Air-conditioning control device

Abstract

An air-conditioning control device (3) for controlling operation of an air conditioner (4), the air-conditioning control device (3) comprising: a data collection section (32) that acquires training data including data at the time of startup and data at the time of shutdown of the air conditioner (4); a learning section (34) that uses the training data to generate a trained model for figuring a startup lead time from startup of the air conditioner (4) until the indoor temperature reaches a set temperature; a calculation section (35) that uses the trained model to output the startup lead time from the data at the time of startup acquired by the data collection section (32); and a control section (36) that causes the air conditioner (4) to start operating the startup lead time before a predetermined set time.

Description

Air conditioning control device

[0001] The present disclosure relates to an air conditioning control device that controls the operation of an air conditioner.

[0002] Conventionally, there is a technology for efficiently controlling an air conditioner using various information such as room temperature, outside air temperature, and heat retention ability of a building. For example, in Patent Document 1, using performance data that is information such as room temperature, target temperature, and temperature and required time to reach the target temperature when the air conditioner was started in the past, the time required to reach the target temperature is estimated, and the optimal startup control for starting the operation of the air conditioner before the target time is described. Further, Patent Document 2 discloses a technology for determining an optimal operation mode from room temperature, outside air temperature, and heat retention ability of a building. Furthermore, Patent Document 2 describes that the heat retention ability of a building is estimated from the indoor temperature change rate, which is the temperature change of the indoor temperature within a certain time after the air conditioner stops.

[0003] Japanese Patent Application Laid-Open No, 20, 16-061487 Japanese Patent Application Laid-Open No, 20, 21-508554

[0004] In the startup control of an air conditioner, the startup preparation time, which is the time required to reach the target temperature, is affected by the performance of the building where the air conditioner is installed, such as the building's structural heat storage amount, in addition to the target temperature and room temperature. By collecting past performance data and learning the startup preparation time as learning data, it is possible to estimate the startup preparation time considering the influence of the building's performance. However, since the learning data used for estimating the startup preparation time is affected by the building, it needs to be collected after the air conditioner is installed in the building, and there is a problem that it takes time to collect the learning data. Furthermore, for example, when the building where the air conditioner is installed is used as an office, the air conditioner starts at the start time and stops at the end time, so data at startup can only be obtained once a day, which is one of the reasons for the time-consuming collection of learning data.

[0005] In order to solve the above problems, the present disclosure aims to obtain an air conditioning control device that collects learning data required for accurately estimating the startup preparation time of an air conditioner considering the performance of a building in a shorter period than when only using startup data.

[0006] The air conditioning control device according to this disclosure is an air conditioning control device for controlling the operation of an air conditioner, and comprises: a data collection unit that acquires training data including data when the air conditioner is started and data when it is stopped; a learning unit that uses the training data to generate a trained model for inferring the start-up buffer time from when the air conditioner is started until the room temperature reaches a set temperature; a calculation unit that uses the trained model to output the start-up buffer time from the start-up conditions acquired by the data collection unit; and a control unit that causes the air conditioner to start operation before the start-up buffer time, which is set to a predetermined time.

[0007] According to this disclosure, an air conditioning control device can be obtained that collects training data for accurately estimating the startup buffer time of an air conditioner, taking into account the building's performance, in a shorter period of time than when using only startup data.

[0008] This is a diagram showing an example configuration of an air conditioning system including an air conditioning control device according to Embodiment 1. This is a block diagram showing the functional configuration of the air conditioning control device according to Embodiment 1. This is a diagram showing an example of a neural network in the inference of the startup buffer time of the air conditioning control device according to Embodiment 1. This is a graph showing the time change of room temperature and outside temperature. This is a flowchart showing the flow of data collection and learning in the air conditioning control device according to Embodiment 1. This is a flowchart showing the data acquisition process flow when the air conditioning control device according to Embodiment 1 starts up. This is a flowchart showing the data acquisition process flow when the air conditioning control device according to Embodiment 1 stops up. This is a flowchart showing the learning flow of the startup buffer time in the air conditioning control device according to Embodiment 1. This is a flowchart showing the inference of the startup buffer time and the control flow of the air conditioner in the air conditioning control device according to Embodiment 1. This is a graph showing the time change of room temperature and outside temperature when temperature difference control processing is performed. This is a flowchart showing the data collection and learning flow in the air conditioning control device according to Embodiment 2. This is a flowchart showing the temperature difference control process in the air conditioning control device according to Embodiment 2. This is a diagram showing an example configuration of an air conditioning system including an air conditioning control device according to Embodiment 3. This is a graph showing the time change of room temperature when stop time difference processing is performed. This is a flowchart showing the data collection and learning flow in the air conditioning control device according to Embodiment 3. This flowchart shows the data collection and learning process for the air conditioning control device according to Embodiment 4. This flowchart also shows the learning process for the startup buffer time in the air conditioning control device according to Embodiment 4.

[0009] The embodiments for carrying out the subject matter of this disclosure will be described with reference to the attached drawings. In each drawing, the same or corresponding parts are denoted by the same reference numerals, and redundant explanations are simplified or omitted as appropriate. However, the subject matter of this disclosure is not limited to the following embodiments, and any modification of any component of the embodiments, or any combination or omission of any component of the embodiments is possible without departing from the spirit of this disclosure.

[0010] Embodiment 1. Figure 1 is a diagram showing an example configuration of an air conditioning system 100 including an air conditioning control device 3 according to this embodiment. The air conditioning control device 3 and the air conditioning system 100 will be described with reference to Figure 1. The air conditioning system 100 comprises a control server 1, a management device 2 installed in a building 10, an air conditioning control device 3, an air conditioner 4 consisting of an outdoor unit 5 and a plurality of indoor units 6, and a handheld remote control 7. Multiple air conditioners 4 may be installed in the building 10. In this embodiment, the building 10 is assumed to be a building used as an office or commercial facility, but it may also be a residence as long as an air conditioner 4 is installed. The control server 1 and the management device 2 are connected to each other so as to be able to communicate with each other via a wide-area network 11 such as the Internet. In the example of Figure 1, the management device 2 and the air conditioning control device 3 are connected, the air conditioning control device 3 and the outdoor unit 5 are connected, and the outdoor unit 5 and the plurality of indoor units 6 and the handheld remote control 7 are connected. For example, communication from the air conditioning control device 3 to the indoor units 6 is performed via the outdoor unit 5. By routing through these devices, the management device 2, the air conditioning control device 3, the outdoor unit 5, the indoor unit 6, and the handheld remote control 7 are configured to communicate with each other. Furthermore, the air conditioning control device 3, the outdoor unit 5, the indoor unit 6, and the handheld remote control 7 communicate with the control server 1 via the management device 2. Note that the connection is not limited to the example in Figure 1, as long as the management device 2, the air conditioning control device 3, the outdoor unit 5, the indoor unit 6, and the handheld remote control 7 can communicate with each other. Also, communication may be performed wirelessly.

[0011] The control server 1 is a computer for servers, and includes components such as a CPU (Central Processing Unit), ROM (Read Only Memory), RAM (Random Access Memory), and RTC (Real Time Clock). The CPU is also called a central processing unit, central computing unit, processor, microprocessor, microcomputer, or DSP (Digital Signal Processor), and functions as a central computing unit that executes each program. The control server 1 also has communication units such as a wired LAN (Local Area Network) or wireless LAN. The control server 1 acquires information about the air conditioner 4 and transmits control signals via the wide-area network 11.

[0012] The management device 2 manages the air conditioners 4 in the building 10. The management device 2 may also be connected to other equipment in the building 10, such as a lighting system (not shown), and manage them together with the air conditioners 4. The management device 2 communicates with the control server 1, transmits information about the air conditioners 4 connected to the management device 2 to the control server 1, and receives control signals for the air conditioners 4 from the control server 1. Note that the air conditioning system 100 may not have a management device 2, and the air conditioning control device 3, described later, may communicate with the control server 1.

[0013] The air conditioning control device 3 is a central controller that controls multiple air conditioners 4 installed in the building 10. The air conditioning control device 3 includes a CPU (Central Processing Unit), ROM (Read Only Memory), RAM (Random Access Memory), RTC (Real Time Clock), etc. The CPU is also called a central processing unit, central computing unit, processor, microprocessor, microcomputer, DSP (Digital Signal Processor), etc., and functions as a central computing unit that executes each program. Furthermore, the air conditioning control device 3 is equipped with non-volatile memory such as flash memory, EPROM (Erasable Programmable ROM), and EEPROM (Electrically Erasable Programmable ROM). The air conditioning control device 3 collects training data from the outdoor unit 5 and the indoor unit 6, and generates a trained model that infers the startup buffer time from the start of the air conditioner 4 until the set temperature is reached. The detailed configuration of the air conditioning control device 3 will be described later.

