a device to be controlled is assigned to a plurality of input / output terminals of each of a plurality of input / output units, and is an input / output unit. Allocation information corresponding to the input / output terminals and the control target device, time zone information indicating the time zone in which the input / output unit and the control target device are operating, the unit power consumption information of the input / output unit, and the control target device. Automatic control calculated based on the power consumption of the input / output unit and the power consumption of the controlled device when the first input information including the device power consumption information of The trained model is input by inputting the system power usage information including the power usage of the system and the information acquisition unit for acquiring the second input information including the allocation information, the first input information, or the second input information. The arithmetic unit functions as an allocation information output unit that outputs the allocation information of the control target device that reduces the power consumption of the automatic control system allocated by the allocation information acquired from the inference unit that performs the inference used. ..

According to the present invention, it is possible to provide an index for reviewing the allocation of the control target device in the automatic control system so that the power consumption of the automatic control system allocated by the allocation information is reduced.

It is a figure which shows the outline of the automatic control system. It is a figure which shows an example of the hardware composition of the allocation support device which concerns on Embodiment 1. FIG. It is a figure which shows typically an example of the functional block of the allocation support device which concerns on Embodiment 1. FIG. It is a figure which shows an example of the allocation information which concerns on Embodiment 1. FIG. It is a figure which shows an example of the time zone information which concerns on Embodiment 1. FIG. It is a figure which shows an example of the unit power consumption information which concerns on Embodiment 1. FIG. It is a figure which shows an example of the device power consumption information which concerns on Embodiment 1. FIG. It is a figure which shows an example of the review allocation information which concerns on Embodiment 1. FIG. It is a figure which shows the daily operation state of the automatic control system based on the review allocation information which concerns on Embodiment 1. FIG. It is a figure which shows typically an example of the functional block of the learning apparatus part which concerns on Embodiment 1. FIG. It is a flowchart which shows the operation which generates the trained model which concerns on Embodiment 1. It is a flowchart which shows the operation of the allocation support program and allocation support device 1 which concerns on Embodiment 1. FIG. It is a figure which shows typically an example of the functional block of the allocation support device 1 which concerns on Embodiment 2. FIG. It is a figure which shows an example of the number-of-times information which concerns on Embodiment 2. It is a figure which shows an example of the review allocation information which concerns on Embodiment 2. It is a figure which shows the daily operation state of the automatic control system based on the review allocation information which concerns on Embodiment 2. FIG. It is a figure which shows typically an example of the functional block of the learning apparatus part which concerns on Embodiment 2. FIG. It is a flowchart which shows the operation which generates the trained model which concerns on Embodiment 2. It is a flowchart which shows the operation of the allocation support program and the allocation support device which concerns on Embodiment 2. FIG. It is a figure which shows typically an example of the functional block of the allocation support device which concerns on Embodiment 3. FIG. It is a figure which shows an example of the functional block of the allocation simulation part which concerns on Embodiment 3. It is a flowchart which shows the operation of the allocation simulation part which concerns on Embodiment 3.

Embodiment 1.
The allocation support program, the allocation support device, the allocation learning program, and the allocation learning device according to the first embodiment of the present invention will be described with reference to the drawings. In the first embodiment, the allocation learning program is included in the allocation support program, and the allocation support device will be described as having the function of the allocation learning device. FIG. 1 is a diagram showing an outline of the automatic control system AS. FIG. 2 is a diagram showing an example of the hardware configuration of the allocation support device 1. FIG. 3 is a diagram schematically showing an example of a functional block of the allocation support device 1. FIG. 4 is a diagram schematically showing an example of a functional block included in the learning unit 21.

As shown in FIG. 1, the automatic control system AS includes a plurality of control target devices 150, and a PLC 200 that executes a control program to control the control target device 150. The automatic control system AS is a system that automatically controls the control target device 150 by communicating with the arithmetic unit 300 in which the control program is stored via the PLC 200.

In FIG. 1, as an example of the automatic control system AS, a PLC 200 having three input / output units 202 having an input / output device unit having three input / output terminals and a control target device 150 are assigned to each input / output terminal. It shows an automatic control system AS equipped with nine controlled devices 150. In the present embodiment, an example in which the number of input / output terminals is three is shown, but it may be two or more.

The PLC 200 includes a CPU unit 201 that reads and executes a control program stored in the arithmetic unit 300, and an input / output unit 202 that is connected to the control target device 150. The CPU unit 201 and the input / output unit 202 are connected via a baseboard that electrically connects the units, or the units provided in each unit are electrically connected to each other without the baseboard. It may be connected by a connector to be connected.

The PLC 200 and the arithmetic unit 300 are communicably connected via a dedicated line or a network. When connected using a dedicated line, for example, it may be connected via a USB (Universal Serial Bus) cable. When connected via a network, for example, it may be connected via an open network such as the Internet, or may be connected via a closed network such as LAN (Local Area Network).

The CPU unit 201 has a CPU unit that reads and executes a control program, an internal memory unit that temporarily stores the control program and data, and an internal power supply unit that supplies electric power to each unit.

The input / output unit 202 is supplied with power from an input / output device unit having a plurality of input / output terminals to which the controlled device 150 is assigned, another unit having a power supply (for example, a CPU unit, a power supply unit, etc.), or It is provided with a power supply unit (for example, a power supply circuit, etc.) that receives power supply from an external power source. Here, the input / output unit 202 inputs and outputs signals when the controlled device 150 is automatically controlled by using a PLC such as an I / O input / output unit, an analog unit, a counter unit, and a network unit. A unit that can do at least one of the above.

The controlled device 150 is a device assigned to the input / output terminals of the input / output unit 202, and is, for example, a sensor, a motor, an actuator, a safety device, and the like.

In addition to storing the above-mentioned control program, the arithmetic unit 300 also functions as an allocation support device 1. FIG. 2 shows an example of the hardware configuration of the arithmetic unit 300 that functions as the allocation support device 1. The allocation support device 1 is a calculation device 300 that installs and executes an allocation support program, a calculation unit 101 that executes the allocation support program, a storage unit 102 that stores the allocation support program, and reads / writes data and instructions, and a keyboard. It is provided with an input unit 103 such as a mouse, a touch panel, etc., a display unit 104 that displays an execution result of an allocation support program, and a communication interface (communication I / F) 105 that communicates with a PLC or the like.

The storage unit 102 includes a non-volatile storage unit in which the installed allocation support program is stored, and a volatile storage unit that serves as a work memory when the allocation support program is executed. As the allocation support device 1, an arithmetic unit 300 such as a notebook computer, a desktop computer, a tablet computer, a smartphone, etc., in which an allocation support program is installed in the storage unit 102, can be used. The allocation support program is stored in a non-temporary arithmetic terminal readable storage medium, and functions when installed in the arithmetic unit 300. As a non-temporary arithmetic terminal readable storage medium, for example, a CD-ROM (Compact Disk Read only memory), a DVD-ROM (Digital Versaille Read only memory), a USB (Universal Serial Bus) flash drive, or the like can be used. can.

FIG. 3 is a diagram schematically showing a functional block of the allocation support device 1 realized by the arithmetic unit 300 including the arithmetic unit 101 and the storage unit 102 shown in FIG. In the first embodiment, the allocation support device 1 has a function as an allocation learning device 2 for performing machine learning, in addition to a function of supporting the review of the allocation of the control target device 150. In the first embodiment, the functional blocks of the allocation learning device 2 will be described together with the allocation support device 1.

The allocation support device 1 is a device that supports the review of the allocation of the control target device 150 in the automatic control system AS that is already in operation. The allocation support device 1 inputs the information acquisition unit 11 that acquires various information in the automatic control system AS and the information acquired by the information acquisition unit 11, and infers the allocation of the control target device 150 in the automatic control system AS. It has a unit 20 and an allocation information output unit 12 that acquires and outputs information output by the arithmetic processing of the inference unit 20. Further, the allocation support device 1 has a learning unit 21 for machine learning the allocation of the control target device 150 in the automatic control system AS. The learning unit 21 is a functional block of the allocation learning device 2.

The information acquisition unit 11 includes allocation information 400 in which the input / output terminals in the plurality of input / output units 202 included in the automatic control system AS correspond to the control target device 150, and the control target device 150 assigned to the plurality of input / output units 202. The first input information 800 including the time zone information 500 indicating the time zone in which the system is operating, the unit power consumption information 600 of the input / output unit 202, and the device power consumption information 700 of the control target device 150 is acquired, or is obtained. Second input information 1000 including system power usage information 900 when the automatic control system AS is operated based on the power consumption of the plurality of input / output units 202 and the power consumption of the controlled device 150, and the above-mentioned allocation information 400. To get. The above-mentioned information acquired by the information acquisition unit 11 will be described later.

The inference unit 20 includes a preprocessing unit 22 and an arithmetic processing unit 26. The preprocessing unit 22 performs preprocessing as a part of the arithmetic processing of the inference unit 20 when data preprocessing is required in order for the information acquired by the information acquisition unit 11 to be arithmetically processed by the arithmetic processing unit 26. .. The pre-processing refers to arithmetic processing including calculation of information acquired by the information acquisition unit 11 and change of format. The arithmetic treatment unit 26 reads out the learned model generated by performing machine learning by the learning unit 21 described later, and allocates the controlled target device 150 in the automatic control system AS based on the information acquired by the information acquisition unit 11. Infer a review. The arithmetic processing in the inference unit 20 will be described later.

The inference unit 20 is input with either the first input information 800 or the second input information 1000, executes each arithmetic process described later, and executes the allocation information 400 included in the first input information 800 or the second input information 1000. The power consumption of the automatic control system AS allocated by the allocation information 400 is reduced, and the power is supplied to at least one input / output unit 202 of the plurality of input / output units 202 of the automatic control system AS allocated by the allocation information 400. The allocation state of the controlled device 150 in which the time to stop the device 150 increases is inferred. The automatic control system AS allocated by the allocation of the allocation information 400 included in the first input information 800 or the second input information 1000 represents the automatic control system AS before the allocation is reviewed.

The allocation information output unit 12 acquires and outputs the allocation information inferred by the inference unit 20 as the direct allocation information 1100. The output review allocation information 1100 is displayed on the display unit 104.

The information acquired by the above-mentioned information acquisition unit 11 and the information output by the allocation information output unit 12 will be described below.

