JP6927350B2 - Air conditioner and air conditioner system - Google Patents

Description

The present invention relates to air conditioners and air conditioners.

Conventionally, various methods for reducing traffic when transmitting and receiving data between electronic devices have been proposed. For example, the data server constantly monitors and accumulates the detected data, extracts the minimum data necessary for analysis from the accumulated large amount of data, compresses and converts it, and then compresses and converts the data. There is a method of transmitting data to an analyzer (for example, Patent Document 1).

Japanese Unexamined Patent Publication No. 2004-199377

In recent years, in air conditioners, for example, it is possible to train AI (Artificial Intelligence) on a server device to generate a learning model by learning operation history data such as a set temperature, and control the operation of air conditioning using the learning model. Proposed. However, when all the operation history data is transmitted to the server device as it is, the amount of operation history data transmitted from the air conditioner to the server device becomes enormous. As a result, the communication traffic between the air conditioner and the server device increases.

On the other hand, if a part of the operation history data is simply deleted in order to reduce the communication traffic, the operation history data required for generating the learning model may be insufficient, and the learning accuracy of AI may be lowered.

In view of such a problem, it is an object of the present invention to provide an air conditioner and an air conditioner system capable of reducing the traffic of data used for learning AI while maintaining the learning accuracy of AI.

One aspect of the air conditioner is an adapter having a control unit that controls the air conditioner using a learning model and a communication unit that communicates with a server device that generates a learning model based on the operation history data of the air conditioner. And have. The adapter has an acquisition unit, a determination unit, an erasing unit, and a transmission unit. The acquisition unit acquires operation history data from the air conditioner at predetermined intervals. The determination unit determines whether or not the amount of change between the temporally continuous data in the operation history data acquired by the acquisition unit is within a predetermined range. When the amount of change between continuous data is within a predetermined range, the erasing unit deletes the other data while leaving at least one of the continuous data. The transmission unit transmits the operation history data after the other data has been deleted by the erasing unit to the server device.

As one aspect, the traffic of data used for learning AI can be reduced while maintaining the learning accuracy of AI.

FIG. 1 is an explanatory diagram showing an example of an air conditioning system of this embodiment. FIG. 2 is a block diagram showing an example of the hardware configuration of the adapter. FIG. 3 is a block diagram showing an example of the functional configuration of the adapter. FIG. 4 is a block diagram showing an example of the hardware configuration of the server device. FIG. 5 is a block diagram showing an example of the functional configuration of the server device. FIG. 6 is an explanatory diagram showing an example of the contents of the operation history data. FIG. 7 is an explanatory diagram showing an example of operation history data used for generating or updating a sensible temperature setting prediction model in summer or winter. FIG. 8 is an explanatory diagram showing an example of the thinning process. FIG. 9 is an explanatory diagram showing an example of a determination operation when determining whether or not the amount of change between operation history data is within a predetermined range. FIG. 10 is a flowchart showing an example of the processing operation of the CPU in the adapter related to the transmission processing. FIG. 11 is a flowchart showing an example of the processing operation of the CPU in the adapter related to the thinning process. FIG. 12 is a flowchart showing an example of the processing operation of the CPU in the server device involved in the update processing. FIG. 13 is a flowchart showing an example of the processing operation of the CPU in the server device involved in the restoration processing.

Hereinafter, examples of the air conditioner, the air conditioner, and the like disclosed in the present application will be described in detail based on the drawings. The disclosed technology is not limited by the present embodiment. In addition, each of the examples shown below may be appropriately modified as long as it does not cause a contradiction.

FIG. 1 is an explanatory diagram showing an example of the air conditioning system 1 of this embodiment. The air conditioning system 1 shown in FIG. 1 includes an indoor unit 2, an adapter 3, an access point 4, a server device 5, a relay device 6, a communication terminal 7, and a communication network 8.

The indoor unit 2 is, for example, a part of an air conditioner that is arranged indoors and heats or cools the air in the room. The user of the indoor unit 2 can remotely control the indoor unit 2 by operating the remote controller 9. The indoor unit 2 has a main body 2A and a control unit 2B that controls the main body 2A. The main body 2A is equipped with an indoor fan and an indoor heat exchanger, and the indoor air that has exchanged heat with the refrigerant by the indoor heat exchanger is blown out from the main body 2A to heat, cool, dehumidify, etc. the room. It is said. Further, the outdoor unit (not shown) is provided with an outdoor fan, a compressor, and the like. The communication terminal 7 is a terminal device such as a user's smartphone.

The adapter 3 has a communication function for wirelessly connecting the indoor unit 2 and the access point 4 and a control function for AI-controlling the indoor unit 2. The adapter 3 is arranged for each indoor unit 2. The access point 4 is a device that connects the adapter 3 and the communication network 8 by wireless communication using, for example, a WLAN (Wireless Local Area Network) or the like. The communication network 8 is, for example, a communication network such as the Internet. The server device 5 has a function of generating an AI learning model that controls the indoor unit 2, a database that stores operation history data, and the like. The server device 5 is located in a data center, for example. The relay device 6 has a function of connecting to the communication network 8 by communication and also connecting to the server device 5 by communication. The relay device 6 transmits the operation history data and the like used for generating or updating the learning model applied to the indoor unit 2 from the adapter 3 to the server device 5 via the communication network 8. Further, the relay device 6 transmits the learning model generated or updated by the server device 5 to the adapter 3 via the communication network 8. The relay device 6 is arranged in, for example, a data center or the like.

The relay device 6 has a first relay unit 6A, a second relay unit 6B, and a third relay unit 6C. The first relay unit 6A transmits various data related to AI control between the adapter 3 and the server device 5. The first relay unit 6A transmits the operation history data and the like used for generating or updating the learning model received from the adapter 3 to the server device 5 via the communication network 8, and the learning model generated or updated by the server device 5. Is transmitted to the adapter 3 via the communication network 8. The second relay unit 6B acquires the operating conditions (operating mode such as cooling / heating, set temperature, etc.) of the indoor unit 2 set by the user using the communication terminal 7 from outside, and obtains the operating conditions (such as the operating mode such as cooling / heating and the set temperature) of the indoor unit 2. Send to. The third relay unit 6C acquires external data such as a weather forecast from a communication network 8 such as the Internet, and transmits the acquired external data to the server device 5. Further, the third relay unit 6C transmits external data to the adapter 3 via the communication network 8.

