AIR-CONDITIONING CONTROLLER

Description

CROSS-REFERENCE TO RELATED APPLICATION

This application is based on and claims the benefits of priority from earlier Japanese Patent Application No. 2023-112038, filed Jul. 7, 2023, the descriptions of which are incorporated herein by reference.

BACKGROUND

1. Technical Field

The present disclosure relates to an air-conditioning controller.

2. Related Art

For example, air-conditioning equipment installed in a building such as a house may include an air-conditioning controller for a user to perform a variety of operations such as temperature setting and an air-conditioning apparatus that performs air conditioning based on instructions from the air-conditioning controller. According to a recently proposed technology, a temperature sensor is housed in an enclosure of an air-conditioning controller to measure an internal temperature of the enclosure and an air-conditioning control is to be performed based on measurement values of the temperature sensor.

Meanwhile, many electronic components mounted on a control board generate heat in operation. In a case where the temperature in the enclosure rises due to the heat from the electronic components, the measurement values of the temperature sensor becomes higher than an actual room temperature. That is to say, the measured internal temperature might not be the same as the room temperature. In consideration of such circumstances, according to a patent literature (JP2022150688A), a temperature sensor is added near a heat source (i.e., a control unit) to measure a temperature of the heat source and the room temperature is to be estimated in view of the temperature of the heat source.

Depending on a positional relationship between an air-conditioning controller and an outlet of an air-conditioning apparatus or an airflow direction, airflow volume, or the like of the air-conditioning apparatus, the airflow from the outlet is likely to blow on the air-conditioning controller. The airflow may lower a temperature of the air-conditioning controller (i.e., an internal temperature). For the air-conditioning controller described in the above-described patent literature, a temperature sensor for measuring an internal temperature and a temperature sensor for measuring a temperature of a heat source are used in combination. The room temperature is calculated simply by subtracting a value given by multiplying a difference between the temperature of the heat source and the internal temperature by an influence coefficient (i.e., a constant calculated by experiment) from the temperature of the heat source. The premise for the influence coefficient is that the temperature of the heat source, the internal temperature, and the actual room temperature are in a relationship of a predetermined temperature gradient and the temperature gradient is constant. It is described that the use of the influence coefficient makes it possible to reduce a deviation between the room temperature and the estimated temperature even in a case where the temperature of the air-conditioning controller is lowered as the airflow blows on the air-conditioning controller.

SUMMARY

However, a further improvement in an air-conditioning controller has been demanded.

An air-conditioning controller according to an embodiment of the present disclosure includes: a control board configured to control an air-conditioning apparatus; an enclosure housing the control board; and a plurality of temperature sensors. The control board includes at least one heat-generating component that is a heat source raising a temperature in the enclosure when powered. The plurality of temperature sensors are located at a plurality of respective positions in the enclosure. Heat from the at least one heat-generating component causes a difference in temperature between the plurality of positions. The control board stores a trained model for room temperature estimation generated by machine learning and is configured to estimate a temperature of a room installed with the air-conditioning controller using temperature data from each of the plurality of temperature sensors and the trained model.

BRIEF DESCRIPTION OF THE DRAWINGS

In the accompanying drawings:

FIG. 1 schematically illustrates a building and air-conditioning equipment in a first embodiment;

FIG. 2 is a perspective view of an air-conditioning controller;

FIG. 3 is a block diagram illustrating a configuration of the air-conditioning controller;

FIG. 4 schematically illustrates a type of an electronic component;

FIG. 5 is a graph illustrating a change in each temperature;

FIG. 6 schematically illustrates a configuration for generation of a trained model;

FIG. 7A schematically illustrates a change in temperature attributed to a heat source;

FIG. 7B schematically illustrates a change in temperature attributed to an airflow;

FIG. 8A is a front view of a control board;

FIG. 8B is a distribution chart illustrating temperature distribution;

FIG. 8C schematically illustrates a comparison between a first region and a second region of the control board;

FIG. 9 schematically illustrates a configuration for estimation of room temperature;