FIG. 4 is a diagram showing an example of the allocation information 400 according to the first embodiment. As shown in FIG. 4, the allocation information 400 is information indicating what kind of control target device 150 is assigned to the input / output terminals of the plurality of input / output units 202 included in the automatic control system AS in the automatic control system AS. Is. In FIG. 4, the three input / output units 202 are shown as input / output units 202A, 202B, and 202C, respectively, and the input / output terminals included in the input / output unit 202A are shown as input / output terminals X0, X1, and X2, respectively. The input / output terminals provided are shown as input / output terminals Y0, Y1, Y2, and the input / output terminals included in the input / output unit 202C are shown as input / output terminals Z0, Z1, Z2. The allocation information 400 is associated with information that identifies the plurality of input / output units 202A, 202B, 202C, the plurality of input / output terminals Xn, Yn, Zn, and the control target device 150. Here, the information that identifies the control target device 150 is device-specific information such as a device name and a model number. In FIG. 4, the device names M1, M2 ... Mn are used as the information for specifying the control target device 150. The allocation information 400 is information indicating the allocation state in the automatic control system AS before the allocation is reviewed.

FIG. 5 is a diagram showing an example of the time zone information 500. Note that FIG. 5 shows the operating hours of each input / output unit 202 and the controlled device 150 of the automatic control system AS shown by the allocation information 400 shown in FIG. As shown in FIG. 5, in the automatic control system AS, the time zone information 500 indicates the time zone in which the control target device 150 assigned to the input / output terminals of the PLC 200 and the input / output unit 202 included in the automatic control system AS. Information indicating whether it is operating. Here, the time zone is defined as 24 hours starting from 0:00 on a day. In FIG. 5, as in FIG. 4, three input / output units are shown as input / output units 202A, 202B, and 202C, and the controlled device 150 is shown by device names M1, M2 ... Mn.

In FIG. 5, the device operation start 1 indicates the time when the controlled device 150 specified by the device name Mn first starts operation in the 24-hour time zone starting from 0:00, and the device operation end 1 Indicates the time when the controlled device 150 specified by the device name Mn that was operated at the device operation start 1 ended its operation. That is, the period from the time of the device operation start 1 to the time of the device operation end 1 indicates that the control target device 150 specified by the device name Mn is in operation. Further, the device operation start 2 indicates the time when the controlled device 150 specified by the device name Mn first starts operating after the device operation end 1, and the device operation end 2 operates at the device operation start 2. Indicates the time when the controlled device 150 specified by the device name Mn has finished operating. That is, it indicates that the period from the time of the device operation start 2 to the time of the device operation end 2 is the time zone in which the controlled device 150 specified by the device name Mn is operating, and the time of the device operation end 1 It can be seen that the period from to the time of the device operation start 2 is a time zone in which the controlled target device 150 specified by the device name Mn is not operating. The same applies to the device operation start 3 and the device operation end 3 and thereafter.

Further, the unit operation start 1 shown in FIG. 5 indicates the time when any of the controlled target devices 150 assigned to the input / output terminals of the input / output unit 202 first started operation in the above-mentioned time zone. The unit operation end 1 indicates the time when all of the controlled devices 150 assigned to the input / output terminals of the input / output unit 202 first end their operation in the above-mentioned time zone. That is, to explain the input / output unit 202A as an example, the control specified by the device names M1, M2, and M3 assigned to the input / output terminals of the input / output unit 202A is from the unit operation start 1 to the unit operation end 1. It indicates the time zone in which any of the target devices 150 is operating and the input / output unit 202A is operating. Further, the unit operation start 2 indicates the time when any of the controlled devices 150 first assigned to the input / output terminals of the input / output unit 202 starts operation after the unit operation end 1, and the unit operation end 2 indicates the time when the operation starts. , All of the control target devices 150 assigned to the input / output terminals of the input / output unit 202 indicate the time when the operation is first terminated after the time indicated in the unit operation start 2 in the above time zone. That is, to explain the input / output unit 202A as an example, from the time of the unit operation start 2 to the time of the unit operation end 2, the device names M1, M2, and M3 assigned to the input / output terminals of the input / output unit 202A are used. It indicates the time zone in which any of the specified controlled devices 150 is operating and the input / output unit 202A is operating. Then, from the time of the unit operation end 1 to the time of the unit operation start 2, the device names M1, M2, and M3 assigned to the input / output terminals of the input / output unit 202A are specified in the above time zone. It can be seen that all of the controlled devices 150 are not operating. The same applies to the unit operation start 3 and the unit operation end 3 and thereafter.

According to FIG. 5, it can be seen that the input / output units 202A, 202B, and 202C in the automatic control system AS indicated by the allocation information 400 are in the operation designated state during the day (24 hours).

FIG. 6 is a diagram showing an example of the unit power consumption information 600. FIG. 6 is information showing the power consumption of each input / output unit 202 of the automatic control system AS shown by the allocation information 400 shown in FIG. Specifically, as shown in FIG. 6, the unit power consumption information 600 indicates the power consumption of the input / output units 202A, 202B, and 202C included in the automatic control system AS as kW (kilowatt) in the automatic control system AS, respectively. Information. For the unit power consumption of each input / output unit 202, the specification value (for example, catalog value) of the input / output unit 202 can be used.

FIG. 7 is a diagram showing an example of the device power consumption information 700. FIG. 7 is information showing the power consumption of the controlled target device 150 of the automatic control system AS shown by the allocation information 400 shown in FIG. Specifically, as shown in FIG. 7, the device power consumption information 700 consumes the controlled device 150 specified by the device name Mn assigned to the input / output terminals of the input / output unit 202 in the automatic control system AS. The information is shown with the power consumption as kW (kilowatt). For the device power consumption of each control target device 150, a specification value (for example, a catalog value) of the control target device can be used.

The first input information 800 is information including the above-mentioned allocation information 400, time zone information 500, unit power consumption information 600, and device power consumption information 700. In the first embodiment, the first input information 800 is composed of four pieces of information: allocation information 400, time zone information 500, unit power consumption information 600, and device power consumption information 700.

The system power usage information 900 is information indicating the power consumption (kWh: kilowatt hour) when the automatic control system AS assigned by the allocation information 400 shown in FIG. 4 is operated for a predetermined time. The electric power consumption in the first embodiment indicates the electric power consumption when the automatic control system AS is operated according to the control program for one day. In addition, one day is 24 hours from 0:00. The power consumption is calculated based on the power consumption of the input / output unit (kW: kW) and the power consumption of the controlled device (kW: kW).

The power consumption when the automatic control system AS is operated for a predetermined time is the power consumption of the input / output unit (kWh: kilowatt hour) and the power usage of the controlled device using the formulas shown in the following equations 1 and 2. The amount (kWh: kilowatt hour) is calculated, and the amount is calculated based on the power consumption of the input / output unit and the power consumption of the controlled device.

That is, in the calculation of the power consumption when the automatic control system AS is operated for a predetermined time, first, the power consumption when the input / output unit provided in the automatic control system AS is operated for a predetermined time according to the control program (kWh: The kilowatt hour) and the power consumption (kWh: kilowatt hour) when the controlled device 150 assigned to the input / output terminals of the input / output unit 202 operates for a predetermined time according to the control program are calculated. These power consumption calculations are performed for all of the plurality of input / output units 202 provided in the automatic control system AS and the controlled target devices 150 assigned to the input / output terminals of the input / output units 202, and each power consumption is calculated. By adding, the power consumption of the automatic control system AS is calculated. The calculated power consumption of the automatic control system AS is the system power usage information 900. Based on the above-mentioned allocation information 400, time zone information 500, unit power consumption information 600, and device power consumption information 700, the power consumption of the automatic control system AS shown in the first embodiment included in the system power usage information 900. Is 2290 kWh.

For the power consumption of the input / output unit used in the formula shown in Equation 1, the specification value of the input / output unit (for example, the catalog value) can be used, and the operating time of the input / output unit is input within one day. The time during which power is being supplied to the output unit can be used. Further, the power consumption of the controlled target device used in the mathematical formula shown in Equation 2 can use the specification value (for example, catalog value) of the controlled target device, and the operating time of the controlled target device is controlled within one day. The time when power is supplied to the target device can be used.

The second input information 1000 is information including the above-mentioned allocation information 400 and system power usage information 900. In the first embodiment, the second input information 1000 is composed of two pieces of information, the allocation information 400 and the system power usage information 900.

FIG. 8 is a diagram showing the review allocation information 1100, and the review allocation information 1100 includes information on the allocation of the control target device 150 in the automatic control system AS. FIG. 9 is a diagram showing a daily operating state of the automatic control system AS when the control target device 150 is assigned to the input / output terminals of the input / output unit 202 in the automatic control system AS based on the review allocation information 1100. be.

As shown in FIG. 8, the review allocation information 1100 includes input / output terminals X0, X1, X2, Y0, Y1, Y2, Z0, Z1 of the input / output units 202A, 202B, 202C and the input / output units 202A, 202B, 202C. , Z2 and the device names M1, M2 ... M9 of the controlled device 150 assigned to each input / output terminal, and the total operating time of the unit operating time and the controlled device 150, which is the total operating time of the input / output unit 202. It shows the operating time of a certain device and the power consumption of the automatic control system AS in correspondence with each other. The input / output units 202A, 202B, 202C and the input / output terminals X0, X1, X2, Y0, Y1, Y2, Z0, Z1, Z2 of the input / output units 202A, 202B, 202C included in the review allocation information 1100. The information associated with the device names M1, M2 ... M9 of the control target device 150 assigned to each input / output terminal is the information of the allocation of the control target device 150 in the automatic control system AS.

Further, as shown in FIG. 9, according to the allocation based on the review allocation information 1100, the input / output units 202A, 202B, and 202C can be allocated with a bias in the operating time zone of the controlled device 150. .. That is, in the input / output units 202A and 202B, since any of the assigned controlled devices 150 is always operating during the daily operation of the automatic control system AS, the input / output units 202A and 202B are relative to the input / output units 202A and 202B. It can be seen that the power is supplied all day long. It can be seen that the input / output unit 202C can stop the power supply between 11:00 and 15:00 and between 21:00 and 24:00 during the daily operation of the automatic control system AS. Further, according to the allocation using the review allocation information 1100, the power consumption of the automatic control system AS is 2255 kWh. That is, according to the allocation based on the review allocation information 1100, the automatic control system AS allocated by the allocation information included in the review allocation information 1100 rather than the power consumption of the automatic control system AS allocated by the allocation information 400. It can be seen that the amount of electricity used is decreasing. Further, it can be seen that the time for stopping the power supply to the input / output unit 202C is longer than the time for stopping the power supply to the input / output unit 202C in the allocation of the automatic control system AS shown in the allocation information 400.

FIG. 10 is a diagram schematically showing a functional block of the learning unit 21. The learning unit 21 is a functional block that realizes the function of the allocation learning device 2, and machine-learns the review of the allocation of the control target device 150 in the automatic control system AS. As shown in FIG. 10, the learning unit 21 includes a preprocessing unit 22a, a learning data acquisition unit 23, a model generation unit 24, and a model storage unit 25.

