JP2024091049A - Control device - Google Patents

Abstract

[Problem] To provide a control device that suppresses the hassle of moving to and operating a setting device and suppresses unintended setting adjustments. [Solution] The control device is characterized by including an acquisition unit 101 that acquires image information including an inhabitant 900, a first estimation unit 109 that estimates a thermal sensation by inputting the image information into a trained first learning model that has been subjected to machine learning to estimate a thermal sensation based on an action A caused by the thermal sensation of the inhabitant 900, a second estimation unit 107 that estimates a thermal sensation by inputting the image information into a trained second learning model that has been subjected to machine learning to estimate a thermal sensation based on an event B performed by the inhabitant 900 that indicates a thermal sensation, and a calculation unit that changes a second setting value that is used when determining a first setting value used to control an air conditioner 500 and is based on the thermal sensation when the thermal sensation is estimated by the first estimation unit 109 and the second estimation unit 107 and the thermal sensation is the same. [Selected Figure] Figure 1

Description

The present invention relates to a control device.

A known method for adjusting the settings of an air conditioner installed in an office is to use a setting device attached to a wall in a common area. By attaching the setting device to a wall in a common area, anyone in the building can adjust the settings.

In addition to the above-mentioned method of adjusting settings, there are also known methods of adjusting the settings of the air conditioner from one's own seat, which is separated from the setting device, by performing a predetermined action toward the camera (see, for example, Patent Document 1), and a method of using camera images to grasp the amount of clothing and activity of occupants, and automatically adjusting the settings of the air conditioner based on the amount of clothing and activity.

JP 2022-062302 A

When adjusting settings using a wall-mounted setting device as described above, it is cumbersome to go to the setting device and operate it, and when settings are adjusted by specific people in the building, adjustments tend to be made less frequently, resulting in unnecessary air conditioning costs.

As mentioned above, when adjusting the settings of an air conditioner using a camera and camera images, people in the building may unintentionally perform predetermined actions or make changes to their clothing, and these actions may be captured by the camera and camera images. In such cases, there is a problem in that settings may be adjusted in a way that is not intended by the people in the building.

The present invention has been made to solve the above-mentioned problems, and aims to provide a control device that reduces the hassle of traveling to and operating the setting device, and prevents unintended setting adjustments by people in the building.

In order to achieve the above object, the present invention provides the following means.
A control device according to one aspect of the present invention is a control device for controlling an air conditioning device, and is characterized in that it is provided with: an acquisition unit that acquires image information from a camera unit that captures images including occupants; a first estimation unit that estimates the thermal sensation of the occupants by inputting the image information acquired by the acquisition unit into a trained first learning model that has been subjected to machine learning to estimate the thermal sensation of the occupants based on actions caused by the thermal sensation of the occupants; a second estimation unit that estimates the thermal sensation of the occupants by inputting the image information acquired by the acquisition unit into a trained second learning model that has been subjected to machine learning to estimate the thermal sensation of the occupants based on events that indicate a thermal sensation performed by the occupants; and a calculation unit that changes a second setting value used when determining a first setting value used to control the air conditioning device when the thermal sensation estimated by the first estimation unit and the thermal sensation estimated by the second estimation unit are the same, the second setting value being based on the thermal sensation of the occupants.

According to a control device according to one aspect of the present invention, the first estimation unit and the second estimation unit estimate the thermal sensation of the occupants, respectively, and the second set value is changed when the thermal sensations estimated by the first estimation unit and the second estimation unit are the same. In other words, the second set value is not changed when the thermal sensation of the occupants is estimated by only one of the first estimation unit and the second estimation unit, or when the thermal sensations estimated by the first estimation unit and the second estimation unit are different.

When the image information acquired by the acquisition unit includes a motion caused by a thermal sensation of an occupant (hereinafter also referred to as a motion caused by a thermal sensation), the first estimation unit estimates the thermal sensation of the occupant using the first learning model.

In addition, if the image information acquired by the acquisition unit includes an event (hereinafter also referred to as an event) performed by an occupant that indicates a thermal sensation and that has been learned in the second learning model, the second estimation unit estimates the thermal sensation of the occupant using the second learning model.

The thermal sensation of the occupants is estimated using different learning models, and the second setting value is changed if the estimated thermal sensations are the same.
For example, if the occupants feel hot, the second setting value of the air conditioning device is changed so that the temperature of the first setting value is lower, and if the occupants feel cold, the second setting value is changed so that the temperature of the first setting value of the air conditioning device is higher.

Actions caused by thermal sensations are actions that occupants make in response to thermal sensations, and differ from events described below in that there is no intention to communicate a thermal sensation. Actions caused by thermal sensations include actions that alleviate a sense of heat or cold, and actions based on biological reactions when feeling hot or cold. Actions caused by feeling hot include changing from long sleeves to short sleeves, taking off clothes, wiping sweat from the forehead, fanning one's face with one's hand or a fan, touching one's head, and stretching one's joints.

Actions that may result from feeling cold include changing from short sleeves to long sleeves, putting on clothes, warming the hands with breath, shivering, and buttoning a shirt.

An event is an action consciously expressed by an inhabitant to communicate a thermal sensation, and is different from the actions caused by thermal sensations described above in that it is an action unrelated to actions that alleviate the feeling of heat or cold, and is also unrelated to actions based on biological reactions. Events include making a specific gesture linked to a thermal sensation, or displaying a specific object linked to a thermal sensation.

Specific gestures include raising the right hand if you feel hot and the left hand if you feel cold.
Specific objects tied to hot and cold sensations include displaying a red card with the word "hot" written on it if you feel hot, and a blue card with the word "cold" written on it if you feel cold.

In one aspect of the above invention, a storage unit is further provided that stores a time limit, which is a time interval of a predetermined length between the timing at which the first estimation unit estimates the thermal sensation and the timing at which the second estimation unit estimates the thermal sensation, and it is preferable that the calculation unit changes the second setting value when the interval between the timing at which the first estimation unit estimates the thermal sensation and the timing at which the second estimation unit estimates the thermal sensation is within the time limit.

By providing a memory unit that stores a time limit in this manner, the first estimation unit and the second estimation unit each estimate the thermal sensation of an occupant within the time limit, and if the thermal sensations estimated by the first estimation unit and the second estimation unit are the same, the second setting value is changed.