FIG. 10 is a graph illustrating an example of a differential between an actual room temperature and an estimated temperature;

FIG. 11A schematically illustrates a configuration for generation of a trained model in a second embodiment;

FIG. 11B schematically illustrates a configuration for estimation of room temperature;

FIG. 12 is a graph illustrating an example of a differential between an actual room temperature and an estimated temperature;

FIG. 13 schematically illustrates a configuration for estimation of room temperature according to a first modification example;

FIG. 14A is a front view of a first control board according to a second modification example;

FIG. 14B is a front view of a second control board according to the second modification example;

FIG. 15A is a front view of a first control board according to a third modification example;

FIG. 15B is a front view of a second control board according to the third modification example; and

FIG. 16 is a hardware configuration diagram of a microcomputer according to an embodiment of the present disclosure.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

First Embodiment

Description will be made below on a first embodiment of the present disclosure with reference to the drawings. An air-conditioning controller according to the present embodiment is provided in air-conditioning equipment installed in a building.

As illustrated in FIG. 1, air-conditioning equipment 20 installed in a building 10 includes an air conditioner 21a, an outdoor unit 21b, piping 24, air-inlet ducts 25 and 26, and delivery ducts 27 and 28. The air conditioner 21a is installed indoors (i.e., in an equipment space 11). The outdoor unit 21b is installed outdoors. The piping 24, which connects the air conditioner 21a and the outdoor unit 21b, is used for heat exchange. Air for air conditioning is to be taken from residential rooms 15 and 16 or the like of the building 10 through the air-inlet ducts 25 and 26. Air-conditioned air is to be delivered to the residential rooms 15 and 16 or the like through the delivery ducts 27 and 28. It should be noted that the building 10 according to the present embodiment is an airtight and super-insulated house. In the building 10, the air-conditioning equipment 20 performs air conditioning of the whole building (so-called central air conditioning).

The residential rooms 15 and 16 (for example, a ceiling portion) are provided with outlets 27a and 28a of the delivery ducts 27 and 28. Movable louvers are disposed on the outlets 27a and 28a. It is possible to change an airflow direction by changing attitudes of the louvers.

An air-conditioning controller 23 that controls an air-conditioning apparatus 21 (i.e., the air conditioner 21a and the outdoor unit 21b) is attached to a partition (i.e., wall surfaces 17 and 18) between the residential rooms 15 and 16. The air-conditioning controller 23 is communicably connected to the air-conditioning apparatus 21. The air-conditioning controller 23 receives operations for a variety of settings such as a setting of a running mode of the air-conditioning apparatus 21 (for example, a cooling mode, a heating mode, a dehumidification mode, and an air-blow mode), a setting of temperature, a setting of airflow volume, and a setting of an airflow direction. The air-conditioning controller 23 sends, in response to a change in such settings, an instruction regarding the changed setting to the air-conditioning apparatus 21. The air-conditioning apparatus 21 switches the running mode on the basis of the received instruction.

As illustrated in FIG. 2, the air-conditioning controller 23 includes a touch-panel display 40 and an enclosure 30 housing the display 40. The display 40 is exposed through an opening formed in a front portion 31 of the enclosure 30. It can also be said that the enclosure 30 and the display 40 form an exterior of the air-conditioning controller 23. The enclosure 30 is in a form of a flat rectangular parallelepiped. The enclosure 30 is sized so that in a case where a flow of air is generated in the vicinity of the air-conditioning controller 23, the air-conditioning controller 23 (excluding a portion of the enclosure 30 facing the wall surface 17 or 18) is assumed to be substantially evenly exposed to the air.

A control board 50 in an elongated rectangular shape is housed in the enclosure 30. The control board 50 faces a back surface of the display 40. Although described later in detail, the control board 50 has one surface (i.e., a surface facing the display 40) provided with a plurality of electronic components and wiring patterns connecting the electronic components to one another.

