Air conditioner control device, air conditioner control method, and program

Abstract

This air conditioner control device comprises: a demand response information acquisition unit that acquires demand response information which includes a start time of a demand response and an end time thereof; a prediction unit that, when a start-time indoor temperature and a start-time outside air temperature at the start time of the demand response have been input, uses a temperature prediction model for outputting a prediction of a transition in indoor temperature by the demand response to predict an end-time indoor temperature at the end time of the demand response; and a setting unit that sets a target temperature to which the start-time indoor temperature has been adjusted so that the end-time indoor temperature is within a preset first temperature range.

Description

Air conditioner control device, air conditioner control method, and program

[0001] This application claims priority to Japanese Patent Application No. 2023-172421, filed on October 4, 2023, the contents of which are incorporated herein by reference.

[0002] To respond to fluctuations in the balance between supply and demand for electricity, a method called demand response (DR) is used, in which consumers adjust their electricity usage. For example, when excess demand is predicted, a supplier issues a DR command (downward DR) to consumers, requesting them to reduce their electricity usage during periods of excess demand. Air conditioners and other appliances comply with this DR command and perform control to cut or reduce power consumption during the DR period. This reduces the air conditioner's heating and cooling capacity during the DR period, which can cause the indoor temperature to deviate from the set temperature and reduce comfort. For this reason, technologies have been developed that, for example, perform supercooling by lowering the set temperature by a certain temperature during cooling and superheating by raising the set temperature by a certain temperature during heating before DR control is performed (see Patent Documents 1 and 2).

[0003] Japanese Patent No. 6375923 Japanese Patent No. 2913584

[0004] However, how the indoor temperature changes during the DR period varies depending on various factors, such as the length of the DR period, the rate of reduction in power consumption, and the outdoor temperature. Therefore, with conventional technology that controls the temperature by a certain amount lower (supercooling) or higher (superheating) than the preset temperature, for example, when the indoor temperature changes significantly due to factors such as the outdoor temperature, the supercooling or superheating may be insufficient, causing the indoor temperature to deviate from the set temperature during the DR period and making it difficult to maintain comfort. On the other hand, when the indoor temperature changes only slightly, for example, the supercooling or superheating may be excessive, resulting in unnecessary power consumption.

[0005] An object of the present disclosure is to provide an air conditioner control device, an air conditioner control method, and a program that can reduce total power consumption throughout all time periods by minimizing overcooling or overheating before demand response while preventing the indoor temperature during the demand response period from deviating from a comfortable temperature range.

[0006] According to one aspect of the present disclosure, an air conditioner control device includes a demand response information acquisition unit that acquires demand response information including a start time and an end time of a demand response; a prediction unit that predicts an end indoor temperature at an end time of the demand response using a temperature prediction model that, when a start indoor temperature and a start outdoor temperature at the start time of the demand response are input, outputs a prediction of the change in indoor temperature due to the demand response; and a setting unit that sets a target temperature by adjusting the start indoor temperature so that the end indoor temperature is within a predetermined first temperature range.

[0007] According to one aspect of the present disclosure, a method for controlling an air conditioner includes the steps of acquiring demand response information including a start time and an end time of a demand response, predicting an end indoor temperature at an end time of the demand response using a temperature prediction model that outputs a prediction of the progress of the indoor temperature due to the demand response when a start indoor temperature and a start outdoor temperature at the start time of the demand response are input, and setting a target temperature by adjusting the start indoor temperature so that the end indoor temperature is within a predetermined first temperature range.

[0008] According to one aspect of the present disclosure, the program causes an air conditioner control device to perform the following steps: acquire demand response information including a start time and an end time of the demand response; predict the end indoor temperature at the end time of the demand response using a temperature prediction model that, when a start indoor temperature and a start outdoor temperature at the start time of the demand response are input, outputs a prediction of the change in indoor temperature due to the demand response; and set a target temperature by adjusting the start indoor temperature so that the end indoor temperature is within a predetermined first temperature range.

[0009] According to the above aspect, it is possible to prevent the indoor temperature during the demand response period from deviating from a comfortable temperature range, while minimizing overcooling or overheating before the demand response, thereby reducing the total power consumption throughout all time periods.

[0010] 1 is a diagram illustrating an example of an air conditioning system according to a first embodiment. FIG. 1 is a flowchart illustrating the flow of a method for learning a temperature prediction model according to the first embodiment. FIG. 2 is a diagram illustrating an example of learning data for a temperature prediction model according to the first embodiment. FIG. 3 is a diagram illustrating an example of a prediction of a change in indoor temperature according to the first embodiment. FIG. 4 is a diagram illustrating an example of learning data for a temperature prediction model according to a second embodiment. FIG. 5 is a diagram illustrating an example of a control method for an air conditioner according to the second embodiment. FIG. 6 is a diagram illustrating an example of an air conditioning system according to a third embodiment. FIG. 7 is a diagram illustrating an example of learning data for a humidity prediction model according to the third embodiment. FIG. 8 is a flowchart illustrating the flow of a control method for an air conditioner according to the third embodiment. FIG. 9 is a diagram illustrating an example of an air conditioning system according to a fourth embodiment. FIG. 10 is a flowchart illustrating the flow of a method for learning a temperature prediction model according to the fourth embodiment. FIG. 11 is a diagram illustrating an example of learning data for a humidity prediction model according to the fourth embodiment.

[0011] First Embodiment Hereinafter, a first embodiment will be described in detail with reference to FIGS.

[0012] (Overall Configuration of Air Conditioning System) FIG. 1 is a diagram illustrating an example of an air conditioning system according to a first embodiment. As illustrated in FIG. 1, the air conditioning system 100 includes an air conditioner 1 and a server 20. The air conditioner 1 includes an outdoor air temperature sensor 2, an indoor air temperature sensor 3, and a control device 10. The server 20 is, for example, a server of an electricity supplier. When an excess demand for electricity is predicted for a certain period in the future, the server 20 distributes a demand response (DR) command to each electricity consumer, instructing them to limit their electricity consumption for that period. The DR command includes a start time (start date and time) and an end time (end date and time) of the DR. The DR command is distributed to each electricity consumer before the DR is implemented. Note that FIG. 1 illustrates a configuration in which the control device 10 is installed inside the air conditioner 1, but in other embodiments, some of the functions of the control device 10 may be installed on an external server, such as the server 20.

[0013] The outdoor air temperature sensor 2 is provided, for example, in the outdoor unit of the air conditioner 1. The indoor temperature sensor 3 is provided, for example, in the indoor unit of the air conditioner 1. The outdoor air temperature sensor 2 and the indoor temperature sensor 3 may be installed in other locations as long as they are capable of measuring the outdoor air temperature and the indoor temperature, respectively.

[0014] The control device 10 is connected to the outdoor air temperature sensor 2 and the indoor temperature sensor 3, and acquires the outdoor air temperature and indoor temperature measured by each sensor. The control device 10 normally controls the air conditioner 1 so that the indoor temperature becomes the set temperature set by the user. The control device 10 is also communicably connected to the server 20 via the network NW. The control device 10 controls the air conditioner 1 by limiting power consumption during the DR period. The control device 10 also controls the air conditioner 1 to supercool or superheat before the DR period so that the indoor temperature during the DR period does not deviate from the set temperature.