FIG. 2 is a block diagram showing an example of the hardware configuration of the adapter 3. The adapter 3 shown in FIG. 2 includes a first communication unit 11, a second communication unit 12, a storage unit 13, and a CPU (Central Processing Unit) 14. The first communication unit 11 is a communication IF (Interface) such as a UART (Universal Asynchronous Receiver Transmitter) that communicates and connects with the control unit 2B in the indoor unit 2. The second communication unit 12 is a communication unit for communicating with the access point 4, for example, a communication IF such as WLAN. The storage unit 13 has, for example, a ROM (Read Only Memory), a RAM (Random Access Memory), and the like, and stores various information such as data and programs. The CPU 14 controls the entire adapter 3.

FIG. 3 is a block diagram showing an example of the functional configuration of the adapter 3. The storage unit 13 in the adapter 3 shown in FIG. 3 includes an operation history memory 13A for temporarily storing operation history data acquired from the indoor unit 2, a model memory 13B for storing a learning model acquired from the server device 5, and the like. It has an external memory 13C for storing external data.

The CPU 14 includes an acquisition unit 14A, a determination unit 14B, an erasing unit 14C, a transmission unit 14D, a reception unit 14E, a setting unit 14F, and a prediction control unit 14G.

The acquisition unit 14A acquires the operation history data and the time stamp every 5 minutes from the indoor unit 2 at the acquisition timing of a predetermined cycle, for example, every 5 minutes. The acquisition unit 14A stores the operation history data acquired every 5 minutes and the time stamp given to the operation history data in the operation history memory 13A.

The determination unit 14B includes the operation history data stored in the operation history memory 13A and the operation history data acquired in the acquired operation history data when the time stamps given to the operation history data are for a predetermined period, which are continuous in time. It is determined whether or not the amount of change between them is within a predetermined range. The predetermined period is, for example, 48 hours when the learning model is a sensible temperature setting prediction model described later. The 48 hours is, for example, a time in consideration of avoiding concentration of communication with the memory capacity of the adapter 3. The determination unit 14B determines whether or not the operation history data stored in the operation history memory 13A is the operation history data for 48 hours acquired in a 5-minute cycle. When the operation history data stored in the determination unit 14B is the operation history data for 48 hours, whether or not the amount of change between the continuous operation history data in the operation history data for 48 hours is within a predetermined range. Is determined. The amount of change between continuous operation history data is the amount of difference in change between the operation history data before and after being continuous in time. The amount of change between the front and rear operation history data is within a predetermined range, for example, the range of the amount of change in which the amount of change between the front and rear operation history data does not affect the learning of AI, for example, no change amount, or This is the case where the amount of change is within a negligible range for learning AI. The range of the amount of change in which the amount of change does not affect the learning of AI is the range in which the indoor unit 2 is not affected by the human experience even if the indoor unit 2 is controlled using the result of learning of AI based on the driving history data before and after. Is. Generally, it is necessary that there is no change between data in order to secure the reproduction of the data, the time stamp attached to the data, and the data deleted at the time of compression. In addition, it is difficult for a person to experience a change in humidity compared to a change in temperature in the environment, and the range considered to be the same value is wider in humidity than in temperature. For example, a person tends to be less likely to perceive a change in humidity of ± 1% than a change in temperature of ± 1 ° C. Therefore, when the operation history data is humidity, even if the amount of change when determining the data is coarser than the temperature, even if the indoor unit 2 is controlled by using the result of AI learning based on these data, a person Considering that it does not affect the sensible temperature, for example, the weighting of humidity with respect to the temperature at the sensible temperature is not a 1: 1 relationship, but a (temperature: humidity) = 1: 5 relationship, for example. The change of ± 5% of the humidity is considered to be the same because it is not a change that affects the human sensation when the indoor unit 2 is controlled by using the learning result of AI, for example. As a result, the amount of data to be transmitted can be reduced.

When the amount of change between the continuous operation history data is within a predetermined range, the determination unit 14B affects the human experience when the continuous operation history data controls the indoor unit 2 using the learning result of AI. Judge that the data is within the range not given. In addition, the data within the range that does not affect the human experience when the indoor unit 2 is controlled by using the learning result of AI may be deleted (in the following description, it may be described as "thinning out"). ) It becomes the target data. When the amount of change between continuous operation history data is not within a predetermined range, the determination unit 14B is data that affects the human experience when the operation history data controls the indoor unit 2 using the learning result of AI. Judge. The data that affects the human experience when the indoor unit 2 is controlled using the AI learning result is not the data to be thinned out.

When the amount of change between each operation history data in the continuous time stamp is within a predetermined range, the erasing unit 14C leaves one operation history data and the time stamp given to the other operation history data. And, the time stamp given to this is thinned out. Then, the erasing unit 14C sequentially stores the operation history data for 48 hours after the thinning out in the operation history memory 13A. For example, when the operation history data in which the amount of change is within a predetermined range is continuous from 12:00 to 12:15, the operation history data at 12:00, which is the oldest time stamp, is left, and the operation history data is 12:05, 10 The operation history data of minutes and 15 minutes and the time stamp attached to the data are thinned out. However, if the amount of change between non-continuous operation history data among the 48 hours of operation history data is within a predetermined range, the operation history data within the predetermined range shall not be thinned out. ..