The enclosure 30 has a side portion 32 formed along a periphery of the control board 50. The side portion 32 is provided with a plurality of vents 34. Some of the electronic components mounted on the control board 50 generate heat when powered. As a temperature in the enclosure 30 rises due to the heat generation of the electronic components, heated air in the enclosure 30 is discharged through the vents 34 and air in the room enters the enclosure 30 through the vents 34.

Now, description will be made on a configuration for the air-conditioning controller 23 with reference to FIG. 3. The control board 50 of the air-conditioning controller 23 includes a power supply unit 51 that supplies an electric power from an external power supply to the above-described display 40 and the plurality of electronic components mounted on the control board 50. The power supply unit 51 includes a connector 52, a polyswitch 53, a rectify smoothing circuit (RSC) 54, a power supply (P/S) 55, and a low drop out (LDO) 56. The connector 52 connects the control board 50 and the external power supply. The RSC 54 converts the electric power supplied from the external power supply from an alternating-current power into a direct-current power. The polyswitch 53 is provided between the connector 52 and the RSC 54. The P/S 55 stably supplies the direct-current power converted by the RSC 54. The LDO 56 is a regulator that lowers a voltage of the electric power from the P/S 55 to a predetermined voltage. It should be noted that the display 40 and the plurality of electronic components require respective different electric powers for operation (i.e., operation powers). Accordingly, a plurality of LDOs 56 corresponding one-to-one to the respective operation powers are disposed.

The control board 50 also includes a microcomputer (MCU) 61, a watchdog (WD) 62, a graphic integrated circuit (IC) 63, a touch control IC 64, an audio output unit 65, a communication unit 66, and temperature sensors 67 and 68. The MCU 61 performs a main control for air conditioning. The WD 62 monitors whether the MCU 61 is in a normal operation. The graphic IC 63 controls a display on the display 40 on the basis of an instruction from the MCU 61. The touch control IC 64 detects a change in electric current (i.e., a change in capacitance) at a portion touched by a finger of a user who operates a touch panel of the display 40 and sends a detection signal to the MCU 61. The audio output unit 65 outputs an operation sound, a message sound, and the like. The communication unit 66 communicates with the air-conditioning apparatus 21 and the like. The temperature sensors 67 and 68 measure a temperature in the enclosure 30. It should be noted that the temperature sensors 67 and 68 may be thermistor temperature sensors. Moreover, the temperature sensors 67 and 68 may be any type of temperature sensors that are able to be housed in the enclosure 30. For example, the temperature sensors 67 and 68 may be resistance-thermometer temperature sensors or linear-resistor temperature sensors. It should be noted that the MCU 61 may include a processor 611 such as a central processing unit (CPU), a system bus 613, an input/output interface 615, and a memory 617 such as a read only memory (ROM) or a random-access memory (RAM) as illustrated in FIG. 16. The processing of the above-described MCU 61 may be implemented by the processor 611 executing a program stored in the memory 617. The memory 617 may also store a later-described trained model.

An electronic component mounted on the control board 50 is categorized as first-type electronic components or second-type electronic components. The first-type electronic component refers to a component that becomes a heat source raising the temperature in the enclosure 30 when powered. The second-type electronic component refers to a component that generates only a so small amount of heat as to be ignorable when powered and thus does not become a heat source substantially raising the temperature in the enclosure 30. In other words, in a case where the quantity of heat at a sufficient level to raise the temperature in the enclosure 30 is defined as a reference amount of heat generation, an electronic component with a larger amount of heat generation than the reference amount of heat generation is the first-type electronic component and an electronic component with a smaller amount of heat generation than the reference amount of heat generation is the second-type electronic component. It should be noted that the first-type electronic component corresponds to a heat-generating component in the present embodiment.

As illustrated in FIG. 4, the temperature of the P/S 55, the LDO 56, and the MCU 61 after generating heat are high and the amounts of heat generation of these components considerably exceed the reference amount of heat generation for a rise in temperature. Moreover, although the temperature of the graphic IC 63 after generating heat is lower than the MCU 61 and the like, the amount of heat generation of the graphic IC 63 exceeds the reference amount of heat generation. The P/S 55, the LDO 56, the MCU 61, and the graphic IC 63 are considered as first-type electronic components.

