Control device and control method

Abstract

A control device (1) comprises: an input / output interface (13) through which data is inputted / outputted; and a processor (11). The processor (11) acquires air-conditioning-related data which is related to execution of a defrosting operation by an air conditioner (2) through the input / output interface (13), and determines, by using at least one trained model, whether or not to prohibit execution of a final defrosting operation scheduled during an operation period of the air conditioner (2), on the basis of the air-conditioning-related data acquired through the input / output interface (13).

Description

[Technical field]

[0001] The present disclosure relates to a control device and a control method for controlling the execution of a defrosting operation by an air conditioner to remove frost from at least one heat exchanger. [Background technology]

[0002] Conventionally, it is known that an air conditioner can perform a defrosting operation to remove frost from a heat exchanger. For example, Japanese Patent Application Laid-Open No. 2017-198430 (Patent Document 1) discloses an air conditioning system configured to perform a defrosting operation when an execution condition is met. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2017-198430 A Summary of the Invention [Problem to be solved by the invention]

[0004] As in the air conditioning system disclosed in JP 2017-198430 A, by performing a defrosting operation, it is possible to avoid a decrease in the heat exchange efficiency in the heating operation performed after the defrosting operation. However, the timing at which the conditions for performing the defrosting operation are met differs depending on the operating state of the air conditioner or the environment in which the air conditioner is placed, so the timing at which the defrosting operation is performed varies. For this reason, when a final defrosting operation (hereinafter also referred to as a "final defrosting operation") is performed during the operation period of the air conditioner, the air conditioner may stop during the execution of the final defrosting operation, or the heating operation after the final defrosting operation may only be performed for a very short period of time despite the execution of the final defrosting operation. In such cases, the power consumption due to the final defrosting operation is wasted, and since the heating operation is not performed during the defrosting operation, the user may feel uncomfortable during the final defrosting operation.

[0005] The present disclosure has been made to solve the above-mentioned problems, and aims to provide a technology that can avoid wasteful power consumption and user discomfort that arise when a final defrosting operation is performed. [Means for solving the problem]

[0006] A control device according to the present disclosure is a control device that controls the execution of a defrosting operation by an air conditioner to remove frost from at least one heat exchanger. The control device includes an input / output interface that inputs and outputs data, and a processor. The air conditioner executes a defrosting operation during an operation period and then executes a heating operation. The processor acquires air conditioning-related data related to the execution of the defrosting operation by the air conditioner from the input / output interface, and uses at least one trained model to determine whether or not to prohibit the execution of a final defrosting operation scheduled during the operation period of the air conditioner based on the air conditioning-related data acquired from the input / output interface.

[0007] A control method according to the present disclosure is a control method in which a control device controls the execution of a defrosting operation by an air conditioner to remove frost from at least one heat exchanger. The air conditioner executes a heating operation after executing a defrosting operation during an operation period. The control method includes a step of acquiring air conditioning-related data related to the execution of the defrosting operation by the air conditioner, and a step of using at least one trained model and based on the acquired air conditioning-related data, determining whether or not to prohibit the execution of a final defrosting operation scheduled during the operation period of the air conditioner. Effect of the Invention

[0008] According to the present disclosure, it is possible to determine whether or not to prohibit the execution of a final defrosting operation scheduled during the operation period of the air conditioner based on air conditioning-related data regarding the execution of a defrosting operation by the air conditioner, thereby avoiding unnecessary power consumption and user discomfort caused by the execution of a final defrosting operation. [Brief description of the drawings]

[0009] [Figure 1] 1 is a diagram showing a configuration of an air conditioning system according to a first embodiment. [Diagram 2] 1 is a diagram showing the configuration of an air conditioner according to a first embodiment. [Diagram 3] 5 is a diagram showing an example of the execution timing of a defrosting operation during an operation period of the air conditioner according to the first embodiment. FIG. [Figure 4] FIG. 2 is a diagram showing a configuration of a control device according to the first embodiment. [Diagram 5] FIG. 2 is a diagram showing a functional configuration of the control device in the utilization phase according to the first embodiment. [Figure 6] FIG. 2 is a diagram for explaining input and output in the control device according to the first embodiment. [Figure 7] 4 is a flowchart showing a process executed by the control device according to the first embodiment in a utilization phase. [Figure 8] FIG. 2 is a diagram for explaining unsupervised learning in a first trained model according to the first embodiment. [Figure 9] FIG. 2 is a diagram for explaining the input and output of unsupervised learning in the first trained model in the first embodiment. [Figure 10] FIG. 2 is a diagram illustrating a functional configuration for generating a first trained model of the learning device according to the first embodiment. [Figure 11] FIG. 1 is a diagram illustrating a configuration of a neural network. [Figure 12] 4 is a flowchart showing processing executed by the learning device according to the first embodiment in a first learning phase. [Figure 13] FIG. 1 is a diagram for explaining an overview of reinforcement learning. [Figure 14] FIG. 11 is a diagram for explaining reinforcement learning in the second trained model according to the first embodiment. [Figure 15] FIG. 11 is a diagram for explaining the input and output of reinforcement learning in the second trained model in the first embodiment. [Figure 16]FIG. 11 is a diagram for explaining an example of reward calculation for reinforcement learning in the second trained model according to embodiment 1. [Figure 17] FIG. 1 is a diagram illustrating a functional configuration for generating a second trained model of a learning device according to embodiment 1. [Figure 18] 6 is a flowchart showing processing executed by the learning device according to the first embodiment in a second learning phase. [Figure 19] FIG. 11 is a diagram showing a functional configuration of a control device according to a second embodiment in a utilization phase. [Figure 20] FIG. 11 is a diagram for explaining input and output in a control device according to a second embodiment. [Figure 21] 11 is a flowchart showing a process executed by a control device according to a second embodiment in a utilization phase. [Figure 22] FIG. 11 is a diagram for explaining supervised learning in a trained model according to the second embodiment. [Figure 23] FIG. 11 is a diagram for explaining the input and output of supervised learning in a trained model according to embodiment 2. [Figure 24] FIG. 11 is a diagram showing an example of the execution timing of a defrosting operation during an operation period of an air conditioner according to a modified example. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Hereinafter, the embodiments of the present disclosure will be described in detail with reference to the drawings. Although several embodiments will be described below, it is planned from the beginning of the application that the configurations described in each embodiment will be appropriately combined. Note that the same or corresponding parts in the drawings are given the same reference numerals, and the description thereof will not be repeated.

[0011] Embodiment 1 [Air conditioning system configuration] Fig. 1 is a diagram showing the configuration of an air conditioning system 1000 according to Embodiment 1. As shown in Fig. 1, the air conditioning system 1000 includes an air conditioner 2 and a control device 1 that controls the air conditioner 2.

[0012] The air conditioner 2 includes an outdoor unit 3 and at least one indoor unit 4. In the air conditioning system 1000, a plurality of indoor units 4 are installed to cool or heat a space to be air-conditioned. In the first embodiment, the air conditioning system 1000 is configured so that a plurality of indoor units 4 are connected to one outdoor unit 3, but it may be configured so that a single indoor unit 4 is connected to a single outdoor unit 3.

[0013] [Air conditioner configuration] Fig. 2 is a diagram showing the configuration of the air conditioner 2 according to embodiment 1. Note that Fig. 2 functionally shows the connection relationships and arrangement of the devices in the air conditioner 2, and does not necessarily show the physical spatial arrangement.

[0014] 2, the air conditioner 2 includes an outdoor unit 3, an indoor unit 4, and an air conditioning device 200. The indoor unit 4 is configured to be connectable to the outdoor unit 3 via an extension pipe 53 and an extension pipe 56.