[0015] (Functional Configuration of Control Device) The control device 10 includes a processor 11, a memory 12, a storage 13, and a communication interface 14. The processor 11 operates according to a predetermined program to cause the control device 10 to perform various functions. The functions of the processor 11 will be described later. The memory 12 has a memory area necessary for the operation of the processor 11. The storage 13 is a so-called auxiliary storage device, such as a hard disk drive (HDD), a solid state drive (SSD), or an EEPROM. The storage 13 stores data that each part of the processor 11 acquires, generates, and references during processing. The communication interface 14 is an interface for transmitting and receiving various data, signals, and the like between the outdoor air temperature sensor 2, the indoor temperature sensor 3, each part of the air conditioner 1, and the server 20.

[0016] Next, we will explain the functions of the processor 11. The processor 11 operates in accordance with a predetermined program to fulfill the functions of a demand-response information acquisition unit 110, a sensor information acquisition unit 111, a prediction unit 112, a setting unit 113, a control unit 114, and a learning unit 115.

[0017] The demand response information acquisition unit 110 (DR information acquisition unit 110) acquires demand response information (DR information) including the start and end times of the DR period. The DR information acquisition unit 110 may also acquire, as one piece of DR information, a demand response rate (DR rate) representing the rate of reduction in power consumption during the DR period. The DR rate is, for example, a reduction rate relative to rated power consumption. When the DR rate is 50%, the air conditioner operates with power consumption reduced to 50% of the rated power consumption, and when the DR rate is 100%, the air conditioner operates with power consumption reduced to zero (stopped). In other words, the larger the DR rate, the smaller the power consumption of the air conditioner 1. The DR rate may be, for example, a fixed value specified in advance by the user. In this case, the DR information acquisition unit 110 acquires the DR rate specified in advance by the user from the storage 13. The DR rate may also be a value specified by the supplier depending on the power supply and demand forecast. In this case, the DR rate is included in the DR command distributed by the supplier.

[0018] The sensor information acquisition unit 111 acquires the outdoor temperature and the indoor temperature from the outdoor temperature sensor 2 and the indoor temperature sensor 3 .

[0019] When the start indoor temperature and start outdoor temperature at the start time of DR are input, the prediction unit 112 predicts the end indoor temperature at the end time of DR using a temperature prediction model M1 that outputs a prediction of the change in indoor temperature due to DR.

[0020] The setting unit 113 sets a target temperature by adjusting the starting indoor temperature so that the ending indoor temperature falls within a predetermined first temperature range. The first temperature range indicates a temperature range in which the user feels comfortable, and is, for example, a range of ±α degrees from the set temperature set by the user.

[0021] The control unit 114 controls the air conditioner 1 (DR control) by reducing power consumption at a predetermined reduction rate during the DR period based on the DR information distributed from the server 20. Furthermore, before the DR period starts, the control unit 114 controls the air conditioner 1 so that the indoor temperature becomes the target indoor temperature set by the setting unit 113 at the start time of the DR period. Furthermore, after the DR period ends, the control unit 114 controls the air conditioner 1 so that the indoor temperature becomes the set temperature set by the user.

[0022] The learning unit 115 learns the temperature prediction model M1 based on the time-series data of the outdoor temperature and the indoor temperature acquired by the sensor information acquisition unit 111 during DR control or when operation is stopped.

[0023] (Method for learning temperature prediction model) Fig. 2 is a flowchart showing the flow of a method for learning a temperature prediction model according to the first embodiment. Fig. 3 is a diagram showing an example of learning data for the temperature prediction model according to the first embodiment. Here, the flow of a method for learning the temperature prediction model M1 by the control device 10 will be described with reference to Figs. 2 and 3 .

[0024] First, the control device 10 collects learning data D1 for the temperature prediction model M1 (step S101). As shown in FIG. 3 , the learning data D1 includes time-series data of indoor temperatures acquired from the indoor temperature sensor 3 during DR control, outdoor temperatures acquired from the outdoor temperature sensor 2, and the DR rate. The time-series data of indoor temperatures is data that associates indoor temperatures with the elapsed time (minutes) since the start of DR control. The outdoor temperature is the outdoor temperature at the start of DR control. For example, the learning unit 115 collects time-series data of outdoor temperatures and indoor temperatures acquired by the sensor information acquisition unit 111 during DR control. Furthermore, upon ending DR control, the learning unit 115 records the learning data D1, including the collected time-series data of outdoor temperatures and indoor temperatures, in the storage 13. Note that if the DR rate is not a fixed value, the learning unit 115 records the learning data D1 for each DR rate (a1%, a2%, a3%, ...). At this time, the learning data D1 also includes information on the DR rate. Furthermore, the learning unit 115 may collect data when the operation is stopped as learning data D1 indicating a DR rate of 100% (power consumption of zero). While FIG. 3 shows an example in which the learning data D1 includes time-series data on the indoor temperature during cooling, in reality, the learning data D1 also includes time-series data on the indoor temperature during heating.

[0025] Next, the learning unit 115 learns a temperature prediction model M1 based on the multiple pieces of learning data D1 recorded in the storage 13 (step S102). When the start outdoor air temperature and start indoor temperature at the start time of DR control are input, the learning unit 115 learns the temperature prediction model M1 so that it outputs a predicted change in the indoor temperature during DR control. Furthermore, if the learning data D1 includes a DR rate, when the DR rate is further input, the learning unit 115 learns the temperature prediction model M1 so that it outputs a predicted change in the indoor temperature when DR control is performed at the specified DR rate. The learned temperature prediction model M1 is recorded in the storage 13.

[0026] The learning unit 115 may include in the learning data D1 not only the outdoor air temperature at the start of DR control, but also data (time-series data) on the outdoor air temperature at each time point during DR control. In this case, when the start time of DR is input, the learning unit 115 predicts whether the outdoor air temperature will tend to increase or decrease during the DR period and trains the temperature prediction model M1 to predict the change in the indoor temperature according to this change in the outdoor air temperature. Alternatively, when the learning unit 115 inputs the predicted value of the outdoor air temperature for each hour during the DR period, the learning unit 115 may train the temperature prediction model M1 to predict the change in the indoor temperature according to changes in the outdoor air temperature (predicted value). The predicted value of the outdoor air temperature may be weather forecast information obtained from an external server (not shown). In this way, the learning unit 115 can further improve the accuracy of indoor temperature prediction by the temperature prediction model M1.

[0027] The learning unit 115 repeatedly collects learning data D1 and learns the temperature prediction model M1 every time the DR control (or operation stop) of the air conditioner 1 is performed.

[0028] (Air conditioner control method) Fig. 4 is a flowchart showing the flow of the air conditioner control method according to the first embodiment. Fig. 5 is a diagram for explaining the air conditioner control method according to the first embodiment. Fig. 6 is a first diagram showing an example of a prediction of the progress of the indoor temperature according to the first embodiment. Fig. 7 is a second diagram showing an example of a prediction of the progress of the indoor temperature according to the first embodiment. Fig. 8 is a third diagram showing an example of a prediction of the progress of the indoor temperature according to the first embodiment. Here, the flow of the control method of the air conditioner 1 by the control device 10 will be described with reference to Figs. 4 to 8.