Contrary to the above, the connector 52, the temperature of the polyswitch 53, the RSC 54, the WD 62, the touch control IC 64, the audio output unit 65, the communication unit 66, and the temperature sensors 67 and 68 after generating heat are low and the amounts of heat generation of these components fall below the reference amount of heat generation. The connector 52, the polyswitch 53, the RSC 54, the WD 62, the touch control IC 64, the audio output unit 65, the communication unit 66, and the temperature sensors 67 and 68 are considered as second-type electronic components. It should be noted that the control board 50 also include other connector, humidity sensor, memory, and the like in addition to the above-described electronic components. These electronic components are, however, also considered as second-type electronic components.

The air-conditioning controller 23 according to the present embodiment measures the temperature in the enclosure 30 (hereinafter, also referred to as internal temperature) using the temperature sensors 67 and 68 built into the enclosure 30. The air-conditioning controller 23 estimates the room temperature from the measured internal temperature and performs air-conditioning control on the basis of the estimated temperature. As described above, a first-type electronic component built into the enclosure 30 along with the temperature sensors 67 and 68 is a heat source that raises the internal temperature of the enclosure 30. Specifically, when the first-type electronic component generates heat, the heat is transferred from the first-type electronic component to the control board 50, causing not only the temperature of the first-type electronic component but also the temperature of surroundings thereof to rise. That is to say, the temperature sensors 67 and 68 mounted on the control board 50 are also influenced to no small extent by the heat source. Measured temperatures by the temperature sensors 67 and 68 become higher than a actual room temperature.

Here, in addition to the heat generation in the enclosure 30, an airflow from the air-conditioning device 21 is also another factor affecting the internal temperature. Specifically, when the airflow from the air-conditioning apparatus 21 blows on the air-conditioning controller 23, the enclosure 30 and the display 40 are cooled and the internal temperature of the enclosure 30 may decrease. In the following description, a state where the airflow from the air-conditioning apparatus 21 blows on the air-conditioning controller 23 and a state where no airflow directly blows on the air-conditioning controller 23 but an airflow causing a decrease in the internal temperature is generated around the air-conditioning controller 23 are referred to as an air-blowing state. Moreover, a state where although an airflow blows from the air-conditioning apparatus 21, the airflow does not blow onto the air-conditioning controller 23 and no airflow causing a decrease in the internal temperature is generated around the air-conditioning controller 23 and a state where blowing is temporarily stopped as the temperature is determined to be suitable are referred to as a non-airflow state.

For example, in a case where a difference between a measured temperature and the room temperature is maintained substantially at a constant level in in the non-airflow state as illustrated in FIG. 5, it is possible to simply calculate an estimated temperature close to an actual room temperature by correcting the measured temperature using a predetermined correction value. Here, when the non-airflow state is switched to the air-blowing state, heat is removed from the air-conditioning controller 23 and the internal temperature decreases. This also causes the measured temperature by the temperature sensor to decrease. In contrast, the actual room temperature does not considerably change as the internal temperature. Thus, in a case where the measured temperature is corrected as in the non-airflow state, a corrected temperature (i.e., an estimated temperature) is deviated from the actual room temperature by an amount corresponding to the decrease in the internal temperature. That is to say, the measured temperature is excessively corrected. Moreover, the airflow from the air-conditioning apparatus 21 has a variety of influences depending on a positional relationship between the air-conditioning controller 23 and an outlet (for example, the outlets 27a and 28a) of the air-conditioning apparatus 21, a running status of the air-conditioning apparatus 21 (for example, the airflow direction, the airflow volume, and the set temperature), or the like. Consequently, it is difficult to deal with the influence of the heat source and the influence of the airflow by a simple correction as described above.