In other words, if the interval between the timing at which the first estimation unit estimates the thermal sensation and the timing at which the second estimation unit estimates the thermal sensation exceeds the time limit, the second setting value is not changed even if the thermal sensations estimated by the first estimation unit and the second estimation unit are the same.

In one aspect of the above invention, it is preferable that a time setting unit that changes the length of the time limit is further provided, and the storage unit stores the time limit changed by the time setting unit.

By providing a time setting unit in this way, the length of the time limit can be changed to any length. For example, the length of the time limit can be adjusted to a length that further reduces operational errors depending on the behavior of people in the building.

In one aspect of the above invention, the first learning model is machine-learned to estimate the thermal sensation of the occupant with a weighting coefficient based on the type of movement caused by the thermal sensation of the occupant, and it is preferable that the calculation unit further changes the second setting value based on the weighting coefficient assigned to the estimated thermal sensation of the occupant.

In this way, by performing machine learning on the first learning model to estimate the thermal sensation of occupants with a weighting coefficient based on the type of movement that causes the occupants' thermal sensation, it is possible to estimate the degree of thermal sensation for each type of movement that causes thermal sensation.

The degree of thermal sensation is related to the weighting factor. For example, the greater the value of the weighting factor, the greater the degree of thermal sensation. In other words, the greater the value of the weighting factor, the greater the degree of hotness felt, or the greater the degree of coldness felt.

For example, the action of wiping sweat from one's forehead is assigned the highest weighting coefficient. The action of changing from long sleeves to short sleeves by rolling up one's sleeves is assigned the second highest weighting coefficient. In this case, if the action of wiping sweat from one's forehead is performed, it is estimated that the occupant feels very hot. If the action of changing from long sleeves to short sleeves is performed by rolling up one's sleeves, it is estimated that the occupant feels somewhat hot.

The largest weighting coefficient is assigned to the movement of shivering. The second largest weighting coefficient is assigned to the movement of changing from short sleeves to long sleeves. In this case, if the movement of shivering is performed, it is estimated that the occupant feels very cold. If the movement of changing from short sleeves to long sleeves is performed, it is estimated that the occupant feels somewhat cold.

The control device of the present invention estimates the thermal sensation of people in the building from actions and events that cause thermal sensation, which has the effect of reducing the hassle of moving to and operating the setting device and making it easier to prevent unintended setting adjustments by people in the building.

FIG. 2 is a block diagram illustrating a configuration of a control device according to the first embodiment. 6 is a flowchart illustrating a process for changing settings of an air conditioner based on a setting value calculated by a target setting unit. 6 is a flowchart illustrating a process of changing settings of an air conditioner based on a setting value calculated by a calculation unit. FIG. 13 is a schematic diagram illustrating an operation caused by a thermal sensation. FIG. 13 is a schematic diagram illustrating an event in which a gesture is used to indicate that it is hot. FIG. 13 is a schematic diagram illustrating an example of an event for indicating that one feels cold by a gesture. FIG. 13 is a schematic diagram illustrating an event in which a card indicates heat. FIG. 13 is a schematic diagram illustrating an event indicating that it is cold using a card. FIG. 11 is a block diagram illustrating a configuration of a control device according to a second embodiment. 10 is a flowchart illustrating a process of a control device according to a second embodiment. FIG. 13 is a block diagram illustrating a configuration of a control device according to a third embodiment. 13 is a flowchart illustrating a process of a control device according to a third embodiment.

First Embodiment
A control device 100 according to a first embodiment of the present invention will be described. The control device 100 of this embodiment is a device for operating an air conditioner 500 installed in a building.

The air-conditioning device 500 is configured to take in indoor air of a building, cool it, and supply the cooled air to the indoors of the building. Specifically, it is configured to cool the air by circulating the refrigerant between the air and the refrigerant and performing heat exchange between the air and the refrigerant between the air-conditioning device 500 and an outdoor unit (not shown). The outdoor unit is configured so that the refrigerant removes heat from the air and releases the heat to the outside air.

The air conditioner 500 is provided with a heat exchanger (not shown) that cools the intake air by heat exchange, and an air conditioner fan (not shown) that blows out the heat-exchanged air. The heat exchanger and the air conditioner fan have known configurations.

The air conditioner 500 is preferably installed in an office. However, the air conditioner 500 may be installed in a building other than an office. For example, it may be installed in a house or a school classroom. The air conditioner 500 may also be installed in a means of transportation other than a building.

As shown in FIG. 1, the control device 100 is an information processing device such as a server or computer having a CPU (central processing unit), ROM, RAM, an input/output interface, etc. The control device 100 is connected to multiple air conditioning devices 500 so that information can be communicated. The information processing device is an information processing device such as a server or a personal computer (also referred to as a PC). The control device 100 may be connected to one or more air conditioning devices 500.

The control device 100 has a ROM or the like that stores programs that cause the CPU, ROM, RAM, and input/output interface to cooperate to function as at least an acquisition unit 101, a memory unit 102, a clothing amount calculation unit 103, a metabolic rate calculation unit 104, a first calculation unit 105, a selection unit 106, a second calculation unit 107, a goal setting unit 108, a first estimation unit 109, a second estimation unit 110, a calculation unit 111, a control unit 112, and an image determination unit 113.

The acquisition unit 101 is connected to the image capture device 300 (described later) so as to be able to communicate and acquire captured image information. The acquisition unit 101 has a function of acquiring environmental information and image information including images of occupants 900 captured by the image capture device 300.

The environmental information is information including the indoor temperature, humidity, air flow velocity, and radiant temperature of the office to be controlled. A thermometer and hygrometer may be used to obtain the indoor temperature and humidity.
An anemometer may be used to obtain the airflow speed, and a radiation thermometer may be used to measure the radiation temperature. Known techniques may be used to obtain each piece of environmental information.

It is preferable that the image information including the occupant 900 includes information for estimating the thermal sensation of the occupant 900. Specifically, it is preferable that the image information includes the amount of clothing worn by the occupant 900, the amount of activity of the occupant 900, surface temperature information of the occupant 900, action A resulting from the thermal sensation, and event B.