[0029] First, the DR information acquisition unit 110 acquires DR information (step S111). The DR information acquisition unit 110 acquires, as DR information, the start time and end time of the DR included in the DR command delivered from the supplier. Furthermore, if the DR command specifies a DR rate, the DR information acquisition unit 110 may further acquire this DR rate as DR information.

[0030] The setting unit 113 also sets the starting indoor temperature and starting outdoor temperature at the start of the DR (step S112). The starting indoor temperature of the DR is, for example, a set temperature set by the user for the first time. The starting outdoor temperature of the DR may be the current outdoor temperature acquired by the sensor information acquisition unit 111 from the outdoor temperature sensor 2, for example, if the start time of the DR is less than a predetermined time from the present. Furthermore, if the start time of the DR is more than a predetermined time from the present, the starting outdoor temperature may be set based on weather forecast information acquired from an external server (not shown).

[0031] Next, the prediction unit 112 uses the temperature prediction model M1 to predict the indoor temperature at the end of the dry-driving cycle (step S113). As shown in FIG. 5 , the prediction unit 112 inputs the starting indoor temperature and the starting outdoor temperature into the temperature prediction model M1. The temperature prediction model M1 then outputs a predicted change in the indoor temperature. For example, as shown in the example of FIG. 5 , the temperature prediction model M1 outputs change prediction data that predicts how the starting indoor temperature will change over time (minutes). The prediction unit 112 references the outputted indoor temperature change prediction and predicts what the indoor temperature (indoor temperature at the end of the dry-driving cycle) will be when the time has passed from the start time t0 to the end time t1.

[0032] If the temperature prediction model M1 is trained for each DR rate, the DR rate may be further input to the temperature prediction model M1 to obtain as an output a predicted change in the indoor temperature when operating at this DR rate. Furthermore, if the temperature prediction model M1 also trains time-series data on the outdoor air temperature, the DR start time may also be input to the temperature prediction model M1. In this case, the temperature prediction model M1 predicts the change trend (increasing or decreasing trend) of the outdoor air temperature from the start time and predicts the change in the indoor temperature according to this change trend of the outdoor air temperature. Alternatively, instead of the DR start time, the predicted value of the outdoor air temperature for each time during the DR period may be input to the temperature prediction model M1. The predicted value of the outdoor air temperature is obtained, for example, as weather forecast information from an external server (not shown). In this case, the temperature prediction model M1 predicts the change in the indoor temperature according to the change in the outdoor air temperature (predicted value).

[0033] Next, the setting unit 113 determines whether the predicted indoor end temperature is within the first temperature range (step S114). FIG. 6 illustrates an example of predicted indoor temperature transitions during cooling operation when the set temperature T0 is the DR start indoor temperature. The first temperature range is a range of ±α degrees from the set temperature T0 (a range from the lower limit temperature TL to the upper limit temperature TU). In the example of FIG. 6, if the air conditioner 1 is operated at the specified DR rate during the DR period (during DR control), the indoor end temperature will exceed the first temperature range (upper limit temperature TU). This may result in the user being uncomfortable during the DR period. If the indoor end temperature exceeds the first temperature range (step S114; NO), the setting unit 113 changes the indoor start temperature (step S116). For example, the setting unit 113 lowers the indoor start temperature by 1 degree during cooling and raises the indoor start temperature by 1 degree during heating.

[0034] Furthermore, when the setting unit 113 changes the starting indoor temperature, the prediction unit 112 uses the changed starting indoor temperature to predict the ending indoor temperature again (step S113). Then, the setting unit 113 re-determines whether the predicted ending indoor temperature is within the first temperature range (step S114). The setting unit 113 and the prediction unit 112 repeat the processes of steps S116, S113, and S114 until the ending indoor temperature is within the first temperature range.

[0035] FIG. 7 shows an example of predicted indoor temperature trends during cooling operation when the starting indoor temperature is lowered (e.g., by 1°C) below the user's set temperature T0. In the example of FIG. 7, by supercooling the indoor temperature before the start of DR, the ending indoor temperature is within the first temperature range (upper limit temperature TU). Therefore, even if the air conditioning capacity of the air conditioner 1 is reduced (or the air conditioner 1 is stopped) due to DR, the user can remain comfortable. If the ending indoor temperature is within the first temperature range (step S114; YES), the setting unit 113 sets the starting indoor temperature used in the prediction as the target temperature before the start of DR (step S115). In this way, the setting unit 113 can minimize supercooling or superheating before the start of DR while preventing the indoor temperature during the DR period from exceeding the first temperature range (the temperature range in which the user can remain comfortable).

[0036] It is desirable that the setting unit 113 sets as the target temperature the one of the multiple starting indoor temperatures that results in an ending indoor temperature closest to the upper limit temperature TU during cooling and that results in an ending indoor temperature closest to the lower limit temperature TL during heating. This prevents excessive overcooling or overheating before the start of DR, thereby preventing an increase in total power consumption throughout the entire operating time period of the air conditioner 1. For comparison, FIG. 8 shows an example in which the temperature reduction during overcooling is constant (e.g., 2 degrees) as in the prior art. As shown in FIG. 8, by significantly reducing the starting indoor temperature, the ending indoor temperature still has a margin before exceeding the comfortable temperature range (upper limit temperature TU). In other words, it is considered that overcooling before the start of DR is excessive. In contrast, in the present embodiment, as shown in FIG. 7, the temperature reduction during overcooling before the start of DR can be set to the minimum necessary value based on the relationship with the outdoor temperature, etc., rather than a fixed value. This reduces power consumption during overcooling compared to the prior art (FIG. 8). Therefore, in this embodiment, the total power consumption throughout all time periods in which the air conditioner 1 is operating can be reduced compared to the prior art.

[0037] The setting unit 113 may also limit the target temperature so that the indoor temperature falls within a second temperature range during supercooling or superheating before the start of DR. For example, the setting unit 113 sets the second temperature range to the same range as the first temperature range (a range of ±α degrees from the set temperature). In steps S113 to S116, the setting unit 113 changes the starting indoor temperature within the second temperature range and searches for a starting indoor temperature (target temperature) that will result in an ending indoor temperature within the first temperature range. In this way, the setting unit 113 can prevent the indoor temperature from becoming too hot or too cold before the start of DR, thereby preventing the user's comfort from being impaired.

[0038] Furthermore, the setting unit 113 may expand the second temperature range if the starting indoor temperature falls within the second temperature range and the ending indoor temperature exceeds the first temperature range. In this case, the allowable range for expanding the second temperature range may be preset so that the user does not feel uncomfortable. For example, if the first temperature range is set to ±2 degrees from the set temperature, the setting unit 113 may set the allowable range for expanding the second temperature range to ±3 degrees from the set temperature. In this way, by allowing the room temperature to be slightly lower or higher to some extent during short-term overcooling or overheating before the dynamic range (DR) period, it is possible to more reliably prevent the room temperature from exceeding the first temperature range during the long-term DR period.

[0039] Furthermore, the control unit 114 controls the air conditioner 1 so that the indoor temperature reaches the target temperature before the DR start time (step S117). Thereafter, during the period from the DR start time to the DR end time, the control unit 114 controls the air conditioner 1 so that the power consumption is reduced by the specified DR rate. In this way, by the control device 10 supercooling or superheating the indoor temperature before the DR start, it is possible to prevent the indoor temperature from deviating from the first temperature range (a comfortable temperature range) during the DR period.