[0014] The outdoor unit 5 is an outdoor unit of the air conditioning system that is installed outside the building. The outdoor unit 5 is equipped with an outdoor temperature sensor 51 that detects the outside temperature. The outdoor unit 5 also stores the heating and cooling capacity of the air conditioner 4 on a per-unit time basis, based on the operating frequency of the compressor (not shown) installed in the outdoor unit 5. Air conditioning indoors means adjusting the temperature, humidity, air purity, etc. of the indoor air. The indoor unit 6 is an indoor unit of the air conditioning system that is installed inside the building. The indoor unit 6 blows air into the room for heating, cooling, dehumidification, ventilation, etc. The indoor unit 6 is equipped with an indoor temperature sensor 61 that detects the indoor temperature. The outdoor unit 5 and the indoor unit 6 are connected by a communication line and refrigerant piping (not shown) for circulating the refrigerant, thus forming the air conditioner 4. At least one indoor unit 6 is connected to each outdoor unit 5 by refrigerant piping and a communication line. The air conditioning control device 3 controls the outdoor unit 5 and the indoor unit 6, and controls the operation of the air conditioner 4.

[0015] The handheld remote control 7 is an operating device installed on the wall of the room where the indoor unit 6 is installed. By operating the handheld remote control 7, the user can turn the air conditioner 4 ON / OFF, change the set temperature, etc. Alternatively, the air conditioning system 100 may be configured so that the air conditioner 4 is controlled only from the air conditioning control device 3, without installing the handheld remote control 7.

[0016] Figure 2 is a block diagram showing the functional configuration of the air conditioning control device 3 according to this embodiment. The configuration of the air conditioning control device 3 will be explained with reference to Figure 2. The air conditioning control device 3 includes a communication unit 31, a data acquisition unit 32, a data processing unit 33, a learning unit 34, a calculation unit 35, a control unit 36, a learning data holding unit 37, and a learned model holding unit 38.

[0017] The communication unit 31 communicates with the management device 2, the outdoor unit 5, and the indoor unit 6. The communication unit 31 manages communications, for example, by forwarding a communication from the management device 2 to the outdoor unit 5 when it receives one.

[0018] The data collection unit 32 collects data for when the air conditioner 4 is started, including the startup conditions and the measured value of the startup buffer time from the start of the air conditioner 4 until the set temperature is reached, via the communication unit 31, as learning data used to learn the startup buffer time. The startup conditions are the outside temperature detected by the outside air temperature sensor 51 of the outdoor unit 5, the room temperature detected by the indoor temperature sensor 61 of the indoor unit 6, and the operating status of the air conditioner 4 at the time the air conditioner 4 is started. The operating status indicates the on / off state of the air conditioner 4, the operating mode such as cooling or heating, and the capacity saving amount of the outdoor unit 5. Furthermore, the data collection unit 32 collects data for when the air conditioner 4 is stopped, including the stopping conditions and the time of room temperature change after stopping, as learning data. The stopping conditions are the outside temperature detected by the outside air temperature sensor 51 of the outdoor unit 5, the room temperature detected by the indoor temperature sensor 61 of the indoor unit 6, and the operating status of the outdoor unit 5 and the indoor unit 6 at the time the air conditioner 4 is stopped. When the air conditioner 4 is stopped, the room temperature changes to approach the outside temperature. The time it takes for the room temperature to change after stopping is the time it takes for the rate of change in the room temperature after the air conditioner 4 is stopped to fall below a predetermined threshold. The rate of change in room temperature is the amount of change in room temperature per unit time. Specifically, the data acquisition unit 32 periodically (for example, every minute) acquires the room temperature and calculates the difference from the previously acquired room temperature as the rate of change in room temperature.

[0019] Figure 4 is a graph showing the time variation of room temperature and outside temperature. Now, referring to Figure 4, the data collected by the data acquisition unit 32 will be explained in detail. Startup time t0 is the time when the air conditioner 4 is started. Setting time tset is the time set in advance by the user on the air conditioning control device 3. Setting time tset is, for example, the start time if the building 10 is an office. The purpose of the air conditioning control device 3 is to control the air conditioner 4 so that the room temperature reaches the set temperature Tset at setting time tset. The set temperature Tset is set by the user on the air conditioning control device 3 or the handheld remote control 7. If sufficient training data has not been collected, startup time t0 will be set to a time a certain amount of time (for example, 1 hour) earlier than setting time tset. If sufficient training data has been collected, startup time t0 will be set to a time an estimated startup reserve time ton_est earlier than setting time tset. The data acquisition unit 32 collects startup data from startup time t0 until the room temperature reaches the set temperature Tset. The startup data includes the startup conditions at startup time t0 and the measured value ton of the startup reserve time. The data acquisition unit 32 acquires the room temperature Ton_in at the startup time t0 of the air conditioner 4 from the indoor temperature sensor 61 as the startup condition. The data acquisition unit 32 acquires the outdoor temperature Ton_out at the startup time t0 of the air conditioner 4 from the outdoor temperature sensor 51 as the startup condition. The data acquisition unit 32 acquires the set temperature Tset as the startup condition. The data acquisition unit 32 also stores the time from the startup time t0 when the air conditioner 4 is started until the time t1 when the room temperature reaches the set temperature Tset as the measured value ton of the startup reserve time. The determination of whether the room temperature has reached the set temperature Tset is made by the data acquisition unit 32 periodically acquiring the room temperature from the indoor temperature sensor 61 and comparing it with the set temperature Tset. Alternatively, the indoor unit 6 may be configured to notify the data acquisition unit 32 when the room temperature reaches the set temperature Tset.

[0020] Furthermore, since the startup reserve time varies depending on the cooling and heating capacity of the air conditioner 4, the data acquisition unit 32 may acquire the amount of heat processed by the air conditioner 4 between the startup time t0 and arrival time t1 as startup data. The amount of heat processed by the air conditioner 4 is determined by the capacity saving amount of the outdoor unit 5 equipped with the air conditioner 4 relative to the cooling and heating capacity of the air conditioner 4. Therefore, the data acquisition unit 32 may acquire the capacity saving amount of the outdoor unit 5 from the startup time t0 to arrival time t1 as startup data. By using the amount of heat processed by the air conditioner 4 or the capacity saving amount as training data, the startup reserve time can be learned with greater accuracy.

[0021] Furthermore, the data acquisition unit 32 may acquire the number of indoor units 6 that are running as startup data. The startup reserve time is learned for each indoor unit 6. If multiple indoor units 6 are installed in the same room for the indoor unit 6 being studied, the startup reserve time will be affected by the operating status of adjacent indoor units 6. Therefore, by using the number of running indoor units 6 as training data, the startup reserve time can be learned with greater accuracy.

[0022] The air conditioning control device 3 stops the air conditioner 4 at a stop time t2 that has been set in advance by the user. For example, if the building 10 is an office, the stop time t2 is set to the end of the workday. Alternatively, the stop time t2 may not be set in the air conditioning control device 3, but may be the time when the user stops the air conditioner 4 by operating the handheld remote control 7. The data acquisition unit 32 collects data at the time of stopping from the stop time t2 until the rate of change of the room temperature falls below a threshold. The data acquisition unit 32 obtains the room temperature Toff_in at the stop time t2 of the air conditioner 4 from the indoor temperature sensor 61 as the stop condition. The data acquisition unit 32 also obtains the outside temperature Toff_out at the stop time t2 of the air conditioner 4 from the outside temperature sensor 51 as the stop condition. When the air conditioner 4 is stopped, the room temperature changes to approach the outside temperature. However, it is known that after a sufficient amount of time has passed, the temperature change in the room temperature becomes gradual, as shown in Figure 4. In particular, if the building 10 is a building, buildings are often more well insulated than houses, and the building structure retains heat, so even if the air conditioner 4 is stopped, the room temperature will not drop to the outside temperature. The data collection unit 32 stores the time from the time t2 when the air conditioner 4 is stopped until the time t3 when the room temperature stabilizes and the rate of change of the room temperature falls below a predetermined threshold as the time of room temperature change after stopping, toff. The data collection unit 32 determines whether the rate of change of the room temperature has fallen below the threshold by periodically acquiring the room temperature from the indoor temperature sensor 61, calculating the rate of change, and comparing it with the threshold. Alternatively, the indoor unit 6 may be configured to notify the data collection unit 32 when the rate of change of the room temperature falls below the threshold. The data collection unit 32 stores the training data collected as described above in the training data storage unit 37.

[0023] The startup and shutdown data collected by the data acquisition unit 32 may also be data showing the relationship between room temperature, ambient temperature, and the passage of time, as shown in Figure 4. That is, the data acquisition unit 32 may collect room temperature and ambient temperature acquired periodically (for example, every minute) from the startup time t0 until the room temperature reaches the set temperature Tset, as startup data. In addition, the data acquisition unit 32 may collect room temperature and ambient temperature acquired periodically (for example, every minute) from the shutdown time t2 until the rate of change of room temperature falls below a threshold, as shutdown data.