The amount of clothing is a parameter indicating the type and number of pieces of clothing worn by the occupant 900. For example, it indicates the sleeve length and number of jackets worn by the occupant 900.
The amount of activity is a parameter related to the intensity of the movement of the occupant 900, and is a parameter that quantifies various amounts of movement based on the amount of movement while sleeping. For example, sitting quietly is 1 MET, and walking normally on flat ground is 4 METs. The larger the MET number, the greater the amount of activity.

The surface temperature information is information about the temperature of the outer surface of the occupants 900. A known technique such as a thermal camera may be used to acquire the surface temperature information.
As shown in Figures 4 and 5, action A caused by a thermal sensation is an action performed by the occupant 900 in response to the thermal sensation, and is different from event B in that there is no intention to communicate the thermal sensation, as shown in Figures 6, 7, and 8. Action A caused by a thermal sensation includes actions that alleviate the hot or cold sensation felt by the occupant 900, and actions based on a biological reaction when the occupant feels hot or cold.

Specifically, actions that result from feeling hot include changing from long sleeves to short sleeves, taking off clothes, wiping sweat from the forehead, fanning the face with the hands or a fan, touching the head, and stretching joints.

Actions that may result from feeling cold include changing from short sleeves to long sleeves, putting on clothes, warming the hands with breath, shivering, and buttoning a shirt.

Event B is an action consciously expressed by occupant 900 to convey a thermal sensation, and is different from action A resulting from the thermal sensation described above in that it is an action unrelated to actions that alleviate the sense of heat or cold, and is also unrelated to actions based on biological reactions. Event B includes making a specific gesture linked to a thermal sensation, or displaying a specific object linked to a thermal sensation.

Specific gestures include raising the right hand if you feel hot and the left hand if you feel cold, and opening the palm if you feel hot and closing the palm if you feel cold.

Specific objects tied to hot and cold sensations include displaying a red card with the word "hot" written on it if you feel hot, and a blue card with the word "cold" if you feel cold.

The storage unit 102 is an information storage medium and has a function of storing information for estimating the thermal sensation of the occupant 900. The stored information preferably includes information acquired by the acquisition unit 101, information of a representative selected by a selection unit 106 described later, a PMV value calculated by a first calculation unit 105 described later, an average PMV value calculated by a second calculation unit 107 described later, the thermal sensation of the occupant 900 estimated by a first estimation unit 109 described later, the thermal sensation of the occupant estimated by a second estimation unit 110 described later, information registered in advance for a clothing amount calculation unit 103 described later to calculate the amount of clothing, and information for a metabolic rate calculation unit 104 described later to calculate the amount of activity. The storage unit 102 may be a flash memory such as an SD memory card, or may be a recording medium of another type.

The clothing amount calculation unit 103 has a function of calculating the amount of clothing worn by the occupants 900 based on image information of the amount of clothing worn by the occupants 900 acquired by the acquisition unit 101. If the image information includes multiple occupants 900, the clothing amount calculation unit 103 can calculate the amount of clothing worn by each occupant 900. The specific calculation contents of the clothing amount calculation unit 103 will be explained later.

The metabolic rate calculation unit 104 has a function of calculating the surface temperature of the occupants 900 based on the surface temperature information of the occupants 900 acquired by the acquisition unit 101, and calculating the activity level based on the surface temperature of the occupants 900. Note that if the image information includes multiple occupants 900, the metabolic rate calculation unit 104 can calculate the activity level for each of the occupants 900.

The selection unit 106 has a function of selecting a representative from among the occupants 900 included in the image information acquired by the acquisition unit 101. Specific selection contents by the selection unit 106 will be described later.
The first calculation unit 105 has a function of changing the second setting value, which is used when determining the first setting value used to control the air conditioner 500, based on the thermal sensation of the occupants 900. Note that in this embodiment, the first setting value will be described as a specific example of the setting value of the air conditioner 500, and the second setting value will be described as a specific example of the PMV value.

The image determination unit 113 has a function of determining whether the image information stored in the memory unit 102 includes an action A and an event B caused by a thermal sensation. As a specific determination process in the image determination unit 113, a publicly known determination process can be used.

The PMV value is a parameter calculated using a comfort equation with six variables: temperature, humidity, radiant temperature, airflow speed, amount of clothing worn, and activity level. The PMV value ranges from -3 (cold) to +3 (hot), and when the PMV value is 0, statistically about 95% of people will feel comfortable, and when the PMV value is +0.5 to -0.5, statistically about 90% of people will feel comfortable. It is preferable to control the air conditioning so that this PMV value falls within a specified range.

In other words, the first calculation unit 105 has a function of calculating the PMV value of the representative selected by the selection unit 106 using a comfort equation based on the indoor temperature, humidity, airflow speed, radiant temperature, amount of clothing worn by the representative, and surface temperature information of the representative acquired by the acquisition unit 101.

The second calculation unit 107 has a function of calculating an average PMV value of the multiple PMV values repeatedly calculated by the first calculation unit 105 at predetermined time intervals. The specific calculation contents in the selection unit 106 will be described later.

The target setting unit 108 has a function of transmitting a signal to change the setting value of the air conditioner 500 based on the average PMV value calculated by the second calculation unit 107. In other words, the target setting unit 108 can transmit a signal to change the setting value of the air conditioner 500 installed at a location away from the control device 100. The means for transmitting the signal may use known technology.

The calculation unit 111 has a function of transmitting a signal to the control unit 112 (described later) to change the setting of the air conditioner 500 changed by the target setting unit 108 when the thermal sensation of the occupants 900 estimated by the first estimation unit 109 (described later) and the second estimation unit 110 (described later) are the same. In other words, the calculation unit 111 can transmit a signal to change the setting value of the air conditioner 500 installed at a position away from the control device 100. A known technology may be used as the means for transmitting the signal.

The first estimation unit 109 has a function of estimating the thermal sensation of the occupant 900 by inputting image information including the action A caused by the thermal sensation acquired by the acquisition unit 101 into the first learning model.

The first learning model is a learning model that performs machine learning to estimate the thermal sensation of the occupant 900 when the occupant performs action A that causes a thermal sensation. A known learning method can be used for the machine learning.