[0040] (Actions and Effects) As described above, the control device 10 of the air conditioner 1 according to this embodiment includes a DR information acquisition unit 110 that acquires DR information including the start time and end time of DR, a prediction unit 112 that predicts the end indoor temperature at the end time of DR using a temperature prediction model M1 that outputs a prediction of the progress of the indoor temperature due to DR when the start indoor temperature and start outdoor air temperature at the start time of DR are input, and a setting unit 113 that sets a target temperature for the indoor temperature to be reached by the start time of DR so that the end indoor temperature is within a predetermined first temperature range.

[0041] In this way, the control device 10 can minimize overcooling or overheating before the dynamic range change while preventing the indoor temperature during the dynamic range change period from deviating from a comfortable temperature range (first temperature range) based on the indoor temperature and outdoor temperature at the start of the dynamic range change. As a result, the total power consumption of the air conditioner 1 throughout the entire time period in which the air conditioner 1 is operating can be reduced.

[0042] Furthermore, the setting unit 113 sets, as the target temperature, the starting indoor temperature that falls within a preset second temperature range, among the starting indoor temperatures that will result in the ending indoor temperature falling within the first temperature range.

[0043] In this way, the control device 10 can suppress the indoor temperature from changing to an extent that would impair comfort not only during the DR period but also during overcooling or overheating before the DR period.

[0044] In addition, the temperature prediction model M1 is trained to predict the change trend of the outdoor air temperature after the start time when the start indoor temperature, start outdoor air temperature, and start time outdoor air temperature are input, and to output a predicted change trend of the indoor temperature according to the predicted change trend of the outdoor air temperature.

[0045] In this way, the control device 10 can predict whether the outdoor temperature will tend to rise or fall during the DR period, thereby improving the accuracy of predicting the indoor temperature at the end of the DR period.

[0046] In addition, the temperature prediction model M1 is trained to output a prediction of the indoor temperature change according to the change in the outdoor temperature at each time, when the starting indoor temperature and starting outdoor temperature as well as the outdoor temperature at each time from the start time to the end time are input.

[0047] In this way, the control device 10 can improve the accuracy of predicting the indoor temperature at the end of the DR period by taking into account changes in the outdoor temperature during the DR period. The outdoor temperature during the DR period may be obtained from weather forecast information provided by an external server.

[0048] In addition, the DR information further includes a DR rate, and the temperature prediction model M1 is trained to output a prediction of the indoor temperature trend according to the DR rate when the DR rate is input in addition to the starting indoor temperature and the starting outdoor temperature.

[0049] In this way, even if a different DR rate is specified by the user or supplier, the control device 10 can appropriately set the target temperature so that the indoor temperature during the DR period does not deviate from a comfortable temperature range (first temperature range) while minimizing overcooling or overheating before DR according to the DR rate.

[0050] Furthermore, if there is no target temperature within the second temperature range that will result in the room temperature at the end being within the first temperature range, the setting unit 113 expands the second temperature range and sets the target temperature.

[0051] By doing this, the control device 10 allows the indoor temperature to become slightly lower or higher during the short period of overcooling or overheating before DR, and can more reliably prevent the indoor temperature from exceeding the first temperature range during the long DR period.

[0052] Second Embodiment A second embodiment of the present disclosure will be described below with reference to Figures 9 and 10. Of the configurations of the second embodiment, configurations that are common to the first embodiment are assigned the same reference numerals and descriptions thereof will be omitted.

[0053] (Method for Learning Temperature Prediction Model) FIG. 9 is a diagram showing an example of learning data for a temperature prediction model according to the second embodiment. As shown in FIG. 9, in step S101 of FIG. 2, the learning unit 115 of this embodiment includes time series data for the indoor temperature before DR control in addition to time series data for the indoor temperature during DR control in the learning data D1 to learn the temperature prediction model M1. While FIG. 9 illustrates learning data D1 including time series data for cooling, in reality, learning data D1 also includes time series data for heating. In the example of FIG. 9, time t0 is the time when DR control is started, and times t-1, t-2, and t-3 are the times when the target temperature for supercooling is reached.

[0054] Furthermore, when the start outdoor temperature and the start indoor temperature are input in step S102 of FIG. 2 , the learning unit 115 trains the temperature prediction model M1 to output a prediction of the indoor temperature transition during the DR period based on the amount of supercooling or superheating performed at the start indoor temperature before the DR period. As shown in FIG. 9 , for example, during cooling, by sufficiently extending the supercooling time before the DR period, the temperature of the building's building structure also drops, allowing the building structure to function as a refrigerant during the DR period. In other words, the heat absorbed by the building structure due to supercooling or superheating can be used as a heat source during the DR period. By further including the supercooling or superheating period before the DR period in the learning data D1, the learning unit 115 can allow the temperature prediction model M1 to learn the impact of the heat absorbed by the building structure on the transition of the indoor temperature during the DR period.

[0055] (Air conditioner control method) Figure 10 is a diagram for explaining an air conditioner control method according to the second embodiment. In step S113 of Figure 4, the prediction unit 112 inputs the start outdoor air temperature and the start indoor temperature into the temperature prediction model M1. As a result, as shown in Figure 10, the temperature prediction model M1 outputs predicted indoor temperature transitions P1, P2, and P3 after the DR start time t0 when supercooling is performed t-1, t-2, and t-3 minutes before DR.

[0056] In steps S114 to S116 of FIG. 4 , the setting unit 113 sets a combination of a target temperature and a target time indicating how long before the DR cycle should reach the target temperature based on a combination of the starting indoor temperature and the supercooling period such that the indoor temperature at the end time t1 falls within a first temperature range (between the lower limit temperature TL and the upper limit temperature TU). In the example of FIG. 10 , for the indoor temperature transition predictions P2 and P3, the ending indoor temperature at the end time t1 is equal to or lower than the upper limit temperature TU. Therefore, the setting unit 113 sets this starting indoor temperature as the target temperature and also sets t-2 minutes or t-3 minutes before the DR cycle start time as the target time for reaching this target temperature. The setting unit 113 sets the target time based on the shortest valid time between t-2 minutes and t-3 minutes. In the example of FIG. 10 , the setting unit 113 sets t-2 minutes before the DR cycle start time as the target time. Although FIG. 10 shows an example of cooling, the same process is performed during heating.

[0057] (Actions and Effects) As described above, in the control device 10 according to this embodiment, the temperature prediction model M1 is further trained so that, when the starting indoor temperature and the starting outdoor temperature are input, it outputs a predicted change in the indoor temperature corresponding to the period during which supercooling or superheating was performed at the starting indoor temperature before demand response, and the setting unit 113 sets the target temperature and the target time indicating how long before the start time the target temperature should be reached, based on the combination of the starting indoor temperature and the period during which the ending indoor temperature falls within the first temperature range.

[0058] By doing this, the control device 10 can more reliably prevent the indoor temperature during the DR period from deviating from a comfortable temperature range (first temperature range) while minimizing overcooling or overheating before the DR period.

[0059] Third Embodiment A second embodiment of the present disclosure will now be described with reference to Figures 11 to 13. Of the configurations of the second embodiment, configurations that are common to the first embodiment are assigned the same reference numerals and descriptions thereof will be omitted.