[0024] The data processing unit 33 processes the training data stored in the training data holding unit 37 into a dataset suitable for training during the startup buffer time. Specifically, the data processing unit 33 creates training data by combining the startup conditions on the same day, the measured value of the startup buffer time, the shutdown conditions, and the time of room temperature change after shutdown. The data processing unit 33 outputs the created training data to the training unit 34.

[0025] The learning unit 34 learns the startup reserve time using the training data created by the data processing unit 33. That is, it generates a trained model that infers the startup reserve time from the room temperature, outside temperature, and set temperature. The learning unit 34 stores the generated training model in the trained model holding unit 38. The time of room temperature change after shutdown included in the training data is influenced by the conditions at the time of shutdown and the performance of the building 10, such as the size of the space in the building 10 where the indoor unit 6 is installed and its thermal conductivity. Therefore, by including the conditions at the time of shutdown and the time of room temperature change after shutdown in the training data, the learning unit 34 can infer the startup reserve time considering the performance of the building 10 even when training the startup reserve time with a small number of days' worth of training data.

[0026] Here, we will explain the method for learning the startup buffer time in the learning unit 34. The learning algorithm used by the learning unit 34 can be any known algorithm such as supervised learning, unsupervised learning, or reinforcement learning. As an example, we will explain the case where a neural network is applied.

[0027] The learning unit 34 learns the startup buffer time, for example, by supervised learning according to a neural network model. Here, supervised learning is a method in which a learning device is given pairs of input and result (label) data, learns features in that training data, and infers the result from the input.

[0028] A neural network consists of an input layer made up of multiple neurons, an intermediate layer (hidden layer) made up of multiple neurons, and an output layer made up of multiple neurons. The intermediate layer may consist of one or more layers.

[0029] Figure 3 shows an example of a neural network used in inferring the startup buffer time of an air conditioning control device according to this embodiment. For example, in a three-layer neural network as shown in Figure 3, when multiple inputs are input to the input layer (X1-X3), these values ​​are multiplied by weights W1 (w11-w16) and input to the hidden layer (Y1-Y2), and the result is further multiplied by weights W2 (w21-w26) and output from the output layer (Z1-Z3). This output result changes depending on the values ​​of weights W1 and W2.

[0030] In this application, the neural network learns the startup reserve time through so-called supervised learning, according to training data created based on the combination of startup conditions and measured startup reserve time acquired by the data acquisition unit 32, and the shutdown conditions and the time of room temperature change after shutdown. That is, the neural network learns by inputting startup conditions into the input layer and adjusting the weights W1 and W2 so that the result output from the output layer approaches the measured startup reserve time.

[0031] The learning unit 34 generates and outputs a trained model by performing the learning described above.

[0032] The calculation unit 35 obtains the room temperature, outside temperature, and set temperature from the data acquisition unit 32, infers the startup buffer time using the trained model stored in the trained model holding unit 38, and outputs it to the control unit 36.

[0033] The control unit 36 ​​controls the air conditioner 4 to start up by a startup buffer time inferred from the scheduled startup time, so that the room temperature reaches the set temperature at the preset startup time.

[0034] Figure 5 is a flowchart showing the flow of data collection and learning in the air conditioning control device 3 according to this embodiment. Using Figure 5, the data collection and learning process for inferring the startup buffer time in the air conditioning control device 3 will be explained.

[0035] First, step S101 shows that the learning process for the startup reserve time is performed periodically. Since the air conditioner 4 is assumed to be started in the morning when the building 10 is used and stopped at night when the building 10 is no longer used, the cycle is one day. However, the cycle is not limited to one day and can be set according to the usage schedule of the air conditioner 4. Next, proceed to step S102.

[0036] In step S102, the air conditioning control device 3 determines whether or not it is time t0 to start up the air conditioner 4. If sufficient training data has not been collected, time t0 is a certain amount of time (e.g., 1 hour) before the set time tset. If sufficient training data has been collected, time t0 is time ton_est, which is the time inferred from the set time tset. If it is time t0, proceed to step S103 (S102: Yes). If it is not time t0, proceed to step S104 (S102: No).

[0037] In step S103, the data acquisition unit 32 of the air conditioning control device 3 performs a process to acquire startup data. Figure 6 is a flowchart showing the flow of the startup data acquisition process of the air conditioning control device 3 according to this embodiment. Figure 6 corresponds to step S103 in Figure 5. The startup data acquisition process in the air conditioning control device 3 will be explained with reference to Figure 6.

[0038] First, in step S10, the data acquisition unit 32 acquires the startup conditions from the air conditioner 4. The startup conditions include the room temperature Ton_in acquired from the indoor temperature sensor 61, the outdoor temperature Ton_out acquired from the outdoor temperature sensor 51, and the set temperature Tset. Next, the process proceeds to step S11.

[0039] In step S11, the air conditioning control device 3 determines whether the room temperature obtained from the indoor temperature sensor 61 has reached the set temperature Tset. The air conditioning control device 3 periodically obtains the room temperature from the indoor temperature sensor 61 and compares it with the set temperature Tset to determine whether the room temperature has reached the set temperature Tset. Alternatively, the indoor unit 6 notifies the air conditioning control device 3 when the value detected by the indoor temperature sensor 61 and the set temperature Tset become equal. When the air conditioning control device 3 receives notification from the indoor unit 6, it determines that the room temperature has reached the set temperature. Note that the air conditioning control device 3 or the indoor unit 6 may also determine that the room temperature has reached the set temperature when the difference between the room temperature obtained from the indoor temperature sensor 61 and the set temperature Tset falls below a threshold. If the air conditioning control device 3 determines that the room temperature has reached the set temperature, this time is set as arrival time t1 and the process proceeds to step S12 (S11: Yes). If the air conditioning control device 3 determines that the room temperature is not the set temperature, the process returns to step S11 (S11: No).

[0040] In step S12, the data acquisition unit 32 stores the time from the start time t0 to the time t1 when the room temperature reaches the set temperature Tset as the measured value ton of the start-up reserve time in the learning data storage unit 37. The data acquisition unit 32 may also acquire the amount of heat processed by the air conditioner 4 during ton from the outdoor unit 5. The outdoor unit 5 knows its heating and cooling capacity from the operating frequency of the compressor, etc., and stores it for each unit time. Therefore, it is possible to determine the amount of heat processed by the air conditioner 4 during ton. If the data acquisition unit 32 acquires the amount of heat processed by the air conditioner 4, it also stores it in the learning data storage unit 37. With this, the data acquisition process at startup is completed, and the process proceeds to step S104 in Figure 5.

[0041] Returning to Figure 5, in step S104, the air conditioning control device 3 determines whether or not it is time t2 for stopping the air conditioner 4. The stopping time t2 is set in advance on the air conditioning control device 3 by the user. Alternatively, the stopping time t2 may be the time when the user stops the air conditioner 4 by operating the handheld remote control 7. If it is time t2, proceed to step S105 (S104: Yes). If it is not time t2, proceed to step S106 (S104: No).

[0042] In step S105, the data collection unit 32 of the air conditioning control device 3 performs a process of acquiring data at the time of stoppage. FIG. 7 is a flowchart showing the flow of the data acquisition process at the time of stoppage of the air conditioning control device 3 according to the present embodiment. FIG. 7 corresponds to step S105 in FIG. 5. With reference to FIG. 7, the data acquisition process at the time of stoppage in the air conditioning control device 3 will be described.

[0043] First, in step S20, the data collection unit 32 acquires the stoppage conditions from the air conditioner 4. The stoppage conditions include the indoor temperature T_off_in acquired from the indoor temperature sensor 61 and the outside air temperature T_off_out acquired from the outside air temperature sensor 51. Next, the process proceeds to step S21.

[0044] In step S21, the air conditioning control device 3 obtains a change rate using the room temperature regularly acquired from the indoor temperature sensor 61 and compares it with a predetermined threshold value, and determines whether or not the change rate of the room temperature has become equal to or less than the threshold value. Alternatively, the indoor unit 6 obtains the change rate of the room temperature from the value detected by the indoor temperature sensor 61, and notifies the air conditioning control device 3 when the change rate of the room temperature has become equal to or less than the threshold value. When the air conditioning control device 3 receives the notification from the indoor unit 6, it determines that the change rate of the room temperature has become equal to or less than the threshold value. If the air conditioning control device 3 determines that the change rate of the room temperature has become equal to or less than the threshold value, the time at this time is set as the room temperature stable time t3, and then the process proceeds to step S22 (S21: Yes). If the air conditioning control device 3 determines that the change rate of the room temperature is not equal to or less than the threshold value, the process returns to step S21 (S21: No).

[0045] In step S22, the data collection unit 32 stores, in the learning data holding unit 37, as the post-stoppage room temperature change time t_off, the time from the stoppage time t2 to the room temperature stable time t3 at which the temperature change rate of the room temperature becomes equal to or less than a predetermined threshold value. With this, the data acquisition process at the time of stoppage is completed, and the process proceeds to step S106 in FIG. 5.