A technology that may address the influence of the heat source and the influence of the airflow has recently been proposed as described in the above-described patent literature. However, there is still a room for improvement in a configuration for estimation of room temperature in order to improve an estimation accuracy under a variety of conditions. In other words, in the technology according to the above-described patent literature, an influence coefficient is calculated in advance, which makes it possible to enhance an estimation accuracy of an estimation processing by a simple processing. However, a further improvement in the estimation accuracy might not be easy due to the nature. In contrast, as for a configuration in which the room temperature is to be estimated using a temperature sensor built into an air-conditioning controller, there is a room for improvement in the estimation accuracy of the room temperature even under a situation with the influence of the heat source and the influence of the airflow. By virtue of an improvement in the estimation accuracy, the properness of an air-conditioning control based on an estimated temperature may be enhanced. Accordingly, it is technically important to improve the estimation accuracy. Moreover, the technology according to the above-described patent literature requires the derivation of the influence coefficient by experiment. However, the derivation of the influence coefficient might not be easy. In particular, the heat source and the airflow, that is, factors causing a difference between the room temperature and a measured temperature, may affect in various manners depending on how heat is generated, how the airflow blows, or the like. Parameters to be considered to derive the influence coefficient are thus increased in number and complicated. Accordingly, machine learning is used, which makes it possible to improve the estimation accuracy of the room temperature without deriving the influence coefficient by experiment. One of the features of the present embodiment is to estimate the room temperature from the internal temperature of the enclosure 30 by using machine learning. Description will be made below on a distinguishing configuration of the present embodiment. It should be noted that the machine learning in the present embodiment is to be performed in a laboratory in a factory. A machine-learning model (hereinafter, referred to as trained model) for room temperature estimation generated using the machine learning is to be stored in the MCU 61. A product equipped with the MCU 61 is then to be shipped.

The MCU 61 includes a model generator 61b that generates a trained model (see, for example, FIG. 3). An algorithm of machine learning to be performed by the model generator 61b may be, for example, supervised learning. Supervised learning is an approach in which a model for estimating a result relative to a new condition is to be generated by recognizing, from a data set of conditions (i.e., training data) and respective corresponding results (i.e., label data), a feature implying a correlation between a condition and a result. For supervised learning, a neural network is used to build a model.

As illustrated in FIG. 6, the model generator 61b according to the present embodiment receives input of a measured temperature (i.e., training data) by the temperature sensor 67, a measured temperature (i.e., training data) by the temperature sensor 68, and a measured temperature (i.e., label data) by a room temperature sensor 100 that measures the room temperature of a laboratory having the air-conditioning controller 23 installed therein. A multi-layer model compliant with a neural network performs computing using the inputted data to update a parameter of a node of each layer. A trained model that enables a proper estimated temperature to be outputted as output data is generated in this manner.

In generating the trained model, an air conditioner installed in the laboratory is used to perform training in the above-described non-airflow state and training in the above-described air-blowing state. In particular, for the training in the air-blowing state, the influence of the airflow is changed in a variety of manners by changing respective control parameters such as the airflow direction and airflow volume of the air conditioner. This produces a simulated environment close to a state where the air conditioner is installed in a house or the like (i.e., an environment close to actual use).

To properly generate the trained model, data correlated with two features in temperature data such as the influence of the heat source and the influence of the airflow is inputted. Here, description will be made on the features with reference to FIGS. 7A and 7B. FIG. 7A illustrates how the heat is to be transferred and FIG. 7B illustrates how the heat is to be released.

In FIG. 7A, a distance from a heat source to a position P1 on the control board is different from a distance from the heat source to a position P2. The heat of the heat source is transferred to the position P1 and the position P2 through the control board in different manners of heat transfer depending on the distances from the heat source. A rise in the internal temperature of the enclosure 30 due to the influence of the heat source may cause a difference between a temperature at the position P1 and a temperature at the position P2. It can be said that the difference shows the influence of the heat source. In a case where the influence of the heat source is considered as a feature, a large difference in temperature is favorable for machine learning. Moreover, a range of a drop in temperature by cooling increases with an increase in temperature due to the influence of the heat source. That is to say, in a case where the influence of the airflow is considered as a feature, a large difference in temperature between the two positions is also favorable.