The pre-processing unit 22a performs pre-processing when pre-processing is required in order to use the data input to the learning unit 21 as learning data. The pre-processing refers to arithmetic processing including calculation of data input to the learning unit 21 and change of format. The learning data acquisition unit 23 acquires learning data necessary for machine learning in the learning unit 21. The learning data in the first embodiment will be described later.

Based on the learning data, the model generation unit 24 reduces the power consumption of the automatic control system AS allocated by the allocation information 400, and has a plurality of input / output units of the automatic control system AS allocated by the allocation information 400. A trained model is generated that infers the allocation state of the controlled device 150 in which the time for stopping the power supply to at least one of the input / output units 202 of 202 increases. In other words, the power consumption of the automatic control system AS is smaller than the power consumption of the automatic control system AS allocated by the allocation information 400, and a plurality of input / output units of the automatic control system AS allocated by the allocation information 400. The time to stop the power supply to at least one input / output unit 202 of the plurality of input / output units 202 of the automatic control system AS is longer than the time to stop the power supply to at least one input / output unit 202 of 202. A trained model that infers the allocation state of the increasing control target device 150 is generated.

The model storage unit 25 stores the trained model generated by the model generation unit 24.

The generation of the trained model performed by the learning unit 21 according to the first embodiment will be described below. As the learning algorithm used by the model generation unit 24, known algorithms such as supervised learning, unsupervised learning, and reinforcement learning can be used. The program that causes the arithmetic unit 300 to function as the learning unit 21 is an allocation learning program.

In the first embodiment, a case where reinforcement learning is applied as an example of machine learning will be described. In reinforcement learning, an agent (behavior) in a certain environment observes the current state (environmental parameters) and decides the action to be taken. The environment changes dynamically depending on the behavior of the agent, and the agent is rewarded according to the change in the environment. The agent repeats this process and learns the action policy that gives the most reward through a series of actions. Q-learning and TD-learning are known as typical methods of reinforcement learning. For example, in the case of Q-learning, the general update formula of the action value function Q (s, a) is represented by Equation 3.

In Equation 3, s _t represents the state of the environment at time t, a _t represents the behavior in time t. By the action _{a t,} the state is changed to _{s t + 1.} rt _{+ 1} represents the reward received by the change of the state, γ represents the discount rate, and α represents the learning coefficient. Note that γ is in the range of 0 <γ ≦ 1 and α is in the range of 0 <α ≦ 1. In the first embodiment, the action a _t becomes "automatic control system AS the operation is power usage information when allowed", "automatic control system allocation information 400 of the control target device 150 in the AS" is next state s _t, the time to learn the best of the action _{a t} in t of the state _{s t.}

The power usage information when the automatic control system AS is operated is the power consumption of the input / output unit 202 and the power consumption of the control target device 150 when the automatic control system assigned by the allocation information 400 is operated. The information including the power consumption of the automatic control system AS calculated based on the above, and the allocation information 400 of the control target device 150 in the automatic control system AS is the input / output terminal of the input / output unit 202 included in the automatic control system AS. This is information corresponding to the control target device 150.

In the update formula represented by the equation 3, if the action value Q of the action a having the highest Q value at time t + 1 is larger than the action value Q of the action a executed at time t, the action value Q is increased. In the opposite case, the action value Q is reduced. In other words, the action value function Q (s, a) is updated so that the action value Q of the action a at time t approaches the best action value at time t + 1. As a result, the best behavioral value in a certain environment is sequentially propagated to the behavioral value in the previous environment.

The reward calculation unit 24R calculates the reward based on the power usage information when the automatic control system AS is operated and the allocation information 400 of the control target device 150 in the automatic control system AS. The reward calculation unit 24R is the power of the automatic control system AS calculated based on the power consumption of the input / output unit 202 and the power consumption of the controlled device 150 when the automatic control system AS assigned by the allocation information 400 is operated. The first reward r1 is calculated based on the difference between the usage amount (kWh) and the predetermined first threshold value. For example, when the power consumption included in the power usage information when the automatic control system AS is operated is smaller than the first threshold value, the first reward r1 is increased (for example, "the value of the power usage amount is increased from the first threshold value". On the other hand, when the power consumption included in the power usage information when the automatic control system AS is operated is larger than the first threshold value, the first reward r1 is reduced (for example, "subtracted positive value"). Negative value obtained by subtracting the value of power consumption from a predetermined threshold value ").

The first threshold value is an arbitrary numerical value for learning, and can be determined with reference to the numerical value of the amount of electric power used. Further, the first reward r1 indicates the renewal type reward rt _{+ 1} represented by the equation 3.

The function update unit 24U reduces the power consumption of the automatic control system AS allocated by the allocation information 400 according to the first reward r1 calculated by the reward calculation unit 24R, and the automatic control system allocated by the allocation information 400. The time to stop the power supply to at least one input / output unit 202 of the plurality of AS input / output units 202 is increased. The function for determining the allocation state of the controlled device 150 is updated and stored in the model storage unit 25. Output. For example, in the case of Q-learning, the number 3 represented by action value function Q (s _{t, a t)} the allocation status of the control target device 150 to power usage is reduced in the automatic control system AS with assigned in allocation information 400 It can be used as a function for calculation. Moreover, action value function Q (s _{t, a t)} power usage of the automatic control system AS is reduced which allocation in allocation information 400, and a plurality of input and output of the automatic control system AS that allocated by the allocation information 400 It can also be used as a function for calculating the allocation state of the control target device 150 in which the time for stopping the power supply to at least one input / output unit 202 of the units 202 increases.

The above learning is repeatedly executed. Model storage unit 25, action value is updated by the function updating unit 24U function _{Q (s t, a t)} , i.e., storing the learned model.

FIG. 11 is a flowchart showing an operation of generating a trained model by machine learning in the learning unit 21 according to the first embodiment.

As shown in FIG. 11, the learning unit 21 has the first input information 800 for learning having the same configuration as the first input information 800 described above, or the second input information 800 for learning having the same configuration as the second input information 1000 described above. Accepts any input of the input information 1000 (step S101). When the learning allocation information 400 of the control target device 150 in the automatic control system AS and the automatic control system AS are operated based on the input learning first input information 800 or learning second input information 1000. Acquire learning data including learning power usage information (step S102). Based on the acquired learning data, the power consumption of the automatic control system AS allocated by the allocation information 400 is reduced, and at least among the plurality of input / output units 202 of the automatic control system AS allocated by the allocation information 400. A trained model for inferring the allocation state of the controlled device 150 in which the time for stopping the power supply to one input / output unit 202 increases is generated (step S103). The generated trained model is stored (step S104). Step S102 and step S103 will be described in detail below.

In step S102, when the input information is the learning first input information 800, the preprocessing unit 22a uses the preprocessing unit 22a to acquire the learning data, and the learning time zone included in the learning first input information 800. Based on the information 500, the learning unit power consumption information 600, and the learning device power consumption information 700, the learning power usage information when the automatic control system AS is operated is calculated. Specifically, the system power usage calculation function corresponding to the mathematical formulas shown in the above-mentioned equations 1 and 2 is stored in the storage unit 102, and the preprocessing unit 22a reads out from the storage unit 102 to read the system power. Information corresponding to usage information 900 is calculated. The learning allocation information 400 included in the learning first input information 800 and the learning power usage information when the automatic control system AS calculated by the preprocessing unit 22a is operated are used as learning data for learning. The data acquisition unit 23 acquires the data.

When the input input information is the learning second input information 1000, the learning allocation information 400 included in the learning second input information 1000 and the learning system power usage information 900 are used as learning data. The learning data acquisition unit 23 acquires the information. The learning system power usage information 900 is learning power usage information.

In step S103, the model generation unit 24 calculates the first reward r1 based on the learning power usage information when the automatic control system AS is operated and the learning allocation information 400 of the controlled target device 150 in the automatic control system AS. do. Specifically, the reward calculation unit 24R acquires the learning power usage information when the automatic control system AS is operated and the learning allocation information 400 of the controlled target device 150 in the automatic control system AS, and is predetermined. Whether to increase the first reward r1 (step S103-A) or decrease the first reward r1 based on the difference between the first threshold value and the power consumption (kWh) included in the learning power usage information (step S103-). B) is judged.

When the reward calculation unit 24R determines that the first reward r1 is to be increased, the reward calculation unit 24R increases the first reward r1 in step S103-A. On the other hand, when the reward calculation unit 24R determines that the first reward r1 is to be reduced, the reward calculation unit 24R reduces the first reward r1 in step S103-B.

Then, in step S103-C, the function updater 24U, based on the first compensation r1 calculated by the compensation calculator 24R, action value function model storage unit 25 is represented by the number 3 for storing Q _{(s t} , to update _{a t).}

The learning unit 21 repeatedly executes the above steps S102 to S103 for different automatic control systems AS of the learning allocation information 400, and generates an action value function Q (st, at) as a learned model. ..

The above-mentioned learning allocation information 400, learning time zone information 500, learning unit power consumption information 600, learning device power consumption information 700, learning first input information 800, learning system power usage information 900, and The learning second input information 1000 (collectively referred to as “information for obtaining learning data”) is information used for machine learning in the learning unit 21. The information for obtaining the learning data is the allocation information 400 and the time zone information 500, which are input when the arithmetic processing for inferring the allocation state of the control target device 150 in the automatic control system AS in the inference unit 20 described later is performed. , Unit power consumption information 600, device power consumption information 700, first input information 800, system power usage information 900, and second input information 1000 (collectively referred to as "information for inference"). This is the term used. The information for obtaining the learning data has the same information structure as the information for inference, and is information having different contents such as specific numerical values indicating each information.

Hereinafter, the operation of the allocation support program and the allocation support device 1 according to the first embodiment will be described with reference to the drawings.

FIG. 12 is a flowchart showing the operation of the allocation support program and the allocation support device 1 according to the first embodiment. The allocation support device 1 is in a state where the operation can be started by activating the installed allocation support program.

The user uses four pieces of information: the allocation information 400 of the automatic control system AS being operated, the time zone information 500, the unit power consumption information 600 of the input / output unit used in the automatic control system AS, and the device power consumption information 700. Either the first input information 800 that is configured, or the second input information 1000 that is composed of two pieces of information, the allocation information 400 of the automatic control system AS in operation and the system power usage information 900, is prepared.

Then, either the prepared first input information 800 or the second input information 1000 is input to the allocation support device 1. The input first input information 800 or second input information 1000 is stored in the storage unit 102. The information acquisition unit 11 reads and acquires the first input information 800 or the second input information 1000 stored in the storage unit 102 (step S201).