[0015] The outdoor unit 3 includes a four-way valve 40, a compressor 25, an outdoor heat exchanger 21, a blower 22, and an expansion valve .

[0016] A connection port 41 of the four-way valve 40 is connected to a suction port 251 of the compressor 25 via a pipe 60. A connection port 42 of the four-way valve 40 is connected to the indoor unit 4 via a pipe 52 and an extension pipe 53. A connection port 43 of the four-way valve 40 is connected to a discharge port 252 of the compressor 25 via a pipe 51. A connection port 44 of the four-way valve 40 is connected to one end side of the outdoor heat exchanger 21 via a pipe 59. The four-way valve 40 is configured to switch an internal communication state according to a control signal from the air conditioner 200.

[0017] The other end of the outdoor heat exchanger 21 is connected to one end of the expansion valve 26 via a pipe 58. The other end of the expansion valve 26 is connected to the indoor unit 4 via a pipe 57 and an extension pipe 56.

[0018] Compressor 25 is configured to operate and stop, and further to change its rotation speed during operation, according to a control signal from air conditioner 200. Air conditioner 200 outputs a control signal to compressor 25 to arbitrarily change the drive frequency of compressor 25. Compressor 25 changes the number of rotations per unit time, i.e., the rotation speed, in response to the change in drive frequency, thereby changing the amount of refrigerant discharged.

[0019] The outdoor heat exchanger 21 exchanges heat between the refrigerant and air drawn in from the outside by a blower 22, that is, outside air.

[0020] The expansion valve 26 is, for example, an electronic expansion valve whose opening is adjusted in accordance with a control signal from the air conditioner 200. The expansion valve 26 reduces the pressure of the refrigerant that has flowed in, and allows the refrigerant obtained by the reduced pressure to flow out.

[0021] The indoor unit 4 includes an indoor heat exchanger 31 and a blower 32. One end of the indoor heat exchanger 31 is connected to the outdoor unit 3 via a pipe 54 and an extension pipe 53. The other end of the indoor heat exchanger 31 is connected to the outdoor unit 3 via a pipe 55 and an extension pipe 56.

[0022] The indoor heat exchanger 31 exchanges heat between the air drawn from the room by the blower 32 and the refrigerant.

[0023] The air conditioner 200 is mounted on the outdoor unit 3. The air conditioner 200 controls each actuator of the compressor 25, the expansion valve 26, the blower 32, the blower 22, the four-way valve 40, etc. The air conditioner 200 may be separate from the outdoor unit 3.

[0024] Furthermore, the air conditioner 2 is provided, in the outdoor unit 3, with a temperature sensor 70 for measuring the temperature T0 of the outdoor heat exchanger 21 (hereinafter also referred to as the "heat exchanger temperature"), a temperature sensor 71 for measuring the temperature outside the air-conditioned space (hereinafter also referred to as the "outdoor air temperature") T1, and a humidity sensor 81 for measuring the humidity outside the air-conditioned space (hereinafter also referred to as the "outdoor air humidity") H1. The air conditioner 2 is provided, in the indoor unit 4, with a temperature sensor 72 for measuring the temperature T2 inside the air-conditioned space (hereinafter also referred to as the "indoor temperature"), and a humidity sensor 82 for measuring the humidity H2 inside the air-conditioned space (hereinafter also referred to as the "indoor humidity"). The heat exchanger temperature T0, the outdoor air temperature T1, the indoor temperature T2, the outdoor air humidity H1, and the indoor humidity H2 measured by each sensor are output to the air conditioner 200.

[0025] The air conditioner 2 configured in this manner is capable of performing a heating operation for heating the space to be air-conditioned. For example, in the heating operation, the internal communication state of the four-way valve 40 is such that the connection port 41 communicates with the connection port 44, and the connection port 42 communicates with the connection port 43, as shown by the solid lines in Fig. 1. In other words, in the heating operation, the intake port 251 of the compressor 25 communicates with the outdoor heat exchanger 21, and the discharge port 252 of the compressor 25 communicates with the indoor unit 4.

[0026] The compressor 25 draws in the low-temperature, low-pressure gas refrigerant that has flowed in from the outdoor heat exchanger 21, and compresses the drawn gas refrigerant to increase the pressure of the gas refrigerant. The compressor 25 discharges the high-temperature, high-pressure gas refrigerant obtained by the compression to the indoor heat exchanger 31.

[0027] In heating operation, the indoor heat exchanger 31 functions as a condenser. The indoor heat exchanger 31 exchanges heat between the high-temperature, high-pressure gas refrigerant from the compressor 25 and the air sucked from the space to be air-conditioned by the blower 32. The gas refrigerant that has released heat to the air through this heat exchange condenses inside the indoor heat exchanger 31 and changes into a high-temperature, high-pressure liquid refrigerant. The high-temperature, high-pressure liquid refrigerant obtained by the indoor heat exchanger 31 flows out to the expansion valve 26. The air that has absorbed heat from the gas refrigerant in the indoor heat exchanger 31 is sent back into the space to be air-conditioned. This heats up the space to be air-conditioned.

[0028] The expansion valve 26 reduces the pressure of the high-temperature, high-pressure liquid refrigerant from the indoor heat exchanger 31. The low-temperature, low-pressure gas-liquid two-phase refrigerant obtained by the expansion valve 26 flows out to the outdoor heat exchanger 21.

[0029] In heating operation, the outdoor heat exchanger 21 functions as an evaporator. The outdoor heat exchanger 21 exchanges heat between the low-temperature, low-pressure two-phase gas-liquid refrigerant from the expansion valve 26 and air drawn in from outside the space to be air-conditioned by the blower 22. The two-phase gas-liquid refrigerant that absorbs heat from the air through this heat exchange evaporates inside the outdoor heat exchanger 21 and changes into a low-temperature, low-pressure gas refrigerant. The low-temperature, low-pressure gas refrigerant obtained by the outdoor heat exchanger 21 flows out to the compressor 25.

[0030] In this manner, in the heating operation, the refrigerant circulates through the compressor 25, the indoor heat exchanger 31 (condenser), the expansion valve 26, and the outdoor heat exchanger 21 (evaporator) in that order.

[0031] For example, in an environment where the outdoor air temperature outside the air-conditioned space in which the outdoor unit 3 is installed is extremely low, the temperature of the outdoor heat exchanger 21 drops, and water vapor in the air condenses on the outdoor heat exchanger 21, which may result in the formation of frost. When frost forms on the outdoor heat exchanger 21, it becomes difficult to send warm air to the air-conditioned space even when the heating operation is being performed, and the heat exchange efficiency during the heating operation decreases. Therefore, the air conditioner 2 is configured to perform a defrosting operation to remove frost from the outdoor heat exchanger 21 before performing the heating operation. The air conditioner 2 according to the first embodiment further includes a heater 5 near the outdoor heat exchanger 21 in the outdoor unit 3, and when the conditions for performing the defrosting operation are met, the heater 5 is operated to remove frost from the outdoor heat exchanger 21 using the heat generated by the heater 5.

[0032] The heater 5 is operated or stopped based on the control of the air conditioner 200. Specifically, the air conditioner 200 checks the heat exchanger temperature T0 acquired from the temperature sensor 70 periodically (for example, every hour), and when the heat exchanger temperature T0 becomes equal to or lower than a predetermined threshold (temperature), outputs a control signal C to the heater 5 to operate the heater 5. This allows the air conditioner 2 to remove frost from the outdoor heat exchanger 21 using heat generated by the heater 5. Alternatively, the air conditioner 200 may check the heat exchanger temperature T0 acquired from the temperature sensor 70 and the outdoor air temperature T1 acquired from the temperature sensor 71 periodically (for example, every hour), and when the difference between the heat exchanger temperature T0 and the outdoor air temperature T1 becomes equal to or higher than a predetermined threshold, output the control signal C to the heater 5 to operate the heater 5. Note that the air conditioner 200 may perform a defrosting operation to remove frost from the outdoor heat exchanger 21 by a known method other than operating the heater 5.