The second estimation unit 110 has a function of estimating the thermal sensation of the occupants 900 by inputting image information including event B acquired by the acquisition unit 101 into the second learning model.
The second learning model is a learning model that performs machine learning to estimate the thermal sensation of the occupant 900 when the occupant 900 performs the event B. A known learning method can be used for the machine learning.

The control unit 112 receives a signal sent by the calculation unit 111 to change the settings of the air conditioner 500. The control unit 112 has a function of changing the settings of the air conditioner 500 based on the signal sent by the calculation unit 111. The means for receiving the signal may use known technology.

The image capturing device 300 is installed in a building and captures images of people 900 in the building. The image capturing device 300 has a camera unit 301 and a communication unit 302. The image capturing device 300 may be an existing camera with camera functions that is installed in the building, may be incorporated in the control device 100, or may be configured independently. In this embodiment, a case will be described in which the image capturing device 300 is configured independently of the control device 100.

The camera unit 301 has a function of photographing the inside of a building including occupants 900 and a function of photographing the surface temperature of the photographed occupants 900. The camera unit 301 can use a publicly known photographing method.

The communication unit 302 has a function of communicating image information captured by the camera unit 301 to the acquisition unit 101 of the control device 100. The means of communication may be a known technology. For example, it may be a wired communication connection or wireless communication.

Next, the operation of the control device 100 configured as described above will be described. First, the air conditioning by the air conditioner 500 in the office will be described, second, the operation of the image capture device 300 will be described, third, the control by the control device 100 will be described, and fourth, the learning of the learning model will be described.

First, the air conditioner 500 rotates the air conditioner fan to draw air from the office into the air conditioner 500. The air that is drawn in is air from which heat has been radiated by the occupants 900 and air that has been exhausted from electronic devices installed in the office.

The drawn-in air is cooled in the heat exchange section (not shown). Specifically, the temperature of the drawn-in air is reduced as heat is absorbed by the refrigerant circulating between the air and the outdoor unit (not shown). The refrigerant from which heat has been absorbed releases the heat to the outside air in the outdoor unit. The refrigerant that has released the heat again absorbs heat from the drawn-in air in the heat exchange section.

The air cooled in the heat exchanger is sent out into the office room by an air conditioner fan (not shown). The air sent out into the office room is conditioned based on the PMV value to provide comfort to occupants 900. The air sent out into the office room is again drawn into the air conditioner 500.

Next, the control in the control device 100 will be described with reference to Figures 2 and 3. First, the control in which the target setting unit 108 changes the settings of the air conditioner 500 will be described with reference to Figure 2.

When control in the control device 100 is started, the acquisition unit 101 acquires image information from the communication unit 302 of the image capture device 300 at any timing (S2). The acquired image information is image information of the inside of the office including occupants 900 captured by the camera unit 301 of the image capture device 300. The image information may include multiple occupants 900 or may include a single occupant 900. The captured image is transmitted to the acquisition unit 101 of the control device 100 by the communication unit 302.

The acquisition unit 101 also acquires environmental information in addition to acquiring image information. The timing of acquiring the environmental information may be the same as the timing of acquiring the image information, or may be a different timing.

The image information and environmental information acquired by the acquisition unit 101 are stored in the memory unit 102. The memory unit 102 also stores data linking clothing to the amount of clothing and data linking movement amount to activity amount. Based on the image information stored in the memory unit 102, the clothing amount calculation unit 103 calculates the amount of clothing of the occupants 900, and the metabolic rate calculation unit 104 calculates the activity amount of the occupants 900 (S3). The clothing amount and activity amount are calculated for all of the occupants 900 included in the image information.

Specifically, the clothing amount calculation unit 103 detects the clothing of the occupant 900 contained in the image information. When clothing is detected, the clothing amount calculation unit 103 obtains clothing data corresponding to the detected clothing from the clothing data stored in the memory unit 102. When the corresponding clothing data is obtained, the clothing amount calculation unit 103 calculates the clothing amount linked to the corresponding clothing as the clothing amount of the occupant 900.

Specifically, the metabolic rate calculation unit 104 detects the amount of exercise of the occupant 900 contained in the image information. When the amount of exercise is detected, the metabolic rate calculation unit 104 obtains data on the amount of exercise corresponding to the detected amount of exercise from the data on the amount of exercise stored in the memory unit 102. When the corresponding amount of exercise is obtained, the metabolic rate calculation unit 104 calculates the amount of activity linked to the corresponding amount of exercise as the amount of clothing worn by the occupant 900.

The selection unit 106 selects a representative from among the occupants 900 included in the image information stored in the storage unit 102 (S4). If the image information contains one occupant 900, that occupant 900 is assumed to be the representative. If the image information contains multiple occupants 900, the representative is selected as described below.

The selection unit 106 selects a representative based on the amount of clothing calculated by the clothing amount calculation unit 103. Specifically, the selection unit 106 divides the occupants 900 into groups based on the amount of clothing calculated, and selects one representative from the group to which the largest number of occupants 900 belong. A publicly known selection method can be used as a method for selecting one representative.

For example, if the image information includes three occupants 900 wearing long-sleeved clothing and one occupant 900 wearing short-sleeved clothing, a representative is selected from the majority group of three occupants 900 wearing long-sleeved clothing. The amount of clothing and the amount of activity of the occupants 900 selected by the selection unit 106 are stored in the storage unit 102 as the amount of clothing and the amount of activity of the representative (S5).

Once the representative's clothing amount and activity amount are stored, the first calculation unit 105 calculates the PMV value (S6). Specifically, the PMV value is calculated based on the representative's clothing amount, the representative's activity amount, image information, and environmental information stored in the storage unit 102. A publicly known method can be used as the calculation method for calculating the PMV value. The calculated PMV value is stored in the storage unit 102.

When the process of S6 ends, the process returns to S2 at every predetermined time and the above process is performed again (in the case of RE). If another predetermined time longer than the predetermined time has elapsed since the control of the control device 100 started and the first process of S6 ended (in the case of GO), the second calculation unit 107 calculates the average PMV value (S7). Also, if another predetermined time has elapsed since the most recent process of S6 that transitioned to the process of S7 ended (in the case of GO), the process transitions from S6 to S7.

In this embodiment, the predetermined time is 10 minutes, and the other predetermined time is 30 minutes. The predetermined time may be longer or shorter than 10 minutes. The other predetermined time may be longer or shorter than 30 minutes.

