Central air conditioning system and its control method

Abstract

To provide a whole building air conditioning system which can control temperatures of rooms to keep the rooms at comfortable temperatures without needing a complicated construction procedure.SOLUTION: A whole building air conditioning system 10 includes: an air conditioner 21; a transfer device 23 including multiple devices disposed within a facility 90 to transfer air whose temperature is adjusted by the air conditioner 21 to multiple rooms 90a to 90c in the facility 90; and a control device 80. The control device 80 acquires room temperatures of the multiple rooms 90a to 90c and uses the acquired room temperatures as input of a reinforcement learning model without using arrangement information of the multiple devices to control the air conditioner 21 and the transfer device 23 in accordance with control values of the air conditioner 21 and the transfer device 23 output from the reinforcement learning model so that the room temperatures of the multiple rooms 90a to 90c approach target temperatures of the multiple rooms 90a to 90c.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a central air conditioning system and a control method thereof. [Background technology]

[0002] A central air conditioning system that uses an air conditioner to condition multiple rooms has been proposed. Patent Document 1 discloses an air conditioning system that can quickly raise room temperatures to a target temperature. This central air conditioning system includes multiple devices, such as dampers, for transporting air to each of the multiple rooms. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 2019-203670 Summary of the Invention [Problem to be solved by the invention]

[0004] Incidentally, in the case of the central air-conditioning system (air-conditioning system) described above, in order to air-condition a plurality of rooms, a complicated installation procedure may be required at the time of installation, in which a correspondence relationship between the plurality of rooms and the arrangement of a plurality of devices is set.

[0005] The present invention provides a central air conditioning system that does not require a complicated installation procedure and that can control temperatures to be comfortable, and a control method for the same. [Means for solving the problem]

[0006] A whole-building air conditioning system according to one embodiment of the present invention comprises an air conditioning unit, a transport device including a plurality of devices arranged within a facility for transporting air whose temperature has been adjusted by the air conditioning unit to each of a plurality of rooms within the facility, and a control device, wherein the control device acquires the room temperature of each of the plurality of rooms, and without using location information of the plurality of devices, uses the acquired room temperatures as input for a reinforcement learning model to control the air conditioning unit and the transport device in accordance with control values ​​of the air conditioning unit and the transport device output from the reinforcement learning model so that the room temperature of each of the plurality of rooms approaches the target temperature of each of the plurality of rooms.

[0007] A control method for a central air conditioning system according to one aspect of the present invention is a control method for a central air conditioning system, wherein the central air conditioning system comprises an air conditioning unit and a transport device including a plurality of devices arranged within a facility for transporting air whose temperature has been adjusted by the air conditioning unit to each of a plurality of rooms within the facility, and the control method includes an acquisition step of acquiring the room temperature of each of the plurality of rooms, and a control step of using the acquired room temperature as input to a reinforcement learning model without using arrangement information of the plurality of devices, and controlling the air conditioning unit and the transport device in accordance with control values ​​of the air conditioning unit and the transport device output from the reinforcement learning model so that the room temperature of each of the plurality of rooms approaches a target temperature of each of the plurality of rooms. [Effects of the Invention]

[0008] The central air conditioning system and the control method thereof of the present invention do not require complicated installation procedures and can control temperatures to be comfortable. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a diagram showing a schematic configuration of a central air-conditioning system according to the first embodiment. [Figure 2] FIG. 2 is a block diagram showing a functional configuration of the central air-conditioning system according to the first embodiment. [Figure 3]FIG. 3 is a flowchart of an operation example 1 of the central air-conditioning system. [Figure 4] FIG. 4 is a flowchart of an operation example 2 of the central air-conditioning system. [Figure 5] FIG. 5 is a diagram showing a schematic configuration of a central air-conditioning system according to the second embodiment. [Figure 6] FIG. 6 is a block diagram showing a functional configuration of the central air-conditioning system according to the second embodiment. [Figure 7] FIG. 7 is a flowchart of an operation example 3 of the central air-conditioning system. [Figure 8] FIG. 8 is a diagram showing a schematic configuration of a central air-conditioning system according to the third embodiment. [Figure 9] FIG. 9 is a block diagram showing a functional configuration of a central air-conditioning system according to the third embodiment. [Figure 10] FIG. 10 is a flowchart of an operation example 4 of the central air-conditioning system. [Figure 11] FIG. 11 is a diagram showing a schematic configuration of a central air-conditioning system according to a modified example of the third embodiment. [Figure 12] FIG. 12 is a block diagram showing a functional configuration of a central air-conditioning system according to a modification of the third embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, the embodiments will be described in detail with reference to the drawings. Note that the embodiments described below are all comprehensive or specific examples. The numerical values, shapes, materials, components, component placement and connection forms, steps, and step order shown in the following embodiments are merely examples and are not intended to limit the present invention. Furthermore, among the components in the following embodiments, components not recited in independent claims will be described as optional components.

[0011] It should be noted that the drawings are schematic diagrams and are not necessarily strict illustrations. In addition, in the drawings, substantially the same components are denoted by the same reference numerals, and overlapping descriptions may be omitted or simplified.

[0012] (Embodiment 1) [composition] The configuration of the central air-conditioning system 10 according to the embodiment will be described below.

[0013] Fig. 1 is a diagram showing a schematic configuration of a central air-conditioning system 10 according to this embodiment. Fig. 2 is a block diagram showing a functional configuration of the central air-conditioning system 10 according to this embodiment.

[0014] As shown in Figures 1 and 2, the central air conditioning system 10 is a system that can adjust the temperature of multiple rooms 90a to 90c in a facility 90 using a single air conditioner 21. The number of rooms is not particularly limited. The central air conditioning system 10 includes a central air conditioner 20, multiple temperature sensors 60a to 60c, and a control device 80. The central air conditioner 20 and each of the multiple temperature sensors 60a to 60c have the function of communicating with the control device 80.

[0015] The central air conditioning unit 20 controls the room temperatures of the plurality of rooms 90a to 90c by adjusting the temperature of air taken in from outside the facility 90 or air taken in from inside the facility 90 and transporting the air to each of the plurality of rooms 90a to 90c. The central air conditioning unit 20 includes an air conditioning unit 21 and a transport unit 23.

[0016] The air conditioner 21 is a so-called air conditioner, and adjusts the temperature of air taken in from outside the facility 90 or air taken in from inside the facility 90 under the control of the control device 80.

[0017] The transport device 23 transports air whose temperature has been adjusted by the air conditioner 21 to each of the multiple rooms 90a to 90c in the facility 90 under the control of the control device 80. As shown in FIG. 1, the transport device 23 is a device disposed in the facility 90. The transport device 23 includes multiple devices. In this embodiment, the multiple devices are multiple transport fans 23a and 23b and multiple VAV (Variable Air Volume) dampers 24a to 24c. The transport device 23 includes a transport fan 23a corresponding to the room 90a, a transport fan 23b corresponding to the rooms 90b and 90c, and one VAV damper corresponding to each room.

[0018] That is, in this embodiment, the transport fan 23a and the VAV damper 24a are arranged to transport air to the room 90a, and the transport fan 23b and the VAV damper 24b are arranged to transport air to the room 90b. Similarly, the transport fan 23b and the VAV damper 24c are arranged to transport air to the room 90c. Also, as shown in Fig. 1, each of the multiple VAV dampers 24a to 24c is arranged in a duct connected to each of the multiple rooms 90a to 90c.

[0019] The temperature sensors 60a-60c are arranged in each of the rooms 90a-90c, one for each room, and measure the room temperature in the room that the temperature sensor is arranged in. Each of the temperature sensors 60a-60c has a communication function and can transmit temperature data indicating the measured room temperature to the control device 80.

[0020] The control device 80 is a control device that controls the central air-conditioning device 20 .

[0021] The control device 80 includes an operation receiving unit 81, a control unit 82, a storage unit 83, and a communication unit 84.

[0022] The operation reception unit 81 is a user interface unit that receives user operations. For example, the operation reception unit 81 receives an operation to specify a target temperature (in other words, a set temperature) for each of the multiple rooms 90a to 90c. The operation reception unit 81 is realized by, for example, a touch panel, but may also include hardware buttons in addition to the touch panel. Although not shown, the control device 80 may also include a display unit that is realized by a display panel such as a liquid crystal panel or an organic EL (Electro Luminescence) panel. Furthermore, the operation reception unit 81 and the display unit may form a GUI (Graphical User Interface).