The transmission unit 14D transmits the operation history data after thinning out stored in the operation history memory 13A and the time stamp given to the operation history data to the server device 5 via the communication network 8. The receiving unit 14E receives the learning model from the server device 5 via the communication network 8, and stores the received learning model in the model memory 13B. The setting unit 14F applies the learning model in storage to the prediction control unit 14G. The prediction control unit 14G controls the control unit 2B in the indoor unit 2 based on the learning model applied by the setting unit 14F. For convenience of explanation, the case where the predictive control unit 14G controls the control unit 2B in the indoor unit 2 based on the learning model is illustrated, but the predictive control unit 14G is the main body of the indoor unit 2 based on the learning model. 2A may be controlled directly. Further, the predictive control unit 14G transmits a control mode based on the learning model to the control unit 2B. That is, the prediction control unit 14G may indirectly control the main body 2A via the control unit 2B, and can be changed as appropriate.

FIG. 4 is a block diagram showing an example of the hardware configuration of the server device 5. The server device 5 shown in FIG. 4 has a communication unit 31, a storage unit 32, and a CPU 33. The communication unit 31 is a communication IF that communicates and connects with the relay device 6. The storage unit 32 has, for example, an HDD (Hard Disk Drive), a ROM, a RAM, or the like, and stores various information such as data and programs. The CPU 33 controls the entire server device 5.

FIG. 5 is a block diagram showing an example of the functional configuration of the server device 5. The storage unit 32 in the server device 5 shown in FIG. 5 has a data memory 32A and a model memory 32B. The data memory 32A stores the operation history data received from each adapter 3 and the time stamp and the like assigned to the operation history data. The model memory 32B stores the learning model generated or updated by the server device 5.

The CPU 33 in the server device 5 includes a model learning unit 33A, a receiving unit 33B, a determination unit 33C, a restoring unit 33D, and a transmitting unit 33E.

The model learning unit 33A connects to each adapter 3 of the plurality of indoor units 2 and receives operation history data for 48 hours from each adapter 3 via the access point 4, the communication network 8, and the relay device 6. Then, the model learning unit 33A learns using the operation history data for 48 hours stored in the data memory 32A from each adapter 3, and generates or updates the learning model of each indoor unit 2 based on the learning result. do. In the learning model, for example, the sensible temperature setting for the indoor user after 5 minutes is predicted according to the operating condition of the air conditioner in each household, and the sensible temperature is controlled according to the predicted sensible temperature. There is a predictive model. Conventionally, since the air conditioner adjusts the temperature so that the room temperature becomes the target temperature, the temperature change may be perceived as unpleasant to the user. On the other hand, the sensible temperature setting prediction model is used when adjusting the air conditioner so that the user feels comfortable according to, for example, time-series operation history data such as indoor temperature, indoor humidity, and outdoor temperature. It is a program to be executed. Then, the model learning unit 33A stores the generated or updated learning model in the model memory 32B. The receiving unit 33B receives the operation history data and the time stamp given to the operation history data from each adapter 3 via the communication network 8. For convenience of explanation, the model learning unit 33A exemplifies, for example, a case where a learning model of the sensible temperature setting prediction model is generated or updated using the operation history data for 48 hours. However, the model learning unit 33A is not limited to the operation history data for 48 hours, and may use the operation history data such as one month or half a year depending on the content of the learning model, as appropriate. It can be changed.

The determination unit 33C refers to the operation history data for 48 hours received from the adapter 3 and the time stamp given to the operation history data, and based on the presence or absence of the time stamp, is there a thinned section from the operation history data for 48 hours? Judge whether or not.

When it is determined that there is a thinning section, the restoration unit 33D determines that the section without the time stamp is the thinning section from which the operation history data has been thinned out because the time stamp has a 5-minute cycle. Further, the restoration unit 33D generates the operation history data of the thinned-out section and the time stamp given to the operation history data of the thinned-out section based on the operation history data immediately before the thinned-out section and the time stamp given thereto for 48 hours. Restore the operation history data and the time stamp attached to it. Then, the restoration unit 33D stores the operation history data for 48 hours after the restoration and the time stamp given to the operation history data in the data memory 32A. The restoration unit 33D refers to the presence / absence of a time stamp, for example, when the amount of change between the operation history data from 12:00 to 12:15 on the adapter 3 side is within a predetermined range. If there is no time stamp for the section from 12:05 to 12:15, the restoration unit 33D determines that the section from 12:05 to 12:15 is a thinned section. For example, when the restoration unit 33D determines that the section from 12:05 to 12:15 is the thinning section, the operation history data at 12:00 becomes the operation history data immediately before the thinning section. Therefore, the restoration unit 33D copies the operation history data at 12:00 as the operation history data at 12:05, 12:10, and 12:15 to obtain the operation history data of the thinned section and the operation history data thereof. Restore the given time stamp. Then, the restoration unit 33D stores the operation history data after restoration and the time stamp given to the operation history data in the data memory 32A. As a result, the restoration unit 33D restores the operation history data for 48 hours received from the adapter 3 and the operation history data of the thinned section among the time stamps given to the operation history data and the time stamps given to the operation history data. As a result, the operation history data for 48 hours and the time stamp given to the operation history data are stored in the data memory 32A. Since the time stamp is restored in units of a predetermined cycle (5-minute cycle), for example, the time stamp of the thinned-out section can be restored by incrementing the time stamp immediately before the thinned-out section by +5 minutes.

The model learning unit 33A generates or updates a learning model corresponding to the adapter 3 based on the operation history data for 48 hours for each adapter 3 stored in the data memory 32A and the time stamp given to the operation history data, and generates the learning model. Alternatively, the updated learning model is stored in the model memory 32B. The transmission unit 33E transmits the learning model generated or updated by the model learning unit 33A to the adapter 3 via the relay device 6, the communication network 8, and the access point 4.

FIG. 6 is an explanatory diagram showing an example of the contents of the operation history data. The operation history data includes, for example, operation state, operation mode, set temperature, indoor temperature, indoor humidity, air volume, wind direction, motion sensor, radiation sensor, indoor heat exchange temperature, outdoor temperature, compressor rotation speed, outdoor air volume, etc. Operating current, outdoor heat exchange temperature, etc. Further, the operation history data includes, for example, discharge temperature, compressor temperature, expansion valve opening degree, radiator temperature, start failure history, abnormal stop history, emergency operation history, time stamp, air conditioner ID, installation location and facility type. And so on.