[0060] (Overall configuration of air conditioning system) Fig. 11 is a diagram showing an example of an air conditioning system according to the third embodiment. As shown in Fig. 11, the air conditioner 1 according to this embodiment further includes a humidity sensor 4. The humidity sensor 4 is provided, for example, in the indoor unit of the air conditioner 1. The control device 10 (sensor information acquisition unit 111) is connected to the humidity sensor 4 and acquires the indoor humidity measured by the humidity sensor 4.

[0061] (Functional Configuration of the Control Device) When the start humidity at the DR start time is input, the prediction unit 112 further predicts the end humidity at the DR end time using a humidity prediction model M2 that outputs a prediction of the change in indoor humidity due to the DR.

[0062] When there is no target temperature within the second temperature range at which the indoor temperature at the end will be within the first temperature range, the setting unit 113 sets a target humidity to be reached by the DR start time so that the humidity at the end will be equal to or lower than a predetermined upper limit humidity.

[0063] (Learning Method for Humidity Prediction Model) FIG. 12 is a diagram illustrating an example of learning data for the humidity prediction model according to the third embodiment. In step S101 of FIG. 2 , the learning unit 115 further collects learning data D2 for the humidity prediction model M2. As illustrated in FIG. 12 , the learning data D2 includes, for example, time-series humidity data acquired from the humidity sensor 4 during DR control. The humidity time-series data is data that associates humidity (%) with the elapsed time (minutes) since the start of DR control. For example, the learning unit 115 collects time-series humidity data acquired by the sensor information acquisition unit 111 during DR control. Furthermore, upon ending DR control, the learning unit 115 records the learning data D2 including the collected time-series humidity data in the storage 13. Data collected when operation is stopped may also be collected as the learning data D2. While FIG. 12 illustrates an example in which the learning data D2 includes time-series indoor temperature data during cooling, in reality, the learning data D2 also includes time-series indoor temperature data during heating.

[0064] 2, the learning unit 115 further learns the humidity prediction model M2 based on the plurality of learning data D2 recorded in the storage 13. When the starting humidity at the start time of the DR is input, the learning unit 115 learns the humidity prediction model M2 so as to output a predicted change in the indoor humidity during DR control.

[0065] The learning unit 115 repeatedly collects learning data D2 and learns the humidity prediction model M2 every time the DR control (or operation stop) of the air conditioner 1 is performed.

[0066] (Air conditioner control method) Fig. 13 is a flowchart showing the flow of an air conditioner control method according to the third embodiment. Here, the flow of the control method for the air conditioner 1 by the control device 10 will be described with reference to Fig. 13. Note that the example in Fig. 13 shows a control method during cooling.

[0067] First, the DR information acquisition unit 110 acquires DR information (step S311). The setting unit 113 sets the starting indoor temperature and the starting outdoor temperature at the start of the DR cycle (step S312). Next, the prediction unit 112 predicts the ending indoor temperature at the end of the DR cycle using the temperature prediction model M1 (step S313). The setting unit 113 determines whether the predicted ending indoor temperature is within the first temperature range (step S314). If the predicted ending indoor temperature is within the first temperature range (step S314: YES), the setting unit 113 sets the starting indoor temperature used in the prediction as the target temperature (step S315). These processes are the same as those in the first embodiment (steps S111 to S115 in FIG. 4). Then, the control unit 114 controls the air conditioner 1 so that the indoor temperature reaches the target temperature before the DR cycle starts (step S318). This process is the same as that in the first embodiment (step S117 in FIG. 4).

[0068] Furthermore, if the predicted end-time indoor temperature exceeds the first temperature range (step S314: NO), the setting unit 113 determines whether the start-time indoor temperature can be changed (step S316). For example, if the start-time indoor temperature is changed (lowered) but still falls within the second temperature range, the setting unit 113 determines that the start-time indoor temperature can be changed (step S316: YES). In this case, the setting unit 113 changes (lowers) the start-time indoor temperature (step S317) and returns to step S313. This process is the same as the process in the first embodiment (step S116 in FIG. 4).

[0069] On the other hand, if the initial indoor temperature is changed (decreased) and does not fall within the second temperature range, the setting unit 113 determines that the initial indoor temperature cannot be changed (step S316: NO). In this case, adjusting the initial indoor temperature makes it difficult to prevent the ending indoor temperature from exceeding the first temperature range. However, even if the indoor temperature is the same, users tend to feel hot when the humidity is high and cool when the humidity is low. Therefore, if it is difficult to maintain a low ending indoor temperature by adjusting the indoor temperature alone, the control device 10 causes the air conditioner 1 to perform dehumidification operation before the start of the dynamic range (DR) period to lower the humidity during the DR period. Specifically, the setting unit 113 first sets the initial indoor temperature (the lower limit of the second temperature range) used in the previous prediction as the target temperature (step S319). The setting unit 113 also sets the starting humidity at the start of the DR period (step S320). For example, if the start time of the DR is less than a predetermined time from the present, the start humidity may be the current humidity acquired by the sensor information acquisition unit 111 from the humidity sensor 4. Alternatively, if the start time of the DR is more than a predetermined time from the present, the start humidity may be set based on weather forecast information acquired from an external server (not shown).

[0070] Next, the prediction unit 112 predicts the ending humidity at the end time of the dry cleaning using the humidity prediction model M2 (step S321). The prediction unit 112 inputs the starting humidity into the humidity prediction model M2. Then, the humidity prediction model M2 outputs a humidity transition prediction.

[0071] Next, the setting unit 113 determines whether the predicted end humidity exceeds a predetermined upper limit humidity (step S322). If the end humidity exceeds the upper limit humidity (step S322; YES), the setting unit 113 changes the start humidity (step S323). For example, the setting unit 113 reduces the start humidity by 10%.

[0072] Furthermore, when the setting unit 113 changes the starting humidity, the prediction unit 112 predicts the ending humidity again using the changed starting humidity (step S321). The setting unit 113 then re-determines whether the predicted ending humidity is equal to or lower than the upper limit humidity (step S322). The setting unit 113 and the prediction unit 112 repeat the processes of steps S321 to S323 until the ending humidity becomes equal to or lower than the upper limit humidity.

[0073] If the ending humidity is equal to or lower than the upper limit humidity (step S322; NO), the setting unit 113 sets the starting humidity used for the prediction as the target humidity before the start of the dry cleaning cycle (step S324).

[0074] The control unit 114 also controls the air conditioner 1 so that the indoor temperature and humidity reach the target temperature and target humidity before the DR start time (step S325). Thereafter, during the period from the DR start time to the DR end time, the control unit 114 controls the air conditioner 1 so that power consumption is reduced by the specified DR rate. This prevents the indoor temperature from falling below the second temperature range due to overcooling before the DR start time, thereby maintaining comfort during the period before the DR start time. On the other hand, even if the indoor temperature exceeds the first temperature range (upper limit temperature TU) during the DR period, comfort can be maintained during the DR period by dehumidifying the room in advance so that the humidity during the DR period does not exceed the upper limit humidity.

[0075] (Actions and Effects) As described above, in the control device 10 according to this embodiment, when the start humidity at the start time is input, the prediction unit 112 further predicts the end humidity at the end time using the humidity prediction model M2 that outputs a prediction of the change in indoor humidity due to demand response, and when there is no target temperature within the second temperature range at which the end indoor temperature is within the first temperature range, the setting unit 113 sets the target humidity to be reached by the start time of the demand response so that the end humidity is equal to or lower than a predetermined upper limit humidity.