[0023] The control unit 82 controls the central air conditioner 20 by causing the communication unit 84 to transmit a control signal. The control unit 82 is realized by, for example, a microcomputer, but may also be realized by a processor.

[0024] The storage unit 83 is a storage device that stores the control program executed by the control unit 82. The storage unit 83 is realized by, for example, a semiconductor memory.

[0025] The communication unit 84 is a communication module (communication circuit) that enables the control device 80 to communicate with the central air-conditioning unit 20 and each of the temperature sensors 60a to 60c via the local communication network. The communication performed by the communication unit 84 is, for example, wireless communication, but may also be wired communication. There are no particular limitations on the communication standard used for the communication.

[0026] As described above, the transport device 23 including a plurality of devices is arranged within the facility 90. The control device 80 (control unit 82) according to this embodiment controls the central air conditioner 20 (here, the air conditioner 21 and the transport device 23) without using arrangement information of the plurality of devices indicating the positions where the plurality of devices are arranged within the facility 90. In other words, the control unit 82 controls the central air conditioner 20 without using information indicating the correspondence between each of the plurality of rooms 90a to 90c and the arrangement of each of the plurality of devices.

[0027] In this embodiment, the control unit 82 controls the air conditioner 21 and the transport device 23 according to the control values ​​of the air conditioner 21 and the transport device 23 output from the machine learning model.

[0028] This reinforcement learning model is a model in which the room temperatures of each of the multiple rooms 90a to 90c are input and the control values ​​of the air conditioner 21 and the transport device 23 are output. The reinforcement learning model is a model constructed by reinforcement learning. In this embodiment, the reinforcement learning model is trained by an algorithm such as deep reinforcement learning (DQN: Deep Q-Network) or the like.

[0029] In addition, the reinforcement learning model is a model that uses the difference between the room temperature and the target temperature as a reward when the control device 80 controls the air conditioning device 21 and the transport device 23 according to the control values ​​of the air conditioning device 21 and the transport device 23 output from the reinforcement learning model.

[0030] A detailed example of the operation will be described below.

[0031] [Example 1] First, a description will be given of an operation example 1 of the central air-conditioning system 10. FIG.

[0032] First, the operation receiving unit 81 of the control device 80 receives an operation to specify a target temperature for each of the plurality of rooms 90a to 90c (step S10). The target temperatures are stored in the storage unit 83. The target temperatures for each of the plurality of rooms are temperatures that are comfortable for the user.

[0033] Next, the communication unit 84 acquires temperature data indicating the current room temperatures of the rooms 90a to 90c from the temperature sensors 60a to 60c installed in the rooms 90a to 90c, respectively (step S12). The acquired temperature data is integrated and stored in the storage unit 83 as temperature information indicating the current room temperatures of the rooms 90a to 90c.

[0034] Next, the control unit 82 determines a control value for the central air conditioner 20 (air conditioner 21 and conveyance device 23) using the stored temperature information and the reinforcement learning model (step S14). This control value for the central air conditioner 20 is a value output from the reinforcement learning model by using the room temperature indicated by the temperature information as an input to the reinforcement learning model. Here, the input and output of this reinforcement learning model will be explained using Table 1.

[0035] [Table 1]

[0036] In this embodiment, the inputs used are the room temperatures of the rooms 90a-90c acquired by the temperature sensors 60a-60c, respectively. Additionally, the outputs used are the heat quantity indicating the degree to which the air is heated or cooled by the air conditioner 21, the air volume transported by the transport fans 23a and 23b, and the opening degrees indicating the degree to which each of the VAV dampers 24a-24c is open or closed.

[0037] Furthermore, the control device 80 (control unit 82) controls the air conditioner 21 and the transport device 23 according to the determined control values ​​(i.e., the control values ​​of the air conditioner 21 and the transport device 23 output from the reinforcement learning model) (step S16). The control unit 82 controls the air conditioner 21 and the transport device 23 so that the room temperature of each of the multiple rooms 90a to 90c approaches the target temperature of each of the multiple rooms 90a to 90c. As described above, the target temperature is a temperature that is comfortable for the user. Therefore, in other words, the control unit 82 controls the air conditioner 21 and the transport device 23 so that the room temperature of each of the multiple rooms 90a to 90c becomes a comfortable temperature.

[0038] Furthermore, after the processing of step S16 is performed and a predetermined period of time has elapsed, the control unit 82 performs learning of the reinforcement learning model (step S18). Note that the predetermined period of time is, for example, several minutes to several tens of minutes, but is not limited to this.

[0039] As described above, the reinforcement learning model outputs control values ​​for the air conditioner 21 and the transport device 23 by using the room temperature indicated by the acquired temperature information as an input to the reinforcement learning model. Here, the control unit 82 uses the difference between the room temperature and the target temperature when controlling the air conditioner 21 and the transport device 23 in accordance with this control value as a reward to train the reinforcement learning model. More specifically, as shown in Table 1, the control unit 82 trains the reinforcement learning model by using the smallness of the difference between the room temperature and the target temperature as a reward. In reinforcement learning, the reinforcement learning model is trained so as to maximize this reward, in other words, so as to reduce the difference between the room temperature and the target temperature.

[0040] When step S18 is completed, the operation example 1 is completed. Here, the processes from step S10 to step S18 may be repeated.

[0041] By repeating the process in this manner, the control unit 82 repeatedly learns the reinforcement learning model. As described above, the reinforcement learning model repeatedly learns so as to maximize the reward. The control unit 82 controls the air conditioner 21 and the transport device 23 in accordance with the control values ​​that are the output of such a reinforcement learning model, so that the room temperatures of the multiple rooms 90a to 90c tend to approach the target temperatures of the multiple rooms 90a to 90c. In other words, by using the reinforcement learning model that has been repeatedly trained, the control unit 82 can control the air conditioner 21 and the transport device 23 so that the room temperatures of the multiple rooms 90a to 90c become more comfortable.

[0042] In summary, the control device 80 controls the central air conditioning device 20 so that the room temperature of each of the multiple rooms 90a to 90c approaches the target temperature of each of the multiple rooms, in other words, so that the room temperature of each of the multiple rooms 90a to 90c becomes a comfortable temperature.

[0043] At this time, the control device 80 controls the central air-conditioning device 20 without using the arrangement information of the plurality of devices.

[0044] Therefore, in order to air-condition the multiple rooms 90a-90c, the installer does not need to go through a complicated installation procedure of setting up correspondences between the multiple rooms 90a-90c and the locations of the multiple devices. This reduces the installer's installation labor. Furthermore, it is less likely that the installer will set up the correspondences incorrectly, resulting in problems with the central air-conditioning system 10, such as "not being able to properly heat or cool the air in the multiple rooms 90a-90c."

[0045] In summary, the central air-conditioning system 10 according to this embodiment does not require a complicated installation procedure and can control the temperature to be comfortable.

[0046] Furthermore, the second operational example will be described below.

[0047] [Example 2] Next, a second operational example of the central air-conditioning system 10 will be described.

[0048] In Operation Example 2, the control device 80 controls the air conditioner 21 and the transport device 23 as follows. As in Operation Example 1, the control device 80 performs control so that the room temperatures of the multiple rooms 90a to 90c approach the target temperatures of the multiple rooms 90a to 90c, respectively, and so that the power consumption of the air conditioner 21 and the transport device 23 is reduced.

[0049] In addition, in operation example 2, the reinforcement learning model is a model that undergoes reinforcement learning using the following as rewards when the control device 80 (control unit 82) controls the air conditioner 21 and the transport device 23 in accordance with the control values ​​of the air conditioner 21 and the transport device 23 output from the reinforcement learning model: In operation example 2, the reinforcement learning model undergoes reinforcement learning using the difference between the room temperature and the target temperature, and the power consumption of the air conditioner 21 and the transport device 23 as rewards.

[0050] FIG. 4 is a flowchart of an operation example 2 of the central air-conditioning system 10. In FIG.