[0046] Returning to FIG. 5, in step S106, the air conditioner control device 3 determines whether it is the timing to learn the startup standby time. It is desirable to set the timing of learning as the time from the end of the data acquisition process at the time of stop to the startup time. For example, when the air conditioner 4 is started at the start time of business and stopped at the end time of business, the time in the middle of the night is set in the air conditioner control device 3 in advance as the timing of learning. If it is the timing of learning, the process proceeds to step S107 (S106: Yes). If it is not the timing of learning, the process proceeds to step S108 (S106: No).

[0047] In step S107, the air conditioner control device 3 learns the startup standby time using the learning data collected by the data collection unit 32. FIG. 8 is a flowchart showing the flow of learning the startup standby time in the air conditioner control device 3 according to the present embodiment. Referring to FIG. 8, the flow of learning the startup standby time will be described.

[0048] First, in step S30, the data processing unit reads the learning data stored in the learning data holding unit 37, and creates learning data by combining the conditions at the time of startup on the same day, the measured value ton of the startup standby time, the conditions at the time of stop, and the room temperature change time toff after stop. The created learning data is output to the learning unit 34. Next, the process proceeds to step S31.

[0049] In step S31, the learning unit 34 learns the startup standby time using the learning data, and generates a learned model for inferring the startup standby time from the conditions at the time of startup. The learning unit 34 stores the generated learned model in the learned model holding unit 38. Thus, the learning of the startup standby time is completed, and the process proceeds to step S108 in FIG. 5.

[0050] Returning to FIG. , in step S108, the process for one cycle of collecting and learning the learning data in the air conditioner control device ends, and the process returns to S101 again. Thus, the flow of collecting and learning the learning data for inferring the startup standby time in the air conditioner control device ends.

[0051] In the above process, the air conditioning control device 3 collects training data for learning the startup buffer time and generates a training model for inferring the startup buffer time. When the learning unit 34 attempts to generate a trained model that infers the startup buffer time with sufficient accuracy, it can generate a trained model in a shorter time than when using only the startup conditions and the measured startup buffer time, as well as the shutdown conditions and the time of room temperature change after shutdown, as training data. Furthermore, for example, if the control server 1 collects training data from air conditioners 4 installed in multiple buildings 10, the influence of each individual building 10 is not considered. In this embodiment, since the training data is actually obtained from air conditioners 4 installed in buildings 10, it is data influenced by the buildings 10, and the learning unit 34 can generate a training model suitable for the air conditioners 4 installed in buildings 10.

[0052] Figure 9 is a flowchart showing the inference of the startup buffer time and the control of the air conditioner 4 in the air conditioning control device 3 according to this embodiment. The inference of the startup buffer time by the air conditioning control device 3 and the control of the air conditioner 4 will be explained with reference to Figure 9. Note that the inference of the startup buffer time and the control of the air conditioner 4 in Figure 9 are performed at any time during the day. However, since the room temperature and outside temperature at the time of processing in Figure 9 are input to the trained model, it is best to perform the inference at a time close to the set time when the difference between the room temperature and outside temperature and the room temperature and outside temperature at the actual startup time is as small as possible, and at a time before the startup buffer time from the set time tset so that the set temperature Tset is reached at the set time tset. Therefore, the inference of the startup buffer time is performed, for example, at a time from the set time tset to the time up to the maximum number of minutes before the scheduled startup time. This reduces the difference between the room temperature and outside temperature input to the trained model and the room temperature and outside temperature at the time the air conditioner 4 is actually started, and enables accurate estimation of the startup buffer time. If there is no data on past scheduled startup times, the system will infer a startup buffer time a certain period of time (for example, 3 hours) before the set time tset.

[0053] First, in step S201, the calculation unit 35 of the air conditioning control device 3 obtains the set time tset and the set temperature Tset that are set in the air conditioning control device 3 or the handheld remote control. Next, the process proceeds to step S202.

[0054] In step S202, the calculation unit 35 obtains the room temperature from the indoor unit 6 and the outside air temperature from the outdoor unit 5 via the data acquisition unit 32. Next, the process proceeds to step S203.

[0055] In step S203, the calculation unit 35 uses the learned model stored in the learned model holding unit 38 as input for room temperature, outside temperature, and set temperature, and infers the startup reserve time. The calculation unit 35 outputs the inferred startup reserve time ton_est to the control unit 36. Next, the process proceeds to step S204.

[0056] In step S204, the control unit 36 ​​determines that the start time t0 is before the start reserve time ton_est, which is inferred from the set time tset, and sends control signals to the outdoor unit 5 and the indoor unit 6 to start the air conditioner 4 at start time t0. Alternatively, in step S202, the control unit 36 ​​may not send control signals to the outdoor unit 5 and the indoor unit 6, but may send control signals to start them when start time t0 arrives. This completes the inference of the start reserve time and the control of the air conditioner 4.

[0057] In this way, the air conditioning control device 3 uses the measured values ​​of the startup conditions and startup buffer time on the same day, along with the shutdown conditions and the time of room temperature change after shutdown, as training data. It then uses the generated trained model to infer the startup buffer time and controls the air conditioner 4 using the inferred startup buffer time. This allows for the inference of a startup buffer time suitable for the building 10 in which the air conditioner 4 is installed in a shorter time than when only startup data is used, and enables the air conditioner 4 to be started at a time suitable for reaching the set temperature at the set time. In particular, when the building 10 is used as an office, there was a problem that the accuracy of the startup buffer time inference decreased when performed after a long holiday. This is because there is a difference in outside temperature before and after a long holiday. In this embodiment, it is possible to generate a trained model that can infer the startup buffer time with higher accuracy in a shorter time than when only startup data is used after a long holiday.

[0058] As described above, the air conditioning control device 3 of this embodiment is an air conditioning control device 3 that controls the operation of an air conditioner 4, and comprises: a data collection unit 32 that acquires learning data including data when the air conditioner 4 is started and data when it is stopped; a learning unit 34 that uses the learning data to generate a trained model for inferring the start-up buffer time from when the air conditioner 4 is started until it reaches a set temperature; a calculation unit 35 that uses the trained model to output the start-up buffer time from the start-up data acquired by the data collection unit 32; and a control unit 36 ​​that causes the air conditioner 4 to start operation before the start-up buffer time, which is set to a predetermined time.

[0059] This allows for the estimation of a suitable startup buffer time for the building 10 in which the air conditioner 4 is installed in a shorter time than when using only startup data, and enables the air conditioner 4 to be started at a suitable time to reach the set temperature at the set time.

[0060] The startup data consists of the startup conditions of the air conditioner 4 and the startup buffer time from startup until the set temperature is reached, while the shutdown data consists of the shutdown conditions of the air conditioner 4 and the time it takes for the room temperature to change after shutdown until the rate of change of room temperature after shutdown falls below a threshold. By using the startup buffer time as training data, a trained model that accurately infers the startup buffer time can be obtained.

[0061] Furthermore, the startup conditions include the room temperature and outside temperature at the time the air conditioner 4 is started, and the shutdown conditions include the room temperature and outside temperature at the time the air conditioner 4 is stopped. Since the startup preparation time and the time of room temperature change after shutdown are affected by the room temperature and outside temperature, the room temperature and outside temperature are used as training data.

[0062] Furthermore, the startup conditions include at least one of the number of indoor units 6 equipped with the air conditioner 4 and the capacity saving amount of the outdoor unit 5 equipped with the air conditioner 4. If multiple indoor units 6 are installed in the same room, the startup reserve time is affected by the operating status of adjacent indoor units 6. Therefore, by using the number of running indoor units 6 as training data, the startup reserve time can be learned with greater accuracy. In addition, the amount of heat processed by the air conditioner 4 is determined by the capacity saving amount of the outdoor unit 5 equipped with the air conditioner 4 relative to the air conditioning and heating capacity of the air conditioner 4. The amount of heat processed by the air conditioner 4 affects the startup reserve time. Therefore, by using the number of indoor units 6 equipped with the air conditioner 4 and the capacity saving amount of the outdoor unit 5 equipped with the air conditioner 4 as training data, a trained model that infers the startup reserve time with greater accuracy can be generated.

[0063] Embodiment 2. Next, Embodiment 2 will be described. In this embodiment, the configuration of the air conditioning system 100 and the air conditioning control device 3 are the same as in Embodiment 1, so their description will be omitted. In Embodiment 1, the air conditioner 4 was stopped at the stop time, and data acquisition processing was performed at the time of stop. In this embodiment, when the data at the time of stop is combined with the data at the time of start-up to be used as training data, the air conditioning control device 3 performs temperature difference control processing to control the air conditioner 4 so that the difference between the room temperature and the outside temperature at the time of stop approaches the difference between the room temperature and the outside temperature at the time of start-up, in order to be able to infer the start-up reserve time that takes the performance of the building 10 into more consideration. The start-up reserve time mainly varies depending on the difference between the room temperature and the outside temperature, the set temperature, the amount of heat processed by the air conditioner 4, the amount of heat stored in the building structure 10, and the thermal insulation performance of the building 10. On the other hand, the time of room temperature change after stop is mainly determined by the difference between the room temperature and the outside temperature and the thermal insulation performance of the building 10. Therefore, by making the difference between the room temperature and the outside temperature the same at startup and shutdown, the air conditioning control device 3 can estimate the effect of the building's heat storage capacity by comparing the startup data with the shutdown data. As a result, the air conditioning control device 3 can generate a trained model that can infer the startup buffer time with sufficient accuracy using less training data.