Next, either the first input information 800 or the second input information 1000 acquired by the information acquisition unit 11 is input to the inference unit 20 (step S202). The learning device unit 21 executes arithmetic processing based on the input information. Here, the arithmetic processing of the inference unit 20 based on the input information will be specifically described.

When the information input to the inference unit 20 is the first input information 800, the preprocessing unit 22 executes the first arithmetic processing on the first input information 800 (step S203). Specifically, the system power consumption calculation function corresponding to the mathematical formulas shown in the above equations 1 and 2 is stored in the storage unit 102, and the preprocessing unit 22 reads from the storage unit 102 and first Based on the time zone information 500 included in the input information 800, the unit power consumption information 600 of the input / output unit used in the automatic control system AS, and the device power consumption information 700, the information corresponding to the system power usage information 900 is calculated. do.

Subsequently, the arithmetic processing unit 26 is based on the allocation information 400 included in the first input information 800 input to the inference unit 20 in step S202 and the information acquired by performing the first arithmetic processing in step S203. , The second arithmetic processing using the trained model is executed (step S204). That is, the arithmetic processing unit 26 reads the learned trained model stored in the model storage unit 25 of the learning unit 21, executes the second arithmetic processing, and allocates the automatic control system AS with the allocation information 400. The allocation state of the controlled device 150 whose power consumption is reduced is inferred, and the result of the inference is output. Alternatively, the arithmetic processing unit 26 executes the second arithmetic processing, reduces the power consumption of the automatic control system AS allocated by the allocation information 400, and has a plurality of automatic control system AS allocated by the allocation information 400. The allocation state of the control target device 150 in which the time for stopping the power supply to at least one of the input / output units 202 is increased is inferred, and the result of the inference is output.

In other words, the arithmetic processing unit 26 executes the second arithmetic processing, and the power consumption of the automatic control system AS is smaller than the power consumption of the automatic control system AS allocated by the allocation information 400. Infers the allocation state of, and outputs the inference result. Alternatively, the arithmetic processing unit 26 executes the second arithmetic processing, and the electric power consumption of the automatic control system AS is smaller than the electric power consumption of the automatic control system AS allocated by the allocation information 400, and the allocation information At least one of the plurality of I / O units 202 of the automatic control system AS is longer than the time to stop the power supply to at least one of the plurality of I / O units 202 of the automatic control system AS assigned by 400. The allocation state of the controlled device 150 in which the time for stopping the power supply to the two input / output units 202 increases is inferred, and the inference result is output.

When the information input to the inference unit 20 is the second input information 1000, the second input information 1000 includes the system power usage information 900, so that the first arithmetic processing performed in step S203 described above can be performed. Since it becomes unnecessary, step S204 is executed after step S202. When the second input information is input, the second arithmetic processing performed in step S204 is executed based on the allocation information 400 and the system power usage information 900 constituting the second input information 1000. That is, the inference performed by the inference unit 20 is the above-mentioned arithmetic processing, and when the first input information 800 is input, it refers to the first arithmetic processing and the second arithmetic processing, and the second input information 1000 is input. If so, it refers to the second arithmetic processing.

Then, the allocation information output unit 12 acquires and outputs the result output by the arithmetic processing unit 26 of the inference unit 20 as the review allocation information 1100 (step S205). By completing step S205, the allocation support program and the allocation support device 1 end the operation. The allocation information output unit 12 is an automatic control system AS that associates the input / output unit, the input / output terminals of the input / output units, and the control target device 150 assigned to each input / output terminal, which are included in the review allocation information 1100. The information on the allocation of the control target device 150 in the above may be acquired and output.

As described above, the allocation support program and the allocation support device 1 according to the first embodiment are the information acquisition unit 11, the first input information 800, or the first input information 800 that acquires either the first input information 800 or the second input information 1000. The second input information 1000 is input, and has an allocation information output unit 12 that acquires and outputs the review allocation information 1100, which is output by inference using the trained model of the inference unit 20. As a result, the allocation of the control target device 150 in the automatic control system AS is an index for reviewing the allocation of the control target device 150 in the automatic control system AS so that the power consumption of the automatic control system AS allocated by the allocation information 400 is reduced. Allocation information can be provided. Alternatively, the power consumption of the automatic control system AS allocated by the allocation information 400 is reduced, and at least one of the plurality of input / output units 202 of the automatic control system AS allocated by the allocation information 400 is transferred to the input / output unit 202. It is possible to provide information on the allocation of the controlled device 150 in the automatic control system AS, which is an index for reviewing so that the time for stopping the power supply is increased.

Therefore, according to the allocation of the automatic control system AS using the review allocation information 1100 in the first embodiment, the operating time zone of the controlled device 150 assigned to the automatic control system AS is set to each input / output unit 202. By biasing it, the power consumption of the automatic control system AS can be reduced. Further, the time for stopping the power supply to at least one input / output unit 202 among the plurality of input / output units 202 of the automatic control system AS allocated by the allocation information 400 increases. Here, stopping the power supply to the input / output unit 202 may mean stopping the power supply so as to stop all the functions of the input / output unit 202, and the input / output device included in the input / output unit 202. The power supply to the whole may be stopped.

Since the learning unit 21 generates the trained model based on the learning data including the allocation information 400 and the power usage information when the automatic control system AS is operated as described above, the allocation information 400 It is possible to infer how the amount of power used when the automatic control system AS is operated changes depending on the allocation state of the control target device 150 in the automatic control system AS specified in. The power consumption when the automatic control system AS is operated tends to be high when the power is supplied to the input / output unit 202 included in the automatic control system AS. It tends to be low when the power supply to the output unit 202 is stopped. Therefore, it is possible to infer the allocation state in which the power consumption is low when the automatic control system AS is operated, and at least one input / output unit out of the plurality of input / output units 202 of the automatic control system AS allocated by the allocation information 400. It is possible to infer the allocation state of the controlled device 150 in which the time for stopping the power supply to the 202 increases.

Embodiment 2.
The allocation support program, the allocation support device, the allocation learning program, and the allocation learning device according to the second embodiment of the present invention will be described with reference to the drawings. Also in the second embodiment, the allocation support program includes the allocation learning program, and the allocation support device will be described as having a function as the allocation learning device. The same configuration as in the first embodiment described above will be described using the same reference numerals, and specific description thereof will be omitted. Hereinafter, a configuration different from that of the first embodiment will be specifically described.

FIG. 13 is a diagram schematically showing an example of the functional block of the allocation support device 1b according to the second embodiment. In the second embodiment as well, as in the first embodiment, the allocation support device 1b serves as an allocation learning device 2b that performs machine learning in addition to the function of supporting the review of the allocation of the controlled device 150. It is a configuration that also has a function. Also in the second embodiment, the functional block of the allocation learning device 2b will be described together with the allocation support device 1b.

The allocation support device 1b is a device that supports the review of the allocation of the controlled device 150 in the automatic control system AS that is already in operation. The allocation support device 1b inputs the information acquisition unit 11b that acquires various information in the automatic control system AS and the information acquired by the information acquisition unit 11b, and infers the allocation of the control target device 150 in the automatic control system AS. It has a unit 20b and an allocation information output unit 12b that acquires and outputs information output by the arithmetic processing of the inference unit 20b. Further, the allocation support device 1b has a learning unit 21b for machine learning the allocation of the control target device 150 in the automatic control system AS. The learning unit 21b is a functional block of the allocation learning device 2b.

The information acquisition unit 11b determines the number of times a signal is transmitted to each input / output terminal of each input / output unit 202 corresponding to the allocation information 400, the time zone information 500, the unit power consumption information 600, the device power consumption information 700, and the allocation information 400. The first input information 800b including the indicated number of times information 401 is acquired, or the second input information 1000b including the system power usage information 900, the allocation information 400, and the number of times information 401 is acquired. The number information 401 will be described later.

The inference unit 20b includes a preprocessing unit 22 and an arithmetic processing unit 26b. The pretreatment unit 22 has the same function as that of the first embodiment. The arithmetic processing unit 26b reads out the learned model generated by performing machine learning by the learning unit 21b described later, and allocates the controlled target device 150 in the automatic control system AS based on the information acquired by the information acquisition unit 11b. Infer a review. The arithmetic processing in the inference unit 20 will be described later.

The inference unit 20b inputs either the first input information 800b or the second input information 1000b, and executes each arithmetic processing described later. The power consumption of the automatic control system AS allocated by the allocation information 400 is reduced, and the power to at least one input / output unit 202 of the plurality of input / output units 202 of the automatic control system AS allocated by the allocation information 400 is reduced. Infer an allocation state that increases the time to stop supply. Further, in this allocation state, the maximum value of the number of transmissions of signals transmitted to a plurality of input / output terminals of at least one input / output unit 202 of the automatic control system AS allocated by the allocation information 400 is reduced. It is inferred by the inference unit 20b.

The allocation information output unit 12b acquires and outputs the allocation state inferred by the inference unit 20b as the direct allocation information 1100b. The output review allocation information 1100b is displayed on the display unit 104.

The information acquired by the above-mentioned information acquisition unit 11b and the information output by the allocation information output unit 12b will be described below. Of the information acquired by the information acquisition unit 11b, the allocation information 400, the time zone information 500, the unit power consumption information 600, the device power consumption information 700, and the system power usage information 900 are the same as those in the above-described first embodiment. Therefore, the description will be omitted.

FIG. 14 is a diagram showing an example of the number-of-times information 401. As shown in FIG. 14, the number-of-times information 401 indicates the number of times a signal is transmitted to each input / output terminal of each input / output unit 202 corresponding to the allocation information 400. Specifically, the number of times information 401 includes three input / output units 202A, 202B, 202C and input / output terminals X0, X1, X2, Y0, Y1 of each input / output unit 202A, 202B, 202C as in FIG. , Y2, Z0, Z1, Z2 are shown in correspondence with each other, and the number of times a signal is transmitted to each input / output terminal is shown in correspondence with each input / output terminal. The number of times the signal is transmitted to each input / output terminal is the total of the number of times the signal is transmitted from the CPU unit 201 to the input / output terminal and the number of times the signal is transmitted from the controlled device 150 to the input / output terminal. The number of times. As shown in FIG. 14, in the second embodiment, the input / output terminal X1 has the largest number of signal transmissions among the input / output terminals of the input / output unit 202A, which is 18429 times. Similarly, the input / output terminal Y0 has the largest number of signal transmissions among the input / output terminals of the input / output unit 202B, which is 15320 times, and the number of signal transmissions among the input / output terminals of the input / output unit 202C. The most common was the input / output terminal Z0, which was 19079 times.