[0051] In the second operational example, steps S10 and S12 shown in FIG. 4 are performed in the same manner as in the first operational example.

[0052] Next, the control unit 82 determines a control value for the central air conditioner 20 (air conditioner 21 and conveyance device 23) using the stored temperature information and the reinforcement learning model (step S14). The control value for the central air conditioner 20 is a value output from the reinforcement learning model by using the room temperature indicated by the temperature information as an input to the reinforcement learning model. Here, the input, output, and reward of this reinforcement learning model will be explained using Table 2.

[0053] [Table 2]

[0054] As shown in Table 2, in this modification, the inputs and outputs of the reinforcement learning model are the same as the inputs and outputs shown in Table 1 of Operation Example 1. However, reinforcement learning is performed using the difference between the room temperature and the target temperature, and the power consumption of the air conditioning device 21 and the transport device 23 as rewards.

[0055] Furthermore, the control device 80 (control unit 82) controls the air conditioner 21 and the transportation device 23 in accordance with the determined control values ​​(i.e., the control values ​​of the air conditioner 21 and the transportation device 23 output from the reinforcement learning model) (step S16a). The control unit 82 controls the air conditioner 21 and the transportation device 23 so that the room temperature of each of the multiple rooms 90a to 90c approaches the target temperature of each of the multiple rooms 90a to 90c and so that the power consumption of the air conditioner 21 and the transportation device 23 is reduced. That is, in Operation Example 2, the control device 80 controls the central air conditioner 20 so that the room temperature of each of the multiple rooms 90a to 90c becomes a comfortable temperature and so that the power consumption of the air conditioner 21 and the transportation device 23 is reduced.

[0056] Furthermore, after the processing of step S16a is performed and a predetermined period of time has elapsed, the control unit 82 performs learning of the reinforcement learning model (step S18a). Note that the predetermined period of time is, for example, several minutes to several tens of minutes, but is not limited to this.

[0057] As described above, the reinforcement learning model outputs control values ​​for the air conditioner 21 and the transport device 23 by using the room temperature indicated by the acquired temperature information as an input to the reinforcement learning model. Here, the reinforcement learning model is trained using the difference between the room temperature and the target temperature when the control unit 82 controls the air conditioner 21 and the transport device 23 according to this control value, and the power consumption of the air conditioner 21 and the transport device 23 as rewards. More specifically, as shown in Table 2, the control unit 82 trains the reinforcement learning model using the smallness of the difference between the room temperature and the target temperature and the smallness of the power consumption of the air conditioner 21 and the transport device 23 as rewards. In reinforcement learning, the reinforcement learning model is trained so as to maximize this reward, that is, so as to reduce the difference between the room temperature and the target temperature and reduce the power consumption of the air conditioner 21 and the transport device 23.

[0058] When step S18a is completed, the operation example 2 is completed. Here, the processes from step S10 to step S18a may be repeated.

[0059] By repeating the process in this manner, the control unit 82 repeatedly learns the reinforcement learning model. As described above, the reinforcement learning model repeatedly learns so as to maximize the reward. By the control unit 82 controlling the air conditioner 21 and the transport device 23 in accordance with the control values ​​that are the output of such a reinforcement learning model, the room temperatures of the multiple rooms 90a to 90c tend to approach the target temperatures of the multiple rooms 90a to 90c. Furthermore, by the control unit 82 controlling the air conditioner 21 and the transport device 23 in accordance with the control values ​​that are the output of such a reinforcement learning model, the power consumption of the air conditioner 21 and the transport device 23 can be further reduced.

[0060] In the second operation example, the control device 80 controls the air conditioner 21 and the transport device 23 without using the arrangement information of the plurality of devices.

[0061] As shown in Operation Example 2, the whole-building air conditioning system 10 according to this embodiment does not require complicated installation procedures, can control the temperature to a comfortable level, and can also reduce power consumption.

[0062] Here, the control unit 82 controls the air conditioner 21 and the transport device 23 so that the room temperature of each of the rooms 90a to 90c approaches the target temperature of each of the rooms 90a to 90c and so that the power consumption of the air conditioner 21 and the transport device 23 is reduced. In other words, a mode in which the room temperature of each of the rooms 90a to 90c is controlled so that the room temperature of each of the rooms 90a to 90c approaches the target temperature of each of the rooms 90a to 90c and a mode in which the power consumption of the air conditioner 21 and the transport device 23 is controlled so that the power consumption of the air conditioner 21 and the transport device 23 is reduced are both possible. However, this is not limited thereto, and the control unit 82 may control the air conditioner 21 and the transport device 23 so that one mode is prioritized. For example, the operation receiving unit 81 may receive an operation instructing priority over one mode, and the control unit 82 may control the air conditioner 21 and the transport device 23 so that one mode is prioritized in accordance with the received operation.

[0063] [Effects, etc.] The central air-conditioning system 10 according to this embodiment includes an air conditioner 21, a transport device 23 including a plurality of devices arranged within a facility 90 for transporting air whose temperature has been adjusted by the air conditioner 21 to each of a plurality of rooms 90a to 90c within the facility 90, and a control device 80. The control device 80 acquires the room temperatures of each of the plurality of rooms 90a to 90c. The control device 80 uses the acquired room temperatures as input for a reinforcement learning model without using location information for the plurality of devices, and controls the air conditioner 21 and the transport device 23 in accordance with the control values ​​for the air conditioner 21 and the transport device 23 output from the reinforcement learning model. The control device 80 controls the air conditioner 21 and the transport device 23 so that the room temperatures of each of the plurality of rooms 90a to 90c approach the target temperatures for each of the plurality of rooms 90a to 90c.

[0064] As shown in Operation Example 1, the control device 80 according to this embodiment controls the air conditioners 21 and the transport devices 23 without using location information for the multiple devices. Therefore, in order to air-condition the multiple rooms 90a-90c, a complex construction procedure in which an installer sets a correspondence between the multiple rooms 90a-90c and the locations of the multiple devices is not required. Furthermore, the control device 80 controls the air conditioners 21 and the transport devices 23 based on a reinforcement learning model so that the room temperatures of the multiple rooms 90a-90c approach the target temperatures of the multiple rooms 90a-90c. In other words, the control device 80 controls the air conditioners 21 and the transport devices 23 based on a reinforcement learning model so that the room temperatures of the multiple rooms 90a-90c are comfortable.

[0065] That is, the central air-conditioning system 10 according to the present embodiment does not require complicated installation procedures and can control the temperature to be comfortable.

[0066] As in operation example 1, the reinforcement learning model performs reinforcement learning using the difference between the room temperature and the target temperature as a reward when the control device 80 controls the air conditioning device 21 and the transport device 23 according to the control values ​​of the air conditioning device 21 and the transport device 23 output from the reinforcement learning model.

[0067] According to the control values ​​output from such a reinforcement learning model, the control device 80 (control unit 82) controls the air conditioner 21 and the conveyance device 23. This makes it easier for the room temperature of each of the multiple rooms 90a to 90c to approach the target temperature of each of the multiple rooms. In other words, it becomes easier for the room temperature of each of the multiple rooms 90a to 90c to become a more comfortable temperature.

[0068] Therefore, the central air-conditioning system 10 according to this embodiment does not require complicated installation procedures and can control the temperature to a more comfortable level.

[0069] As in the second operation example, the control device 80 controls the air conditioner 21 and the transport device 23 so that the power consumption of the air conditioner 21 and the transport device 23 is reduced.

[0070] As a result, the central air conditioning system 10 according to this embodiment does not require complicated installation procedures, can control the temperature to a comfortable level, and can also reduce power consumption.

[0071] As in operation example 2, the reinforcement learning model is reinforced using the power consumption of the air conditioning device 21 and the transport device 23 when the control device 80 controls the air conditioning device 21 and the transport device 23 according to the control values ​​of the air conditioning device 21 and the transport device 23 output from the reinforcement learning model as a reward.

[0072] The control device 80 (control unit 82) controls the air conditioner 21 and the transport device 23 in accordance with the control values ​​output from such a reinforcement learning model, which makes it easier for the power consumption of the air conditioner 21 and the transport device 23 to be reduced.