The operating state is an ON-OFF state of the operation of the indoor unit 2. The operation mode is an operation mode such as cooling or heating of the indoor unit 2. The set temperature is an indoor target temperature in the room where the indoor unit 2 is used. The room temperature is the actual temperature in the room where the indoor unit 2 is used. The indoor humidity is the actual humidity in the room where the indoor unit 2 is used. The air volume is the air volume of the conditioned air blown out from the indoor unit 2. The wind direction is the wind direction of the conditioned air blown out from the indoor unit 2. The motion sensor is a detection result of the presence or absence of a person in the room and the amount of activity. The radiation sensor is a detection result of the temperature of the floor or wall in the room. The indoor heat exchange temperature is the temperature of an indoor heat exchanger (not shown) that forms a part of the main body 2A of the indoor unit 2. The outdoor temperature is the actual outdoor temperature. The compressor rotation speed is the operating rotation speed of the compressor provided in the outdoor unit (not shown) connected to the indoor unit 2 by the refrigerant pipe. The outdoor air volume is the air volume of the outdoor fan provided in the outdoor unit. The operating current is, for example, the operating current of the entire air conditioner such as the indoor unit 2 and the outdoor unit. The outdoor heat exchange temperature is the temperature of the outdoor heat exchanger provided in the outdoor unit. The discharge temperature is the temperature of the refrigerant discharged from the compressor. The compressor temperature is the temperature at the bottom of the compressor. The expansion valve opening degree is the opening degree of the electronic expansion valve provided in the outdoor unit. The radiator temperature is the temperature of the power semiconductor that drives and controls the compressor. The startup failure history is the history of compressor startup failure. The abnormal stop history is the history of abnormal stop. The emergency operation history is an implementation history of emergency operation. The time stamp is the year, month, day, hour, minute, and second at the time of data acquisition. The air conditioner ID is an ID given to the indoor unit 2 in order to identify the air conditioner such as the indoor unit 2. The installation location is the address of the location where the air conditioner is installed. The facility type is a type of facility such as a store, a restaurant, or a factory in which an air conditioner such as an indoor unit 2 is installed.

The operation history data includes data used for home air conditioners and data used for commercial air conditioners. The operation history data used in the air conditioner for home use includes, for example, the operation state, the operation mode, the set temperature, the room temperature, the room humidity, the air volume, the wind direction, the value detected by the motion sensor, and the radiation sensor. Value, time stamp, air conditioner ID, installation location, etc. In a home air conditioner, AI is used to perform operations and proposals in pursuit of comfort and energy saving. Therefore, for example, set temperature, operation mode, indoor and ambient environment, etc. are necessary data for home use.

The operation history data used in the commercial air conditioner includes, for example, the operation state, the operation mode, the set temperature, the room temperature, the room humidity, the air volume, the wind direction, the value detected by the human sensor, and the radiation sensor. These are the detected values, indoor heat exchange temperature, outdoor temperature, compressor rotation speed, outdoor air volume, operating current, outdoor heat exchange temperature, and the like. Other operation history data includes, for example, discharge temperature, compressor temperature, expansion valve opening, radiator temperature, start failure history, abnormal stop history, emergency operation history, time stamp, air conditioner ID, installation location, and the like. Facility type, etc. In commercial air conditioners, AI predicts the failure of each device and the need for maintenance. Therefore, for example, the operating status and history of each component level in the air conditioner used for failure prediction are accumulated to predict a failure. Is determined by AI. In addition to the above, the operation history data required for a commercial air conditioner is, for example, data used for failure prediction. The number of revolutions of the compressor and fan motor used in the air conditioner is stopped while the air conditioner is stopped, and each operation history data is not generated. Therefore, for example, the number of revolutions of the compressor, the amount of outdoor air, and the operating current. Data such as outdoor heat exchange temperature shall not be acquired while the air conditioner is stopped.

When the learning model is, for example, the above-mentioned sensible temperature setting prediction model, time-series operation history data such as set temperature, indoor temperature, indoor humidity, and outdoor temperature are used to generate or update the sensible temperature setting prediction model. FIG. 7 is an explanatory diagram showing an example of operation history data used for generating or updating a sensible temperature setting prediction model in summer or winter. As shown in FIG. 7, the operation history data used in the sensible temperature setting prediction model differs depending on, for example, the season. For example, in the winter sensible temperature setting prediction model, the set temperature, indoor temperature, indoor humidity, outdoor temperature, and the like are used. On the other hand, in the summer sensible temperature setting prediction model, in addition to the set temperature, indoor temperature, indoor humidity, outdoor temperature, etc., for example, air volume and detection data of the motion sensor (presence or absence of people and activity amount) are added. ..

As described above, the operation history data is acquired every 5 minutes, for example. If the operation history data corresponding to the continuous time stamps are the same, the time stamps and the operation history data are thinned out. At this time, a predetermined range of the amount of change during thinning is set according to the characteristics of each operation history data. The predetermined range of the amount of change when determining the operation history data corresponds to the width of the threshold value for determining the presence or absence of a change between continuous operation history data, and changes depending on, for example, summer or winter. ..

For example, for each data when determining the operating state, operating mode, set temperature, indoor temperature, indoor humidity, air volume, wind direction, value detected by the human sensor, value detected by the radiation sensor, indoor heat exchange temperature and outdoor temperature. The predetermined range of the change amount is not limited to, for example, summer or winter, and may be changed according to other seasons, the installation environment of the air conditioner, or the setting history of the air conditioner. It may be the same.