[0064] Figure 10 is a graph showing the time change of room temperature and outside temperature when temperature difference control processing is performed. Referring to Figure 10, the data collected by the data acquisition unit 32 when temperature difference control processing is performed will be explained. The data acquisition process at startup is the same as in Embodiment 1, so the explanation will be omitted. As shown in Figure 10, the difference between the room temperature and the outside temperature at startup time t0 is defined as the startup temperature difference Tdiff. In this embodiment, the time set in advance by the user to stop the air conditioner 4 in the air conditioning control device 3 is defined as the scheduled stop time t4'. When the scheduled stop time t4' arrives, the air conditioning control device 3 controls the air conditioner 4 so that the difference between the room temperature and the outside temperature becomes the startup temperature difference Tdiff. The air conditioning control device 3 sets the time when the difference between the room temperature and the outside temperature becomes the startup temperature difference Tdiff as the stop control time t4, and stops the air conditioner 4. The data acquisition unit 32 acquires the room temperature Toff_in at the stop control time t4 of the air conditioner 4 from the indoor temperature sensor 61 as the stop condition. Furthermore, the data acquisition unit 32 acquires the outside temperature Toff_out at the stop control time t4 of the air conditioner 4 from the outside temperature sensor 51 as the stop condition. In addition, the data acquisition unit 32 stores the time from the time t4 when the air conditioner 4 is stopped until the room temperature stabilization time t3 when the rate of change of the room temperature falls below a predetermined threshold as the room temperature change time after stop toff. The determination of whether the rate of change of the room temperature has fallen below the threshold is made by the data acquisition unit 32 periodically (for example, every minute) acquiring the room temperature from the indoor temperature sensor 61, calculating the difference from the previously acquired room temperature as the rate of change, and comparing it with the threshold. Alternatively, the indoor unit 6 may be configured to notify the data acquisition unit 32 when the rate of change of the room temperature falls below the threshold. The data acquisition unit 32 stores the learning data collected as described above in the learning data holding unit 37.

[0065] Figure 11 is a flowchart showing the flow of data collection and learning in the air conditioning control device 3 according to this embodiment. Using Figure 11, the data collection and learning process for inferring the startup buffer time in the air conditioning control device 3 will be explained. In Figure 11, processes similar to those in Figure 5 of Embodiment 1 are denoted by the same reference numerals, and the explanation is simplified.

[0066] First, step S101 shows that the learning process for the startup reserve time is executed periodically. Next, we proceed to step S102.

[0067] In step S102, the air conditioning control device 3 determines whether or not it is time t0 to start the air conditioner 4. If it is time t0, proceed to step S103 (S102: Yes). If it is not time t0, proceed to step S110 (S102: No).

[0068] In step S103, the data acquisition unit 32 of the air conditioning control device 3 performs a process to acquire startup data. The startup data acquisition process is the same as in Embodiment 1, so the explanation is omitted. Next, proceed to step S110.

[0069] In step S110, the air conditioning control device 3 determines whether or not the scheduled stop time t4' for the air conditioner 4 has arrived. The scheduled stop time t4' is set in advance on the air conditioning control device 3 by the user. Alternatively, the scheduled stop time t4' may be the time when the user stops the air conditioner 4 by operating the handheld remote control 7. If it is the scheduled stop time t4', proceed to step S111 (S110: Yes). If it is not the scheduled stop time t4', proceed to step S106 (S110: No).

[0070] In step S111, the control unit 36 ​​of the air conditioning control device 3 performs a temperature difference control process to control the air conditioner 4 so that the difference between the room temperature and the outside temperature is the same as the startup temperature difference Tdiff, which is the difference between the room temperature and the outside temperature at the startup time t0. Figure 12 is a flowchart showing the flow of the temperature difference control process in the air conditioning control device 3 according to this embodiment. Figure 12 corresponds to step S111 in Figure 11. The temperature difference control process in the air conditioning control device 3 will be explained with reference to Figure 12.

[0071] First, in step S40, the control unit 36 ​​controls the air conditioner 4 so that the difference between the room temperature and the outside temperature is equal to the startup temperature difference Tdiff. Specifically, when the operating mode of the air conditioner 4 is heating, the control unit 36 ​​sends a control signal to the air conditioner 4 to reduce its operating capacity if the difference between the room temperature and the outside temperature is greater than the startup temperature difference Tdiff. If the difference between the room temperature and the outside temperature is less than the startup temperature difference Tdiff, the control unit 36 ​​sends a control signal to the air conditioner 4 to increase its operating capacity. Also, when the operating mode of the air conditioner 4 is cooling, the control unit 36 ​​sends a control signal to the air conditioner 4 to reduce its operating capacity if the difference between the room temperature and the outside temperature is greater than the startup temperature difference Tdiff. If the difference between the room temperature and the outside temperature is less than the startup temperature difference Tdiff, the control unit 36 ​​sends a control signal to the air conditioner 4 to increase its operating capacity. Next, the process proceeds to step S41.

[0072] In step S41, the control unit 36 ​​calculates the difference between the room temperature obtained from the indoor unit 6 and the outside temperature obtained from the outdoor unit 5 and compares it with the startup temperature difference Tdiff to determine whether the difference between the room temperature and the outside temperature is equal to the startup temperature difference Tdiff. The control unit 36 ​​may also determine that the difference between the room temperature and the outside temperature is equal to the startup temperature difference Tdiff if the difference between the difference between the room temperature and the outside temperature and the startup temperature difference Tdiff is less than or equal to a threshold. If the control unit 36 ​​determines that the difference between the room temperature and the outside temperature is equal to the startup temperature difference Tdiff, this time is set as the stop control time t4 and the process proceeds to step S42 (S41: Yes). If the control unit 36 ​​determines that the difference between the room temperature and the outside temperature is not equal to the startup temperature difference Tdiff, the process returns to step S40 (S41: No) and controls the air conditioner 4 again.

[0073] In step S42, the control unit 36 ​​stops the air conditioner 4. This completes the temperature difference control process in the air conditioning control device 3, and the process proceeds to step S105 in Figure 11.

[0074] Returning to Figure 11, in step S105, the data acquisition unit 32 of the air conditioning control device 3 performs the process of acquiring data at the time of shutdown. The data at the time of shutdown is the same as in Embodiment 1, consisting of the conditions at the time of shutdown and the time of change in room temperature after shutdown, Toff. The conditions at the time of shutdown include the room temperature Toff_in acquired from the indoor temperature sensor 61 and the outdoor temperature Toff_out acquired from the outdoor temperature sensor 51. Next, proceed to step S106.

[0075] In step S106, the air conditioning control device 3 determines whether or not it is time to learn the startup reserve time. If it is time to learn, proceed to step S107 (S106: Yes). If it is not time to learn, proceed to step S108 (S106: No).

[0076] In step S107, the air conditioning control device 3 learns the startup reserve time using the learning data collected by the data acquisition unit 32. The startup reserve time learning is performed in the same manner as in Embodiment 1. Next, the process proceeds to step S108.

[0077] In step S108, the collection of training data and processing for one cycle of training in the air conditioning control device 3 are completed, and the process returns to S101. This concludes the process of collecting training data and training to infer the startup buffer time in the air conditioning control device 3.

[0078] In the above process, the air conditioning control device 3 collects training data for learning the startup buffer time and generates a training model for inferring the startup buffer time. In this embodiment, before stopping the air conditioner 4, the air conditioning control device 3 controls the air conditioner 4 so that the difference between the room temperature and the outside temperature is the same as the startup temperature difference Tdiff, which is the temperature difference at the startup time t0. As a result, the air conditioning control device 3 can generate a training model for inferring the startup buffer time considering the heat storage capacity of the building 10 by learning the measured values ​​of the startup conditions and startup buffer time, as well as the shutdown conditions and the time of room temperature change after shutdown.

[0079] The inference of the startup buffer time using the trained model by the air conditioning control device 3 and the control of the air conditioner 4 using the inferred startup buffer time are the same as in Embodiment 1, so an explanation will be omitted.