[0073] Therefore, the whole-building air conditioning system 10 according to this embodiment does not require complicated installation procedures, can control the temperature to a comfortable level, and can further reduce power consumption.

[0074] A central air-conditioning system 10 according to this embodiment includes an air conditioner 21 and a transport device 23 including a plurality of devices that are arranged within a facility 90 to transport air whose temperature has been adjusted by the air conditioner 21 to each of a plurality of rooms 90a to 90c within the facility 90. A control method for the central air-conditioning system 10 includes an acquisition step of acquiring the room temperatures of each of the plurality of rooms 90a to 90c. The control method includes a control step of using the acquired room temperatures as input for a reinforcement learning model without using location information for the plurality of devices, and controlling the air conditioner 21 and the transport device 23 in accordance with control values ​​for the air conditioner 21 and the transport device 23 output from the reinforcement learning model. In the control step, the air conditioner 21 and the transport device 23 are controlled so that the room temperatures of each of the plurality of rooms 90a to 90c approach the target temperatures for each of the plurality of rooms 90a to 90c.

[0075] As shown in Operation Example 1, the control method according to this embodiment controls the air conditioner 21 and the transport device 23 without using information about the arrangement of the multiple devices. Therefore, in order to air-condition the multiple rooms 90a-90c, a complex construction procedure in which an installer sets a correspondence between the multiple rooms 90a-90c and the arrangement of the multiple devices is not required. Furthermore, this control method controls the air conditioner 21 and the transport device 23 based on a reinforcement learning model so that the room temperature of each of the multiple rooms 90a-90c approaches the target temperature of each of the multiple rooms 90a-90c. In other words, this control method controls the air conditioner 21 and the transport device 23 based on a reinforcement learning model so that the room temperature of each of the multiple rooms 90a-90c becomes a comfortable temperature.

[0076] In other words, the control method according to this embodiment does not require a complicated installation procedure and can control the temperature to a comfortable level.

[0077] (Embodiment 2) [composition] The configuration of a central air-conditioning system 10a according to the second embodiment will be described below.

[0078] Fig. 5 is a diagram showing a schematic configuration of the central air-conditioning system 10a according to this embodiment. Fig. 6 is a block diagram showing a functional configuration of the central air-conditioning system 10a according to this embodiment.

[0079] The central air-conditioning system 10a differs from the central air-conditioning system 10 in that it further includes a plurality of humidity sensors 61b and 61c and a humidifier 50. That is, the central air-conditioning system 10a includes a central air-conditioning unit 20, a plurality of temperature sensors 60a to 60c, a plurality of humidity sensors 61b and 61c, a control device 80, and a humidifier 50. Each of the plurality of humidity sensors 61b and 61c and the humidifier 50 has a function of communicating with the control device 80.

[0080] The humidifier 50 is a device that adjusts the humidity of the air transported by the central air-conditioning device 20. Here, the humidifier 50 increases the humidity of the air inside the facility 90.

[0081] The humidity sensors 61b and 61c are arranged in each of the rooms 90b and 90c, respectively, and measure the room humidity in the room in which the humidity sensor is arranged. Each of the humidity sensors 61b and 61c has a communication function and can transmit humidity data indicating the measured room humidity to the control device 80. The rooms in which the humidity sensors 61b and 61c are arranged correspond to first rooms, and in this embodiment, the two first rooms correspond to rooms 90b and 90c, respectively. It is sufficient that there is at least one first room in which a humidity sensor is arranged.

[0082] In this embodiment, the control unit 82 controls the air conditioner 21, the transport device 23, and the humidifier 50 according to the control values ​​of the air conditioner 21, the transport device 23, and the humidifier 50 output from the machine learning model.

[0083] This reinforcement learning model is a model in which the room temperatures of the plurality of rooms 90a to 90c and the room humidities of the two first rooms are input, and the control values ​​of the air conditioner 21, the conveying device 23, and the humidifier 50 are output.

[0084] Furthermore, the reinforcement learning model is a model that undergoes reinforcement learning using the following as rewards when the control device 80 controls the air conditioner 21, the transport device 23, and the humidifier 50 in accordance with the control values ​​of the air conditioner 21, the transport device 23, and the humidifier 50 output from the reinforcement learning model: In this embodiment, the reinforcement learning model is a model that undergoes reinforcement learning using the difference between the room temperature and the target temperature, and the difference between the room humidity and the target humidity as rewards.

[0085] [Example 3] Next, a third operational example of the central air-conditioning system 10a will be described.

[0086] FIG. 7 is a flowchart of an operation example 3 of the central air-conditioning system 10a.

[0087] First, the operation receiving unit 81 of the control device 80 receives an operation to specify a target temperature for each of the plurality of rooms 90a to 90c and a target humidity for each of the two first rooms (step S20). The target temperature and target humidity are stored in the storage unit 83. The target humidity for each of the two first rooms is a humidity that is comfortable for the user.

[0088] Next, the communication unit 84 acquires temperature data indicating the current room temperatures of the rooms 90a to 90c from the temperature sensors 60a to 60c installed in each of the rooms 90a to 90c (step S22). The acquired temperature data is integrated and stored in the memory unit 83 as temperature information indicating the current room temperatures of each of the rooms 90a to 90c. Also, in step S22, the communication unit 84 acquires humidity data indicating the current room humidity of each of the rooms from the humidity sensors 61b and 61c installed in the two first rooms. The acquired humidity data is integrated and stored in the memory unit 83 as humidity information indicating the current room humidity of each of the two first rooms.

[0089] Next, the control unit 82 determines control values ​​for the air conditioner 21, the transport device 23, and the humidifier 50 using the stored temperature information and humidity information and the reinforcement learning model (step S24). The control values ​​for the air conditioner 21, the transport device 23, and the humidifier 50 are values ​​output from the reinforcement learning model by using the room temperature indicated by the temperature information and the room humidity indicated by the humidity information as inputs to the reinforcement learning model. Here, the inputs and outputs of this reinforcement learning model will be explained using Table 3.

[0090] [Table 3]

[0091] In this embodiment, the inputs used are the room temperatures of the multiple rooms 90a-90c acquired by the multiple temperature sensors 60a-60c, respectively. The inputs also include the room humidities of the two first rooms (rooms 90b and 90c) acquired by the multiple humidity sensors 61b and 61c, respectively. The outputs also include the amount of heat indicating the degree to which air is heated or cooled by the air conditioner 21 and the air volume transported by each of the multiple transport fans 23a and 23b. The outputs also include the opening degrees indicating the degree to which each of the multiple VAV dampers 24a-24c is open or closed, and the humidification amount indicating the degree to which air is humidified by the humidifier 50.

[0092] Furthermore, the control device 80 (control unit 82) controls the air conditioner 21, the transport device 23, and the humidifier 50 in accordance with the determined control values ​​(i.e., the control values ​​of the air conditioner 21, the transport device 23, and the humidifier 50 output from the reinforcement learning model) (step S26). The control unit 82 controls the air conditioner 21, the transport device 23, and the humidifier 50 so that the room temperature of each of the multiple rooms 90a to 90c approaches the target temperature of each of the multiple rooms 90a to 90c. As described above, the target temperature is a temperature that is comfortable for the user. Therefore, in other words, the control unit 82 controls the air conditioner 21, the transport device 23, and the humidifier 50 so that the room temperature of each of the multiple rooms 90a to 90c becomes a comfortable temperature.

[0093] Furthermore, the control unit 82 controls the air conditioner 21, the transport device 23, and the humidifier 50 so that the room humidity of each of the two first rooms approaches the target humidity of each of the two first rooms. As described above, the target humidity is a humidity that is comfortable for the user. Therefore, in other words, the control unit 82 controls the air conditioner 21, the transport device 23, and the humidifier 50 so that the room humidity of each of the two first rooms becomes a comfortable humidity.

[0094] Furthermore, after the processing of step S26 is performed and a predetermined period of time has elapsed, the control unit 82 performs learning of the reinforcement learning model (step S28). Note that the predetermined period of time is, for example, several minutes to several tens of minutes, but is not limited to this.