In the second embodiment, the first input information 800b is information including the above-mentioned allocation information 400, time zone information 500, unit power consumption information 600, device power consumption information 700, and number of times information 401. In the second embodiment, the first input information 800b is composed of five pieces of information: allocation information 400, time zone information 500, unit power consumption information 600, device power consumption information 700, and number of times information 401.

Further, in the second embodiment, the second input information 1000b is information including the above-mentioned allocation information 400, system power usage information 900, and number of times information 401. In the second embodiment, the second input information 1000b is composed of three pieces of information: allocation information 400, system power usage information 900, and number of times information 401.

FIG. 15 is a diagram showing the review allocation information 1100b, and the review allocation information 1100b includes the allocation information of the control target device 150 in the automatic control system AS. FIG. 16 is a diagram showing a daily operating state of the automatic control system AS when the control target device 150 is assigned to the input / output terminals of the input / output unit 202 in the automatic control system AS based on the review allocation information 1100b. be.

As shown in FIG. 15, the review allocation information 1100b includes the input / output units 202A to 202C, the input / output terminals X0 to Z2 provided in the input / output units 202A to 202C, and the control target device 150 assigned to each input / output terminal. The device names M1 to M9, the number of transmissions, which is the number of times signals are transmitted to the input / output terminals, the unit operating time, which is the total operating time of the input / output unit 202, and the total operating time of the controlled device 150. It shows the device operating time and the power consumption of the automatic control system AS in correspondence with each other. The input / output units 202A to 202C, the input / output terminals X0 to Z2 of the input / output units 202A to 202C, and the device names M1 to the control target device 150 assigned to each input / output terminal included in the review allocation information 1100b. The information associated with M9 is the information on the allocation of the control target device 150 in the automatic control system AS. According to the allocation based on the review allocation information 1100b, the input / output units 202A, 202B, and 202C are biased in the operating time zone of the controlled device 150, and the input / output units 202A, 202B, and 202C are assigned. Therefore, the number of signal transmissions can be biased and assigned.

That is, as shown in FIG. 16, the input / output units 202A and 202B are turned on because any of the assigned controlled devices 150 is always operating during the daily operation of the automatic control system AS. It can be seen that power is supplied to the output units 202A and 202B all day long. It can be seen that the input / output unit 202C can stop the power supply between 12:00 and 15:00 and between 22:00 and 24:00 during the daily operation of the automatic control system AS. Further, as shown in FIG. 15, according to the allocation using the review allocation information 1100b, the power consumption of the automatic control system AS is 2265 kWh. That is, according to the allocation based on the review allocation information 1100b, it can be seen that the power consumption of the automatic control system AS is reduced. Further, it can be seen that the time for stopping the power supply to the input / output unit 202C is longer than the time for stopping the power supply to the input / output unit 202C in the allocation of the automatic control system AS shown in the allocation information 400.

Further, as shown in FIG. 15, the input / output unit 202A and the input / output unit 202C each input / output out of the number of times of transmission of signals transmitted to a plurality of input / output terminals of each of the input / output units 202A and 202C. The maximum value for the units 202A and 202C is the number of times the signals transmitted to the plurality of input / output terminals of the input / output units 202A and 202C in the allocation of the automatic control system AS shown in the allocation information 400 shown in FIG. 14 are transmitted. It is smaller than the maximum value for the input / output units 202A and 202C, respectively.

Specifically, the maximum number of transmissions of the signal transmitted to each input / output terminal of the input / output unit 202A of the automatic control system AS allocated by the allocation information 400 shown in FIG. 14 was 18429 times. However, the maximum number of transmissions of the signal transmitted to each input / output terminal of the input / output unit 202A of the automatic control system AS allocated by the allocation information included in the review allocation information 1100b is 7959. Further, the maximum number of transmissions of the signal transmitted to each input / output terminal of the input / output unit 202C of the automatic control system AS allocated by the allocation information 400 was 19079 times, but the review allocation information 1100b The maximum number of transmissions of the signal transmitted to each input / output terminal of the input / output unit 202C allocated by the included allocation information is 2615 times.

FIG. 17 is a diagram showing an example of the functional block of the learning unit 21b according to the second embodiment. The learning unit 21b is a functional block that realizes the function of the allocation learning device 2b, and machine-learns the review of the allocation of the control target device 150 in the automatic control system AS. As shown in FIG. 16, the learning unit 21b includes a preprocessing unit 22a that functions in the same manner as in the first embodiment, a learning data acquisition unit 23b that acquires learning data in the second embodiment, and a trained model. It includes a weighted information acquisition unit 27 that acquires information that affects the generation of the trained model, a model generation unit 25b that generates a trained model, and a model storage unit 25b that stores the trained model.

The weighted information acquisition unit 27 acquires weighted information that weights the reward at the time of generating the learned model, which will be described later. The learning data acquisition unit 23b acquires learning data necessary for machine learning in the learning unit 21b. The learning data in the second embodiment will be described later.

The model generation unit 24b reduces the power consumption of the automatic control system AS allocated by the allocation information 400 based on the learning data and the allocation information, and has a plurality of automatic control system ASs allocated by the allocation information 400. Generates a trained model that infers an allocation state that increases the time to stop the power supply to at least one of the input / output units 202 of the above. Further, in this trained model, the maximum value of the number of transmissions of signals transmitted to a plurality of input / output terminals of at least one input / output unit 202 of the automatic control system AS allocated by the allocation information 400 is reduced. Is generated in.

The generation of the trained model performed by the learning unit 21b according to the second embodiment will be described below. As the learning algorithm used by the model generation unit 24b, known algorithms such as supervised learning, unsupervised learning, and reinforcement learning can be used. The program that causes the arithmetic unit 300 to function as the learning unit 21b is an allocation learning program.

In the second embodiment, the case where the reinforcement learning of machine learning is applied as in the first embodiment will be described. When Q-learning, which is a typical method of reinforcement learning, is used, the general update formula of the action value function Q (s, a) is represented by the above equation 3.

In the second embodiment, in Equation 3, "power usage information when the automatic control system AS is operated" and "plurality of input / output terminals of each of the plurality of input / output units 202 included in the automatic control system AS". times information 401 "behavioral a _t next showing the number of times of transmission of the signal transmitted to the" automatic control system of the control target device 150 in the aS allocation information 400 "is next state s _t, the time t at state s _t best to learn the action _{a t.} In the second embodiment, learning is performed in consideration of the weighted information acquired by the weighted information acquisition unit 27, which will be described later.

The power usage information when the automatic control system AS is operated is the power consumption of the input / output unit 202 and the power consumption of the control target device 150 when the automatic control system assigned by the allocation information 400 is operated. The information including the power consumption of the automatic control system AS calculated based on the above, and the allocation information 400 of the control target device 150 in the automatic control system AS is the input / output terminal of the input / output unit 202 included in the automatic control system AS. This is information corresponding to the control target device 150.

The weighted information acquisition unit 27 acquires weighted information that weights the reward when the trained model is generated. The weighted information is a coefficient that acts on the reward calculated by the reward calculation unit 24Rb, which will be described later, and can be set in the range of 0.0 to 0.1. The weighted information acquisition unit 27 acquires the coefficient set by the user via the input unit 103 as the weighted information. In the second embodiment, two coefficients acting on each of the first reward r1 and the second reward r2, which will be described later, are acquired as weighted information. For example, when it is desired to increase the influence of the first reward r1 for learning, a coefficient acting on the first reward r1 is set large (for example, "0.8"), and the coefficient acts on the second reward r2. The coefficient is set to be smaller than the coefficient acting on the first reward r1 (for example, "0.6").

The reward calculation unit 24Rb is the power usage information when the automatic control system AS is operated, and the number of transmissions of signals transmitted to a plurality of input / output terminals of each of the plurality of input / output units 202 included in the automatic control system AS. The reward is calculated based on the number of times information 401 indicating the above, the allocation information 400 of the control target device 150 in the automatic control system AS, and the weighting information.

The reward calculation unit 24Rb is a predetermined threshold value in the power consumption amount, and is a difference between a predetermined first threshold value (hereinafter, referred to as “power threshold value” for convenience in the second embodiment) and the power consumption amount (kWh). The first reward r1 is calculated based on the above, and the first reward r1 is multiplied by a coefficient based on the weighting information. For example, when the power consumption included in the power usage information when the automatic control system AS is operated is smaller than the power threshold, the first reward r1 is increased (for example, "the value of the power usage is subtracted from the power threshold". (Positive value), and multiply the increased first reward r1 by a coefficient. On the other hand, when the power consumption included in the power usage information when the automatic control system AS is operated is larger than the power threshold, the first reward r1 is reduced (for example, "the value of the power usage is subtracted from the power threshold". Negative value "), and multiply the reduced first reward r1 by a coefficient. The power threshold is an arbitrary numerical value for learning, and can be determined with reference to the numerical value of the amount of electric power used.

Further, the reward calculation unit 24Rb is a predetermined threshold value for the number of times the signal transmitted to the input / output terminal is transmitted, and is a predetermined second threshold value (hereinafter, in the second embodiment, it is referred to as a “number of times threshold value” for convenience). The second reward r2 is calculated based on the difference between the total number of times the signal is transmitted to the respective input / output terminals of the plurality of input / output units 202 and the total number of times the maximum value is transmitted to the respective input / output terminals of the plurality of input / output units 202.

The total number of times can be obtained by, for example, the following method. First, among the number of times the signal transmitted to each input / output terminal of the input / output unit 202 when the automatic control system AS is operated, the number of transmissions corresponding to the input / output terminal having the largest number of transmissions is input / output. It is the maximum value of the number of transmissions in the unit 202. This maximum value is extracted for each I / O unit. Next, the plurality of extracted maximum values are summed to obtain the total number of times.

In the reward calculation unit 24Rb, the total number of times included in the number-of-times information 401 indicating the number of times of transmission of signals transmitted to a plurality of input / output terminals of each of the plurality of input / output units 202 included in the automatic control system AS is from the number-of-times threshold value. If it is small, the second reward r2 is increased (for example, “a positive value obtained by subtracting the total number of times from the number threshold value”), and the increased second reward r2 is multiplied by a coefficient. On the other hand, when the total number of times included in the number-of-times information 401 indicating the number of times of transmission of signals transmitted to a plurality of input / output terminals of each of the plurality of input / output units 202 included in the automatic control system AS is larger than the number-of-times threshold value. The second reward r2 is reduced (for example, “a negative value obtained by subtracting the power consumption value from the power threshold value”), and the reduced second reward r2 is multiplied by a coefficient. The number-of-times threshold value is an arbitrary numerical value at the time of learning, and can be determined with reference to the numerical value of the number of times of signal transmission to the input / output terminal.