The second calculation unit 107 acquires multiple PMV values stored in the memory unit 102 during another predetermined time period, and calculates an average PMV value based on the multiple acquired PMV values. The average PMV value is the sum of the multiple PMV values divided by the number of the multiple PMV values. In other words, first, the PMV values calculated by the first calculation unit 105 for every 10 minutes are stored in the memory unit 102. The second calculation unit 107 calculates an average PMV value based on the three PMV values stored for every 30 minutes. Thereafter, every time the first calculation unit 105 calculates a PMV value for every 10 minutes, the second calculation unit 107 calculates an average PMV value for the most recent 30 minutes. The average PMV value is stored in the memory unit 102.

Based on the average PMV value stored in the memory unit 102, the control unit 112 changes the settings of the air conditioner 500 (S8). Specifically, if the occupants 900 feel hot, the settings of the air conditioner 500 are changed so that the calculated average PMV value becomes +0.5. For example, the set temperature of the air conditioner 500 is changed to a lower value. If the occupants 900 feel cold, the settings of the air conditioner 500 are changed so that the calculated average PMV value becomes -0.5. For example, the set temperature of the air conditioner 500 is changed to a higher value. After the settings of the air conditioner 500 have been changed, the process returns to S2 and the above-mentioned process is repeated.

Next, the control by the calculation unit 111 to change the settings of the air conditioner 500 will be described with reference to FIG. 3. Note that the process by the acquisition unit 101 to acquire image information and the process by the imaging device 300 are similar to the control by the target setting unit 108 shown in FIG. 2 to change the settings of the air conditioner 500, and therefore the description will be omitted.

When control in the control device 100 is started, the acquisition unit 101 acquires image information from the communication unit 302 of the imaging device 300 at any timing (S11). The image information acquired by the acquisition unit 101 is stored in the storage unit 102.

The image determination unit 113 performs a process of determining whether the image information stored in the storage unit 102 includes the action A that causes a thermal sensation (S12). If the action A that causes a thermal sensation is included (YES), the first estimation unit 109 estimates the thermal sensation of the occupant 900 (S
13).

Specifically, the first estimation unit 109 inputs the image information stored in the memory unit 102 in the processing of S11 into the first learning model, and performs processing to estimate the thermal sensation of the occupant 900. The first learning model is a trained model that has undergone machine learning to estimate the thermal sensation of the occupant 900 based on the action A that causes a thermal sensation. The machine learning for the first learning model will be described later.

If the action A causing a thermal sensation is not included (if NO), the process proceeds to step S20, which will be described later.
In the process of S12, the estimated thermal sensation of the occupant 900 (hereinafter also referred to as a first thermal sensation) is stored in the storage unit 102. After the first thermal sensation is stored in the storage unit 102, the acquisition unit 101 acquires image information (S14). The acquired image is stored in the storage unit 102.

The image determination unit 113 performs a process of determining whether event B is included in the image information stored in the memory unit 102 (S15). If event B is included in the image information (YES), the second estimation unit 110 estimates the thermal sensation of the occupant 900 (S16).

Specifically, the second estimation unit 110 inputs the image information stored in the memory unit 102 in the process of S14 into the second learning model, and performs a process of estimating the thermal sensation of the occupant 900. The second learning model is a trained model that has undergone machine learning to estimate the thermal sensation of the occupant 900 based on event B. The machine learning for the second learning model will be described later. Note that the case where event B is not included in the image information (cases NO1 and NO2) will be described later.

In the process of S16, the estimated thermal sensation of the occupant 900 (hereinafter also referred to as the second thermal sensation) is stored in the memory unit 102. If the first thermal sensation and the second thermal sensation are similar, the calculation unit 111 changes the PMV value based on the first thermal sensation and the second thermal sensation (S17). Similar thermal sensations mean that the first thermal sensation and the second thermal sensation are both hot sensations or cold sensations. Even if the degree of the hot sensation or the degree of the cold sensation differs, the thermal sensations are considered to be similar. A publicly known method can be used as the calculation method for changing the PMV value.

The PMV value changed by the calculation unit 111 is stored in the memory unit 102. Based on the PMV value stored in the memory unit 102, the control unit 112 changes the settings of the air conditioner 500 (S18). After that, the control device 100 returns to S11 and repeats the above-mentioned process.

When processing S15, if event B is not included in the image information, or if event B indicating a thermal sensation similar to the thermal sensation estimated from action A caused by a thermal sensation is not included in the image information (case NO1), the control device 100 returns to S14 and performs the above-mentioned processing. The process of returning from S15 to S14 is performed for a predetermined time.

The measurement of the predetermined time period starts from the point when an image is acquired in the process of S14, which is performed following the process of S13. In other words, the measurement of the predetermined time period does not start when the image is acquired in S14, which is performed after returning from S15.

If the process of returning from S15 to S14 is repeated for a predetermined time (if NO2), the control device 100 returns to S2 and repeats the above-mentioned process. Note that in this embodiment, an example in which the predetermined time is one minute will be described. Also, the predetermined time may be longer or shorter than one minute.

If the image information does not include action A caused by a thermal sensation during the process of S12 (NO), the acquisition unit 101 acquires the image information (S19). The image information acquired by the acquisition unit 101 is stored in the storage unit 102.

The image determination unit 113 performs a process of determining whether event B is included in the image information stored in the memory unit 102 (S20). If event B is included (YES), the second estimation unit 110 estimates the thermal sensation of the occupants 900 (S21).

The specific method by which the second estimation unit 110 estimates the second thermal sensation is the same as the specific method by which the second estimation unit 110 estimates the thermal sensation of the occupants 900 in the processing of S16, so a description thereof will be omitted.

If event B is not included (NO), the control device 100 returns to S11 and performs the above-mentioned processing.
In the process of S21, the estimated second thermal sensation is stored in the storage unit 102. After the second thermal sensation is stored in the storage unit 102, the acquisition unit 101 acquires image information (S22). The acquired image is stored in the storage unit 102.

The image determination unit 113 performs a process of determining whether the image information stored in the memory unit 102 includes action A that causes a thermal sensation (S23). If action A that causes a thermal sensation is included (YES), the first estimation unit 109 estimates the thermal sensation of the occupant 900 (S24).