[0076] This prevents the indoor temperature from falling below the second temperature range due to excessive cooling before the start of DR, thereby maintaining comfort during the period before the start of DR. On the other hand, even if the indoor temperature exceeds the first temperature range (upper limit temperature TU) during the DR period, comfort can be maintained during the DR period by dehumidifying the room in advance so that the humidity during the DR period does not exceed the upper limit humidity.

[0077] Fourth Embodiment A second embodiment of the present disclosure will be described below with reference to Figures 14 to 17. Of the configurations of the second embodiment, configurations that are common to the first embodiment are assigned the same reference numerals and descriptions thereof will be omitted.

[0078] (Overall configuration of air conditioning system) Fig. 14 is a diagram showing an example of an air conditioning system according to the fourth embodiment. As shown in Fig. 14, the air conditioning system 100 according to this embodiment includes multiple air conditioners 1A and 1B. Each of the multiple air conditioners 1 conditions a different room. In the example of Fig. 14, the air conditioner 1A conditions room A on the second floor, and the air conditioner 1B conditions room B on the first floor.

[0079] In this embodiment, the control device 10 controls each of the multiple air conditioners 1A and 1B. The control device 10 adjusts the DR rate of each air conditioner 1A and 1B according to the air conditioning load of each air conditioner 1A and 1B. In the example of FIG. 14 , the room temperature of room A on the second floor is easily affected by the sun during the day, while the room temperature of room B on the first floor is less affected by the sun and is less likely to rise. Since the room temperature of room A on the second floor is easily increased, the air conditioning load to lower the room temperature during cooling is large and the air conditioning load to raise the room temperature during heating is small. On the other hand, the room temperature of room B on the first floor is less likely to increase, so the air conditioning load to lower the room temperature during cooling is small and the air conditioning load to raise the room temperature during heating is large. The control device 10 adjusts the DR rate according to the air conditioning load of each air conditioner to achieve the specified DR rate for the entire air conditioning system 100. Note that, for simplicity of explanation, FIG. 14 illustrates an example in which the control device 10 controls two air conditioners 1A and 1B, but the present invention is not limited to this. The control device 10 may control three or more air conditioners 1 .

[0080] (Method for learning temperature prediction model) Fig. 15 is a flowchart showing the flow of a method for learning a temperature prediction model according to the fourth embodiment. Fig. 16 is a diagram showing an example of learning data for the temperature prediction model according to the fourth embodiment. Here, the flow of a method for learning the temperature prediction model M1 by the control device 10 will be described with reference to Figs. 15 and 16 .

[0081] First, the control device 10 collects learning data D3 for the temperature prediction model M1 (step S401). As shown in FIG. 16 , the learning data D3 is data collected for each of the air conditioners 1A and 1B, and includes time-series data of indoor temperatures acquired from the indoor temperature sensor 3 during DR control, outdoor temperatures acquired from the outdoor temperature sensor 2, and the DR rate. The time-series data of indoor temperatures is data that associates indoor temperatures with the elapsed time (minutes) since the start of DR control. Each piece of learning data D3 is labeled to indicate which air conditioner the learning data corresponds to. Furthermore, the learning data D3 may further include a capacity (class) indicating the air conditioning capacity of each of the air conditioners 1A and 1B.

[0082] Next, the learning unit 115 learns the temperature prediction models M1a and M1b for each of the air conditioners 1A and 1B (step S402) based on the multiple pieces of learning data D3 recorded in the storage 13. Specifically, when the start indoor temperature, start outdoor temperature, DR start time, and DR rate are input, the learning unit 115 learns each of the temperature prediction models M1a and M1b so as to output a prediction of the change in the indoor temperature after the start time.

[0083] In addition, the learning unit 115 predicts the changes in the air conditioning load for each of the air conditioners 1A and 1B at each time based on the predicted changes in the indoor temperature using the temperature prediction models M1a and M1b, and sets weights for the air conditioners 1A and 1B for each DR rate and time period required for the entire air conditioning system 100 (step S403).

[0084] FIG. 17 is a diagram showing an example of weighting of air conditioners according to the fourth embodiment. The weighting is a weighting of the air conditioning capacity exerted by each air conditioner 1A, 1B during DR control. The learning unit 115 sets a higher weighting as the air conditioning load decreases and a lower weighting as the air conditioning load increases. FIG. 17 shows an example of weighting of the DR rates of each air conditioner 1A, 1B for each time period when a DR rate of 50% is required for the entire air conditioning system 100. For example, in a certain time period 1, there is no difference in the air conditioning load between the air conditioners 1A and 1B, and both are weighted to "1." When a DR rate of 50% is required for time period 1, the DR rates of both air conditioners 1A, 1B are 50%. Furthermore, in time period 2, the air conditioning load of air conditioner 1B is smaller than that of air conditioner 1A, and therefore a higher weighting is set for air conditioner 1B than for air conditioner 1A. In the example of Fig. 17, the weighting of air conditioner 1A is "0.6", and the weighting of air conditioner 1B is "1.4". In this case, when a DR rate of 50% is requested in time slot 2, the DR rate of air conditioner 1A will be 30%, and the DR rate of air conditioner 1B will be 70%. Fig. 17 only shows an example in which a DR rate of 50% is requested for the entire air conditioning system 100, but weightings are set similarly for the other DR rates. A weighting table D4 in which weightings are set for each of air conditioners 1A and 1B is recorded in storage 13.

[0085] The learning unit 115 repeatedly collects learning data D3, learns the temperature prediction models M1a and M1b, and updates the weighting table D4 every time DR control is performed on each of the air conditioners 1A and 1B.

[0086] (Air conditioner control method) Fig. 18 is a flowchart showing the flow of an air conditioner control method according to the fourth embodiment. Here, the flow of the method of controlling the air conditioner 1 by the control device 10 will be described with reference to Fig. 18 .

[0087] First, the DR information acquisition unit 110 acquires DR information (step S411). The setting unit 113 sets the initial indoor temperature and the initial outdoor temperature at the start of DR for each of rooms A and B (step S412). These processes are the same as those in the first embodiment (steps S111 and S112 in FIG. 4).

[0088] Next, the setting unit 113 refers to the weighting table D4 corresponding to the DR rate (overall DR rate) required for the entire air conditioning system 100, and reads out the weighting of each air conditioner 1A, 1B for the time period that includes the DR start time (step S413).

[0089] The prediction unit 112 also uses the temperature prediction models M1a and M1b to predict the indoor temperatures at the end of the DR for each of rooms A and B (step S414). At this time, the prediction unit 112 inputs, for example, the starting indoor temperature, the starting outdoor temperature, and the DR rate for each of air conditioners 1A and 1B, calculated by multiplying the overall DR rate by a weighting, into each of the temperature prediction models M1a and M1b, and obtains as output a predicted transition of the indoor temperatures for each of rooms A and B. If the overall DR rate is 50%, the weighting for air conditioner 1A is 0.6, and the weighting for air conditioner 1B is 1.4, the prediction unit 112 inputs 30% as the DR rate for air conditioner 1A into the temperature prediction model M1a for air conditioner 1A, and inputs 70% as the DR rate for air conditioner 1B into the temperature prediction model M1b for air conditioner 1B. The prediction unit 112 refers to the output predicted changes in the indoor temperatures of each of rooms A and B, and predicts what the indoor temperatures (indoor temperatures at the end) of each of rooms A and B will be when the DR start time and end time have elapsed.