[0033] [Defrosting and heating operation timing] 3 is a diagram showing an example of the execution timing of a defrosting operation during the operation period of the air conditioner 2 according to embodiment 1. The air conditioner 2 repeatedly executes the heating operation and the defrosting operation during the operation period from the start of a series of heating controls for the heating operation to the end of the series of heating controls.

[0034] For example, in the example of FIG. 3(A), the air conditioner 2 starts a heating operation at a timing t1, and then periodically (for example, every hour) during the heating operation determines whether or not the execution condition for the defrosting operation is satisfied. When the execution condition for the defrosting operation is satisfied at a timing t2, the air conditioner 2 starts the defrosting operation, and restarts the heating operation at a timing t3 when the defrosting operation ends. After that, the air conditioner 2 periodically determines whether or not the execution condition for the defrosting operation is satisfied during the heating operation, and when the execution condition for the defrosting operation is satisfied at a timing t4, the air conditioner 2 starts the defrosting operation, and restarts the heating operation at a timing t5 when the defrosting operation ends. After that, the air conditioner 2 periodically determines whether or not the execution condition for the defrosting operation is satisfied during the heating operation, and when the execution condition for the defrosting operation is satisfied at a timing t6, the air conditioner 2 starts the defrosting operation, and restarts the heating operation at a timing t7 when the defrosting operation ends. After that, the air conditioner 2 ends a series of heating controls for the heating operation at a timing t8. In this case, the defrosting operation performed during the period from timing t6 to t7 is the “final defrosting operation,” and the heating operation performed during the period from timing t7 to t8 is the “final heating operation.” Also, the period from timing t1 to t8 is the “operation period.”

[0035] In this way, the air conditioner 2 can avoid a decrease in heat exchange efficiency during heating operation by performing a heating operation after performing a defrosting operation during the operation period. However, the timing at which the conditions for performing the defrosting operation are met varies depending on the operation state of the air conditioner 2 or the environment in which the air conditioner 2 is placed, such as the heat load, the date and time, the day of the week, the presence or absence of holidays, the temperature, and the humidity. That is, the operation period of the air conditioner 2 varies depending on the environment of the property in which it is installed, and further, the execution timing of the defrosting operation and the heating operation during the operation period varies. For this reason, there are cases in which the air conditioner 2 stops during the execution of the final defrosting operation, or in which the heating operation after the final defrosting operation is executed is executed only for a very short period of time, even though the final defrosting operation has been executed.

[0036] For example, when the air conditioner 2 is installed in an office building, it operates from 8:00 to 20:00 on weekdays from Monday to Friday, from 8:00 to 17:00 on weekends on Saturday and Sunday, and does not operate on holidays. In this way, the tendency of the operation period differs depending on the date, the day of the week, and the presence or absence of holidays. In addition, the operation period is not necessarily fixed as described above, and the tendency of the operation period may differ depending on values ​​that vary depending on the environment, such as the outdoor air temperature T1, the indoor temperature T2, the outdoor air humidity H1, the indoor humidity H2, and the power consumption of the air conditioner 2 during the operation period. In this way, even though the operation period varies depending on the environment in which the air conditioner 2 is placed, the air conditioner 2 may perform a final defrosting operation at the end of the operation period because it determines whether or not to perform a defrosting operation based on the heat exchanger temperature T0 or the difference between the heat exchanger temperature T0 and the outdoor air temperature T1. As a result, in the example of Figure 3 (A), the air conditioner 2 may stop and the operation period may end midway through the period from timing t6 to t7, which is the period during which the final defrosting operation is performed, or the period from timing t7 to t8, which is the period during which the final heating operation is performed, may become extremely short.

[0037] In such a case, the power consumption due to the final defrosting operation is wasted, and since the heating operation is not performed during the defrosting operation, the user experiences discomfort during the final defrosting operation. Therefore, in the air conditioning system 1000 according to the first embodiment, the control device 1 controls the execution of the defrosting operation by the air conditioner 2, thereby making it possible to avoid the wasteful power consumption and the user's discomfort caused by the execution of the final defrosting operation.

[0038] Specifically, the control device 1 acquires air conditioning-related data related to the execution of a defrosting operation by the air conditioner 2, and uses AI (Artificial Intelligence) to infer the execution time of the final defrosting operation and the execution time of the final heating operation scheduled after the final defrosting operation based on the air conditioning-related data. Furthermore, the control device 1 uses AI to infer the power consumption when the final defrosting operation and the final heating operation are executed, that is, the power consumption when the final defrosting operation and the final heating operation are executed in the period from timing t6 to t8 shown in FIG. 3(A). Furthermore, the control device 1 uses AI to infer the power consumption when the final defrosting operation is not executed, that is, the power consumption when the heating operation started at timing t5 is continued until timing t8 and only the heating operation (final heating operation) is executed in the period from timing t6 to t8 as shown in FIG. 3(B). The control device 1 compares the power consumption of the air conditioner 2 during the period from timing t6 to t8 when the final defrosting operation is performed with when the final defrosting operation is not performed, and based on the comparison result, decides whether or not to prohibit the air conditioner 2 from performing the final defrosting operation.

[0039] When the control device 1 decides to prohibit the execution of the final defrosting operation, it outputs a prohibition signal for prohibiting the execution of the final defrosting operation to the air conditioner 200 that controls the air conditioner 2. As a result, the air conditioner 2 is controlled based on the prohibition signal from the control device 1 so as not to execute the final defrosting operation.

[0040] [Control device configuration] 4 is a diagram showing the configuration of the control device 1 according to embodiment 1. As shown in FIG. 4, the control device 1 includes a processor 11, a storage device 12, and an input / output interface 13.

[0041] The processor 11 is a computing entity (computer) that executes various processes by executing various programs. The processor 11 is composed of at least one of a central processing unit (CPU), a field-programmable gate array (FPGA), and a graphics processing unit (GPU), for example. The processor 11 may also be composed of a processing circuitry such as an application specific integrated circuit (ASIC). The processor 11 may also include a volatile memory such as a dynamic random access memory (DRAM) and a static random access memory (SRAM) that temporarily stores program code or a work memory when executing various programs.

[0042] The storage device 12 includes a non-volatile memory such as a ROM (Read Only Memory) and a flash memory that stores various data required for the processor 11 to execute various programs. The storage device 12 may be a solid state drive (SSD) or a hard disk drive (HDD).

[0043] The input / output interface 13 receives input of air conditioning-related data regarding the execution of a defrosting operation by the air conditioner 2. The input / output interface 13 is also capable of outputting to the air conditioner 2 a prohibition signal for prohibiting the execution of a final defrosting operation.

[0044] The processor 11 includes, as its main functional components, a first inference unit 111, a second inference unit 112, a first model generation unit 113, a second model generation unit 114, a data acquisition unit 115, and an output unit .

[0045] The first inference unit 111 uses a first trained model 121 described later to infer final operation data for identifying the execution time of a final defrosting operation and the execution time of a final heating operation scheduled after the final defrosting operation, based on the air conditioning-related data. Details of the final operation data will be described later with reference to FIG. 6. The second inference unit 112 uses a second trained model 122 described later to determine whether or not to prohibit the execution of a final defrosting operation, based on the final operation data inferred by the first inference unit 111.