[0095] As described above, by using the room temperature indicated by the acquired temperature information and the room humidity indicated by the humidity information as inputs to the reinforcement learning model, the reinforcement learning model outputs control values ​​for the air conditioner 21, the transport device 23, and the humidifier 50. Here, the reinforcement learning model is trained using the following as rewards when the control unit 82 controls the air conditioner 21, the transport device 23, and the humidifier 50 in accordance with these control values. The rewards are the difference between the room temperature and the target temperature, and the difference between the room humidity and the target humidity. More specifically, as shown in Table 3, the control unit 82 trains the reinforcement learning model using the smallness of the difference between the room temperature and the target temperature and the smallness of the difference between the room humidity and the target humidity as rewards. In reinforcement learning, the reinforcement learning model is trained to maximize this reward, that is, to reduce the difference between the room temperature and the target temperature and to reduce the difference between the room humidity and the target humidity.

[0096] When step S28 is completed, the operation example 3 is completed. Here, the processes from step S20 to step S28 may be repeated.

[0097] By repeating the process in this manner, the control unit 82 repeatedly learns the reinforcement learning model. As described above, the reinforcement learning model repeatedly learns so as to maximize the reward. The control unit 82 controls the air conditioner 21, the transport device 23, and the humidifier 50 in accordance with the control values ​​that are the output of such a reinforcement learning model, which makes it easier for the room temperature of each of the multiple rooms 90a to 90c to approach the target temperature of each of the multiple rooms 90a to 90c. In other words, by using the reinforcement learning model that has been repeatedly trained, the control unit 82 can control the air conditioner 21, the transport device 23, and the humidifier 50 so that the room temperature of each of the multiple rooms 90a to 90c becomes a more comfortable temperature.

[0098] Furthermore, by having the control unit 82 control the air conditioner 21, the transport device 23, and the humidifier 50 in accordance with the control values ​​that are the output of such a reinforcement learning model, the room humidity of each of the two first rooms becomes more likely to approach the target humidity of each of the two first rooms. In other words, by using the reinforcement learning model that has been repeatedly trained, the control unit 82 can control the air conditioner 21, the transport device 23, and the humidifier 50 so that the room humidity of each of the two first rooms becomes a comfortable humidity.

[0099] In the third operation example, the control device 80 controls the air conditioner 21, the transport device 23, and the humidifier 50 without using the arrangement information of the plurality of devices.

[0100] As shown in Operation Example 3, the central air-conditioning system 10a according to this embodiment does not require a complicated installation procedure and can control the temperature and humidity to be comfortable.

[0101] Here, the control unit 82 controlled the air conditioner 21, the conveying device 23, and the humidifier 50 as follows. The control unit 82 controlled the air conditioner 21, the conveying device 23, and the humidifier 50 so that the room temperature of each of the multiple rooms 90a to 90c approached the target temperature of each of the multiple rooms 90a to 90c and the room humidity of each of the two first rooms approached the target humidity of each of the two first rooms. In other words, control was performed to achieve both of the two modes. One of the two modes is a mode in which the room temperature of each of the multiple rooms 90a to 90c is controlled to approach the target temperature of each of the multiple rooms 90a to 90c. The other of the two modes is a mode in which the room humidity of each of the two first rooms is controlled to approach the target humidity of each of the two first rooms. However, this is not a limitation, and the control unit 82 may control the air conditioner 21, the conveying device 23, and the humidifier 50 so that one mode is prioritized. For example, the operation receiving unit 81 may receive an operation instructing priority to one mode, and the control unit 82 may control the air conditioning device 21, the conveying device 23, and the humidifying device 50 in accordance with the received operation so that one mode is prioritized.

[0102] In Operation Example 3, the control unit 82 controls the air conditioner 21, the transport device 23, and the humidifier 50 so that the room temperature approaches the target temperature and the room humidity approaches the target humidity. However, this is not limited to this. For example, in Operation Example 3, the control unit 82 may further control the air conditioner 21, the transport device 23, and the humidifier 50 so that the power consumption of the air conditioner 21, the transport device 23, and the humidifier 50 is reduced. This allows the central air conditioning system 10a according to this embodiment to control the temperature and humidity to a comfortable level without requiring a complicated installation procedure, and further reduces power consumption. In this case, the control unit 82 may train the reinforcement learning model using a small difference between the room temperature and the target temperature, a small power consumption of the air conditioner 21, the transport device 23, and the humidifier 50, and a small difference between the room humidity and the target humidity as rewards.

[0103] Furthermore, even in this case, it is preferable to repeatedly perform learning of the reinforcement learning model. By having the control unit 82 control the air conditioner 21, the transport device 23, and the humidifier 50 according to the control values ​​that are the outputs of such a reinforcement learning model, the power consumption of the air conditioner 21, the transport device 23, and the humidifier 50 can be further reduced.

[0104] [Effects, etc.] The central air-conditioning system 10a according to this embodiment further includes a humidifier 50 that humidifies the air delivered to each of the multiple rooms 90a to 90c. The control device 80 acquires the room temperature of each of the multiple rooms 90a to 90c and the room humidity of a first room, which is at least one of the multiple rooms 90a to 90c. The control device 80 uses the acquired room temperature and the acquired room humidity as inputs to a reinforcement learning model, and controls the air conditioner 21, the transport device 23, and the humidifier 50 in accordance with control values ​​for the air conditioner 21, the transport device 23, and the humidifier 50 output from the reinforcement learning model. The control device 80 controls the air conditioner 21, the transport device 23, and the humidifier 50 so that the room humidity of the room approaches the target humidity of the first room.

[0105] As a result, as shown in Operation Example 3, the central air-conditioning system 10a according to this embodiment does not require a complicated installation procedure and can control the temperature and humidity to be comfortable.

[0106] As in Operation Example 3, the reinforcement learning model performs reinforcement learning using, as a reward, the difference between the room humidity and the target humidity when the control device 80 controls the air conditioner 21, the transport device 23, and the humidifier 50 in accordance with the control value. The control value is the control value of the air conditioner 21, the transport device 23, and the humidifier 50 output from the reinforcement learning model.

[0107] According to the control values ​​output from such a reinforcement learning model, the control device 80 (controller 82) controls the air conditioner 21, the conveying device 23, and the humidifier 50. This makes it easier for the room humidity of each of the two first rooms to approach the target humidity of each of the two first rooms. In other words, it becomes easier for the room humidity of each of the two first rooms to become a more comfortable humidity.

[0108] Therefore, the central air-conditioning system 10a according to this embodiment does not require a complicated installation procedure and can control the temperature and humidity to be more comfortable.

[0109] (Embodiment 3) [composition] The configuration of a central air-conditioning system 10b according to the third embodiment will be described below.

[0110] Fig. 8 is a diagram showing a schematic configuration of a central air-conditioning system 10b according to this embodiment. Fig. 9 is a block diagram showing a functional configuration of the central air-conditioning system 10b according to this embodiment.

[0111] The central air-conditioning system 10b differs from the central air-conditioning system 10 in that it includes a carbon dioxide sensor 62a and a ventilation device 30 instead of multiple humidity sensors 61b and 61c and a humidifier 50. That is, the central air-conditioning system 10b includes a central air-conditioning device 20, multiple temperature sensors 60a to 60c, a carbon dioxide sensor 62a, a control device 80, and a ventilation device 30. The carbon dioxide sensor 62a and the ventilation device 30 each have a function of communicating with the control device 80.

[0112] The ventilation device 30 is a device that, under the control of the control device 80, takes in air from outside the facility 90 into the facility 90 and conveys the air to the central air-conditioning device 20. The ventilation device 30 is, for example, a heat exchange ventilation device.

[0113] One carbon dioxide sensor 62a is placed in room 90a and measures the carbon dioxide concentration in room 90a. The carbon dioxide sensor 62a has a communication function and can transmit carbon dioxide data indicating the measured carbon dioxide concentration to the control device 80. The room in which the carbon dioxide sensor 62a is placed corresponds to the second room, and in this embodiment, the second room corresponds to room 90a. At least one second room in which the carbon dioxide sensor 62a is placed is sufficient. A high carbon dioxide concentration in a room can have harmful effects on the human body, such as fatigue, headaches, or tinnitus. Therefore, reducing the carbon dioxide concentration can improve the air quality in the room.