[0090] Next, the setting unit 113 determines whether the predicted end-time indoor temperatures for each of rooms A and B are within the first temperature range (step S415). If the predicted end-time indoor temperatures for each of rooms A and B are within the first temperature range (step S415: YES), the setting unit 113 sets the start-time indoor temperatures used in the prediction as the target temperatures for each of the air conditioners 1A and 1B (step S416). If the predicted end-time indoor temperatures for either room exceed the first temperature range (step S415: NO), the setting unit 113 changes the start-time indoor temperature for that room (step S417). The prediction unit 112 then re-predicts the end-time indoor temperatures using the changed start-time indoor temperatures (step S414). The setting unit 113 then re-determines whether the predicted end-time indoor temperatures are within the first temperature range (step S415). The setting unit 113 and the prediction unit 112 repeat the processes of steps S414, S415, and S417 until the room temperatures at the end of all the rooms are within the first temperature range.

[0091] The control unit 114 controls the air conditioners 1A and 1B so that the indoor temperatures of the rooms A and B reach the target temperatures before the DR start time (step S418). After that, during the period from the DR start time to the DR end time, the control unit 114 controls the air conditioners 1A and 1B so that the power consumption is reduced by the DR rate assigned to each of the air conditioners 1A and 1B.

[0092] (Actions and Effects) As described above, in the control device 10 according to this embodiment, the learning unit 115 sets weights for the demand response rates of the air conditioners 1A and 1B based on the predicted changes in the indoor temperatures of the rooms A and B output by the temperature prediction models M1a and M1b of the plurality of air conditioners 1A and 1B that are to air condition different rooms A and B. Furthermore, the prediction unit 112 inputs, into the temperature prediction models M1a and M1b of the air conditioners 1A and 1B, the demand response rates obtained by multiplying the overall demand response rate required for the plurality of air conditioners as a whole by the weights for the air conditioners 1A and 1B, to predict the ending indoor temperatures of the rooms A and B, and the setting unit 113 sets target temperatures for the air conditioners 1A and 1B so that the ending indoor temperatures of the rooms A and B are within the first temperature range.

[0093] In this way, the control device 10 can increase or decrease the DR rate according to the air conditioning load of each air conditioner 1A, 1B, thereby achieving the required overall DR rate overall while more reliably preventing the indoor temperatures of each room A, B from deviating from a comfortable temperature range (first temperature range) during the DR period.

[0094] Other Embodiments Although the embodiments have been described in detail above with reference to the drawings, the specific configuration is not limited to the above, and various design changes are possible. That is, in other embodiments, the order of the above-described processes may be changed as appropriate. Furthermore, some processes may be executed in parallel.

[0095] <Additional Notes> The air conditioner control device, air conditioner control method, and program described in the above-described embodiment can be understood, for example, as follows.

[0096] (1) According to a first aspect, the control device 10 of the air conditioner 1 includes a demand response information acquisition unit 110 that acquires demand response information including the start time and end time of the demand response, a prediction unit 112 that predicts the end indoor temperature at the end time of the demand response using a temperature prediction model M1 that outputs a prediction of the change in the indoor temperature due to the demand response when the start indoor temperature and the start outdoor air temperature at the start time of the demand response are input, and a setting unit 113 that sets a target temperature by adjusting the start indoor temperature so that the end indoor temperature is within a predetermined first temperature range.

[0097] In this way, the control device 10 can minimize overcooling or overheating before the dynamic range change while preventing the indoor temperature during the dynamic range change period from deviating from a comfortable temperature range (first temperature range) based on the indoor temperature and outdoor temperature at the start of the dynamic range change. As a result, the total power consumption of the air conditioner 1 throughout the entire time period in which the air conditioner 1 is operating can be reduced.

[0098] (2) According to the second aspect, in the control device 10 of the air conditioner 1 according to the first aspect, the setting unit 113 sets the starting indoor temperature that falls within a predetermined second temperature range as the target temperature, from among the starting indoor temperatures that will result in the ending indoor temperature falling within the first temperature range.

[0099] In this way, the control device 10 can suppress the indoor temperature from changing to an extent that would impair comfort not only during the DR period but also during overcooling or overheating before the DR period.

[0100] (3) According to the third aspect, in the control device 10 of the air conditioner 1 relating to the first or second aspect, the temperature prediction model M1 is trained to predict the trend in change in the outdoor air temperature after the start time when the start time is further input, and to output a predicted change in the indoor temperature according to the predicted trend in change in the outdoor air temperature.

[0101] In this way, the control device 10 can predict whether the outdoor temperature will tend to rise or fall during the DR period, thereby improving the accuracy of predicting the indoor temperature at the end of the DR period.

[0102] (4) According to the fourth aspect, in the control device 10 of the air conditioner 1 relating to the first or second aspect, the temperature prediction model M1 is trained to output a prediction of the change in the indoor temperature according to the change in the outdoor air temperature at each time, when the outdoor air temperature at each time from the start time to the end time is further input.

[0103] In this way, the control device 10 can improve the accuracy of predicting the indoor temperature at the end of the DR period by taking into account changes in the outdoor temperature during the DR period. The outdoor temperature during the DR period may be obtained from weather forecast information provided by an external server.

[0104] (5) According to a fifth aspect, in the control device 10 of the air conditioner 1 relating to any one of the first to fourth aspects, the demand response information further includes a demand response rate indicating the rate of reduction in power consumption required from the start time to the end time, and the temperature prediction model M1 is trained to output a prediction of the indoor temperature trend according to the demand response rate when the demand response rate is further input.

[0105] In this way, even if a different DR rate is specified by the user or supplier, the control device 10 can appropriately set the target temperature so that the indoor temperature during the DR period does not deviate from a comfortable temperature range (first temperature range) while minimizing overcooling or overheating before DR according to the DR rate.

[0106] (6) According to the sixth aspect, in the control device 10 of the air conditioner 1 relating to any one of the first to fifth aspects, the temperature prediction model M1 is further trained to output a prediction of the indoor temperature trend corresponding to the period during which supercooling or superheating was performed at the starting indoor temperature before demand response when the starting indoor temperature and the starting outdoor temperature are input, and the setting unit 113 sets the target temperature and the target time indicating how long before the start time the target temperature should be reached based on the combination of the starting indoor temperature and the period during which the ending indoor temperature is within the first temperature range.

[0107] By doing this, the control device 10 can more reliably prevent the indoor temperature during the DR period from deviating from a comfortable temperature range (first temperature range) while minimizing overcooling or overheating before the DR period.

[0108] (7) According to the seventh aspect, in the control device 10 of the air conditioner 1 relating to any one of the first to sixth aspects, when the start humidity at the start time is input, the prediction unit 112 further predicts the end humidity at the end time using a humidity prediction model M2 that outputs a prediction of the change in indoor humidity due to demand response, and when there is no target temperature within the second temperature range at which the end indoor temperature is within the first temperature range, the setting unit 113 sets the target humidity to be reached by the start time of the demand response so that the end humidity is equal to or lower than a predetermined upper limit humidity.