The specific method by which the first estimation unit 109 estimates the first thermal sensation is the same as the specific method by which the first estimation unit 109 estimates the thermal sensation of the occupant 900 in the process of S13, so a description thereof will be omitted. In addition, the case where the action A resulting from the thermal sensation is not included in the image information (cases NO1 and NO2) will be described later.

In the process of S24, the estimated first thermal sensation is stored in the memory unit 102. If the first thermal sensation and the second thermal sensation are similar, the calculation unit 111 changes the PMV value based on the first thermal sensation and the second thermal sensation (S25).

The PMV value changed by the calculation unit 111 is stored in the memory unit 102. Based on the PMV value stored in the memory unit 102, the control unit 112 changes the settings of the air conditioner 500 (S26). After that, the control device 100 returns to S2 and repeats the above-mentioned process.

Next, the machine learning of the first learning model and the second learning model will be described.
In this embodiment, the present invention will be described by applying it to an example in which machine learning of the first learning model and the second learning model is performed in an information processing device different from the control device 100. The first learning model and the second learning model on which machine learning has been performed are stored in the storage unit 102 before being controlled by the control device 100.

Furthermore, after control by the control device 100, the first learning model and the second learning model on which machine learning has been further performed may be stored in the memory unit 102. In this case, the first learning model and the second learning model previously stored are replaced with the first learning model and the second learning model on which machine learning has been further performed.

The machine learning of the first learning model and the second learning model may be performed in different information processing devices as described above, or may be performed in the control device 100. When the machine learning is performed in the control device 100, a machine learning unit that performs machine learning is provided in the control device 100. Furthermore, machine learning for one of the first learning model and the second learning model may be performed in different information processing devices, and machine learning for the other may be performed in the control device 100.

In the machine learning of the first learning model, training data is used that links image information including action A that causes a thermal sensation to a thermal sensation such as hot or cold that corresponds to action A that causes the thermal sensation. In the machine learning of the second learning model, training data is used that links image information including event B to a thermal sensation such as hot or cold that corresponds to event B.

The specific machine learning in the first learning model and the second learning model can use well-known supervised learning, and the specific content of the calculation processing in supervised learning is not limited.

In addition, the method of creating the teacher data for the first learning model and the teacher data for the second learning model can be any known method, and there are no limitations on the specific creation method.

According to the control device 100 configured as described above, the first and second thermal sensations are estimated, and the second set value is changed if the first and second thermal sensations are the same. In other words, the second set value is not changed if only one of the first and second thermal sensations is estimated, or if the first and second thermal sensations are different.

The second setting value and the air conditioning device 500 are controlled based on the first thermal sensation and the second thermal sensation estimated by the first estimation unit 109 and the second estimation unit 110, so that it is possible to provide a control device 100 that reduces the hassle of traveling to and operating the setting device and also prevents unintended setting adjustments by the occupants 900.

Second Embodiment
Next, a second embodiment of the present invention will be described with reference to Fig. 9 and Fig. 10. The basic configuration of a control device 100A of this embodiment is similar to that of the first embodiment, but differs from the first embodiment in a storage unit 102A, a calculation unit 111A, a time determination unit 114A, and a time setting unit 115A. Therefore, in this embodiment, only the configurations different from the first and second embodiments will be described, and a description of the same configurations will be omitted.

As shown in FIG. 9, in the control device 100A of the second embodiment, the ROM etc. stores programs that cause the CPU, ROM, RAM, and input/output interface to cooperate to function as at least an acquisition unit 101, a memory unit 102A, a clothing amount calculation unit 103, a metabolic rate calculation unit 104, a first calculation unit 105, a selection unit 106, a second calculation unit 107, a goal setting unit 108, a first estimation unit 109, a second estimation unit 110, a calculation unit 111A, a control unit 112, an image determination unit 113, a time determination unit 114A, and a time setting unit 115A.

The basic information stored in the storage unit 102A includes the same information as that in the storage unit 102 of the first embodiment, as well as a time limit, which will be described later.
The time limit is the interval between the timing when the first estimation unit 109 estimates the first thermal sensation and the timing when the second estimation unit 110 estimates the second thermal sensation, and is a time limit that is a predetermined time interval of a specified length.

In other words, the time limit is a time interval of a predetermined length from the start of estimation of either the first or second thermal sensation to the start of estimation of the other thermal sensation. The time limit may be longer or shorter than one minute.

The time determination unit 114A has a function of determining whether the time interval from when estimation of either the first or second thermal sensation is started to when estimation of the other is started falls within the time limit stored in the memory unit 102A. A publicly known determination method can be used as a method for determining whether the time falls within the time limit.

The calculation unit 111A has a function of transmitting a signal to change the settings of the air conditioner 500 when the first thermal sensation and the second thermal sensation are the same, and when the interval between the timing at which the first estimation unit 109 estimates the first thermal sensation and the timing at which the second estimation unit 110 estimates the second thermal sensation is within a time limit.

The time setting unit 115A has a function to change the value of the time limit used in the control by the control device 100A. The changed time limit value may be input from outside the control device 100A, or may be a value selected from multiple different time limit values stored in the memory unit 102A. For example, the time setting unit 115A may have a function to store in the memory unit 102A a time limit value input by an operator who operates the control device 100A. Also, the time setting unit 115A may have a function to allow the operator to select a predetermined value from multiple time limit values stored in the memory unit 102A.

Next, the operation of the control device 100A having the above configuration will be described with reference to FIG. 10. Note that in this embodiment, only the processing that differs from the first and second embodiments will be described. The processing from S11 to S16 is the same as in the first embodiment, so the description will be omitted.

When the process of S16 is completed, the time determination unit 114A performs a determination process (S30). Specifically, the time determination unit 114A determines whether the interval between when the first estimation unit 109 estimates the first thermal sensation in the process of S13 and when the second estimation unit 110 estimates the second thermal sensation in the process of S16 is within the time limit. Specifically, if the first thermal sensation and the second thermal sensation are similar, it is determined whether they are within the time limit. The time limit is stored in the memory unit 102A before control is performed by the control device 100A.