[0046] The first model generation unit 113 generates a first trained model 121 for inferring the final operation data based on the air conditioning-related data by performing unsupervised learning on the relationship between the air conditioning-related data and the final operation data. The second model generation unit 114 generates a second trained model 122 for determining whether or not to prohibit the execution of the final defrosting operation based on the final operation data by performing reinforcement learning using a reward based on the power consumption when the final defrosting operation and the final heating operation are performed according to the final operation data.

[0047] The data acquisition unit 115 acquires air conditioning-related data from the air conditioner 2 via the input / output interface 13. The output unit 116 outputs a prohibition signal via the input / output interface 13 to the air conditioner 2 to prohibit the execution of the final defrosting operation.

[0048] [Utilization phase] An application example of the control device 1 in the utilization phase will be described with reference to Fig. 5 to Fig. 7. Fig. 5 is a diagram showing a functional configuration of the control device 1 according to the first embodiment in the utilization phase.

[0049] As shown in FIG. 5, the data acquisition unit 115 acquires air conditioning-related data as an input 1. The first inference unit 111 infers final operation data as an output 1 based on the input 1 using the first learned model 121 stored by the first learned model storage unit 1221. The second inference unit 112 acquires the output 1 of the first inference unit 111 as an input 2 using the second learned model 122 stored by the second learned model storage unit 1222, and infers a decision result of whether or not to prohibit the execution of the final defrosting operation as an output 2 based on the input 2. The output unit 116 outputs a prohibition signal to the air conditioner 2 based on the output 2 of the second inference unit 112. The first learned model storage unit 1221 and the second learned model storage unit 1222 correspond to the functional units of the storage device 12.

[0050] FIG. 6 is a diagram for explaining input and output in the control device 1 according to the first embodiment.

[0051] As shown in Fig. 6, in the control device 1, air conditioning-related data is used as input 1 of the first inference unit 111. The air conditioning-related data is data related to the execution of defrosting operation by the air conditioner 2, and specifically includes at least one of the operation history of the air conditioner 2 (for example, the history of heating operation and defrosting operation), the transition of the heat load in the space to be air conditioned, the date and time, the day of the week, the presence or absence of holidays, the heat exchanger temperature T0, the outdoor temperature T1, the room temperature T2, the outdoor humidity H1, the room humidity H2, and the power consumption of the air conditioner 2 during the operation period. Note that the air conditioning-related data of input 1 may include all of the multiple parameters described above, or may include some of the multiple parameters.

[0052] In the control device 1, the final operation data is used as the output 1 of the first inference unit 111, i.e., the input 2 of the second inference unit 112. The final operation data is data for specifying the execution time of the final defrosting operation and the execution time of the final heating operation, and specifically includes the start time of the final defrosting operation, the end time of the final defrosting operation and the start time of the final heating operation, and the end time of the final heating operation.

[0053] In the control device 1, the result of whether or not to prohibit the execution of the final defrosting operation is used as the output 2 of the second inference unit 112.

[0054] Fig. 7 is a flowchart relating to the processing executed by the control device 1 according to the first embodiment in the utilization phase. The processing shown in Fig. 7 is executed by the processor 11 of the control device 1. In Fig. 7, "S" is used as an abbreviation for "STEP."

[0055] 7, the control device 1 acquires air-conditioning-related data as input 1 from the air conditioner 200 of the air conditioner 2 by the data acquisition unit 115 (S1). The control device 1 inputs the acquired input 1 to the first trained model 121 (S2). The control device 1 uses the first trained model 121 to infer final operation data as output 1 based on the air-conditioning-related data that is input 1 (S3).

[0056] The control device 1 inputs the inferred output 1 as input 2 to the second trained model 122 (S4). The control device 1 uses the second trained model 122 to determine, based on the final operation data that is input 2, whether or not to prohibit the execution of the final defrosting operation as output 2 (S5).

[0057] When the control device 1 determines to prohibit the execution of the final defrosting operation (YES in S6), it outputs a prohibition signal to the air conditioner 2 (S7) and ends this process. As a result, the air conditioner 2 is controlled not to execute the final defrosting operation.

[0058] When the control device 1 determines not to prohibit the execution of the final defrosting operation (NO in S6), it ends this process without outputting a prohibition signal to the air conditioner 2. As a result, the air conditioner 2 is controlled to execute the final defrosting operation.

[0059] [First learning phase] An application example of the control device 1 in the first learning phase for generating the first trained model 121 will be described with reference to Figs. 8 to 12. The first trained model 121 is generated by performing unsupervised learning on the relationship between the air conditioning-related data and the final operation data. Unsupervised learning is a method of learning features in the first training data 131 using the first training data 131 that does not include a result (label). An example of an unsupervised learning method is a grouping method such as the K-means method, but the first model generation unit 113 may use any known unsupervised learning method as long as it is a method that can output the final operation data as the output 1 from the air conditioning-related data as the input 1.

[0060] Fig. 8 is a diagram for explaining unsupervised learning in the first trained model 121 according to embodiment 1. As shown in Fig. 8, in the first learning phase, the control device 1 (first model generation unit 113) executes the first learning program 141 to generate (update) the first trained model 121 based on the first training data 131 including the input 1. As a result, in the utilization phase, the control device 1 can obtain the output 1 based on the input 1 using the first trained model 121.

[0061] 9 is a diagram for explaining the input and output of unsupervised learning in the first trained model 121 according to embodiment 1. As shown in Fig. 9, in the first learning phase, air conditioning-related data is used as input 1, and final operation data is used as output 1.

[0062] The operating history of the air conditioner 2, the changes in the heat load in the air conditioned space, the date and time, day of the week, whether or not there are holidays, the heat exchanger temperature T0, the outdoor temperature T1, the indoor temperature T2, the outdoor humidity H1, the indoor humidity H2, and the power consumption of the air conditioner 2 during the operating period, which are contained in the air conditioning-related data, are all correlated with the start time of the final defrosting operation, the end time of the final defrosting operation and the start time of the final heating operation, and the end time of the final heating operation, which are contained in the final operation data.

[0063] 10 is a diagram showing a functional configuration for generating a first trained model 121 of a learning device according to embodiment 1. The learning device is realized by a processor 11 of a control device 1. The learning device is capable of transferring data to and from a first learning program storage unit 1231 and a first trained model storage unit 1221. The first learning program storage unit 1231 and the first trained model storage unit 1221 are realized by a storage device 12 of the control device 1.

[0064] As shown in Figure 10, the learning device generates a first trained model 121 based on first training data 131 including input 1 by executing a first training program 141 stored in a first training program memory unit 1231.

[0065] Specifically, the data acquiring unit 115 acquires first learning data 131 including air conditioning-related data as input 1. The first model generating unit 113 generates a first trained model 121 for inferring final operation data based on the air conditioning-related data, using the first learning data 131 acquired by the data acquiring unit 115. The first model generating unit 113 stores the generated first trained model 121 in the first trained model storage unit 1221.

[0066] 11 is a diagram illustrating a configuration of a neural network. The first model generation unit 113 generates the first trained model 121 by unsupervised learning according to, for example, a neural network model.

[0067] A neural network is composed of an input layer consisting of multiple neurons, an intermediate layer (hidden layer) consisting of multiple neurons, and an output layer consisting of multiple neurons. The intermediate layer may be one layer, or two or more layers.

[0068] A three-layered neural network is shown in Fig. 11. A configuration with three inputs and three outputs is shown in Fig. 11. When multiple inputs are input to the input layers X1, X2, and X3, the values ​​multiplied by weights w11 to w16 are input to the intermediate layers Y1 and Y2, and the results are further multiplied by weights w21 to w26 and output from the output layers Z1, Z2, and Z3. This output result changes depending on the values ​​of the weights w11 to w16, and w21 to w26.