[0080] As described above, the control unit 36 ​​of the air conditioning control device 3 in this embodiment calculates the startup temperature difference, which is the difference between the room temperature and the outside temperature when the air conditioner 4 is started. When it is time for the air conditioner 4 to be stopped, it performs temperature difference control to control the air conditioner 4 so that the difference between the room temperature and the outside temperature approaches the startup temperature difference. When the difference between the difference between the room temperature and the outside temperature and the startup temperature difference becomes smaller than a threshold, the operation of the air conditioner 4 is stopped. By stopping the air conditioner 4 when the difference between the difference between the room temperature and the outside temperature and the startup temperature difference becomes smaller than a threshold through temperature difference control, the difference between the room temperature and the outside temperature can be matched under the conditions for starting and stopping, and the startup reserve time can be estimated taking into account the performance of the building 10.

[0081] The temperature difference control by the control unit 36 ​​controls the operating capacity of the air conditioner 4 to be lower than before the scheduled shutdown time if the difference between the room temperature and the outside temperature is greater than the startup temperature difference, and controls the operating capacity of the air conditioner to be higher than before the scheduled shutdown time if the difference between the room temperature and the outside temperature is less than the startup temperature difference. By controlling the air conditioner 4 in this way, the control unit 36 ​​can control the difference between the room temperature and the outside temperature to be closer to the startup temperature difference.

[0082] Embodiment 3. Next, Embodiment 3 will be described. In this embodiment, components identical to those in Embodiment 1 are denoted by the same reference numerals, and their descriptions will be omitted as appropriate. Figure 13 is a diagram showing an example configuration of an air conditioning system 100 including an air conditioning control device 3 according to this embodiment. In this embodiment, it is assumed that a second indoor unit 6b is installed in the same room 8 as the first indoor unit 6a which is the target of learning. Note that there may be more than two indoor units 6 installed in room 8. Also, hereafter, when indoor unit 6 is referred to, it includes the first indoor unit 6a and the second indoor unit 6b. When multiple indoor units 6 are installed in room 8, the startup reserve time is affected by the operating state of adjacent indoor units 6. In this embodiment, the air conditioning control device 3 controls the stopping time difference of the air conditioner 4 so that the difference in startup times between the first indoor unit 6a and the second indoor unit 6b is the same as the difference in stopping times between the first indoor unit 6a and the second indoor unit 6b. As a result, by combining startup data and shutdown data, the influence of the second indoor unit 6b adjacent to the first indoor unit 6a can be inferred, and the air conditioning control device 3 can generate a trained model that can infer the startup buffer time with sufficient accuracy as expected, using less training data.

[0083] The configurations of the first indoor unit 6a and the second indoor unit 6b are the same as those of the indoor unit 6 in Embodiment 1. The first indoor unit 6a and the second indoor unit 6b are connected to the outdoor unit 5 by communication lines and refrigerant piping (not shown) to constitute the air conditioner 4. The first indoor unit 6a is equipped with a first indoor temperature sensor 61a for detecting the indoor temperature. The second indoor unit 6b is equipped with a second indoor temperature sensor 61b for detecting the indoor temperature.

[0084] Figure 14 is a graph showing the time change of room temperature when stop time difference processing is performed. The room temperature shown is the first room temperature detected by the first indoor temperature sensor 61a of the first indoor unit 6a and the second room temperature detected by the second indoor temperature sensor 61b of the second indoor unit 6b. The outside temperature is not shown. In this embodiment, the time set in advance by the user to stop the air conditioner 4 in the air conditioning control device 3 is set as the scheduled stop time t5'. Note that the learning of the start-up reserve time is performed for each indoor unit 6. In Figure 14, the target of learning is the first indoor unit 6a, and values ​​related to the learning data of the first indoor unit 6a are shown. Referring to Figure 14, the learning data of the first indoor unit 6a when stop time difference processing is performed will be explained. The set time tset, set temperature Tset, and scheduled stop time t5' are set once for multiple indoor units 6 installed in the same room 8.

[0085] The first startup time t0_1 is the time when the first indoor unit 6a is started. The second startup time t0_2 is the time when the second indoor unit 6b is started. Similar to the startup time t0 in Embodiment 1, the first startup time t0_1 and the second startup time t0_2 are set to a time a certain amount of time (for example, 1 hour) earlier than the set time tset if sufficient training data has not been collected. In other words, the air conditioning control device 3 aims to control the air conditioner 4 so that the first room temperature and the second room temperature become the set temperature Tset at the set time tset. The first indoor unit 6a and the second indoor unit 6b differ in distance from the window, distance from the heat source machine installed in the room 8, etc., so even if the set time tset and set temperature Tset are the same, the startup reserve time will be different. The startup reserve time inferred for the first indoor unit 6a is ton_est1, and the startup reserve time inferred for the second indoor unit 6b is ton_est2 (not shown). If sufficient training data has been collected, the first startup time t0_1 is set to a time that is inferred to start up by ton_est1 from the set time tset. If sufficient training data has been collected, the second startup time t0_2 is set to a time that is inferred to start up by ton_est2 from the set time tset. The air conditioning control device 3 calculates the startup time difference tdiff, which is the difference between the first startup time t0_1 and the second startup time t0_2. In Figure 14, the first startup time t0_1 is earlier than the second startup time t0_2. That is, the first indoor unit 6a is started up before the second indoor unit 6b.

[0086] The data acquisition unit 32 acquires the room temperature Ton_in_1 at the first start time t0_1 from the first indoor temperature sensor 61a as a start condition. The data acquisition unit 32 acquires the outside temperature Ton_out_1 (not shown) at the first start time t0_1 of the air conditioner 4 from the outside temperature sensor 51 as a start condition. The data acquisition unit 32 acquires the set temperature Tset as a start condition. The data acquisition unit 32 also stores the time from the first start time t0_1 when the first indoor unit 6a is started up to the time t1_1 when the room temperature reaches the set temperature Tset as the measured value ton_1 of the start reserve time for the first indoor unit 6a. The determination of whether the room temperature has reached the set temperature Tset is made by the data acquisition unit 32 periodically acquiring the room temperature from the first indoor temperature sensor 61a and comparing it with the set temperature Tset. Alternatively, the first indoor unit 6a may notify the data acquisition unit 32 when the room temperature reaches the set temperature Tset.

[0087] The air conditioning control device 3 stops the first indoor unit 6a at the scheduled stop time t5', which has been set in advance by the user. The first stop time t5_1, which is the time when the first indoor unit 6a is stopped, is the scheduled stop time t5'. Next, after a start time difference tdiff has elapsed, the air conditioning control device 3 stops the second indoor unit 6b. The second stop time t5_2, which is the time when the second indoor unit 6b is stopped, is the time when the start time difference tdiff has elapsed from the scheduled stop time t5'. This control performed by the air conditioning control device 3 to make the operating state indicating the start or stop of the first indoor unit 6a and the second indoor unit 6b the same during the start-up reserve time and the time of room temperature change after stopping, by setting the time difference between stopping the first indoor unit 6a and the second indoor unit 6b to the start time difference tdiff, is called stop time difference control. The data acquisition unit 32 acquires the room temperature Toff_in_1 at the first stop time t5_1 from the first indoor temperature sensor 61a as the stop condition. Furthermore, the data acquisition unit 32 acquires the outside temperature Toff_out_1 (not shown) at the first stop time t5_1 when the first indoor unit 6a is stopped from the outside temperature sensor 51 as the stop condition. In addition, the data acquisition unit 32 stores the time from the first stop time t5_1, which is the time when the first indoor unit 6a is stopped, to the first room temperature stabilization time t3_1, when the rate of change of room temperature falls below a predetermined threshold, as the room temperature change time toff_1 after the first indoor unit 6a is stopped.

[0088] As shown in Figure 14, by stopping the second indoor unit 6b after a start time difference tdiff has elapsed from the first stop time t5, the operating states indicating start or stop in terms of the start-up reserve time and the time of room temperature change after stopping for both the first indoor unit 6a and the second indoor unit 6b can be made the same. In other words, the influence of the second indoor unit 6b on the start-up reserve time of the first indoor unit 6a is equivalent to the influence of the second indoor unit 6b on the time of room temperature change after stopping the first indoor unit 6a. Therefore, by combining the data at the time of start-up and the data at the time of stop-up and using them as training data, it is possible to infer the start-up reserve time while considering the influence of adjacent indoor units 6.

[0089] As described above, the learning data for the first indoor unit 6a has been explained. The data acquisition unit 32 also stores the room temperature Ton_in_2, the outside temperature Ton_out_2, and the measured value ton_2 of the startup reserve time at the second startup time t0_2 as startup data for the second indoor unit 6b. The data acquisition unit 32 also stores the room temperature Toff_in_2, the outside temperature Toff_out_2, and the time toff_2 of the room temperature change after stopping at the second stop time t5_2 as shutdown data for the second indoor unit 6b. The data acquisition unit 32 stores the learning data collected as described above in the learning data holding unit 37.

[0090] Figure 15 is a flowchart showing the flow of data collection and learning in the air conditioning control device 3 according to this embodiment. Using Figure 15, the data collection and learning process for inferring the startup buffer time in the air conditioning control device 3 will be explained. In Figure 15, processes similar to those in Figure 5 of Embodiment 1 are denoted by the same reference numerals, and the explanation is simplified.