[0114] In this embodiment, the control unit 82 controls the air conditioner 21, the transport device 23, and the ventilation device 30 according to the control values ​​of the air conditioner 21, the transport device 23, and the ventilation device 30 output from the machine learning model.

[0115] This reinforcement learning model is a model in which the room temperatures of the plurality of rooms 90a to 90c and the carbon dioxide concentration of the second room are input, and the control values ​​of the air conditioner 21, the conveying device 23, and the ventilation device 30 are output.

[0116] The reinforcement learning model is a model that undergoes reinforcement learning using the following as rewards when the control device 80 controls the air conditioner 21, the transport device 23, and the ventilation device 30 in accordance with the control values ​​for the air conditioner 21, the transport device 23, and the ventilation device 30 output from the reinforcement learning model: In this embodiment, this reinforcement learning model is a model that undergoes reinforcement learning using the difference between the room temperature and the target temperature, and the carbon dioxide concentration as rewards.

[0117] [Example 4] Next, a fourth operational example of the central air-conditioning system 10b will be described.

[0118] FIG. 10 is a flowchart of an operation example 4 of the central air-conditioning system 10b.

[0119] First, as in the first operational example, the operation receiving unit 81 of the control device 80 receives an operation to specify a target temperature for each of the rooms 90a to 90c (step S10).

[0120] Next, the communication unit 84 acquires temperature data indicating the current room temperatures of the rooms 90a-90c from the temperature sensors 60a-60c installed in each of the rooms 90a-90c (step S32). The acquired temperature data is integrated and stored in the memory unit 83 as temperature information indicating the current room temperatures of each of the rooms 90a-90c. Also in step S32, the communication unit 84 acquires carbon dioxide data indicating the current carbon dioxide concentration of the second room from the carbon dioxide sensor 62a installed in the second room. The acquired carbon dioxide data is stored in the memory unit 83 as carbon dioxide information indicating the current carbon dioxide concentration of the second room.

[0121] Next, the control unit 82 determines control values ​​for the air conditioner 21, the transport device 23, and the ventilation device 30 using the stored temperature information and carbon dioxide concentration information and the reinforcement learning model (step S34). The control values ​​for the air conditioner 21, the transport device 23, and the ventilation device 30 are values ​​output from the reinforcement learning model by using the room temperature indicated by the temperature information and the carbon dioxide concentration indicated by the carbon dioxide information as inputs to the reinforcement learning model. Here, the inputs and outputs of this reinforcement learning model will be explained using Table 4.

[0122] [Table 4]

[0123] In this embodiment, the room temperatures of the multiple rooms 90a-90c acquired by the multiple temperature sensors 60a-60c, respectively, are used as inputs. The carbon dioxide concentration of the second room (room 90a) acquired by the carbon dioxide sensor 62a is also used as inputs. The heat quantity indicating the degree to which air is heated or cooled by the air conditioner 21 and the air volume transported by each of the multiple transport fans 23a and 23b are also used as outputs. The opening degree indicating the degree to which each of the multiple VAV dampers 24a-24c is open or closed, and the intake volume indicating the amount of air taken in by the ventilation device 30 are also used as outputs.

[0124] Furthermore, the control device 80 (control unit 82) controls the air conditioner 21, the transport device 23, and the ventilation device 30 according to the determined control values ​​(i.e., the control values ​​of the air conditioner 21, the transport device 23, and the ventilation device 30 output from the reinforcement learning model) (step S36). The control unit 82 controls the air conditioner 21, the transport device 23, and the ventilation device 30 so that the room temperatures of the multiple rooms 90a to 90c approach the target temperatures of the multiple rooms 90a to 90c. As described above, the target temperatures are temperatures that are comfortable for the user. Therefore, in other words, the control unit 82 controls the air conditioner 21, the transport device 23, and the ventilation device 30 so that the room temperatures of the multiple rooms 90a to 90c become comfortable. Furthermore, the control unit 82 controls the air conditioner 21, the transport device 23, and the ventilation device 30 so that the carbon dioxide concentration in the second room is reduced. As described above, reducing the carbon dioxide concentration can improve the air quality in the rooms. Therefore, in other words, the control unit 82 controls the air conditioner 21, the conveying device 23, and the ventilation device 30 so as to improve the air quality in the second room.

[0125] Furthermore, after the processing of step S36 has been performed and a predetermined period has elapsed, the control unit 82 performs learning of the reinforcement learning model (step S38). Note that the predetermined period is, for example, several minutes to several tens of minutes, but is not limited to this.

[0126] As described above, by using the room temperature indicated by the acquired temperature information and the carbon dioxide concentration indicated by the carbon dioxide information as inputs to the reinforcement learning model, the reinforcement learning model outputs control values ​​for the air conditioner 21, the transport device 23, and the ventilation device 30. Here, the reinforcement learning model is trained using the following as rewards when the control unit 82 controls the air conditioner 21, the transport device 23, and the ventilation device 30 in accordance with these control values. The rewards are the difference between the room temperature and the target temperature and the carbon dioxide concentration. More specifically, as shown in Table 4, the control unit 82 trains the reinforcement learning model using a small difference between the room temperature and the target temperature and a low carbon dioxide concentration as rewards. In reinforcement learning, the reinforcement learning model is trained to maximize this reward, that is, to reduce the difference between the room temperature and the target temperature and reduce the carbon dioxide concentration.

[0127] When step S38 is completed, the operation example 4 is completed. Here, the processes from step S10 to step S38 may be repeated.

[0128] By repeating this process, the control unit 82 repeatedly learns the reinforcement learning model. As described above, the reinforcement learning model repeatedly learns to maximize the reward. By controlling the air conditioner 21, the transport device 23, and the ventilation device 30 according to the control values ​​that are the output of this reinforcement learning model, the control unit 82 can more easily bring the room temperatures of the multiple rooms 90a-90c closer to the target temperatures of the multiple rooms 90a-90c. In other words, by using the reinforcement learning model that has been repeatedly trained, the control unit 82 can control the air conditioner 21, the transport device 23, and the ventilation device 30 so that the room temperatures of the multiple rooms 90a-90c become more comfortable. Furthermore, by controlling the air conditioner 21, the transport device 23, and the ventilation device 30 according to the control values ​​that are the output of this reinforcement learning model, the carbon dioxide concentration in the second room is more likely to decrease. That is, by using the reinforcement learning model that has been repeatedly trained, the control unit 82 can control the air conditioner 21, the conveying device 23, and the ventilation device 30 so as to further improve the air quality in the second room.

[0129] In addition, in the fourth operation example, the control device 80 controls the air conditioner 21, the conveyance device 23, and the ventilation device 30 without using the arrangement information of the plurality of devices.

[0130] As shown in Operation Example 4, the central air-conditioning system 10b according to this embodiment does not require complicated installation procedures and can control the air to have high quality and a comfortable temperature.

[0131] Here, the control unit 82 controls the air conditioner 21, the transport device 23, and the ventilation device 30 so that the room temperatures of the rooms 90a-90c approach the target temperatures of the rooms 90a-90c and so that the carbon dioxide concentration in the second room is reduced. In other words, a mode in which the room temperatures of the rooms 90a-90c are controlled so that the target temperatures of the rooms 90a-90c are controlled, and a mode in which the carbon dioxide concentration in the second room is reduced are both possible. However, this is not a limitation. The control unit 82 may control the air conditioner 21, the transport device 23, and the ventilation device 30 so that one mode is prioritized. For example, the operation receiving unit 81 may receive an operation instructing priority over one mode, and the control unit 82 may control the air conditioner 21, the transport device 23, and the ventilation device 30 so that one mode is prioritized in accordance with the received operation.

[0132] In Operation Example 4, the control unit 82 controls the air conditioner 21, the transport device 23, and the ventilation device 30 so that the room temperature approaches the target temperature and the carbon dioxide concentration decreases. However, this is not limited to this. For example, in Operation Example 4, the control unit 82 may further control the air conditioner 21, the transport device 23, and the ventilation device 30 so that the power consumption of the air conditioner 21, the transport device 23, and the ventilation device 30 decreases. As a result, the central air conditioning system 10b according to this embodiment can control the air to a comfortable temperature with high air quality without requiring a complicated installation procedure, and can also reduce power consumption. In this case, the control unit 82 may train the reinforcement learning model using a small difference between the room temperature and the target temperature, small power consumption of the air conditioner 21, the transport device 23, and the ventilation device 30, and a low carbon dioxide concentration as rewards.