[0069] In the first learning phase, the neural network performs unsupervised learning to obtain output 1 from input 1 by adjusting weights based on first learning data 131 including input 1 acquired by data acquisition unit 115.

[0070] Fig. 12 is a flowchart of the process executed by the learning device according to the first embodiment in the first learning phase. The process shown in Fig. 12 is executed by the processor 11 of the control device 1 corresponding to the learning device. In Fig. 12, "S" is used as an abbreviation for "STEP."

[0071] 12, the learning device acquires first learning data 131 including an input 1 by the data acquisition unit 115 (S11). The learning device generates a first trained model 121 by performing unsupervised learning based on the first learning data 131 by the first model generation unit 113 (S12). The learning device stores the generated first trained model 121 in the first trained model storage unit 1221 (S13) and ends this process.

[0072] [Second learning phase] 13 to 18, an application example of the control device 1 in the second learning phase for generating the second trained model 122 will be described. The second trained model 122 is generated by performing reinforcement learning using a reward based on the power consumption when the final defrosting operation and the final heating operation are performed according to the final operation data.

[0073] FIG. 13 is a diagram for explaining an overview of reinforcement learning. As shown in FIG. 13, in reinforcement learning, an agent (acting subject) in a certain environment observes the current state (environmental parameters) and determines the action to be taken based on a policy. A policy is a rule for determining an action, and optimizing the policy optimizes the selection of an action. The environment changes dynamically due to the agent's actions, and the agent is given a reward based on a reward criterion according to the change in the environment. The agent repeats such actions, and learns the action that will obtain the most reward through a series of actions. In FIG. 13, the current state is represented by B2 (state), the agent's action is represented by B1 (action), and the reward criterion is represented by D (reward criterion).

[0074] Q-learning and TD-learning are well known as representative methods of reinforcement learning. For example, in the case of Q-learning, the general update formula for the action value function Q(s,a) is expressed as Equation (1).

[0075]

number

[0076] In formula (1), S t represents the state of the environment at time t, and a t represents the action at time t. Action a t Therefore, the condition is S t+1 Changes to r t+1 represents the reward that the agent can receive depending on the change in 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. B1 (action) is action a t Then, B2 (state) is in state S t Then, the state S at time t t Best Practices in a t is learned.

[0077] The update formula expressed by formula (1) increases the action value function Q if the action value function Q of the action a with the highest Q value at time t+1 is greater than the action value function Q of the action a executed at time t, and decreases the action value function Q in the opposite case. In other words, the action value function Q(s, a) is updated so that the action value function Q of the action a at time t approaches the best action value at time t+1. This allows the best action value in a certain environment to be propagated sequentially to the action value in the previous environment.

[0078] Fig. 14 is a diagram for explaining reinforcement learning in the second trained model 122 according to the first embodiment. As shown in Fig. 14, in the second learning phase, the control device 1 (second model generation unit 114) executes the second learning program 142 to generate (update) the second trained model 122 based on the second training data 132 including B1 (action) and B2 (state) corresponding to the input 2, and D (reward criterion). As a result, in the utilization phase, the control device 1 can obtain C (output) corresponding to the output 2 based on B2 (state) corresponding to the input 2, using the second trained model 122.

[0079] Fig. 15 is a diagram for explaining the input and output of reinforcement learning in the second trained model 122 according to embodiment 1. As shown in Fig. 15, in the second learning phase, an action of whether or not to prohibit the execution of the final defrosting operation is used as B1 (action), the final operation data is used as B2 (state) corresponding to input 2, the result of whether or not to prohibit the execution of the final defrosting operation is used as C (output) corresponding to output 2, and the power consumption when the final defrosting operation and the final heating operation are performed and the power consumption when only the final heating operation is performed are applied as D (reward criterion).

[0080] FIG. 16 is a diagram for explaining an example of the reward calculation of the reinforcement learning in the second trained model 122 according to the first embodiment. As shown in FIG. 16, the reward in the reinforcement learning is calculated based on the power consumption when the final defrosting operation and the final heating operation are performed according to the final operation data. Specifically, when the second trained model 122 can select as the output result the operation with the lower power consumption between the case where the final defrosting operation and the final heating operation are performed according to the final operation data (for example, the operation from timing t6 to t8 shown in FIG. 3(A)) and the case where only the final heating operation is performed without performing the final defrosting operation (for example, the operation from timing t6 to t8 shown in FIG. 3(B)), the reward is increased, and when the second trained model 122 selects as the output result the operation with the higher power consumption, the reward is decreased.

[0081] 17 is a diagram showing a functional configuration for generating the second trained model 122 of the learning device according to the first embodiment. The learning device is realized by the processor 11 of the control device 1. The learning device can exchange data with each of the second learning program storage unit 1232 and the second trained model storage unit 1222. The second learning program storage unit 1232 and the second trained model storage unit 1222 are realized by the storage device 12 of the control device 1.

[0082] As shown in FIG. 17, the learning device executes the second learning program 142 stored in the second learning program memory unit 1232 to generate a second trained model 122 based on second learning data 132 including B1 (action) and B2 (state) corresponding to input 2, and D (reward criterion).

[0083] Specifically, the data acquisition unit 115 acquires an action of whether or not to prohibit the execution of a final defrosting operation as B1 (action), and acquires the final operation data as B2 (state) corresponding to input 2. The second model generation unit 114 uses the second learning data 132 acquired by the data acquisition unit 115 to generate a second trained model 122 for determining whether or not to prohibit the execution of a final defrosting operation based on the final operation data. The second model generation unit 114 stores the generated second trained model 122 in the second trained model storage unit 1222.

[0084] The second model generation unit 114 includes a remuneration calculation unit 1141 and a function update unit 1142. The remuneration calculation unit 1141 calculates the remuneration based on D (remuneration standard). Specifically, the remuneration calculation unit 1141 calculates the remuneration based on the power consumption when the final defrosting operation and the final heating operation are executed and the power consumption when only the final heating operation is executed as D (remuneration standard).

[0085] The function update unit 1142 updates the function for determining whether or not to prohibit the execution of the final defrosting operation, based on the reward calculated by the reward calculation unit 1141. For example, in the case of Q-learning, the function update unit 1142 updates the action value function Q(S t ,a t ) is used as a function for calculating the result of whether or not to prohibit the execution of the final defrosting operation.

[0086] The second trained model storage unit 1222 stores the action value function Q(S t ,a t ) is stored as the second trained model 122.

[0087] Fig. 18 is a flowchart of the process executed by the learning device according to the first embodiment in the second learning phase. The process shown in Fig. 18 is executed by the processor 11 of the control device 1 corresponding to the learning device. In Fig. 18, "S" is used as an abbreviation for "STEP".

[0088] As shown in FIG. 18, the learning device acquires second learning data 132 including B1 (action) and B2 (state) corresponding to input 2 by the data acquisition unit 115 (S21). The learning device calculates a reward based on the second learning data 132 (S22). Specifically, the learning device determines whether or not to prohibit the execution of the final defrosting operation based on the second learning data 132, and increases the reward if a result with lower power consumption can be output as a result of the determination (S23), and decreases the reward if a result with higher power consumption is output (S24). The learning device updates the action value function Q(st,at) represented by Equation 1 and stored in the second learned model storage unit 1222 based on the calculated reward (S25).

[0089] The learning device repeatedly executes the processes of S21 to S25 described above, stores the generated action value function Q(st,at) as the second trained model 122 in the second trained model storage unit 1222 (S26), and terminates this process.