[0091] First, step S101 indicates that the learning of the startup reserve time is performed periodically. Next, the process proceeds to steps S301 and S301'. Steps S301 and S301' are processed in parallel by the air conditioning control device 3.

[0092] In step S301, the air conditioning control device 3 determines whether or not the first start time t0_1 of the first indoor unit 6a has arrived. If it has arrived at the first start time t0_1, the process proceeds to step S302 (S301: Yes). If it has not arrived at the first start time t0_1, the process proceeds to step S304 (S301: No).

[0093] In step S302, the data acquisition unit 32 of the air conditioning control device 3 performs a process to acquire data when the first indoor unit 6a is started up. The process for acquiring data at startup is the same as in Embodiment 1, so the explanation will be omitted. Next, proceed to step S303.

[0094] In step S301', the same processing as in step S301 is performed on the second indoor unit 6b. That is, the air conditioning control device 3 determines whether or not it has reached the second start time t0_2 for the second indoor unit 6b. If it has reached the second start time t0_2, the process proceeds to step S302' (S301': Yes). If it has not reached the second start time t0_2, the process proceeds to step S304 (S301': No). Note that the first start time t0_1 is set to be earlier than the second start time t0_2. That is, the indoor unit 6 that starts up first is designated as the first indoor unit 6a.

[0095] In step S302', the same processing as in step S302 is performed on the second indoor unit 6b. That is, the data acquisition unit 32 of the air conditioning control device 3 performs the process of acquiring startup data for the second indoor unit 6b. The startup data acquisition process is the same as in Embodiment 1, so the explanation is omitted. Next, proceed to step S303.

[0096] In step S303, the air conditioning control device 3 calculates the difference between the first start time t0_1 and the second start time t0_2 as the start time difference tdiff. Next, the process proceeds to step S304.

[0097] In step S304, the air conditioning control device 3 determines whether or not it has reached the scheduled stop time t5' for the air conditioner 4. If it has reached the scheduled stop time t5', proceed to step S305 (S304: Yes). If it has not reached the scheduled stop time t5', proceed to step S106 (S304: No).

[0098] In step S305, the control unit 36 ​​of the air conditioning control device 3 stops the first indoor unit 6a. Next, the process proceeds to steps S306 and S308. Steps S306 and S308 are processed in parallel by the air conditioning control device 3.

[0099] In step S306, the air conditioning control device 3 determines whether a start time difference tdiff has elapsed from the scheduled stop time t5'. If a start time difference tdiff has elapsed, the device proceeds to step S307 (S306: Yes). If a start time difference tdiff has not elapsed, the device returns to step S306 (S306: No).

[0100] In step S307, the control unit 36 ​​stops the second indoor unit 6b. Next, the process proceeds to step S308'. Note that the process of stopping the first indoor unit 6a at the scheduled stopping time t5', and then stopping the second indoor unit 6b after the start time difference tdiff has elapsed (corresponding to S304, S305, S306, and S307) is called stop time difference control.

[0101] In step S308, the data acquisition unit 32 of the air conditioning control device 3 performs the process of acquiring data when the first indoor unit 6a is stopped. The process of acquiring data when stopped is the same as in Embodiment 1, so the explanation will be omitted. Next, proceed to step S106.

[0102] In step S308', the data acquisition unit 32 of the air conditioning control device 3 performs the process of acquiring data when the second indoor unit 6b is stopped. The process of acquiring data when stopped is the same as in Embodiment 1, so the explanation is omitted. Next, proceed to step S106.

[0103] In step S106, the air conditioning control device 3 determines whether or not it is time to learn the startup reserve time. If it is time to learn, proceed to step S107 (S106: Yes). If it is not time to learn, proceed to step S108 (S106: No).

[0104] In step S107, the air conditioning control device 3 learns the startup reserve time using the training data collected by the data acquisition unit 32. The air conditioning control device 3 learns the startup reserve time for each of the first indoor unit 6a and the second indoor unit 6b using the collected training data and generates a trained model. The learning of the startup reserve time is performed in the same manner as in Embodiment 1. Next, the process proceeds to step S108.

[0105] In step S108, the collection of training data and processing for one cycle of training in the air conditioning control device 3 are completed, and the process returns to S101. This concludes the process of collecting training data and training to infer the startup buffer time in the air conditioning control device 3.

[0106] In the above configuration, the air conditioning control device 3 collects training data for learning the startup reserve time and generates a training model for inferring the startup reserve time. In this embodiment, the air conditioning control device 3 controls the first indoor unit 6a and the second indoor unit 6b so that the time difference between stopping the first indoor unit 6a and the second indoor unit 6b is equal to the startup time difference tdiff. As a result, the operating states indicating startup or shutdown are equal during the startup reserve time and the time of room temperature change after shutdown for the first indoor unit 6a and the second indoor unit 6b. Therefore, by combining the startup data and the shutdown data and using them as training data, a training model can be generated that infers the startup reserve time considering the influence of adjacent indoor units 6.

[0107] The inference of the startup buffer time using the trained model by the air conditioning control device 3 and the control of the air conditioner 4 using the inferred startup buffer time are the same as in Embodiment 1, so an explanation will be omitted.

[0108] As described above, the air conditioner 4 controlled by the air conditioning control device 3 of this embodiment has a first indoor unit 6a and a second indoor unit 6b. The control unit 36 ​​stores the time from when the first indoor unit 6a starts up until when the second indoor unit 6b starts up as a start-up time difference, and stops the second indoor unit 6b when the start-up time difference has elapsed since the first indoor unit 6a stopped operating. This makes it possible to match the start-up conditions and the stop-down conditions. Therefore, by combining the start-up data and the stop-down data and using them as training data, it is possible to consider the influence of adjacent indoor units 6 and generate a trained model that infers a more accurate start-up buffer time.

[0109] Embodiment 4. Next, Embodiment 4 will be described. In this embodiment, the temperature difference control processing described in Embodiment 2 is performed, and the amount of heat stored in the building structure is estimated using the acquired training data and used for inferring the startup reserve time. Note that the configuration of the air conditioning system 100 and the air conditioning control device 3 in this embodiment is the same as in Embodiment 2, so a description will be omitted.

[0110] Figure 16 is a flowchart showing the flow of data collection and learning in the air conditioning control device 3 according to this embodiment. In this embodiment, the operation of the data acquisition process at startup, the temperature difference control process, and the data acquisition process at shutdown are the same as those shown in Figure 11 of Embodiment 2, so their explanation is omitted. In this embodiment, after the data acquisition process at shutdown (step S105) is performed, the process of estimating the amount of heat stored in the building structure in step S401 is added.

[0111] In step S401, the air conditioning control device 3 estimates the amount of heat stored in the building structure 10 using startup data and shutdown data. An example of a method for estimating the amount of heat stored in the building structure is described below.

[0112] The rate of temperature change after the air conditioner 4 is started can be expressed, for example, by equation (1), using a thermal characteristic model with typical influencing factors. In equation (1), a' and b' are coefficients, and C is the indoor heat capacity [kWh / K] targeted by the air conditioner 4. R is the window thermal conductivity [kW / K], and α is the amount of heat due to other influencing factors of the heat load. W(t) is the amount of heat [kW] that the air conditioner 4 processes at time t. The amount of heat W(t) is the amount of heat removed in the case of cooling operation, and the amount of heat supplied in the case of heating operation. b'Q(t) is the amount of heat [kW] released from the building structure at time t. Q(t) is the amount of heat stored in the building structure of the building 10 by the time the air conditioner 4 is started, and represents the amount released into the room at time t.

[0113]

[0114] Using equation (1), the startup reserve time ton from the start of the air conditioner 4 until the room temperature reaches the set temperature Tset is determined. Assuming that the outside temperature Tout and the heat quantity Q(t) are constant during the startup reserve time ton, it can be expressed as shown in equation (2). In equation (2), a and b are coefficients. W is the average amount of heat [kW] that the air conditioner 4 processes during the startup reserve time ton, and Q is the amount of heat stored in the building structure 10 by the time of startup [kWh]. bQ represents the amount of heat per unit time released into the room from the heat stored in the building structure. Note that while the air conditioner 4 is running, the amount of heat bQ released per unit time into the room from the heat stored in the building structure can be considered constant regardless of time t. R is the average value [kW] of the amount of heat flowing into the room per unit time during the startup reserve time ton, and is determined by the window thermal conductivity Rwin and the temperature difference between the room temperature and the outside temperature (Tset - Tin). Tin is the room temperature at the time the air conditioner 4 is started. Note that the air conditioning heat quantity W is a negative value during cooling. Also, the amount of heat flowing into the room R is a negative value if the outside temperature is lower than the room temperature.