In the processing of S30, if it is determined that the time interval from when the first estimation unit 109 estimates the first thermal sensation (S13) until the second estimation unit 110 estimates the second thermal sensation similar to the first thermal sensation (S16) is within the time limit (if YES), and if the first thermal sensation and the second thermal sensation are similar, the calculation unit 111A changes the PMV value based on the first thermal sensation and the second thermal sensation (S31).

The PMV value changed by the calculation unit 111A is stored in the memory unit 102A. Based on the PMV value stored in the memory unit 102A, the control unit 112 changes the settings of the air conditioner 500 (S32). After that, the control device 100A returns to S11 and performs the above-mentioned processing.

If the first thermal sensation and the second thermal sensation are not the same, or if the interval between when the first estimation unit 109 estimates the first thermal sensation in the process of S13 and when the second estimation unit 110 estimates the second thermal sensation in the process of S16 is not within the time limit (NO in S30), the control device 100A returns to S11 and performs the above-mentioned process.

If the image information does not include the action A caused by the thermal sensation during the process of S12 (NO), the acquisition unit 101 acquires the image information (S19). The processes from S19 to S24 are the same as those in the first embodiment, and therefore will not be described.

When the process of S24 is completed, the time determination unit 114A performs a determination process (S33). Specifically, the time determination unit 114A determines whether the interval between when the second estimation unit 110 estimates the second thermal sensation in the process of S20 and when the first estimation unit 109 estimates the first thermal sensation in the process of S23 is within the time limit. Specifically, when the first thermal sensation and the second thermal sensation are similar, it is determined whether the interval is within the time limit.

In the process of S33, if it is determined that the interval between when the second estimation unit 110 estimates the second thermal sensation (S20) and when the first estimation unit 109 estimates the first thermal sensation (S23) is within the time limit (if YES), and the first thermal sensation and the second thermal sensation are similar, the calculation unit 111A changes the PMV value based on the first thermal sensation and the second thermal sensation (S34).

The PMV value changed by the calculation unit 111A is stored in the storage unit 102A. Based on the PMV value stored in the storage unit 102A, the control unit 112 changes the settings of the air conditioner 500 (S34). After that, the control device 100A returns to S11 and performs the above-mentioned processing.

If the first thermal sensation and the second thermal sensation are not the same, or if the interval between when the second estimation unit 110 estimates the second thermal sensation in the process of S20 and when the first estimation unit 109 estimates the first thermal sensation in the process of S23 is not within the time limit (NO in S33), the control device 100A returns to S11 and performs the above-mentioned process.

Next, the operation of the time setting unit 115A configured in the control device 100A having the above configuration will be described.
The time setting unit 115A acquires a time limit value input by, for example, an operator. The acquired time limit value is stored in the memory unit 102A. The acquired time limit value may be stored in a format that overwrites an already stored time limit value, or may be stored in an additional format. If stored in an overwriting format, the time limit value stored in the memory unit 102A is used for control by the control device 100A. If stored in an additional format, the time limit value that was most recently stored is used for control by the control device 100A.

Alternatively, the time setting unit 115A receives an input to select a predetermined value from a plurality of different values of the time limit stored in the memory unit 102A. The memory unit 102A stores the selected time limit value in an identifiable format, and the selected time limit value is used for control by the control device 100A.

By providing a memory unit 102A that stores the time limit in this way, the first thermal sensation and the second thermal sensation are estimated within the time limit, and the second set value is changed if the first thermal sensation and the second thermal sensation are the same. Also, by providing a time setting unit 115A in this way, the length of the time limit can be changed to any length.

If both the first and second thermal sensations are included in the image information within a time limit set to an arbitrary length, the air conditioning device 500 is controlled, thereby providing a control device 100A that prevents unintentional adjustment of settings by the occupants 900.

Third Embodiment
Next, a third embodiment of the present invention will be described with reference to Fig. 11. The basic configuration of a control device 100B of this embodiment is similar to that of the first embodiment, but differs from the first embodiment in a storage unit 102B, a first estimation unit 109B, and a learning unit 116B. Therefore, in this embodiment, only the configurations different from the first and third embodiments will be described, and a description of the same configurations will be omitted.

The basic information stored in the memory unit 102B includes information similar to that in the memory unit 102 of the first embodiment, as well as a learning model for estimating the thermal sensation of the occupant 900 to which a weighting coefficient is assigned based on the type of action A resulting from the thermal sensation, which will be described later.

The learning unit 116B has a function of performing machine learning on the first learning model to estimate the thermal sensation (hereinafter also referred to as the third thermal sensation) of the occupant 900 based on the action A that causes the thermal sensation of the occupant 900, and a weighting coefficient (hereinafter also referred to as the weighting coefficient) based on the type of action A that causes the thermal sensation of the occupant 900. The specific content of the weighting coefficient will be explained later.

The learning unit 116B may be provided in the control device 100B, or in an information processing device other than the control device 100B.
The first estimation unit 109B has the function of estimating the third thermal sensation and weighting coefficient of the occupant 900 when the occupant 900 performs action A caused by a thermal sensation by inputting image information including action A caused by a thermal sensation acquired by the acquisition unit 101 into a learning model (hereinafter also referred to as the third learning model) that has undergone machine learning to estimate the third thermal sensation and weighting coefficient of the occupant 900 when the occupant performs action A caused by a thermal sensation.

Next, the operation of the control device 100B configured as described above will be described. Note that in this embodiment, the processing program of the control device 100B is the same as in the first embodiment, so the explanation will be omitted. Only the different operations will be described.

First, the machine learning of the third learning model will be described.
The machine learning of the third learning model is performed by the learning unit 116B, and the teacher data used in the machine learning is data linked to image information including an action A that causes the thermal sensation of the occupant 900, the thermal sensation such as hot or cold that corresponds to the action A, and a weighting coefficient assigned corresponding to the action A.

For example, for image information including the action of turning on an electric fan, the thermal sensation of "hot" is associated with the weighting coefficient with the largest value. For image information including the action of wiping sweat from one's forehead, the thermal sensation of "hot" is associated with the second largest weighting coefficient. For image information including the action of changing from long sleeves to short sleeves by rolling up one's sleeves, the thermal sensation of "hot" is associated with the third largest weighting coefficient.