Moreover, the temperature rapidly rises at the position P1 due to the influence of the heat source, whereas the temperature moderately rises at the position P2 as compared with at the position P1. That is to say, not only a difference in peak of temperature but also a difference in rate of temperature rise arises between the position P1 and the position P2.

Contrary to the above, in a case where there is no difference in temperature between portions of the enclosure 30, the temperature of the enclosure 30 decreases with a certain uniformity when the airflow blows on the enclosure 30 (i.e., the air-conditioning controller 23) as illustrated in FIG. 7B. The heat of the control board is released through air and a contact portion (not illustrated) with the enclosure 30. However, a large difference in rate of heat release is unlikely to arise between the position P1 and the position P2. In other words, a difference between the temperature at the position P1 and the temperature at the position P2 is unlikely to change. In contrast, in a case where there is a difference in temperature between portions of the enclosure 30, the temperature of the enclosure 30 does not evenly decrease. For example, a range of a drop of the temperature at the position P1 relatively close to the heat source (i.e., relatively high in temperature) is larger than that of the temperature at the position P2 relatively distant from the heat source (i.e., relatively low in temperature).

For the above reasons, a relationship between the temperature at the position P1 and the temperature at the position P2 is different between a case where the temperature rises due to the influence of the heat source and a case where the temperature drops due to the influence of the airflow. In other words, even though the temperature at the position P1 is the same, a difference is likely to arise between the temperature at the position P2 when the temperature rises due to the influence of the heat source (for example, in the non-airflow state) and the temperature at the position P2 when the temperature drops due to the influence of the airflow (i.e., in the air-blowing state).

It should be noted that a case where the position P1 is close to the position P2 is assumed. In this case, a behavior of the rising temperature is similar to a behavior of the dropping temperature. This means that a difference in temperature between the position P1 and the position P2 is small. Further, a difference in temperature between the position P1 and the position P2 may be unclear due to noise, measurement error, and the like. Here, the room temperature itself may be inconstant and vary up and down. This may make it difficult to distinguish between a variation in the measured temperature due to a variation in the room temperature and a variation in the measured temperature due to the influence of the heat source and the influence of the airflow.

In view of the above circumstances, measured temperatures by the two temperature sensors 67 and 68 built into the air-conditioning controller 23 are used as inputted data in the present embodiment. Moreover, the temperature sensor 67 and the temperature sensor 68 are located in different regions so that a difference in temperature is increased under the influence of the heat source. Description will be made below on the locations of the temperature sensors 67 and 68 with reference to FIGS. 8A, 8B, and 8C. It should be noted that FIG. 8A is a plan view of the control board 50. It should be noted that the electronic components and wiring patterns mounted on the control board 50 are omitted for convenience.

The above-described first-type electronic components, the P/S 55, the LDO 56, the MCU 61, and the graphic IC 63, are collectively located at a specific region (hereinafter, referred to as first region 91) in the control board 50. Since the heat sources, i.e., first-type electronic components, are gathered, the first region 91 is higher in temperature than a region (hereinafter, referred to as second region 92) around the first region 91 in the control board 50. In the first region 91, the second-type electronic components are also located and wiring patterns are densely arranged as compared with in the second region 92. The wiring patterns are likely to transfer heat from the heat sources and the heat is to be released from the wiring patterns. That is to say, the densely routed wiring patterns are also a factor that increases the temperature of the first region 91.

All the first-type electronic components are not always maintained in a powered state. An operation of a part of the first-type electronic components may stop depending on the situation. Since the first-type electronic components are gathered in the first region 91, a range of variation in the temperature of the first region 91 is small even when the operation of a part of the first-type electronic components stops.