[0115]

[0116] Furthermore, the time toff for the change in room temperature after shutdown is calculated using equation (1). While the air conditioner 4 is running, the amount of heat stored in the building can be considered to be 0. It is assumed that the outside temperature Tout is constant during the time toff for the change in room temperature after shutdown. Also, due to the temperature difference control process, the temperature difference between the room temperature at shutdown and the outside temperature is the same as the temperature difference at startup (Tset - Tin). Therefore, the average amount of heat flowing into the room per unit time during the time toff for the change in room temperature after shutdown can be considered equivalent to the average amount of heat R flowing into the room per unit time during the startup preparation time ton. Furthermore, due to the temperature difference control process, the difference between the room temperature at shutdown and the outside temperature is the same as the temperature difference at startup (Tset - Tin). Therefore, the temperature change at the time toff for the change in room temperature after shutdown can be considered equivalent to the difference between Tset and Tin. Thus, the time toff for the change in room temperature after shutdown can be expressed as shown in equation (3).

[0117]

[0118] The amount of heat W processed by the air conditioner 4 during the startup reserve time ton can be obtained from the air conditioner 4. The indoor heat capacity C is a value determined by the material of the building 10 and the size of the space targeted by the air conditioner 4. By obtaining the amount of heat processed by the air conditioner 4 and the change in room temperature obtained from the indoor temperature sensor 61 over a long period of time, the indoor heat capacity C and the coefficients a, b, and α can be estimated. Therefore, using equations (2) and (3), the amount of heat stored in the building structure Q can be estimated from the startup reserve time ton and the time toff of the change in room temperature after shutdown.

[0119] The air conditioning control device 3 stores the estimated building heat storage amount Q, along with the startup data and shutdown data, in the learning data storage unit 37 as learning data for inferring the startup reserve time. Note that the model equations used to estimate the building heat storage amount Q are not limited to equations (2) and (3). If the building heat storage amount Q is estimated using the startup reserve time ton and the time toff of the room temperature change after shutdown when temperature difference control processing is performed, a different thermal characteristic model may be used. Next, the air conditioning control device 3 proceeds to step S106.

[0120] In step S106, the air conditioning control device 3 determines whether or not it is time to learn the startup reserve time. If it is time to learn, proceed to step S402 (S106: Yes). If it is not time to learn, proceed to step S108 (S106: No).

[0121] In step S402, the air conditioning control device 3 learns the startup reserve time using learning data, including the amount of heat stored in the building structure, collected by the data acquisition unit 32. Figure 17 is a flowchart showing the flow of learning the startup reserve time in the air conditioning control device 3 according to this embodiment. The flow of learning the startup reserve time will be explained with reference to Figure 17.

[0122] First, in step S50, the data processing unit 33 reads the learning data stored in the learning data holding unit 37, and creates learning data by combining the measured values ​​of the startup conditions and startup buffer time on the same day, the shutdown conditions and the time of room temperature change after shutdown, and the amount of heat stored in the building body estimated in step S401. The created learning data is output to the learning unit 34. Next, the process proceeds to step S51.

[0123] In step S51, the learning unit 34 learns the startup buffer time using the training data and generates a trained model that infers the startup buffer time from the startup conditions. The learning unit 34 stores the generated trained model in the trained model holding unit 38. With this, the learning of the startup buffer time is completed, and the process proceeds to step S108 in Figure 16.

[0124] Returning to Figure 16, step S108 completes the collection of training data and the processing of one cycle of training in the air conditioning control device 3, and the process returns to step S101. This concludes the process of collecting training data and training to infer the startup buffer time in the air conditioning control device 3.

[0125] As described above, in this embodiment, the estimated building structure heat storage amount Q is used as training data for learning the startup reserve time. The startup reserve time is affected by the building structure heat storage amount Q in addition to the outside temperature, room temperature, and set temperature. By estimating the building structure heat storage amount Q in advance and using it as training data, the learning unit 34 can generate a trained model that infers a startup reserve time suitable for the building 10 in which the air conditioner 4 is installed.

[0126] The inference of the startup buffer time using the trained model by the air conditioning control device 3 and the control of the air conditioner 4 using the inferred startup buffer time are the same as in Embodiment 1, so an explanation will be omitted.

[0127] As described above, the learning unit 34 of the air conditioning control device 3 in this embodiment estimates the amount of heat stored in the building structure using the startup reserve time of the air conditioner 4 and the time of room temperature change after shutdown, and generates a trained model using the estimated amount of heat stored in the building structure as a feature. As a result, the learning unit 34 can generate a trained model that infers a startup reserve time suitable for the building 10 in which the air conditioner 4 is installed.

[0128] Furthermore, the learning unit 34 estimates the amount of heat stored in the building structure using the room temperature obtained from the indoor temperature sensor 61 installed on the air conditioner 4 and the amount of heat processed by the air conditioner 4. By estimating the amount of heat stored in the building structure using values ​​obtained by actually operating the air conditioner 4, it is possible to estimate an appropriate amount of heat stored in the building structure for the building 10 in which the air conditioner 4 is installed.

[0129] The configurations shown in the above embodiments are merely examples of the content of the present invention, and can be combined with other known technologies. It is also possible to omit or modify parts of the configuration without departing from the spirit of the present invention.

[0130] According to this disclosure, an air conditioning control device can be obtained that collects training data for accurately estimating the startup buffer time of an air conditioner, taking into account the building's performance, in a shorter period of time than when using only startup data.

[0131] 1 Control server, 2 Management device, 3 Air conditioning control device, 4 Air conditioner, 5 Outdoor unit, 6 Indoor unit, 6a First indoor unit, 6b Second indoor unit, 7 Handheld remote control, 8 Room, 10 Building, 11 Wide area network, 31 Communication unit, 32 Data acquisition unit, 33 Data processing unit, 34 Learning unit, 35 Calculation unit, 36 Control unit, 37 Learning data storage unit, 38 Learned model storage unit, 51 Outdoor temperature sensor, 61 Indoor temperature sensor, 61a First indoor temperature sensor, 61b Second indoor temperature sensor.

Claims

1. An air conditioning control device for controlling the operation of an air conditioner, comprising: a data collection unit that acquires learning data including data at the start and stop of the air conditioner; a learning unit that uses the learning data to generate a trained model for inferring the start-up buffer time from the start of the air conditioner until the room temperature reaches a set temperature; a calculation unit that uses the trained model to output the start-up buffer time from the start-up conditions acquired by the data collection unit; and a control unit that causes the air conditioner to start operation before the start-up buffer time, which is set to a predetermined time.

2. The air conditioning control device according to claim 1, characterized in that the startup data is the startup conditions of the air conditioner and the startup buffer time from startup until the set temperature is reached, and the shutdown data is the shutdown conditions of the air conditioner and the time of room temperature change after shutdown until the rate of change of room temperature after shutdown falls below a threshold.

3. The air conditioning control device according to claim 2, characterized in that the startup conditions include the room temperature and outside temperature at the time the air conditioner is started, and the shutdown conditions include the room temperature and outside temperature at the time the air conditioner is stopped.

4. The air conditioning control device according to claim 2 or 3, characterized in that the startup conditions include at least one of the number of indoor units provided by the air conditioner and the amount of capacity saved by the outdoor unit provided by the air conditioner.

5. The control unit determines the startup temperature difference, which is the difference between the room temperature and the outside temperature when the air conditioner is started, and when the scheduled time for stopping the air conditioner arrives, it performs temperature difference control to control the air conditioner so that the difference between the room temperature and the outside temperature approaches the startup temperature difference, and stops the operation of the air conditioner when the difference between the difference between the room temperature and the outside temperature and the startup temperature difference becomes smaller than a threshold, as described in any one of claims 1 to 4.

6. The air conditioning control device according to claim 5, characterized in that the temperature difference control by the control unit controls the operating capacity of the air conditioner to be lower than before the scheduled stop time when the difference between the room temperature and the outside temperature is greater than the startup temperature difference, and controls the operating capacity of the air conditioner to be higher than before the scheduled stop time when the difference between the room temperature and the outside temperature is less than the startup temperature difference.

7. The air conditioning control device according to any one of claims 1 to 4, wherein the air conditioner controlled by the air conditioning control device has a first indoor unit and a second indoor unit, and the control unit stores the time from when the first indoor unit is started up until when the second indoor unit is started up as a start time difference, and stops the second indoor unit when the start time difference has elapsed since the first indoor unit stopped operating.

8. The air conditioning control device according to claim 5 or 6, characterized in that the learning unit estimates the amount of heat stored in the building structure using the startup preparation time of the air conditioner and the time of change in room temperature after shutdown until the rate of change in room temperature after shutdown falls below a threshold, and generates the trained model using the estimated amount of heat stored in the building structure as a feature quantity.

9. The air conditioning control device according to claim 8, characterized in that the learning unit estimates the amount of heat stored in the building structure using the room temperature obtained from the indoor temperature sensor installed in the air conditioner and the amount of heat processed by the air conditioner.

Images