[0090] [Examples of control device applications] With the air conditioner 2 installed in a property such as a building, the control device 1 executes the above-described machine learning in the first learning phase to generate the first trained model 121, and executes the machine learning in the second learning phase to generate the second trained model 122. For example, when a user inputs an ON command for the final heating operation stop function for the air conditioner 2 installed in a certain property, the control device 1 starts machine learning to generate the first trained model 121 and the second trained model 122. That is, the control device 1 generates the first trained model 121 and the second trained model 122 that can be applied to the environment in which the air conditioner 2 is placed. Such machine learning may also be executed in the utilization phase after the first trained model 121 and the second trained model 122 are generated, and thus the control device 1 can update the first trained model 121 and the second trained model 122 according to the environment that changes daily.

[0091] Then, in the utilization phase, the control device 1 determines whether or not to prohibit the execution of the final defrosting operation based on the air conditioning-related data that varies depending on the environment, using the first trained model 121 and the second trained model 122. The air conditioner 2 performs the final defrosting operation and the final heating operation, or performs the final heating operation without performing the final defrosting operation, according to the control of the control device 1.

[0092] In this way, the control device 1 according to the first embodiment uses the first trained model 121 and the second trained model 122 suited to the environment to determine whether or not to prohibit the execution of a final defrosting operation scheduled during the operation period of the air conditioner 2, based on the air conditioning-related data acquired from the air conditioner 2 via the input / output interface 13. Specifically, the control device 1 uses the first trained model 121 to infer final operation data for identifying the execution time of the final defrosting operation and the execution time of the final heating operation, based on the air conditioning-related data, and uses the second trained model 122 to determine whether or not to prohibit the execution of the final defrosting operation, based on the final operation data.

[0093] This allows the control device 1 to avoid a situation in which the air conditioner 2 stops during the execution of the final defrosting operation and the operation period ends, or the execution period of the final heating operation after the final defrosting operation becomes extremely short despite the execution of the final defrosting operation, due to the final defrosting operation being executed at the end of the operation period of the air conditioner 2. This allows the control device 1 to avoid wasting power consumption caused by the final defrosting operation, or causing discomfort to the user during the final defrosting operation.

[0094] Embodiment 2 A control device 500 according to the second embodiment will be described with reference to Fig. 19 to Fig. 23. Note that, in the following, only the parts of the control device 500 according to the second embodiment that are different from the control device 1 according to the first embodiment will be described.

[0095] [Utilization phase] An application example of the control device 500 in the utilization phase will be described with reference to Fig. 19 to Fig. 21. Fig. 19 is a diagram showing a functional configuration of the control device 500 according to the second embodiment in the utilization phase.

[0096] As shown in Figure 19, the control device 500 of embodiment 2 uses one learned model 520 to determine whether or not to prohibit the execution of a final defrost operation scheduled during the operation period of the air conditioner 2, based on air conditioning-related data acquired from the air conditioner 2 via the input / output interface 13.

[0097] Specifically, the processor 511 of the control device 500 includes, as main functional components, an inference unit 510, a data acquisition unit 515, and an output unit 516.

[0098] The data acquisition unit 515 acquires air conditioning-related data as input 11. The inference unit 510 uses the learned model 520 stored in the learned model storage unit 5220 to infer, as output 11, a decision result as to whether or not to prohibit execution of a final defrosting operation based on the input 11. The output unit 516 outputs a prohibition signal to the air conditioner 2 based on the output 11 of the inference unit 510. The learned model storage unit 5220 corresponds to a functional unit of the storage device 512 included in the control device 500.

[0099] Fig. 20 is a diagram for explaining input and output in the control device 500 according to embodiment 2. As shown in Fig. 20, in the control device 500, air conditioning-related data is used as input 11 of the inference unit 510. Also, in the control device 500, the result of whether or not to prohibit the execution of the final defrosting operation is used as output 11 of the inference unit 510.

[0100] Fig. 21 is a flowchart relating to the processing executed in the utilization phase by the control device 500 according to the second embodiment. The processing shown in Fig. 21 is executed by the processor 511 of the control device 500. In Fig. 21, "S" is used as an abbreviation for "STEP."

[0101] 21, the control device 500 acquires air conditioning-related data as input 11 from the air conditioner 200 of the air conditioner 2 by the data acquisition unit 515 (S51). The control device 500 inputs the acquired input 11 to the trained model 520 (S52). The control device 500 uses the trained model 520 to determine, based on the air conditioning-related data that is the input 11, whether or not to prohibit the execution of a final defrosting operation as output 11 (S53).

[0102] When it is determined that the execution of the final defrosting operation is to be prohibited (YES in S54), the control device 500 outputs a prohibition signal to the air conditioner 2 (S55) and ends this processing. As a result, the air conditioner 2 is controlled not to execute the final defrosting operation.

[0103] When the control device 500 determines not to prohibit the execution of the final defrosting operation (NO in S54), it ends this processing without outputting a prohibition signal to the air conditioner 2. As a result, the air conditioner 2 is controlled to execute the final defrosting operation.

[0104] [Learning Phase] 22 and 23, an application example of the control device 500 in the learning phase for generating the trained model 520 will be described. The trained model 520 is generated by performing supervised learning on the relationship between the air conditioning-related data and whether or not to prohibit the execution of the final defrosting operation. Supervised learning is a method of using a data set of factors and results (labels) to learn the features in the learning data 530 and infer the results from the input.

[0105] Fig. 22 is a diagram for explaining supervised learning in a trained model 520 according to embodiment 2. As shown in Fig. 22, in the learning phase, the control device 1 (model generation unit) executes a learning program 540 to generate (update) the trained model 520 based on training data 530 including an input 11 and an input 12 which is correct answer data. As a result, in the utilization phase, the control device 500 can obtain an output 11 based on the input 11 using the trained model 520.

[0106] Fig. 23 is a diagram for explaining input and output of supervised learning in trained model 520 according to embodiment 2. As shown in Fig. 23, in the learning phase, air conditioning-related data is used as input 11, an action as to whether or not to prohibit the execution of a final defrosting operation is used as input 12 corresponding to the correct answer data, and an action as to whether or not to prohibit the execution of a final defrosting operation is used as output 11.

[0107] That is, the trained model 520 according to the second embodiment decides whether or not to prohibit the execution of a final defrosting operation based on the air conditioning-related data which is the input 11, compares the decision result with the action of whether or not to prohibit the execution of a final defrosting operation which is the input 12 corresponding to the correct answer data, and if the two match, maintains the weights of the neural network shown in Fig. 8, and if the two do not match, adjusts the weights of the neural network so that the two match. This allows the trained model 520 to output a result (whether or not to prohibit the execution of a final defrosting operation) which matches the input 12 corresponding to the correct answer data, based on the air conditioning-related data which is the input 11.

[0108] As described above, the control device 500 according to the second embodiment uses the trained model 520 to determine whether or not to prohibit the execution of a final defrosting operation scheduled during the operation period of the air conditioner 2, based on the air conditioning-related data acquired from the air conditioner 2 via the input / output interface 13. This enables the control device 500 to avoid wasting power consumption caused by the final defrosting operation and causing discomfort to the user during the final defrosting operation.

[0109] [Variations] The present disclosure is not limited to the above-described embodiment, and various modifications and applications are possible. Modifications that can be applied to the present disclosure will be described below.