[0133] Furthermore, even in this case, it is preferable to repeatedly perform learning of the reinforcement learning model. By having the control unit 82 control the air conditioner 21, the transport device 23, and the ventilation device 30 according to the control values ​​that are the outputs of such a reinforcement learning model, the power consumption of the air conditioner 21, the transport device 23, and the ventilation device 30 can be further reduced.

[0134] [Effects, etc.] The central air-conditioning system 10b according to this embodiment further includes a ventilation device 30 that ventilates the air in each of the multiple rooms 90a to 90c. The control device 80 acquires the room temperature of each of the multiple rooms 90a to 90c and the carbon dioxide concentration of a second room, which is at least one of the multiple rooms 90a to 90c. The control device 80 uses the acquired room temperature and the acquired carbon dioxide concentration as inputs to a reinforcement learning model, and controls the air conditioner 21, the transport device 23, and the ventilation device 30 in accordance with control values ​​for the air conditioner 21, the transport device 23, and the ventilation device 30 output from the reinforcement learning model. The control device 80 controls the air conditioner 21, the transport device 23, and the ventilation device 30 so as to reduce the carbon dioxide concentration in the second room.

[0135] As a result, as shown in Operation Example 4, the central air-conditioning system 10b according to this embodiment does not require a complicated installation procedure and can control the air to have high quality and a comfortable temperature.

[0136] As in Operation Example 4, the reinforcement learning model performs reinforcement learning using the carbon dioxide concentration as a reward when the control device 80 controls the air conditioner 21, the transport device 23, and the ventilation device 30 in accordance with the control value. The control value is the control value of the air conditioner 21, the transport device 23, and the ventilation device 30 output from the reinforcement learning model.

[0137] The control device 80 (control unit 82) controls the air conditioner 21, the conveying device 23, and the ventilation device 30 in accordance with the control values ​​output from such a reinforcement learning model, which makes it easier to lower the carbon dioxide concentration in the second room. In other words, it makes it easier to improve the air quality in the second room.

[0138] Therefore, the central air-conditioning system 10b according to this embodiment does not require complicated installation procedures and can control the air quality to be higher and at a more comfortable temperature.

[0139] (Modification of the third embodiment) The configuration of a central air-conditioning system 10c according to a modification of the third embodiment will be described below.

[0140] Fig. 11 is a diagram showing a schematic configuration of a central air-conditioning system 10c according to a modified example of the present embodiment. Fig. 12 is a block diagram showing a functional configuration of the central air-conditioning system 10c according to a modified example of the present embodiment. The central air-conditioning system 10c includes a central air-conditioning unit 20, a plurality of temperature sensors 60a to 60c, a plurality of humidity sensors 61b and 61c, a carbon dioxide sensor 62a, a control device 80, a humidifier 50, and a ventilation device 30.

[0141] In this modification, the control unit 82 controls the air conditioner 21, the transport device 23, the humidifier 50, and the ventilation device 30 according to the control values ​​of the air conditioner 21, the transport device 23, the humidifier 50, and the ventilation device 30 output from the machine learning model. For simplicity, the air conditioner 21, the transport device 23, the humidifier 50, and the ventilation device 30 may be referred to as controlled devices. The reinforcement learning model is a model that receives as input the room temperatures of each of the multiple rooms 90a to 90c, the carbon dioxide concentration of the second room, and the room humidity of each of the two first rooms, and outputs the control values ​​of the controlled devices.

[0142] In this modified example, the reinforcement learning model is a model that undergoes reinforcement learning using the following as rewards when the control device 80 (control unit 82) controls the controlled device in accordance with the control value of the controlled device output from the reinforcement learning model: This reinforcement learning model is a model that undergoes reinforcement learning using the difference between the room temperature and the target temperature, the carbon dioxide concentration, and the difference between the room humidity and the target humidity as rewards.

[0143] The control unit 82 according to this modification determines a control value of the controlled device using the temperature information, carbon dioxide information, and humidity information, as well as a reinforcement learning model. The control value of the controlled device is a value output from the reinforcement learning model by using the room temperature indicated by the temperature information, the room humidity indicated by the humidity information, and the carbon dioxide concentration indicated by the carbon dioxide information as inputs to the reinforcement learning model.

[0144] Here, the inputs and outputs of this reinforcement learning model are explained using Table 5.

[0145] [Table 5]

[0146] In this modification, the inputs used are the room temperatures of the multiple rooms 90a-90c acquired by the multiple temperature sensors 60a-60c, respectively. The inputs used are the room humidities of the two first rooms (rooms 90b and 90c) acquired by the multiple humidity sensors 61b and 61c, respectively. The inputs used are the carbon dioxide concentration of the second room (room 90a) acquired by the carbon dioxide sensor 62a. The outputs used are the amount of heat indicating the degree to which air is heated or cooled by the air conditioner 21, the air volume transported by each of the multiple transport fans 23a and 23b, and the opening degrees indicating the degree to which each of the multiple VAV dampers 24a-24c is open or closed. The outputs used are the humidification amount indicating the degree to which air is humidified by the humidifier 50 and the intake amount indicating the amount of air taken in by the ventilation device 30.

[0147] Furthermore, the control device 80 (control unit 82) controls the controlled device in accordance with the determined control value (i.e., the control value of the controlled device output from the reinforcement learning model). Note that the control unit 82 controls the controlled device so that the room temperature of each of the multiple rooms 90a to 90c approaches the target temperature of each of the multiple rooms 90a to 90c.

[0148] Furthermore, the control unit 82 controls the controlled devices according to the control values ​​of the controlled devices so that the room humidity of each of the two first rooms approaches the target humidity of each of the two first rooms. Furthermore, the control unit 82 controls the controlled devices according to the control values ​​of the controlled devices so that the carbon dioxide concentration in the second room decreases.

[0149] In this modification, the control device 80 controls the controlled device without using the layout information of the plurality of devices.

[0150] Therefore, the central air-conditioning system 10c according to this modified example does not require a complicated installation procedure, and can control the temperature and humidity to be high quality and comfortable.

[0151] Furthermore, as shown in Table 5, the control unit 82 may perform training of the reinforcement learning model using a small difference between the room temperature and the target temperature, a small difference between the room humidity and the target humidity, and a low carbon dioxide concentration as rewards.

[0152] (Other embodiments) Although the embodiment and modifications have been described above, the present invention is not limited to the above embodiment and modifications.

[0153] In the above-described embodiment, the processing executed by a specific processing unit may be executed by another processing unit.

[0154] Furthermore, the communication method between the devices in the above-described embodiments is not particularly limited. Wireless communication or wired communication may be performed between the devices. Furthermore, wireless communication and wired communication may be combined between the devices. Furthermore, when two devices communicate in the above-described embodiments and modifications, a relay device (not shown) may be interposed between the two devices.

[0155] The order of the processes described in the flowcharts of the above embodiments is merely an example. The order of the processes may be changed, or the processes may be executed in parallel.

[0156] In the above-described embodiments, each component may be realized by executing a software program suitable for that component, or by a program execution unit such as a CPU or processor reading and executing a software program recorded on a recording medium such as a hard disk or semiconductor memory.

[0157] Furthermore, each component may be realized by hardware. For example, each component may be a circuit (or integrated circuit). These circuits may form a single circuit as a whole, or each may be a separate circuit. Furthermore, each of these circuits may be a general-purpose circuit or a dedicated circuit.

[0158] Furthermore, the general or specific aspects of the present invention may be realized as a system, an apparatus, a method, an integrated circuit, a computer program, or a computer-readable recording medium such as a CD-ROM, or as any combination of a system, an apparatus, a method, an integrated circuit, a computer program, and a recording medium.

[0159] For example, the present invention may be realized as a control method for a central air conditioning system executed by a computer, or as a program for causing a computer to execute such a control method. Furthermore, the present invention may be realized as a computer-readable non-transitory recording medium on which such a program is recorded.