The temperature sensor 67 is located in the first region 91. Superficially, the temperature sensor 67 is located at a slight distance from the first-type electronic components. However, since the heat sources, i.e., first-type electronic components, are gathered in the first region 91, a temperature of a measurement position for the temperature sensor 67 is sufficiently high. To properly generate a trained model, it is favorable that a difference between respective measured temperatures by the temperature sensors 67 and 68 be large. However, it is not desirable to force the temperature sensor 67 to be extremely close to the first-type electronic components so as to increase a difference in temperature. The first region 91 does not require such a location of the temperature sensor 67. This is also favorable for coexistence of the temperature sensor 67 and other electronic components on the control board 50.

The heat from the heat sources is transferred mainly through the control board 50 or air in the enclosure 30. An influence of the heat sources transferred through the control board 50 is larger than an influence of the heat sources transferred through the air in the enclosure 30. Here, description is given of the location of the temperature sensor 67. A position of the temperature sensor 67 is in a range where heat from the P/S 55, the LDO 56, the MCU 61, and the graphic IC 63 is to be transferred through the control board 50 (including the wiring patterns). A temperature at the position of the temperature sensor 67 is thus likely to rise considerably due to the influence of the heat sources. In contrast, the temperature sensor 68 is located at a position where less heat is transferred through the control board 50 than in the first region 91 where the temperature sensor 67 is located. However, since the heat is transferred through air, a temperature at the position of the temperature sensor 68 is higher than at least the room temperature.

The second region 92 is broader than the first region 91. The second-type electronic components are mainly located in the second region 92. Since the second region 92 is continuous with the first region 91, the heat generated in the first region 91 is transferred to the second region 92 thorough the control board 50. However, since the second region 92 is at a larger distance from the heat sources than the first region 91, a temperature of the second region 92 is basically lower than that of the first region 91. A difference in temperature between the first region 91 and the second region 92 increases with an increase in distance from the first region 91. It should be noted that electronic components with a low heat resistance are located in the second region 92, so that an influence of the heat on the electronic component is reduced.

A linear slit 75 is formed in the control board 50. The slit 75 surrounds a part (specifically, a bottom-right corner 72) of the second region 92. A transfer path of the heat is partly disrupted by the formation of the slit 75. This reduces transfer of the heat (i.e., the influence of the heat sources) to the corner 72. As a result, a temperature of the corner 72 surrounded by the slit 75 is lower than that of any other portion of the second region 92. It should be noted that the above-described vents 34 are provided near the corner 72.

The temperature sensor 68 is located at the corner 72. Description will be made on a relationship among the temperature sensor 68, the first-type electronic components, and the slit 75. The slit 75 intersects an imaginary straight-line FL connecting the temperature sensor 68 to each of the first-type electronic components in a front view of the control board 50. This is favorable for efficiently reducing the influence of the heat sources.

The temperature sensor 67 is located in the first region 91 and the temperature sensor 68 is located in the second region 92 (in particular, at the corner 72), which causes a measured temperature by the temperature sensor 67 to be considerably different from a measured temperature by the temperature sensor 68 under the situation where being affected by the heat sources. This is beneficial in properly generating a trained model (i.e., in improving the estimation accuracy of the room temperature).

As illustrated in FIG. 9, a trained model 81 generated using inputted data from the temperature sensors 67 and 68 is stored in the MCU 61 (i.e., an estimation unit 61a). The estimation unit 61a inputs the respective measured temperatures by the temperature sensors 67 and 68 to the trained model 81 and obtains an estimated temperature as output data from the trained model 81. The MCU 61 performs the air-conditioning control on the basis of the estimated temperature.

FIG. 10 shows an example of an estimation result of the room temperature using the trained model 81. According to the result shown in FIG. 10, a differential between the estimated temperature and the actual room temperature exceeds a range of ±2° F. at the timing of switching from the non-airflow state to the air-blowing state. However, the differential is within the range of ±2° F. at any other time point. Further, the differential falls within a targeted range of ±1° F. during the majority of an estimation period in both the non-airflow state and the air-blowing state. A substantially similar results were obtained even when conditions such as a position of the air-conditioning controller 23 was changed. As seen from the above, it is possible to ensure a high estimation accuracy under a variety of conditions. This contributes to enhancement of the suitability of the air-conditioning control.