Then, the reward calculation unit 24Rb adds the first reward r1 and the second reward r2 to determine the final reward. The final reward here indicates the renewal reward rt _{+ 1} represented by Equation 3.

The function update unit 24Ub reduces the power consumption of the automatic control system AS allocated by the allocation information 400 according to the final reward calculated by the reward calculation unit 24Rb, and reduces the power consumption of the automatic control system AS allocated by the allocation information 400. The time for stopping the power supply to at least one input / output unit 202 among the plurality of input / output units 202 is increased, and the plurality of the plurality of the at least one input / output unit 202 of the automatic control system AS allocated by the allocation information 400 The function for determining the allocation state of the control target device 150 in which the maximum value of the number of transmissions of the signal transmitted to the input / output terminal of is decreased is updated and output to the model storage unit 25.

For example, in the case of Q-learning, power usage of the automatic control system AS that action value represented by the number 3 function Q (s _{t, a t)} was allocated in allocation information 400 is reduced, and the allocation in allocation information 400 It is used as a function for calculating an allocation state such that the time for stopping the power supply to at least one input / output unit 202 among the plurality of input / output units 202 of the automatic control system AS increases. Further, the allocation state calculated by this function is the maximum value of the number of transmissions of signals transmitted to a plurality of input / output terminals of at least one input / output unit 202 of the automatic control system AS allocated by the allocation information 400. Is in an allocation state where is decreasing.

The above learning is repeatedly executed. Model storage unit 25b, action value is updated by the function updating unit 24Ub function _{Q (s t, a t)} , i.e., storing the learned model.

FIG. 18 is a flowchart showing an operation of generating a trained model by machine learning in the learning unit 21b according to the second embodiment.

As shown in FIG. 18, the learning unit 21 has the first input information 800b for learning having the same configuration as the first input information 800b described above, or the second input information 800b for learning having the same configuration as the second input information 1000b described above. Accepts any input of the input information 1000b (step S301). The weighted information acquisition unit 27 acquires the weighted information (step S302). When the learning allocation information 400 of the control target device 150 in the automatic control system AS and the automatic control system AS are operated based on the input learning first input information 800b or learning second input information 1000b. The number of times of learning indicating the number of times of transmission of signals transmitted to a plurality of input / output terminals of each of the plurality of input / output units 202 included in the automatic control system AS corresponding to the learning power usage information and the learning allocation information 400. Acquire learning data including information 401 (step S303). Based on the acquired learning data, a trained model that infers the allocation state of the controlled device 150 is generated (step S304). The generated trained model is stored (step S305). Step S303 and step S304 will be described in detail below.

In step S303, when the input information is the learning first input information 800b, the preprocessing unit 22 uses the preprocessing unit 22 to acquire the learning data in the same manner as in the first embodiment described above. Based on the learning time zone information 500, the learning unit power consumption information 600, and the learning device power consumption information 700 included in the information 800b, the learning power usage information when the automatic control system AS is operated is calculated. The learning allocation information 400 and the learning frequency information 401 included in the learning first input information 800b, and the learning power usage information when the automatic control system AS calculated by the preprocessing unit 22 is operated. The learning data acquisition unit 23b acquires the learning data.

When the input input information is the learning second input information 1000b, the learning allocation information 400 included in the learning second input information 1000b, the learning system power usage information 900, and the learning frequency information 401 and are acquired by the learning data acquisition unit 23b as learning data. The learning system power usage information 900 is learning power usage information.

In step S304, the model generation unit 23b includes learning power usage information when the automatic control system AS is operated, learning allocation information 400 of the control target device 150 in the automatic control system AS, and a plurality of information included in the automatic control system AS. The reward is calculated based on the number of times information 401 indicating the number of times of learning transmission of the signal transmitted to the plurality of input / output terminals of each of the input / output units 202 of the above. Specifically, the reward calculation unit 24Rb calculates the first reward r1 and the second reward r2, and calculates the final reward in consideration of the weighting information. First, the reward calculation unit 24Rb uses the learning power usage information and the learning allocation information 400, and is based on the difference between the predetermined power threshold and the power usage amount (kWh) included in the learning power usage information. It is determined whether to increase the first reward r1 (step S304-A) or decrease the first reward r1 (step S304-B).

When the reward calculation unit 24Rb determines that the first reward r1 is to be increased, the reward calculation unit 24Rb increases the first reward r1 in step S304-A. On the other hand, when the reward calculation unit 24Rb determines that the first reward r1 is to be reduced, the reward calculation unit 24Rb reduces the first reward r1 in step S304-B. Then, in step S304-C, the coefficient included in the weighting information acquired in step S302 is multiplied by the increased or decreased first reward r1.

Subsequently, the reward calculation unit 24Rb uses the learning allocation information 400 and the learning number of times information 401 to increase the second reward r2 based on the difference between the predetermined number of times threshold value and the above-mentioned total number of times. (Step S304-A) or (step S304-B) to reduce the second reward r2 is determined.

When the reward calculation unit 24Rb determines that the second reward r2 is to be increased, the reward calculation unit 24Rb increases the second reward r2 in step S304-A. On the other hand, when the reward calculation unit 24Rb determines that the second reward r2 is to be reduced, the reward calculation unit 24Rb reduces the second reward r2 in step S304-B. Then, in step S304-C, the coefficient included in the weighting information acquired in step S302 is multiplied by the increased or decreased second reward r2.

Then, in step S304-D, the reward calculation unit 24Rb calculates the final reward by adding the calculated first reward r1 and second reward r2.

In step S304-E, function updating unit 24Ub, based on the final compensation calculated by the compensation calculator 24Rb, activation level model storage unit 24b is represented by the number 3 for storing function _Q (s _{t, a} t) To update.

The learning unit 21b repeatedly executes the above steps S303 to S304 for different automatic control systems AS of the allocation information 400 for learning, and generates an action value function Q (st, at) as a learned model. do.

The above-mentioned learning allocation information 400, learning frequency information 401, learning time zone information 500, learning unit power consumption information 600, learning device power consumption information 700, learning first input information 800b, learning system. The power usage information 900 and the learning second input information 1000b (collectively referred to as "information for obtaining learning data") are information used for machine learning in the learning unit 21b. The information for obtaining the learning data is input when the arithmetic processing for inferring the allocation state of the control target device 150 in the automatic control system AS in the inference unit 20b described later is performed, the allocation information 400, the number of times information 401, This term is used to distinguish the time zone information 500, the unit power consumption information 600, the device power consumption information 700, the first input information 800, the system power usage information 900, and the second input information 1000. The information for obtaining the learning data is information having the same information structure as each information with respect to the information for inference, and different contents such as specific numerical values indicating each information.

Hereinafter, the operation of the allocation support program and the allocation support device 1b according to the second embodiment will be described with reference to the drawings.

FIG. 19 is a flowchart showing the operation of the allocation support program and the allocation support device 1b according to the second embodiment. The allocation support device 1b is in a state where the operation can be started by activating the installed allocation support program.

The user can use the allocation information 400 of the automatic control system AS, the time zone information 500, the unit power consumption information 600 of the input / output unit used in the automatic control system AS, the device power consumption information 700, and the allocation information 400. The first input information 800b composed of five pieces of information 401 indicating the number of times a signal transmitted to a plurality of input / output terminals of each of the plurality of input / output units 202 included in the corresponding automatic control system AS is transmitted. Alternatively, one of the second input information 1000b composed of three pieces of information, that is, the allocation information 400, the system power usage information, and the number of times information 401 of the automatic control system AS that is in operation, is prepared.

Then, either the prepared first input information 800b or the second input information 1000b is input to the allocation support device 1. The input first input information 800b or second input information 1000b is stored in the storage unit 102. The information acquisition unit 11b reads out and acquires the first input information 800b or the second input information 1000b stored in the storage unit 102 (step S401).

Next, either the first input information 800b or the second input information 1000b acquired by the information acquisition unit 11b is input to the inference unit 20b (step S402). The inference unit 20b executes arithmetic processing based on the input information. Here, the arithmetic processing of the inference unit 20b based on the input information will be specifically described.

When the information input to the inference unit 20b is the first input information 800b, the preprocessing unit 22 executes the first arithmetic processing on the first input information 800b (step S403). Specifically, the system power consumption calculation function corresponding to the above-mentioned equations 1 and 2 is stored in the storage unit 102, and the preprocessing unit 22 reads it from the storage unit 102 and stores it in the first input information 800b. Based on the time zone information 500, the unit power consumption information 600 of the input / output unit used in the automatic control system AS, and the device power consumption information 700, the information corresponding to the system power usage information 900 is calculated.

Subsequently, a calculation is performed based on the allocation information 400 and the number of times information 401 included in the first input information 800b input to the inference unit 20b in step S402, and the information acquired by performing the first calculation process in step S403. The processing unit 26b executes the second arithmetic processing using the trained model (step S404). That is, the arithmetic processing unit 26b reads the learned trained model stored in the model storage unit 25b of the learning unit 21b, executes the second arithmetic processing, and allocates the automatic control system AS with the allocation information 400. Allocation so that the amount of power used in the system is reduced, and the time for stopping the power supply to at least one of the plurality of input / output units 202 of the automatic control system AS allocated in the allocation information 400 is increased. Infer the state. Further, in this allocation state, the maximum value of the number of transmissions of signals transmitted to a plurality of input / output terminals of at least one input / output unit 202 of the automatic control system AS allocated by the allocation information 400 is reduced. It is inferred by the inference unit 20b. Then, the result of the inference is output from the inference unit 20b.

Further, when the information input to the inference unit 20b is the second input information 1000b, the second input information 1000b includes the system power usage information 900, so that the first arithmetic processing performed in the above step S403 can be performed. Since it becomes unnecessary, step S404 is executed after step S402. When the second input information 1000b is input, the second arithmetic processing performed in step S404 is executed based on the allocation information 400, the number of times information 401, and the system power usage information 900 constituting the second input information 1000b. That is, the inference performed by the inference unit 20b is the above-mentioned arithmetic processing, and when the first input information 800b is input, it refers to the first arithmetic processing and the second arithmetic processing, and the second input information 1000b is input. If so, it refers to the second arithmetic processing.

Then, the allocation information output unit 12b acquires and outputs the result output by the arithmetic processing unit 26b of the inference unit 20b as the review allocation information 1100b (step S405). By completing step S405, the allocation support program and the allocation support device 1b end their operations. The allocation information output unit 12b is an automatic control system AS that associates the input / output unit, the input / output terminals of the input / output units, and the control target device 150 assigned to each input / output terminal, which are included in the review allocation information 1100b. The information on the allocation of the control target device 150 in the above may be acquired and output.