FIG. 8 is an explanatory diagram showing an example of the thinning process. As shown in FIG. 8, the CPU 14 of the adapter 3 acquires, for example, operation history data from data 1 to data 11 from the indoor unit 2 at a cycle of 5 minutes. For example, when the amount of change between each operation history data of the time stamp (181001113000) and the time stamp (1810011113500) data 1 to data 11 is within a predetermined range, the CPU 14 has each operation history data of the time stamp (181001113500). Is the same as each operation history data of the time stamp (181001113000) and is thinned out. On the other hand, when the CPU 14 has at least one operation history data whose change amount is not within a predetermined range among the operation history data of the time stamp (181001113500) and the time stamp (181001114000), the data 1 to the data 11. All the operation history data is transmitted to the server device 5 without thinning out the operation history data. As a result, when the operation history data is uploaded to the server device 5, the compression efficiency is increased because the original data is thinned out, so that the compression efficiency is increased and the communication traffic can be reduced.

FIG. 9 is an explanatory diagram showing an example of a determination operation when determining whether or not the amount of change between the operation history data is within a predetermined range, and here, among the operation history data shown in FIG. Take indoor humidity as an example. The accuracy of the humidity sensor used in the air conditioner is about ± 5 ° C. Relative humidity in a room may fluctuate greatly depending on the opening and closing of doors in the room. As described above, the change in humidity is harder to experience than the change in temperature. Therefore, by utilizing such a difference in experience, a predetermined range of the amount of change in room temperature and a predetermined range of the amount of change in indoor humidity can be obtained. change. Further, not only the predetermined range of the change amount of the room temperature and the indoor humidity may be changed, but also the predetermined range of the change amount may be changed according to the seasons such as winter and summer. In summer, dehumidification operation is selected in addition to cooling, and dehumidification operation controls humidity, whereas in winter, there is no operation mode to control humidity. Therefore, the predetermined range of the amount of change in humidity in winter is made larger than the predetermined range of the amount of change in humidity in summer. For example, in the case of summer, the amount of change in indoor humidity is in a predetermined range, for example, ± 3. If it is within%, it may be determined that there is no change, and in winter, if the amount of change in indoor humidity is within a predetermined range, for example, ± 5%, it may be determined that there is no change.

The explanatory diagram shown in FIG. 9 is indoor humidity data acquired at a 5-minute cycle. At the start point shown in FIG. 9, the indoor humidity (RH) is 45%, after which the amount of change continues within a predetermined range, for example ± 3%. However, when the indoor humidity reaches 53%, the change amount changes from 45% to + 8%, that is, the change amount is no longer within the predetermined range. Therefore, it is judged that there is a change, and the standard is set to 53%. After that, when the indoor humidity reached 49%, the amount of change changed from 53% to -4%, that is, the amount of change was no longer within the predetermined range, so it was judged that there was a change, and the standard was set to 49%. Further, when the indoor humidity reaches 45%, the change amount changes from 49% to -5%, and the change amount is no longer within the predetermined range. Therefore, it is judged that there is a change, and the standard is set to 45%. Further, when the indoor humidity reaches 50%, the amount of change changes from 45% to + 5%, and the amount of change is no longer within the predetermined range. Therefore, it is judged that there is a change, and the standard is set to 50%. As for other operation history data, as in the case of humidity, a predetermined range for determining whether or not there is a change is determined according to the characteristics of each data, and whether or not there is a change is determined, and each data in all the operation history data is determined. If the amount of change in is the same time stamp within a predetermined range, it will be thinned out.

For example, in the sensible temperature setting prediction model or the like, the acquisition unit 14A in the adapter 3 acquires operation history data such as the set temperature, the indoor temperature, the indoor humidity, and the outdoor temperature from the indoor unit 2 in a 5-minute cycle, and acquires the operation. The history data is stored in the operation history memory 13A. The determination unit 14B in the adapter 3 determines whether or not the operation history data stored in the operation history memory 13A is for 48 hours.

When the stored operation history data is for 48 hours, the determination unit 14B determines the amount of change between the continuous operation history data, for example, each operation history data at 12:00 and each operation history data at 12:05. It is determined whether or not the amount of change between the two is within a predetermined range. The amount of change between continuous operation history data does not include, for example, a time stamp. When the amount of change between the operation history data is within a predetermined range, the erasing unit 14C in the adapter 3 thins out the operation history data within the predetermined range. Further, the transmission unit 14D in the adapter 3 transmits the operation history data for 48 hours after the thinning out to the server device 5 via the access point 4, the communication network 8, and the relay device 6.

When the determination unit 33C in the server device 5 receives the operation history data for 48 hours, it determines whether or not there is a thinning section in the operation history data. When the restoration unit 33D in the server device 5 has a thinning section in the operation history data for 48 hours, the restoration unit 33D refers to the time stamp and determines that the section without the time stamp is the thinning section. Further, the restoration unit 33D restores the operation history data of the thinning section and the time stamp given to the operation history data of the thinning section based on the operation history data immediately before the thinning section and the time stamp given to the operation history data. The restoration unit 33D stores the operation history data after restoration and the time stamp given to the operation history data in the data memory 32A. For example, when the restoration unit 33D determines that the section from 12:05 to 12:15 is the thinning section, the operation history data at 12:00 becomes the operation history data immediately before the thinning section. Therefore, the restoration unit 33D copies the operation history data at 12:00 as the operation history data at 12:05, 12:10, and 12:15 to obtain the operation history data of the thinned section and the operation history data thereof. Restore the given time stamp. The model learning unit 33A in the server device 5 generates or updates a learning model such as a sensible temperature setting prediction model based on the operation history data stored in the data memory 32A and the time stamp given to the operation history data. Then, the transmission unit 33E in the server device 5 transmits the learning model such as the sensible temperature setting prediction model generated or updated by the model learning unit 33A to the adapter 3 via the relay device 6, the communication network 8, and the access point 4.

When the setting unit 14F in the adapter 3 receives a learning model such as a sensible temperature setting prediction model from the server device 5, the setting unit 14F stores the learning model in the model memory 13B. Then, the prediction control unit 14G in the adapter 3 is an indoor unit based on a learning model such as a sensible temperature setting prediction model stored in the model memory 13B when the room temperature is stably controlled at a set temperature of ± 1.5 degrees. Controls the control unit 2B of 2. For convenience of explanation, the case where the predictive control unit 14G controls the control unit 2B in the indoor unit 2 based on the learning model is illustrated, but the predictive control unit 14G is the main body of the indoor unit 2 based on the learning model. 2A may be controlled directly. Further, the predictive control unit 14G may transmit the control mode based on the learning model to the control unit 2B and control the main body 2A via the control unit 2B, which can be changed as appropriate. That is, the predictive control unit 14G may directly control the main body 2A or indirectly control the main body 2A using the control unit 2B, and can be changed as appropriate.