In this case, the action of turning on the electric fan is evaluated as being very hot by the occupants 900, and is associated with the largest weighting coefficient.
Specifically, the action of wiping sweat from the forehead is evaluated as a behavior in which the occupant 900 feels hot, and the second largest weighting coefficient is associated with this action. The action of changing clothing from long sleeves to short sleeves by rolling up the sleeves is evaluated as a behavior in which the occupant 900 feels slightly hot, and the third largest weighting coefficient is associated with this action. The third learning model trained by the learning unit 116B is stored in the storage unit 102B.

The third learning model on which machine learning has been performed is stored in the memory unit 102B before control is performed by the control device 100B. Also, after control is performed by the control device 100B, the third learning model on which machine learning has been further performed may be stored in the memory unit 102B. In this case, the previously stored third learning model is replaced with the third learning model on which machine learning has been further performed.

Next, the operation of the control device 100B configured as described above will be described with reference to FIG. 12. Note that in this embodiment, only the processing that differs from the first and third embodiments will be described. The processing from S11 to S12 is the same as in the first embodiment, so the description will be omitted.

When the processing of S12 is completed, the first estimation unit 109B estimates the third thermal sensation and weighting coefficient of the occupant 900 (S40). Specifically, the first estimation unit 109B estimates the third thermal sensation, etc. of the occupant 900. Furthermore, the first estimation unit 109B inputs the image information stored in the memory unit 102B in the processing of S11 into a third learning model, and performs processing to estimate the third thermal sensation and weighting coefficient of the occupant 900.

Next, when the process of S40 is completed, the process of S14 is performed. The processes from S14 to S16 are the same as those in the first embodiment, and therefore the description thereof will be omitted.
When the process of S16 is completed, the calculation unit 111 performs a process of changing the PMV value (S41). Specifically, when the second thermal sensation and the third thermal sensation are similar, the calculation unit 111 performs a process of changing the PMV value based on the second thermal sensation, the third thermal sensation, and the value of the weighting coefficient. In this embodiment, the amount of change in the PMV value is also based on the magnitude of the estimated weighting coefficient.

Specifically, in the process of changing the PMV value by the calculation unit 111, the change range of the PMV value is the largest for action A caused by thermal sensation associated with the largest weighting coefficient. The change range of the PMV value is the second largest for action A caused by thermal sensation associated with the second largest weighting coefficient. The change range of the PMV value is smaller than in other cases for action A caused by thermal sensation associated with the third largest weighting coefficient. The change range of the PMV value may be proportional to the magnitude of the estimated weighting coefficient.

In this embodiment, the weighting coefficients are classified into three levels, but the weighting coefficients may be classified into more than three levels, or may be classified into two levels, which is less than three levels.
After completing the process of S41, the control device 100B performs the process of S 18. Note that the specific process of S 18 is the same as that of the first embodiment, and therefore a description thereof will be omitted.

Next, the process from S42 to S43 will be described. The process from S19 to S23 is the same as in the first embodiment, so the description will be omitted.
When the process of S23 is completed, the first estimation unit 109B estimates the third thermal sensation and weighting coefficient of the occupant 900 (S42). Note that the specific method by which the first estimation unit 109B estimates the third thermal sensation and weighting coefficient of the occupant 900 is the same as the specific method by which the first estimation unit 109B estimates the third thermal sensation and weighting coefficient of the occupant 900 in the process of S40, so a description thereof will be omitted.

The calculation unit 111 performs a process of changing the PMV value (S43). Note that the specific method by which the calculation unit 111 changes the PMV value is the same as the process by which the calculation unit 111 changes the PMV value in the process of S41, and therefore the explanation is omitted.

According to the control device 100B configured as described above, the air conditioner 500 is controlled by estimating the third thermal sensation of the occupant 900 and the value of the weighting coefficient. Therefore, the air conditioner 500 can be controlled according to the value of the weighting coefficient. In other words, the air conditioner 500 can be controlled according to the degree of thermal sensation felt by the occupant 900.

100, 100A, 100B... control device, 101... acquisition unit, 102, 102A, 102B... storage unit, 103... clothing amount calculation unit, 104... metabolic rate calculation unit, 105... first calculation unit, 10
6...selection unit, 107...second calculation unit, 108...target setting unit, 109, 109B...first estimation unit, 110...second estimation unit, 111, 111A...calculation unit, 112...control unit, 113...image determination unit, 114A...time determination unit, 115A...time setting unit, 116B...evaluation unit, 300...imaging device, 301...camera unit, 302...communication unit, 500...air conditioning device, 900...occupants, A...action caused by thermal sensation, B...event

Claims (4)

A control device for controlling an air conditioning device,
an acquisition unit that acquires image information from a camera unit that captures images including people in the building;
a first estimation unit that estimates the thermal sensation of the occupant by inputting the image information acquired by the acquisition unit into a trained first learning model that has been subjected to machine learning to estimate the thermal sensation of the occupant based on a movement caused by the thermal sensation of the occupant;
a second estimation unit that estimates the thermal sensation of the occupant by inputting the image information acquired by the acquisition unit into a trained second learning model that has been subjected to machine learning to estimate the thermal sensation of the occupant based on an event indicating a thermal sensation performed by the occupant;
a calculation unit that changes a second setting value used when calculating a first setting value used for controlling the air conditioner when the thermal sensation of the occupants is estimated by the first estimation unit and the second estimation unit, and the thermal sensation estimated by the first estimation unit and the thermal sensation estimated by the second estimation unit are the same, the second setting value being based on the thermal sensation of the occupants;
A control device comprising: a storage unit configured to store a time limit, which is an interval between a timing at which the first estimating unit estimates the thermal sensation and a timing at which the second estimating unit estimates the thermal sensation, and which is a time interval of a predetermined length,
The control device according to claim 1, characterized in that the calculation unit changes the second set value when an interval between a timing at which the first estimation unit estimates the thermal sensation and a timing at which the second estimation unit estimates the thermal sensation is within the time limit. A time setting unit is further provided for changing the length of the time limit,
3. The control device according to claim 2, wherein the storage unit stores the time limit changed by the time setting unit. The first learning model is machine-learned to estimate the thermal sensation of the occupant to which a weighting coefficient is assigned based on the type of movement caused by the thermal sensation of the occupant,
The control device according to claim 1 , wherein the calculation unit further changes the second set value based on the weighting coefficient assigned to the estimated thermal sensation of the occupants.