[0109] This prevents the indoor temperature from falling below the second temperature range due to excessive cooling before the start of DR, thereby maintaining comfort during the period before the start of DR. On the other hand, even if the indoor temperature exceeds the first temperature range (upper limit temperature TU) during the DR period, comfort can be maintained during the DR period by dehumidifying the room in advance so that the humidity during the DR period does not exceed the upper limit humidity.

[0110] (8) According to the eighth aspect, the control device 10 of the air conditioner 1 according to any one of the first to seventh aspects further includes a learning unit 115 that sets a weighting for the demand response rate for each of the air conditioners 1A, 1B based on a predicted change in the indoor temperature for each of the rooms A, B output by the temperature prediction models M1a, M1b of each of the plurality of air conditioners 1A, 1B that are to condition different rooms A, B. The prediction unit 112 inputs, into the temperature prediction models M1a, M1b of each of the air conditioners 1A, 1B, a demand response rate obtained by multiplying the overall demand response rate required for the plurality of air conditioners as a whole by the weighting for each of the air conditioners 1A, 1B, to predict the ending indoor temperatures for each of the rooms A, B. The setting unit 113 sets a target temperature for each of the air conditioners 1A, 1B so that the ending indoor temperatures for each of the rooms A, B are within a first temperature range.

[0111] In this way, the control device 10 can increase or decrease the DR rate according to the air conditioning load of each air conditioner 1A, 1B, thereby achieving the required overall DR rate overall while more reliably preventing the indoor temperatures of each room A, B from deviating from a comfortable temperature range (first temperature range) during the DR period.

[0112] (9) According to a ninth aspect, the control method for the air conditioner 1 includes the steps of: acquiring demand response information including a start time and an end time of the demand response; predicting an end indoor temperature at an end time of the demand response using a temperature prediction model M1 that outputs a prediction of the progress of the indoor temperature due to the demand response when a start indoor temperature and a start outdoor temperature at the start time of the demand response are input; and setting a target temperature by adjusting the start indoor temperature so that the end indoor temperature is within a predetermined first temperature range.

[0113] (10) According to the tenth aspect, the program causes the control device 10 of the air conditioner 1 to execute the following steps: acquiring demand response information including the start time and end time of the demand response; predicting the end indoor temperature at the end time of the demand response using a temperature prediction model M1 that, when the start indoor temperature and start outdoor temperature at the start time of the demand response are input, outputs a prediction of the change in indoor temperature due to the demand response; and setting a target temperature by adjusting the start indoor temperature so that the end indoor temperature is within a predetermined first temperature range.

[0114] According to the above aspect, it is possible to prevent the indoor temperature during the demand response period from deviating from a comfortable temperature range, while minimizing overcooling or overheating before the demand response, thereby reducing the total power consumption throughout all time periods.

[0115] REFERENCE SIGNS LIST 100 Air conditioning system 1, 1A, 1B Air conditioner 2 Outdoor air temperature sensor 3 Indoor temperature sensor 4 Humidity sensor 10 Control device 11 Processor 110 Demand response information acquisition unit 110 DR information acquisition unit 111 Sensor information acquisition unit 112 Prediction unit 113 Setting unit 114 Control unit 115 Learning unit 12 Memory 13 Storage 14 Communication interface 20 Server D1, D2, D3 Learning data D4 Weighting table M1, M1a, M1b Temperature prediction model M2 Humidity prediction model

Claims

1. An air conditioner control device comprising: a demand response information acquisition unit that acquires demand response information including a start time and an end time of a demand response; a prediction unit that predicts an end indoor temperature at an end time of the demand response using a temperature prediction model that outputs a prediction of the change in indoor temperature due to demand response when a start indoor temperature and a start outdoor air temperature at the start time of the demand response are input; and a setting unit that sets a target temperature by adjusting the start indoor temperature so that the end indoor temperature is within a predetermined first temperature range.

2. The air conditioner control device according to claim 1, wherein the setting unit sets as the target temperature a starting indoor temperature that is within a predetermined second temperature range, out of starting indoor temperatures at which the ending indoor temperature is within the first temperature range.

3. The air conditioner control device of claim 1, wherein the temperature prediction model is trained to predict the trend of change in outdoor air temperature after the start time when the start time is further input, and to output a prediction of the progress of indoor temperature according to the predicted trend of change in outdoor air temperature.

4. The air conditioner control device of claim 1, wherein the temperature prediction model is trained to output a prediction of the change in indoor temperature in response to changes in the outdoor air temperature at each time point from the start time to the end time when the outdoor air temperature at each time point is further input.

5. The air conditioner control device of claim 1, wherein the demand response information further includes a demand response rate indicating a rate of reduction in power consumption required from the start time to the end time, and the temperature prediction model is trained to output a prediction of the indoor temperature trend according to the demand response rate when the demand response rate is further input.

6. The air conditioner control device of claim 1, wherein the temperature prediction model is further trained to output a prediction of the indoor temperature trend corresponding to a period during which supercooling or superheating was performed at the start indoor temperature before the demand response when the start indoor temperature and the start outdoor temperature are input, and the setting unit sets the target temperature and a target time indicating how long before the start time the target temperature should be reached based on a combination of the start indoor temperature and the period during which the end indoor temperature falls within the first temperature range.

7. The air conditioner control device of claim 2, wherein the prediction unit, when receiving input of the start humidity at the start time, further predicts the end humidity at the end time using a humidity prediction model that outputs a prediction of the change in indoor humidity due to the demand response, and the setting unit, when there is no target temperature within the second temperature range that will cause the end indoor temperature to be within the first temperature range, sets the target humidity to be reached by the start time of the demand response so that the end humidity is below a predetermined upper limit humidity.

8. An air conditioner control device as described in any one of claims 1 to 7, further comprising a learning unit that sets a weighting for the demand response rate for each air conditioner based on a predicted trend in the indoor temperature for each room output by a temperature prediction model for each of a plurality of air conditioners that condition different rooms, wherein the prediction unit inputs a demand response rate obtained by multiplying an overall demand response rate required for the entire plurality of air conditioners by the weighting for each air conditioner into the temperature prediction model for each air conditioner, and predicts the ending indoor temperature for each of the rooms, and the setting unit sets the target temperature for each air conditioner so that the ending indoor temperature for each of the rooms is within the first temperature range.

9. A method for controlling an air conditioner comprising the steps of: acquiring demand response information including a start time and an end time of a demand response; predicting an end indoor temperature at an end time of the demand response using a temperature prediction model which, when a start indoor temperature and a start outdoor temperature at the start time of the demand response are input, outputs a prediction of the progress of the indoor temperature due to the demand response; and setting a target temperature by adjusting the start indoor temperature so that the end indoor temperature is within a predetermined first temperature range.

10. A program that causes an air conditioner control device to execute the following steps: acquiring demand response information including a start time and an end time of a demand response; predicting an end indoor temperature at an end time of the demand response using a temperature prediction model that outputs a prediction of the change in indoor temperature due to demand response when a start indoor temperature and a start outdoor temperature at the start time of the demand response are input; and setting a target temperature by adjusting the start indoor temperature so that the end indoor temperature is within a predetermined first temperature range.

Images