Next, the operation of the air conditioning system 1 of this embodiment will be described. FIG. 10 is a flowchart showing an example of the processing operation of the CPU 14 in the adapter 3 involved in the transmission processing. The acquisition unit 14A in the CPU 14 in the adapter 3 determines whether or not the current time is the acquisition timing of the operation history data (step S11). The acquisition timing is, for example, a 5-minute cycle. When the current time is the acquisition timing of the operation history data (step S11Yes), the acquisition unit 14A acquires the operation history data from the indoor unit 2 (step S12). Further, the acquisition unit 14A adds a time stamp to the operation history data (step S13).

The acquisition unit 14A stores the acquired operation history data and the time stamp given to the acquired operation history data in the operation history memory 13A (step S14). The determination unit 14B in the CPU 14 determines whether or not the operation history data stored in the operation history memory 13A is the operation history data for 48 hours (step S15). The determination unit 14B determines whether or not there is operation history data for 48 hours by counting the time stamps of the 5-minute cycle, for example.

When the operation history data stored in the determination unit 14B is the operation history data for 48 hours (step S15Yes), the determination unit 14B executes the thinning process shown in FIG. 11 to be described later (step S16). The erasing unit 14C in the CPU 14 stores the operation history data for 48 hours after the thinning process is executed and the time stamp given to the operation history data in the operation history memory 13A, and then stores the operation history data for 48 hours and the operation history data stored in the memory. The time stamp given to this is transmitted to the server device 5 (step S17), and the processing operation shown in FIG. 10 ends. After transmitting the operation history data for 48 hours and the time stamp assigned to the operation history data to the server device 5, the acquisition unit 14A does not overlap with the data after transmission, and the operation history for the next 48 hours is not duplicated. The data and the time stamp given to the data will be acquired in sequence.

If the acquisition timing is not the current operation history data acquisition timing (step S11No), the acquisition unit 14A returns to the process of step S11. If the operation history data stored in the acquisition unit 14A is not the operation history data for 48 hours (step S15No), the acquisition unit 14A returns to the process of step S11.

As described above, the adapter 3 acquires the operation history data from the indoor unit 2 every 5 minutes, executes the thinning process when the operation history data for 48 hours is stored, and 48 hours after the thinning process. The operation history data for the minute is transmitted to the server device 5. As a result, the server device 5 can acquire the operation history data every 48 hours and use them to generate or update the learning model. Moreover, since the adapter 3 transmits the operation history data after the thinning process to the server device 5, it is possible to reduce the communication traffic related to the transmission of the operation history data.

FIG. 11 is a flowchart showing an example of the processing operation of the CPU 14 in the adapter 3 involved in the thinning process. In FIG. 11, the determination unit 14B in the CPU 14 determines whether or not the operation history data for which the amount of change between the operation history data continuous in the order of the oldest time stamp is within a predetermined range is selected from the operation history data for 48 hours. Determine (step S21).

When the erasing unit 14C in the CPU 14 selects operation history data in which the amount of change is within a predetermined amount (step S21Yes), the erasing unit 14C operates the selected operation history data other than the operation history data having the oldest time stamp. The historical data is thinned out (step S22).

After thinning out the operation history data in step S22, the determination unit 14B determines whether or not there is unselected operation history data whose change amount is within a predetermined range (step S23). When there is unselected operation history data whose change amount is within a predetermined range (step S23Yes), the determination unit 14B processes in step S21 in order to select unselected operation history data whose change amount is within a predetermined range. Return to.

When there is no unselected operation history data whose amount of change is within a predetermined range (step S23No), the determination unit 14B ends the processing operation shown in FIG. When the determination unit 14B does not select the operation history data whose change amount is within a predetermined range (step S21No), the determination unit 14B ends the processing operation shown in FIG.

As described above, the adapter 3 does not transmit all the operation history data for 48 hours and the time stamp given to the operation history data to the server device 5, but among the continuous operation history data for 48 hours, the adapter 3 All of the plurality of operation history data whose amount of change is within a predetermined range are thinned out except for the oldest operation history data. As a result, it is possible to reduce the communication traffic when transmitting the operation history data between the adapter 3 and the server device 5 and the time stamp given to the operation history data.

FIG. 12 is a flowchart showing an example of the processing operation of the CPU 33 in the server device 5 involved in the update processing. The update process is a process on the server device 5 side that updates the learning model using the operation history data for 48 hours received from the adapter 3. Here, a case where the server device 5 updates the sensible temperature setting prediction model using the operation history data for three weeks will be illustrated. The receiving unit 33B in the CPU 33 in the server device 5 determines whether or not the operation history data for 48 hours and the time stamp given to the operation history data have been received from the adapter 3 (step S31). When the determination unit 33C in the CPU 33 receives the operation history data for 48 hours and the time stamp given to the operation history data (step S31Yes), determines whether or not the received operation history data has a thinning section (step S31Yes). Step S32).

When the restored unit 33D in the CPU 33 has a thinned-out section in the received 48-hour operation history data (step S32Yes), the thinned-out section in the 48-hour operation history data and the time stamp given to the thinned-out section. The operation history data and the time stamp given to the operation history data of the above are restored by the restoration process shown in FIG. 13 described later (step S33). The restoration unit 33D determines that the section without the time stamp is the thinned section based on the presence or absence of the time stamp.