[0110] Fig. 24 is a diagram showing an example of the execution timing of a defrosting operation during the operation period of an air conditioner according to a modified example. As shown in Fig. 24, the air conditioner according to the modified example may include a plurality of outdoor heat exchangers, such as a first heat exchanger 21A and a second heat exchanger 21B, in the outdoor unit 3. Furthermore, the air conditioner may alternately execute a defrosting operation on the first heat exchanger 21A and the second heat exchanger 21B.

[0111] In the example of FIG. 24(A), the air conditioner starts a heating operation at a timing t1, and then when the execution condition of the defrosting operation is satisfied at a timing t2, the air conditioner starts a defrosting operation for the first heat exchanger 21A, and starts the heating operation again at a timing t3 when the defrosting operation is completed. Note that, during the execution of the defrosting operation for the first heat exchanger 21A, the heating operation may be performed using the second heat exchanger 21B. Thereafter, when the execution condition of the defrosting operation is satisfied at a timing t4, the air conditioner 2 starts a defrosting operation for the second heat exchanger 21B, and starts the heating operation again at a timing t5 when the defrosting operation is completed. Note that, during the execution of the defrosting operation for the second heat exchanger 21B, the heating operation may be performed using the first heat exchanger 21A. Thereafter, when the execution condition of the defrosting operation is satisfied at a timing t6, the air conditioner 2 starts a final defrosting operation for the first heat exchanger 21A, and starts the final heating operation again at a timing t7 when the defrosting operation is completed. During the execution of the defrosting operation for the first heat exchanger 21A, the heating operation may be executed using the second heat exchanger 21B. The final defrosting operation may be executed not only for the first heat exchanger 21A but also for the second heat exchanger 21B.

[0112] In this way, the air conditioner will perform a final defrosting operation on either the first heat exchanger 21A or the second heat exchanger 21B, but as shown in Figure 24(B), the control device 1 may use at least one trained model (for example, the first trained model 121 and the second trained model 122) to determine whether to prohibit the performance of the final defrosting operation.

[0113] The first heat exchanger 21A and the second heat exchanger 21B are not limited to being included in one outdoor unit 3. For example, the first heat exchanger 21A and the second heat exchanger 21B may be included in different outdoor units, such as the first heat exchanger 21A being included in the first outdoor unit and the second heat exchanger 21B being included in the second outdoor unit.

[0114] Furthermore, by the control device 1 outputting a prohibition signal, the air conditioner 2 ends the operation period without performing the final defrosting operation. In this case, there is a possibility that frost will be attached to the outdoor heat exchanger 21 when the air conditioner 2 operates next time. For example, if the air conditioner 2 is left alone during a winter night when the outdoor temperature is low, and the air conditioner 2 starts heating operation the next morning, the heat exchange efficiency of the outdoor heat exchanger 21 will decrease due to the frost attached to the outdoor heat exchanger 21. Therefore, when the processor 11 of the control device 1 prohibits the execution of the final defrosting operation by the prohibition signal, it may output a reservation signal for causing the air conditioner 2 to perform the defrosting operation at the start of the next operation period of the operation period. The control device 1 may output such a reservation signal to the air conditioner 2 during the operation period in which the execution of the final defrosting operation is prohibited, or may output such a reservation signal to the air conditioner 2 at the start of the next operation period. The air conditioner 2 may perform the defrosting operation at the start of the next operation period based on the reservation signal received from the control device 1. As a result, in the next operating period, the defrosting operation is performed and then the heating operation is performed, so that the control device 1 can prevent the heat exchange efficiency of the outdoor heat exchanger 21 from decreasing.

[0115] In addition, the air conditioner 2 may determine whether or not the above-mentioned conditions for executing the defrosting operation are met at the start of the next operating period based on the reservation signal received from the control device 1, and perform the defrosting operation if the conditions for execution are met.

[0116] Furthermore, when the processor 11 of the control device 1 has prohibited the execution of the final defrosting operation by the prohibition signal, the processor 11 may output an permission signal to the air conditioner 2 for permitting the execution of the final defrosting operation when the duration of the final heating operation by the air conditioner 2 exceeds a threshold value. The threshold value of the duration may be set in advance by the user. The air conditioner 2 may perform the final defrosting operation when the duration of the final heating operation exceeds a threshold value after prohibiting the execution of the final defrosting operation based on the permission signal received from the control device 1. As a result, if the duration of the final heating operation after the prohibition of the execution of the final defrosting operation is long, the final defrosting operation will be performed, and the control device 1 can prevent the heat exchange efficiency of the outdoor heat exchanger 21 from decreasing.

[0117] The air conditioning system 1000 is not limited to the air conditioner 2 installed in a property such as a building, but can also be applied to an in-vehicle freezer, a heat pump type in-vehicle air conditioner, or a heat pump type hot water supply system. In particular, in the case of the air conditioning system 1000 applied to an in-vehicle freezer, the vehicle's behavior history (for example, GPS data) may be used as the air conditioning related data. In this way, the control device 1 can detect the completion of the transportation of luggage based on the vehicle's behavior history and infer the start time of the final defrosting operation by empirical machine learning.

[0118] The control device 1 is not necessarily separate from the air conditioner 200. The air conditioner 200 may have the functions of the control device 1.

[0119] [summary] The present disclosure provides a control device 1 that controls the execution of a defrosting operation by an air conditioner 2 to remove frost from at least one heat exchanger. The control device 1 includes an input / output interface 13 that inputs and outputs data, and a processor 11. The processor 11 acquires air conditioning-related data related to the execution of a defrosting operation by the air conditioner 2 from the input / output interface 13, and uses at least one trained model to determine whether or not to prohibit the execution of a final defrosting operation scheduled during the operation period of the air conditioner 2, based on the air conditioning-related data acquired from the input / output interface 13.

[0120] By providing such a configuration, the control device 1 can avoid wasting power consumed by the final defrosting operation and avoid the user feeling uncomfortable during the final defrosting operation.

[0121] The processor 11 uses at least one learned model to infer final operation data for identifying the execution time of the final defrosting operation and the execution time of the final heating operation scheduled after the final defrosting operation based on the air conditioning-related data, and determines whether to prohibit the execution of the final defrosting operation based on the final operation data.

[0122] By having such a configuration, the control device 1 can infer the execution time of the final defrosting operation and the execution time of the final heating operation using at least one learned model, and can determine whether or not to prohibit the execution of the final defrosting operation based on the inference result.

[0123] At least one of the learned models includes a first learned model 121 that infers final operating data based on air conditioning-related data, and a second learned model 122 that determines whether to prohibit the execution of a final defrosting operation based on the final operating data inferred by the first learned model 121.

[0124] By having such a configuration, the control device 1 can use the first trained model 121 and the second trained model 122 to determine whether or not to prohibit the execution of final defrosting operation based on air conditioning-related data.

[0125] The first trained model 121 is generated by performing unsupervised learning on the relationship between the air conditioning-related data and the final operation data.

[0126] With such a configuration, the control device 1 can use the first trained model 121 to infer the final operating data based on the air conditioning-related data.

[0127] The second trained model 122 is generated by performing reinforcement learning using a reward based on the power consumption when the final defrosting operation and the final heating operation are performed according to the final operation data.

[0128] With such a configuration, the control device 1 can use the second learned model 122 to determine whether or not to prohibit the execution of the final defrosting operation based on the final operation data.

[0129] The air conditioning-related data includes at least one of the operating history of the air conditioner 2, the change in heat load in the space to be air conditioned, the date and time, the day of the week, whether or not there are public holidays, the heat exchanger temperature T0 of at least one heat exchanger, the outdoor air temperature T1 outside the space to be air conditioned, the indoor temperature T2 in the space to be air conditioned, the outdoor air humidity H1 outside the space to be air conditioned, the indoor humidity H2 in the space to be air conditioned, and the power consumption of the air conditioner 2 during the operating period.