As described above, the allocation support program and the allocation support device 1b according to the second embodiment are the information acquisition unit 11b, the first input information 800b, or the information acquisition unit 11b that acquires either the first input information 800b or the second input information 1000b. , The second input information 1000b is input, and the allocation information for outputting the allocation information of the controlled target device 150 included in the review allocation information 1100b acquired from the inference unit 20b that performs the inference using the trained model. It has an output unit 12b. As a result, it is possible to provide information on the allocation of the control target device 150 in the automatic control system AS described in the second embodiment, which is an index for reviewing the allocation of the control target device 150 in the automatic control system AS.

Therefore, according to the allocation of the automatic control system AS using the review allocation information 1100b in the second embodiment, the operating time zone of the controlled device 150 assigned to the automatic control system AS is set to each input / output unit 202. By biasing it, the power consumption of the automatic control system AS can be reduced as in the first embodiment described above.

Further, the input / output unit 202 usually includes consumable parts such as a relay and a capacitor in order to transmit a signal transmitted to the input / output terminals to a predetermined circuit. If one of the consumable parts corresponding to the plurality of input / output terminals provided in the input / output unit 202 is consumed and damaged, the entire input / output unit 202 needs to be replaced or repaired even if the other input / output terminals are normal. According to the allocation of the automatic control system AS using the review allocation information 1100b in the second embodiment, the plurality of input / output terminals of at least one input / output unit 202 of the automatic control system AS allocated by the allocation information 400 The maximum value of the number of times the transmitted signal is transmitted can be reduced.

That is, in the input / output unit 202 after reviewing the allocation, the control target device 150 having a smaller number of signal transmissions is allocated to the input / output terminals to which the most signals have been transmitted as compared with before the review. On the other hand, the control target device 150 assigned to the input / output terminal to which the most signals are transmitted will be assigned to the input / output terminal of the other input / output unit 202. In other words, the control target device 150 can be assigned to the plurality of input / output units 202 included in the automatic control system AS so that the number of transmissions of signals transmitted to the plurality of input / output terminals of the input / output unit 202 increases. The control target device 150 is assigned so that the number of transmissions of signals transmitted to the plurality of input / output terminals of the input / output unit 202 is reduced, and the control target device 150 is assigned to the plurality of input / output units 202 so as to occur. The allocation can be such that the number of times the signal is transmitted is biased.

Therefore, the input / output unit 202 allocated so that the number of signal transmissions to the plurality of input / output terminals of one input / output unit 202 is reduced can reduce the frequency of replacement or repair. Further, even if the input / output unit 202 is assigned so that the number of signal transmissions increases to the plurality of input / output terminals of one input / output unit 202, the input / output is performed because one input / output terminal is abnormal. Since the situation of replacing or repairing the entire unit 202 can be reduced, the maintainability of the automatic control system AS can be improved.

The learning unit 21b generates a learned model based on the learning data including the allocation information 400, the number of times information 401, and the power usage information when the automatic control system AS is operated as described above. Therefore, how the power consumption when the automatic control system AS is operated changes depending on the allocation state of the control target device 150 in the automatic control system AS specified by the allocation information 400, and the automatic control system AS. It is possible to infer how the number of transmissions of signals transmitted to each input / output terminal of each input / output unit 202 when the system is in operation changes.

The power consumption when the automatic control system AS is operated tends to be high when the power is supplied to the input / output unit 202 included in the automatic control system AS. It tends to be low when the power supply to the output unit 202 is stopped. Further, the number of times the signal is transmitted to the input / output terminal depends on the controlled device 150 to which the signal is transmitted. That is, the input / output terminal to which the frequently operated controlled device 150 is assigned increases the number of times the signal is transmitted. Therefore, by inferring the allocation state in which the power consumption when the automatic control system AS is operated is low and the maximum value of the number of times of signal transmission is small for at least one input / output unit, the allocation information 400 is used. The power consumption of the allocated automatic control system AS is reduced, and the power supply to at least one input / output unit 202 of the plurality of input / output units 202 of the automatic control system AS allocated by the allocation information 400 is stopped. A control target in which the time increases and the maximum value of the number of times of transmission of signals transmitted to a plurality of input / output terminals of at least one input / output unit 202 of the automatic control system AS allocated by the allocation information 400 decreases. The allocation state of the device 150 can be inferred.

Embodiment 3.
The allocation support program, the allocation support device, the allocation learning program, and the allocation learning device according to the third embodiment of the present invention will be described with reference to the drawings. Also in the third embodiment, the allocation support program includes the allocation learning program, and the allocation support device will be described as having a function as the allocation learning device. The same configurations as those in the first and second embodiments described above will be described using the same reference numerals, and specific description thereof will be omitted. Further, in the third embodiment, similarly to the first embodiment described above, three input / output units 202 having an input / output device unit having three input / output terminals are provided, and each input / output terminal is a controlled device. An explanation will be given by taking as an example an automatic control system AS in which 150 is assigned and nine controlled devices 150 are provided. Hereinafter, configurations different from those of the first embodiment and the second embodiment will be specifically described.

FIG. 20 is a diagram schematically showing an example of a functional block of the allocation support device 1c according to the third embodiment. FIG. 21 is a diagram showing an example of a functional block of the allocation simulation unit according to the third embodiment. The allocation support device 1c according to the third embodiment has an allocation simulation unit 31 in addition to the configurations described in the first and second embodiments described above.

As shown in FIG. 21, the allocation simulation unit 31 has a function of creating a plurality of learning information based on one allocation information 400. The allocation simulation unit 31 includes a control information acquisition unit 32, an allocation calculation unit 33, and an information calculation unit 34. The functional blocks of the allocation simulation unit 31 will be described below.

The control information acquisition unit 32 acquires a control program for operating the automatic control system AS from the arithmetic unit 300. The control program acquired in the third embodiment is a ladder program for operating the automatic control system AS illustrated in the first and second embodiments described above.

The allocation calculation unit 33 creates a plurality of learning allocation information 400 based on one allocation information 400. The learning allocation information 400 is a term used to distinguish it from the above-mentioned allocation information 400, and the information structure is the same as that of the allocation information 400. Specifically, the input / output units 202A, 202B, 202C are associated with the input / output terminals X0, X1, X2, Y0, Y1, Y2, Z0, Z1, Z2, and the input / output terminals Xn, Yn, The correspondence between Zn and the device name Mn that specifies the device to be controlled 150 is different for each learning allocation information 400.

The information calculation unit 34 receives the learning time zone information 500 and the learning for each of the plurality of learning allocation information 400 created by the allocation calculation unit 33 based on the control program acquired by the control information acquisition unit 32. The number of times information 401 and the learning system power usage information 900 are calculated and created. The learning time zone information 500 and the learning system power usage information 900 are terms used to distinguish the time zone information 500, the unit power consumption information 600, and the system power usage information 900, and are used as each information. The information structure of the above is the same, and the information has different contents such as specific numerical values indicating each information.

Hereinafter, the arithmetic processing executed by the allocation simulation unit 31 according to the third embodiment will be described with reference to the drawings.

FIG. 22 is a flowchart showing the operation of the allocation simulation unit 31. The allocation simulation unit 31 acquires the control program stored in the arithmetic unit 300 (step S501). Then, the allocation simulation unit 31 accepts inputs of the allocation information 400, the unit power consumption information 600, and the device power consumption information 700 (step S502). The allocation simulation unit 31 creates the learning allocation information 400 based on the allocation information 400 that has received the input by using the allocation calculation unit 33 (step S503). The allocation simulation unit 31 uses the information calculation unit 34 to use the learning time zone information 500, the learning frequency information 401, and the learning system power based on the acquired control program and the created learning allocation information 400. Information 900, is created (step S504). The arithmetic processing in each step will be specifically described below.

In step S501, the control information acquisition unit 32 reads and acquires the control program stored in the storage unit 102 of the arithmetic unit 300.

In step S502, the allocation simulation unit 31 controls the allocation information 400, the unit power consumption information 600 of the input / output unit 202 included in the allocation information 400, and the device power consumption of the controlled device 150 included in the allocation information 400. Accepts the input of information 700.

In step S503, the allocation calculation unit 33 creates a plurality of learning allocation information 400 based on the allocation information 400 input in step S502. Specifically, the correspondence between the input / output unit 202 and the input / output terminal is not changed, but the arithmetic processing for changing the correspondence between the input / output terminal and the control target device 150 is performed, and each input / output terminal and the control target device are subjected to arithmetic processing. A plurality of learning allocation information 400s having different combinations with 150 are created. The created plurality of learning allocation information 400 is stored in the storage unit 102.

In step S504, the information calculation unit 34 reads one of the plurality of learning allocation information 400 created in step S503 from the storage unit 102, and constructs a simulated automatic control system AS based on the learning allocation information 400. do. The constructed simulated automatic control system AS is subjected to arithmetic processing for executing the control program acquired in step S501 for a predetermined time. The information calculation unit 34 acquires the learning time zone information 500 and the learning number of times information 401 as a result of the calculation process. The information calculation unit 34 calculates the learning system power usage information 900 based on the acquired learning time zone information 500, learning allocation information 400, unit power consumption information 600 and device power consumption information 700 input in S502. do. The calculation of the learning system power usage information 900 is performed by reading the function represented by the above-mentioned equations 1 and 2 stored in the storage unit 102 from the storage unit 102. The information calculation unit 34 acquires the learning system power usage information 900 as a result of the calculation processing. The acquired information for each learning is stored in the storage unit 102 together with the learning allocation information 400.

When step S504 is completed for one learning allocation information 400, the allocation simulation unit 31 performs step S504 for the other learning allocation information 400 created in step S503. Thereby, a plurality of learning first input information 800, 800b and a plurality of learning second input information 1000, 1000b can be created.

Regarding the generation of the trained model performed by the learning unit 21c according to the third embodiment, any one of the learning first input information 800 and 800b and the learning second input information 1000 and 1000b created by the simulation unit 31 described above. Can be used as training data to generate a trained model. Since the specific contents are the same as those in the first or second embodiment described above, the description thereof will be omitted. Also in the third embodiment, the learning unit 21c is a functional block representing the function of the allocation learning device 2c.

As described above, the allocation support program and the allocation support device 1c according to the third embodiment include the simulation unit 31, so that the first learning input information 800, 800b and the second learning unit input to the learning units 21 and 21b are provided. It becomes easy to prepare a large number of input information 1000 and 1000b, and the learning efficiency in the learning units 21 and 21b can be improved.

The configuration shown in the above-described embodiment shows an example of the content of the present invention, can be combined with another known technique, and is one of the configurations without departing from the gist of the present invention. It is also possible to omit or change the part.