After executing the restoration process, the restoration unit 33D stores the operation history data for 48 hours and the time stamp given to the operation history data in the data memory 32A (step S34). The model learning unit 33A in the CPU 33 stores the operation history data for 48 hours and the time stamp given to the operation history data, and then the stored operation history data and the time stamp given to the operation history data for 3 weeks or more. (Step S35). The model learning unit 33 determines, for example, whether or not the stored operation history data is for 3 weeks or more according to the number of counts of the time stamps having a 5-minute cycle. The model learning unit 33A is based on the operation history data and the time stamp given to the operation history data for 3 weeks or more (step S35Yes), based on the operation history data for 3 weeks or more and the time stamp given to the operation history data. A learning model such as a sensible temperature setting prediction model is generated or updated (step S36).

After generating or updating the learning model, the transmission unit 33E in the CPU 33 transmits the learning model to the adapter 3 (step S37), and ends the processing operation shown in FIG.

Further, when the receiving unit 33B does not receive the operation history data for 48 hours from the adapter 3 and the time stamp assigned to the operation history data (step S31No), the receiving unit 33B ends the processing operation shown in FIG. When there is no thinning section in the received 48 hours of operation history data and the time stamp given to the CPU 33 (step S32No), the CPU 33 obtains the 48 hours of operation history data and the time stamp given to the thinned out section. The process returns to step S34 in order to store the data.

Further, when the stored operation history data for 48 hours and the time stamp given thereto are not for 3 weeks or more (step S35No), the CPU 33 provides the operation history data for 48 hours from the adapter 3 and the operation history data thereof. The process returns to step S31 in order to determine whether or not the given time stamp has been received.

As described above, when the server device 5 that executes the update process has a thinned-out section in the 48-hour operation history data received from the adapter 3, the server device 5 restores the thinned-out section and operates for 48 hours after the restoration. The history data is stored in the data memory 32A. Further, when the operation history data for 48 hours after restoration in storage is for 3 weeks, the server device 5 generates or updates the sensible temperature setting prediction model, and transmits the generated or updated learning model to the adapter 3. .. As a result, the server device 5 can maintain the accuracy of the learning model to be generated or updated even if the operation history data thinned out from the adapter 3 is acquired.

For convenience of explanation, for example, the case where the sensible temperature setting prediction model is generated or updated when the operation history data for 48 hours is for 3 weeks is illustrated, but it is limited to the sensible temperature setting prediction model for 3 weeks. The learning model can be changed as appropriate according to the amount of storage of the operation history data.

FIG. 13 is a flowchart showing an example of the processing operation of the CPU 33 in the server device 5 involved in the restoration processing. In FIG. 13, the restoration unit 33D in the CPU 33 in the server device 5 is a thinned-out section as a section without a time stamp based on the operation history data for 48 hours and the presence or absence of the time stamp among the time stamps given to the operation history data. Is certified (step S41). After the thinning section is certified, the restoration unit 33D selects the operation history data immediately before the thinning section and the time stamp given to the time stamp based on the presence or absence of the time stamp (step S42). In addition, the restoration unit 33D refers to the presence / absence of the time stamp, and selects the operation history data with the time stamp immediately before the thinning section and the time stamp given to the operation history data. The restoration unit 33D restores the operation history data of the thinned section and the time stamp given to the operation history data of the thinned section based on the immediately preceding operation history data and the time stamp given thereto (step S43). For example, when the restoration unit 33D recognizes the period from 12:05 to 12:15 as the thinning section, the operation history data at 12:00 becomes the operation history data immediately before the thinning section. Therefore, the restoration unit 33D copies the operation history data at 12:00 as the operation history data at 12:05, 12:10, and 12:15 to obtain the operation history data of the thinned section and the operation history data thereof. Restore the given time stamp.

Further, the determination unit 33C restores the operation history data of the thinning section and the time stamp given to the operation history data, and then determines whether or not there is an uncertified thinning section (step S44). When there is an uncertified thinning section (step S44Yes), the CPU 33 returns to the process of step S41 in order to certify the uncertified thinning section in the order of time stamps. When there is no uncertified thinning section (step S44No), the determination unit 33C ends the processing operation shown in FIG.

As described above, the server device 5 that executes the restoration process recognizes the thinning section from the operation history data for 48 hours, and the thinning section from the operation history data immediately before the thinning section and the time stamp given to the thinning section. Restore the operation history data and the time stamp given to it. As a result, the server device 5 can generate or update a highly accurate learning model using the operation history data after thinning out.

The adapter 3 of the present embodiment acquires operation history data temporally from the indoor unit 2 at predetermined intervals, and when the acquired operation history data is for a predetermined period, the time of the acquired operation history data is included. It is determined whether or not the amount of change between the continuously continuous operation history data is within a predetermined range. When the amount of change between continuous operation history data is within a predetermined range, the adapter 3 thins out other operation history data by leaving the oldest operation history data among the continuous operation history data. Further, the adapter 3 transmits data for a predetermined period after thinning out to the server device 5. As a result, the adapter 3 can reduce the communication traffic from the adapter 3 to the server device 5 by thinning out the amount of operation history data.

The adapter 3 adds a time stamp for each acquisition timing to the operation history data acquired at each predetermined timing. As a result, the adapter 3 can manage the operation history data in chronological order with reference to the time stamp.

The adapter 3 retains the continuous operation history data in which the amount of change is within a predetermined range, the oldest operation history data among the time stamps given to the data, and the time stamp given to the data, and performs other operations. The historical data and the time stamp attached to it are thinned out. As a result, the adapter 3 can refer to the time stamp among the continuous operation history data in which the amount of change is within a predetermined range, and can thin out the oldest operation history data.

When the amount of change between continuous operation history data is within a predetermined range, the adapter 3 affects the human experience when the continuous operation history data controls the indoor unit 2 using the learning result of AI. Judge that the data is within the range that does not exist. As a result, the continuous operation history data is transmitted from the adapter 3 to the server device 5 in order to thin out the operation history data in a range that does not affect the human experience when the indoor unit 2 is controlled by using the learning result of AI. It is possible to reduce the communication traffic of the operation history data at the time of operation.

The adapter 3 determines whether or not the amount of change between continuous operation history data is zero within a predetermined range. As a result, the communication traffic of the operation history data when being transmitted from the adapter 3 to the server device 5 can be reduced.