[0130] With this configuration, the control device 1 can determine whether or not to prohibit the execution of the final defrosting operation, based on air conditioning-related data that varies depending on the environment in which the air conditioner 2 is placed.

[0131] The final operation data includes the start time of the final defrosting operation, the end time of the final defrosting operation and the start time of the final heating operation, and the end time of the final heating operation.

[0132] With such a configuration, the control device 1 can infer, based on the air conditioning-related data, the start time of the final defrosting operation, the end time of the final defrosting operation and the start time of the final heating operation, and the end time of the final heating operation as final operation data.

[0133] The at least one heat exchanger includes a first heat exchanger 21A and a second heat exchanger 21B. The air conditioner alternately performs defrosting operations on the first heat exchanger 21A and the second heat exchanger 21B during an operation period. The processor 11 determines whether or not to prohibit the execution of a final defrosting operation scheduled during the operation period of the air conditioner, out of the defrosting operation on the first heat exchanger 21A and the defrosting operation on the second heat exchanger 21B.

[0134] By having such a configuration, the control device 1 can prohibit the execution of a final defrosting operation on either the first heat exchanger 21A or the second heat exchanger 21B, even when defrosting operations are performed alternately on the first heat exchanger 21A and the second heat exchanger 21B.

[0135] When the processor 11 prohibits the execution of the final defrosting operation, it outputs a prohibition signal to the air conditioner 2 to prohibit the execution of the final defrosting operation.

[0136] By providing such a configuration, the control device 1 can prohibit the air conditioner 2 from executing the final defrosting operation.

[0137] When the processor 11 prohibits the execution of the final defrosting operation, when the duration of the heating operation by the air conditioner 2 exceeds a threshold value, the processor 11 permits the execution of the final defrosting operation.

[0138] By having such a configuration, even if the control device 1 prohibits the execution of the final defrosting operation, it can permit the execution of the final defrosting operation if the heating operation by the air conditioner 2 continues for a long period of time, for example. This makes it possible to prevent a decrease in the heat exchange efficiency of the outdoor heat exchanger 21 due to the prohibition of the execution of the final defrosting operation.

[0139] The present disclosure provides a control method for controlling the execution of a defrosting operation for removing frost from at least one heat exchanger by an air conditioner 2 using a control device 1. The control method includes a step (S1) of acquiring air conditioning-related data related to the execution of a defrosting operation by the air conditioner 2, and a step (S2) of using at least one trained model and based on the acquired air conditioning-related data, determining whether or not to prohibit the execution of a final defrosting operation scheduled during the operation period of the air conditioner.

[0140] By providing such a configuration, the control device 1 can avoid wasting power consumed by the final defrosting operation and avoid the user feeling uncomfortable during the final defrosting operation.

[0141] The embodiments disclosed herein should be considered to be illustrative and not restrictive in all respects. The scope of the present disclosure is defined by the claims, not by the description of the embodiments described above, and is intended to include all modifications within the meaning and scope of the claims. [Explanation of symbols]

[0142] 1,500 control device, 2 air conditioner, 3 outdoor unit, 4 indoor unit, 5 heater, 11,511 processor, 12,512 storage device, 13 input / output interface, 21 outdoor heat exchanger, 21A first heat exchanger, 21B second heat exchanger, 22,32 blower, 25 compressor, 26 expansion valve, 31 indoor heat exchanger, 40 four-way valve, 41,42,43,44 connection port, 51,52,54,55,57,58,59,60 piping, 53,56 extension piping, 70,71,72 temperature sensor, 81,82 humidity sensor, 111 first inference unit, 112 second inference unit, 113 first model generation unit, 114 second model generation unit, 115,515 data acquisition unit, 116,516 Output unit, 121 first trained model, 122 second trained model, 131 first training data, 132 second training data, 141 first training program, 142 second training program, 200 air conditioning device, 251 intake port, 252 exhaust port, 510 inference unit, 520 trained model, 530 training data, 540 training program, 1000 air conditioning system, 1141 reward calculation unit, 1142 function update unit, 1221 first trained model memory unit, 1222 second trained model memory unit, 1231 first training program memory unit, 1232 second training program memory unit, 5220 trained model memory unit.

Claims

1. A control device that controls execution of a defrosting operation for removing frost from at least one heat exchanger by an air conditioner, an input / output interface for inputting and outputting data; a processor; The processor, Acquire air conditioning-related data regarding the execution of the defrosting operation by the air conditioner from the input / output interface; A control device that uses at least one learned model to infer final operation data for specifying the execution time of the final defrosting operation and the execution time of the final heating operation scheduled after the final defrosting operation based on the air conditioning-related data acquired from the input / output interface, and determines whether to prohibit the execution of the final defrosting operation scheduled during the operation period of the air conditioner based on the final operation data.

2. The control device according to claim 1, wherein the at least one learned model includes a first learned model that infers the final operating data based on the air conditioning-related data, and a second learned model that determines whether to prohibit execution of the final defrosting operation based on the final operating data inferred by the first learned model.

3. The control device according to claim 2 , wherein the first learned model is generated by performing unsupervised learning on the relationship between the air conditioning-related data and the final operation data.

4. The control device according to claim 2 or 3, wherein the second learned model is generated by performing reinforcement learning using a reward based on power consumption when the final defrosting operation and the final heating operation are performed in accordance with the final operation data.

5. A control device as described in any one of claims 1 to 3, wherein the air conditioning-related data includes at least one of the operating history of the air conditioner, the change in heat load in the air conditioned space, the date and time, the day of the week, the presence or absence of holidays, the temperature of the at least one heat exchanger, the outdoor air temperature outside the air conditioned space, the indoor temperature in the air conditioned space, the outdoor air humidity outside the air conditioned space, the indoor humidity in the air conditioned space, and the power consumption of the air conditioner during the operating period.

6. The control device according to any one of claims 1 to 3, wherein the final operation data includes a start time of the final defrosting operation, an end time of the final defrosting operation and a start time of the final heating operation, and an end time of the final heating operation.

7. the at least one heat exchanger includes a first heat exchanger and a second heat exchanger; The air conditioner performs the defrosting operation alternately on the first heat exchanger and the second heat exchanger during the operation period, The control device according to any one of claims 1 to 3, wherein the processor determines whether or not to prohibit the execution of the final defrosting operation scheduled during the operation period of the air conditioner, between the defrosting operation for the first heat exchanger and the defrosting operation for the second heat exchanger.

8. The control device according to any one of claims 1 to 3, wherein when the processor prohibits execution of the final defrosting operation, the processor outputs a signal to the air conditioner to prohibit execution of the final defrosting operation.

9. The control device according to any one of claims 1 to 3, wherein when the processor prohibits execution of the final defrosting operation, it outputs a signal to cause the air conditioner to execute the defrosting operation at the start of the next operating period following the operating period.

10. The control device according to any one of claims 1 to 3, wherein when the processor prohibits execution of the final defrosting operation, the processor permits execution of the final defrosting operation when a duration of the heating operation by the air conditioner exceeds a threshold value.

11. A control method for controlling execution of a defrosting operation for removing frost from at least one heat exchanger by an air conditioner using a control device, comprising: The control method includes: acquiring air conditioning-related data regarding execution of the defrosting operation by the air conditioner; a step of inferring final operation data for specifying the execution time of the final defrosting operation and the execution time of a final heating operation scheduled after the final defrosting operation based on the acquired air conditioning-related data using at least one learned model, and determining whether to prohibit the execution of a final defrosting operation scheduled during the operation period of the air conditioner based on the final operation data.

Images