In the above-described first to third embodiments, the allocation learning program is incorporated in the allocation support program, and the allocation support devices 1, 1b and 1c have the functions of the allocation learning devices 2, 2b and 2c. However, the allocation support program and the allocation learning program may be divided into two programs as separate programs. That is, a program including at least an information acquisition unit and an allocation information output unit may be an allocation support program, and a program including at least a learning unit may be an allocation learning program. Further, the separately created allocation support program and the allocation learning program may be stored in separate arithmetic unit readable storage media, or may be stored in one arithmetic unit readable storage medium.

Further, the separately created allocation support program and the allocation learning program may be installed in separate arithmetic units, and the allocation support device and the allocation learning device may be prepared separately. Further, the separately created allocation support program and allocation learning program may be installed in one arithmetic unit, and one arithmetic unit may function as an allocation support device and an allocation learning device.

In the above-described first to third embodiments, the automatic control system AS has been described as having three input / output units 202 and nine control target devices. However, the input / output unit 202 and the control target device are different from each other. It suffices to have multiple units.

In the first to third embodiments, the time zone information 500 and the system power usage information 900 show data when the day is operated for 24 hours from 0:00, but the operation is performed for a predetermined time. It suffices if it is the data when it is made. For example, one day may be operated for three days with 24 hours from 0:00, and the data for three days may be averaged as data for one day, or data for a predetermined time less than one day may be used. However, in order to obtain more effective review allocation information, it is preferable to use the data when the product has been operated for one day or more.

In the above-described second embodiment, the weighted information acquisition unit 27 has been described as a functional block of the learning unit 21b, but it is sufficient if the weighted information can be acquired when the learning unit 21b performs learning. For example, when the allocation learning program and the allocation support program are separately configured, a program that causes the arithmetic unit to function as the allocation information acquisition unit 27 may be incorporated into the allocation support program. In such a case, the learning data acquisition unit 23b of the learning unit 21b may acquire the weighted information input by the user as one of the learning data.

In the third embodiment described above, the allocation simulation unit 31 has been described as a functional block of the allocation support device 1c, but the allocation simulation unit 31 may be configured as a functional block of the learning units 21 and 21b.

Further, although the allocation simulation unit 31 has explained that the learning time zone information 500, the learning frequency information 401, and the learning system power usage information 900 are created, it is also possible that unnecessary information is not created as the learning data. good. For example, in the learning of the learning unit 21 according to the first embodiment described above, the learning count information 401 is not required. Therefore, in such a case, the learning count information 401 can be omitted.

AS automatic control system, 150 control target device, 200 PLC, 201 CPU unit, 202 input / output unit, 300 arithmetic unit, 1 1b 1c allocation support device, 11 information acquisition unit, 12 allocation information output unit, 101 arithmetic unit, 102 storage Unit, 103 input unit, 104 display unit, 105 communication interface, 22b allocation learning device, 20 20b inference unit, 21 21b learning unit, 22 22a preprocessing unit, 23 23b learning data acquisition unit, 24 24b model generation unit, 24R 24Rb Reward calculation unit, 24U 24Ub function update unit, 25 25b model storage unit, 26 26b calculation processing unit, 27 multiple information acquisition unit, 31 allocation simulation unit, 32 control information acquisition unit, 33 allocation calculation unit, 34 information calculation Department, 400 allocation information, 401 frequency information, 500 time zone information, 600 unit power consumption information, 700 equipment power consumption information, 800 800b first input information, 900 system power usage information, 1000 1000b second input information, 1100 1100b allocation Review information

Claims (19)

This is an allocation support program that supports the review of allocation in an automatic control system in which controlled devices are assigned to multiple input / output terminals of each of multiple input / output units.
Allocation information corresponding to the input / output terminal in the input / output unit and the control target device, time zone information indicating a time zone in which the input / output unit and the control target device are operating, a unit of the input / output unit. The power consumption information and the power consumption of the input / output unit and the control when the first input information including the device power consumption information of the control target device or the automatic control system allocated by the allocation information is operated. An information acquisition unit that acquires system power usage information including the power consumption of the automatic control system calculated based on the power consumption of the target device, and second input information including the allocation information.
The power consumption of the automatic control system allocated by the allocation information acquired from the inference unit that performs inference using the learned model by inputting the first input information or the second input information. Allocation information output unit that outputs the decreasing allocation information of the controlled device,
An allocation support program that makes the arithmetic unit function as. The allocation information is information that increases the time for stopping the power supply to at least one input / output unit of the plurality of input / output units of the automatic control system allocated by the allocation information.
The allocation support program according to claim 1. The first input information and the second input information further include number information indicating the number of times the signal transmitted to the input / output terminal is transmitted.
The allocation information is information in which the maximum value of the number of transmissions of signals transmitted to the plurality of input / output terminals of at least one input / output unit of the automatic control system allocated by the allocation information is reduced. ,
The allocation support program according to claim 2. It is an allocation learning program that learns to review the allocation in the automatic control system in which the control target device is assigned to the multiple input / output terminals of each of the multiple input / output units.
The allocation information corresponding to the input / output terminal and the control target device, the power consumption of the input / output unit when the automatic control system allocated by the allocation information is operated, and the power consumption of the control target device. The learning data acquisition unit that acquires the power consumption information including the power consumption of the automatic control system calculated based on the above and the learning data including the power consumption.
Using the learning data, a learned model that infers the allocation state of the controlled device in which the power consumption of the automatic control system allocated by the allocation information decreases from the allocation information and the power usage information. Model generator to generate,
To make the arithmetic unit function as
Allocation learning program. The model generation unit determines the allocation state of the control target device in which the time for stopping the power supply to at least one input / output unit of the plurality of input / output units of the automatic control system allocated by the allocation information increases. Generate the trained model to infer,
The allocation learning program according to claim 4. The learning data further includes count information indicating the number of transmissions of the signal transmitted to the input / output terminal.
The model generation unit uses the learning data to transmit signals to the plurality of input / output terminals of at least one input / output unit of the automatic control system allocated by the allocation information. Generate the trained model that infers the allocation state of the controlled device whose maximum value decreases.
The allocation learning program according to claim 5. The model generation unit, compensation calculation unit for calculating a first compensation based on the difference between the first predetermined threshold and the power usage, and on the basis of the first compensation, assignment before Symbol split with information A function update unit for updating a function for inferring the allocation state of the controlled device for which the power consumption of the automatic control system is reduced is included.
The allocation learning program according to claim 4. Based on the first reward, the function update unit increases the time to stop the power supply to at least one input / output unit of the plurality of input / output units of the automatic control system allocated by the allocation information. The allocation learning program according to claim 7, wherein the function for inferring the allocation state of the controlled device is updated. The model generation unit calculates a first reward based on the difference between the power consumption and a predetermined first threshold value, and transmits the first reward to the plurality of input / output terminals in each of the plurality of input / output units. The second reward is calculated based on the difference between the total number of times the maximum value of the signal transmissions is totaled and the predetermined second threshold value, and the final reward is calculated based on the first reward and the second reward. The remuneration calculation unit to calculate, the power consumption of the automatic control system allocated by the allocation information is reduced based on the final remuneration, and the plurality of inputs / outputs of the automatic control system allocated by the allocation information. The time to stop the power supply to at least one input / output unit among the units is increased, and the signal is transmitted to the plurality of input / output terminals of at least one input / output unit of the automatic control system allocated by the allocation information. A function update unit for updating a function for inferring the allocation state of the controlled device, which reduces the maximum value of the number of times the signal is transmitted, is included.
The allocation learning program according to claim 5. The reward calculation unit multiplies each of the first reward and the second reward by a coefficient for weighting the first reward and the second reward, and then calculates the first reward and the second reward. The allocation learning program according to claim 9, wherein the final reward is calculated by adding them. Further includes an allocation simulation unit that calculates the learning data used for generating the trained model based on the allocation information.
The allocation learning program according to any one of claims 4 to 10. An arithmetic unit readable storage medium that stores the allocation support program according to any one of claims 1 to 3. An arithmetic unit readable storage medium that stores the allocation learning program according to any one of claims 4 to 11. It is an allocation support device that supports the review of allocation in an automatic control system in which a device to be controlled is assigned to a plurality of input / output terminals of each of a plurality of input / output units.
Allocation information corresponding to the input / output terminal in the input / output unit and the control target device, time zone information indicating a time zone in which the input / output unit and the control target device are operating, a unit of the input / output unit. The power consumption information and the power consumption of the input / output unit and the control when the first input information including the device power consumption information of the control target device or the automatic control system allocated by the allocation information is operated. An information acquisition unit that acquires system power usage information including the power consumption of the automatic control system calculated based on the power consumption of the target device, and second input information including the allocation information.
The power consumption of the automatic control system allocated by the allocation information acquired from the inference unit that performs inference using the learned model by inputting the first input information or the second input information. The allocation information output unit that outputs the allocation information of the controlled device,
Allocation support device equipped with. The allocation information is information that increases the time for stopping the power supply to at least one input / output unit of the plurality of input / output units of the automatic control system allocated by the allocation information.
The allocation support device according to claim 14. The first input information and the second input information further include number information indicating the number of times the signal transmitted to the input / output terminal is transmitted.
The allocation information is information in which the maximum value of the number of transmissions of signals transmitted to the plurality of input / output terminals of at least one input / output unit of the automatic control system allocated by the allocation information is reduced. ,
The allocation support device according to claim 15. It is an allocation learning device that learns to review the allocation in an automatic control system in which a device to be controlled is assigned to a plurality of input / output terminals of each of a plurality of input / output units.
The allocation information corresponding to the input / output terminal and the control target device, the power consumption of the input / output unit when the automatic control system allocated by the allocation information is operated, and the power consumption of the control target device. A learning data acquisition unit that acquires power usage information including the power usage amount of the automatic control system calculated based on the above, and learning data including the power usage amount of the automatic control system.
Using the learning data, a learned model that infers the allocation state of the controlled device in which the power consumption of the automatic control system allocated by the allocation information decreases from the allocation information and the power usage information. The model generator to generate and
Allocation learning device equipped with. The model generation unit determines the allocation state of the control target device in which the time for stopping the power supply to at least one input / output unit of the plurality of input / output units of the automatic control system allocated by the allocation information increases. Generate the trained model to infer,
The allocation learning device according to claim 17. The learning data further includes count information indicating the number of transmissions of the signal transmitted to the input / output terminal.
The model generation unit uses the learning data to transmit signals to the plurality of input / output terminals of at least one input / output unit of the automatic control system allocated by the allocation information. Generate the trained model that infers the allocation state of the controlled device whose maximum value decreases.
The allocation learning device according to claim 18.

Images