The adapter 3 sets a predetermined range that differs for each season such as summer and winter, and determines whether or not the amount of change between continuous operation history data is within the predetermined range according to the current season. .. As a result, the server device 5 can generate or update a learning model according to the season by using a predetermined range of the amount of change according to the season.

The adapter 3 sets a predetermined range of a different amount of change for each use of the air conditioner for home or business use, for example, and the amount of change between time-continuous operation history data is the use of the air conditioner. It is determined whether or not it is within a predetermined range according to. As a result, the server device 5 can generate or update the learning model according to the application by using a predetermined range of the amount of change according to the application.

When the server device 5 receives the operation history data for a predetermined period from the adapter 3 and the time stamp assigned to the operation history data, the server device 5 thins out sections in the operation history data for the predetermined period based on the time stamp of each operation history data. Determine if there is. When it is determined that there is a thinning section, the server device 5 is assigned to the operation history data of the thinning section and the operation history data of the thinning section based on the operation history data immediately before the thinning section and the time stamp given to the operation history data. A time stamp is generated to restore the operation history data for a predetermined period and the time stamp given to the operation history data. The server device 5 updates the learning model based on the restored operation history data for a predetermined period and the time stamp given to the restored operation history data. As a result, the server device 5 can generate or update a highly accurate learning model using the operation history data for a predetermined period.

In this embodiment, the sensible temperature setting prediction model is illustrated as a learning model, but the present invention is not limited to this, and can be changed as appropriate. Further, since the driving history data required for learning differs depending on the learning model, the contents can be changed as appropriate. Further, since the acquisition cycle and the collection amount required for learning differ depending on the learning model, the predetermined cycle and the predetermined period can be changed as appropriate.

In this embodiment, when the amount of change between at least one data group among the plurality of continuous operation history data is not within a predetermined range, all the data groups including the data are transmitted to the server device 5. Was illustrated. However, even when the amount of change between at least one data in the plurality of data groups is not within the predetermined range, the data in the plurality of data groups whose amount of change is within the predetermined range are thinned out and the amount of change is predetermined. Only data that is not within the range of may be transmitted to the server device 5, and can be changed as appropriate.

Although the adapter 3 illustrates the case where the operation history data of the indoor unit 2 is transmitted to the server device 5 via the relay device 6, the operation history data is transmitted to the server device 5 as it is without passing through the relay device 6. It may be changed as appropriate.

Further, each component of each of the illustrated parts does not necessarily have to be physically configured as shown in the figure. That is, the specific form of distribution / integration of each part is not limited to the one shown in the figure, and all or part of them are functionally or physically distributed / integrated in arbitrary units according to various loads and usage conditions. Can be configured.

Further, various processing functions performed by each device are performed on a CPU (Central Processing Unit) (or a microcomputer such as an MPU (Micro Processing Unit) or an MCU (Micro Controller Unit)) in whole or in any part thereof. You may try to do it. Further, various processing functions may be executed in whole or in any part on a program analyzed and executed by a CPU (or a microcomputer such as an MPU or MCU) or on hardware by wired logic. Needless to say.

1 Air conditioning system 2 Indoor unit 2A Main unit 2B Control unit 3 Adapter 5 Server device 14A Acquisition unit 14B Judgment unit 14C Erase unit 14D Transmission unit 14E Reception unit 14F Setting unit 14G Prediction control unit 33A Model learning unit 33B Reception unit 33C Judgment unit 33D Restoration unit 33E Transmitter unit

Claims (5)

An air conditioner having a control unit that controls an air conditioner using a learning model and an adapter having a communication unit that communicates with a server device that generates the learning model based on the operation history data of the air conditioner. There,
The adapter
An acquisition unit that acquires the operation history data from the air conditioner at predetermined intervals, and
A predetermined range in which the amount of change between temporally continuous data in the operation history data acquired by the acquisition unit is set according to the degree of influence of the amount of change in the operation history data on the experience. A judgment unit that determines whether or not it is inside,
An erasing unit that deletes other data while leaving at least one of the continuous data when the amount of change between the continuous data is within a predetermined range.
It has a transmission unit that transmits the operation history data after the other data is deleted by the erasing unit to the server device.
An air conditioner characterized in that a time stamp is added to the operation history data acquired at each predetermined cycle. The air conditioner according to claim 1, wherein the erasing unit also deletes a time stamp added to the other data when the other data is deleted. The acquisition unit
With different characteristics, the air conditioner according to claim 1 or 2, wherein the obtaining at least two operation history data. The operation history data includes
The air conditioner according to any one of claims 1 to 3, wherein at least temperature and humidity are included. Air conditioning having an air conditioner, an adapter equipped with a control unit that controls the air conditioner using a learning model, and a server device that generates the learning model based on the operation history data of the air conditioner. It ’s a system,
The adapter
An acquisition unit that acquires the operation history data from the air conditioner at predetermined intervals, and
Among the operation history data acquired by the acquisition unit, a first determination unit for determining whether or not the amount of change between temporally continuous data is within a predetermined range, and
An erasing unit that deletes other data while leaving at least one of the continuous data when the amount of change between the continuous data is within a predetermined range.
It has a first transmission unit that transmits the operation history data after the deleted data is deleted by the erasing unit to the server device.
A time stamp is added to the operation history data acquired at each predetermined cycle, and a time stamp is added.
The server device
A second method of determining whether or not there is a deletion section in each of the received operation history data based on the time stamp added to each operation history data when the operation history data and the time stamp are received from the adapter. Judgment unit and
When it is determined that there is the deleted section, the operation history data and the time stamp of the deleted section are generated based on the operation history data and the time stamp immediately before the deleted section, and the received operation history data and the time stamp are used. Restoration part to restore and
A learning unit that generates the learning model based on the operation history data and the time stamp restored by the restoration unit, and
It has a second transmission unit that transmits the learning model generated by the learning unit to the adapter.
The adapter
An air conditioning system comprising a setting unit to which the learning model received from the server device is applied.

Images