[0160] Furthermore, in the above embodiment, the central air-conditioning system is realized by a plurality of devices, but it may also be realized as a single device. The central air-conditioning system may be realized, for example, as a single device corresponding to a control device, or as a single device corresponding to a server device. When the central air-conditioning system is realized by a plurality of devices, the components of the central air-conditioning system may be distributed among the plurality of devices in any manner.

[0161] In addition, the present invention also includes forms obtained by applying various modifications to each embodiment that a person skilled in the art would think of, or forms realized by arbitrarily combining the components and functions of each embodiment within the scope of the present invention. [Explanation of symbols]

[0162] 10, 10a, 10b, 10c Whole building air conditioning system 21 Air conditioner 23 Conveyor equipment 30 Ventilation equipment 50 Humidifier 80 Control device 90 facilities Rooms 90a, 90b, and 90c

Claims

1. An air conditioning device, a transport device including a plurality of devices arranged within the facility to transport the air whose temperature has been adjusted by the air conditioning device to each of a plurality of rooms within the facility; a control device; The control device Acquire the room temperatures of all of the rooms among the plurality of rooms; using all of the acquired room temperatures as inputs to a reinforcement learning model without using the location information of the plurality of devices, and controlling the air conditioning devices and the transportation devices in accordance with control values ​​of the air conditioning devices and the transportation devices output from the reinforcement learning model so that the room temperatures of all of the rooms approach target temperatures of all of the rooms; The arrangement information is information indicating to which room among all the rooms each of the plurality of devices included in the transport device is associated with transporting air. Air conditioning system throughout the building.

2. An air conditioning device; a transport device including a plurality of devices arranged within the facility to transport the air whose temperature has been adjusted by the air conditioning device to each of a plurality of rooms within the facility; a control device; The control device acquiring room temperatures for each of the plurality of rooms; using the acquired room temperatures as an input for a reinforcement learning model without using the arrangement information of the plurality of devices, and controlling the air conditioning devices and the transportation devices in accordance with control values ​​of the air conditioning devices and the transportation devices output from the reinforcement learning model so that the room temperatures of each of the plurality of rooms approach target temperatures of each of the plurality of rooms; the location information is information indicating locations where the plurality of devices are located within the facility, The reinforcement learning model performs reinforcement learning using, as a reward, a difference between the room temperature and the target temperature when the control device controls the air conditioner and the transportation device in accordance with the control values ​​of the air conditioner and the transportation device output from the reinforcement learning model. Air conditioning system throughout the building.

3. An air conditioning device; a transport device including a plurality of devices arranged within the facility to transport the air whose temperature has been adjusted by the air conditioning device to each of a plurality of rooms within the facility; a control device; The control device acquiring room temperatures for each of the plurality of rooms; using the acquired room temperatures as an input for a reinforcement learning model without using the arrangement information of the plurality of devices, and controlling the air conditioning devices and the transportation devices in accordance with control values ​​of the air conditioning devices and the transportation devices output from the reinforcement learning model so that the room temperatures of each of the plurality of rooms approach target temperatures of each of the plurality of rooms; the location information is information indicating locations where the plurality of devices are located within the facility, The control device controls the air conditioning device and the transport device so as to reduce power consumption of the air conditioning device and the transport device. Air conditioning system throughout the building.

4. The reinforcement learning model performs reinforcement learning using, as a reward, power consumption of the air conditioning device and the transportation device when the control device controls the air conditioning device and the transportation device in accordance with the control values ​​of the air conditioning device and the transportation device output from the reinforcement learning model. The central air conditioning system according to claim 3.

5. An air conditioning device; a transport device including a plurality of devices arranged within the facility to transport the air whose temperature has been adjusted by the air conditioning device to each of a plurality of rooms within the facility; a control device; a humidifier that humidifies the air delivered to each of the plurality of rooms, The control device acquiring room temperatures for each of the plurality of rooms; using the acquired room temperatures as an input for a reinforcement learning model without using the arrangement information of the plurality of devices, and controlling the air conditioning devices and the transportation devices in accordance with control values ​​of the air conditioning devices and the transportation devices output from the reinforcement learning model so that the room temperatures of each of the plurality of rooms approach target temperatures of each of the plurality of rooms; the location information is information indicating locations where the plurality of devices are located within the facility, The control device acquiring the room temperature of each of the plurality of rooms and the room humidity of a first room that is at least one of the plurality of rooms; Using the acquired room temperature and the acquired room humidity as inputs to the reinforcement learning model, the air conditioner, the transport device, and the humidifier are controlled in accordance with control values ​​for the air conditioner, the transport device, and the humidifier output from the reinforcement learning model so that the room humidity in the first room approaches a target humidity for the first room. Air conditioning system throughout the building.

6. The reinforcement learning model performs reinforcement learning using, as a reward, a difference between the room humidity and the target humidity when the control device controls the air conditioner, the transport device, and the humidifier in accordance with the control values ​​of the air conditioner, the transport device, and the humidifier output from the reinforcement learning model. The central air conditioning system according to claim 5.

7. An air conditioning device; a transport device including a plurality of devices arranged within the facility to transport the air whose temperature has been adjusted by the air conditioning device to each of a plurality of rooms within the facility; a control device; a ventilation device that ventilates the air in each of the plurality of rooms, The control device acquiring room temperatures for each of the plurality of rooms; using the acquired room temperatures as an input for a reinforcement learning model without using the arrangement information of the plurality of devices, and controlling the air conditioning devices and the transportation devices in accordance with control values ​​of the air conditioning devices and the transportation devices output from the reinforcement learning model so that the room temperatures of each of the plurality of rooms approach target temperatures of each of the plurality of rooms; the location information is information indicating locations where the plurality of devices are located within the facility, The control device acquiring the room temperature of each of the plurality of rooms and the carbon dioxide concentration of a second room that is at least one room among the plurality of rooms; Using the acquired room temperature and the acquired carbon dioxide concentration as inputs to the reinforcement learning model, the air conditioning device, the transport device, and the ventilation device are controlled in accordance with control values ​​for the air conditioning device, the transport device, and the ventilation device output from the reinforcement learning model so as to reduce the carbon dioxide concentration in the second room. Air conditioning system throughout the building.

8. The reinforcement learning model performs reinforcement learning using the carbon dioxide concentration as a reward when the control device controls the air conditioning device, the transport device, and the ventilation device in accordance with the control values ​​of the air conditioning device, the transport device, and the ventilation device output from the reinforcement learning model. The central air conditioning system according to claim 7.

9. A method for controlling a central air conditioning system, comprising: The whole-building air conditioning system is An air conditioning device, a transport device including a plurality of devices arranged within the facility to transport the air whose temperature has been adjusted by the air conditioning device to each of a plurality of rooms within the facility; Equipped with The control method includes: an acquisition step of acquiring room temperatures of all of the rooms among the plurality of rooms; a control step of using all of the acquired room temperatures as inputs to a reinforcement learning model without using arrangement information of the plurality of devices, and controlling the air conditioning devices and the transportation devices in accordance with control values ​​of the air conditioning devices and the transportation devices output from the reinforcement learning model so that the room temperatures of all of the rooms approach target temperatures of all of the rooms, The arrangement information is information indicating to which room among all the rooms each of the plurality of devices included in the transport device is associated with transporting air. How to control a whole-building air conditioning system.

10. A method for controlling a central air conditioning system, comprising: The whole-building air conditioning system is An air conditioning device, a transport device including a plurality of devices arranged within the facility to transport the air whose temperature has been adjusted by the air conditioning device to each of a plurality of rooms within the facility; Equipped with The control method includes: an acquisition step of acquiring room temperatures of each of the plurality of rooms; a control step of using the acquired room temperatures as an input of a reinforcement learning model without using arrangement information of the plurality of devices, and controlling the air conditioning devices and the transportation devices in accordance with control values ​​of the air conditioning devices and the transportation devices output from the reinforcement learning model so that the room temperatures of each of the plurality of rooms approach target temperatures of each of the plurality of rooms, the location information is information indicating locations where the plurality of devices are located within the facility, The reinforcement learning model performs reinforcement learning using, as a reward, a difference between the room temperature and the target temperature when the control method controls the air conditioning device and the transportation device in accordance with the control values ​​of the air conditioning device and the transportation device output from the reinforcement learning model. Air conditioning system throughout the building.

Images