In a case where the heat sources generate heat, differences in measured temperature and the amount of change of the measured temperature may arise between the temperature sensor 67 and the temperature sensor 68 built into the air-conditioning controller 23 due to a difference between the influences of the heat sources received by the temperature sensor 67 and the temperature sensor 68. Moreover, in a case where a difference in temperature arises between the temperature sensors 67 and 68 depending on the respective positions thereof and the temperature in the enclosure 30 decreases due to the airflow from the air-conditioning apparatus 21 blowing on the air-conditioning controller 23, differences in measured temperature and the amount of change of the measured temperature may also arise. That is to say, the influence of the heat sources and the influence of the airflow are shown in the measured temperatures by the temperature sensors 67 and 68. Specifically, the influence of the heat sources and the influence of the airflow are shown as a difference between the measured temperatures by the temperature sensors 67 and 68 and a change in the difference. In contrast, a variation in the room temperature causes neither a difference in measured temperature between the temperature sensors 67 and 68 nor a change in the difference. Here, since there are a variety of possible manners in which the airflow blows on the air-conditioning controller 23, the temperatures of the heat sources are not always stable. That is to say, the influence of the heat sources and the influence of the airflow may change depending on a variety of factors in an environment where the air-conditioning controller 23 is actually used. However, the features of the influence of the heat sources and the influence of the airflow on the measured temperatures are shown as a difference between the measured temperatures and a change in the difference as described above. Accordingly, in the present embodiment, the trained model 81 is generated on the basis of the training data indicating the measured temperatures by the temperature sensors 67 and 68 and the label data (i.e., Ground Truth) indicating the room temperature. Then, temperature data from each of the temperature sensors 67 and 68 is inputted to the trained model 81 for room temperature estimation and the room temperature is estimated. This makes it possible to accurately estimate the room temperature even though the heat sources and the airflow affect the measured temperatures.

For the generation of the trained model 81, it is favorable that the influence of the heat sources and the influence of the airflow be considerably different between positions where the temperature sensors 67 and 68 are located. Accordingly, in a case where a plurality of heat-generating components are mounted, the temperature sensors 67 and 68 are located in such a way that a distance from each of the plurality of heat-generating components to the temperature sensor 67 is smaller than a distance from each of the plurality of heat-generating components to the temperature sensor 68. This makes it possible to suitably improve the estimation accuracy of the trained model 81 generated using the training data according to the temperature sensors 67 and 68.

In the air-conditioning controller 23, components included in the power supply unit 51 and the MCU 61 that implements the functions of a control unit may generate a large amount of heat. Accordingly, it is possible to implement a favorable configuration for practical use by locating the temperature sensors 67 and 68 with reference to the positions of these heat-generating components.

In addition to the heat-generating components (i.e., the first-type electronic components), other electronic components (i.e., the second-type electronic components) are also mounted on the control board 50. Further, the wiring patterns connecting these components is also formed. Consequently, in a case where the temperature sensor 67 is located near a specific heat-generating component, the temperature sensor 67 may be difficult to coexist with the other components and the wiring patterns. It is possible to increase the temperature of the first region 91 by gathering the plurality of heat-generating components in a specific region (i.e., the first region 91) as described in the present embodiment. This makes it possible to locate the temperature sensor 67 within the first region 91 as desired. This means that flexibility in locating the temperature sensor 67 is increased. As a result, it is possible to reduce an increase in restriction on the locations of the other components and the wiring patterns.

In the small-sized control board 50 as described in the present embodiment, it might be difficult to ensure a difference in temperature between the temperature sensors 67 and 68, since the distance from the heat sources to the temperature sensor 68 on a low-temperature side is short. Accordingly, the slit 75 may be formed as described in the present embodiment. The slit 75 reduces transfer of the heat to the temperature sensor 68. This makes it possible to ensure a difference in temperature between the temperature sensors 67 and 68 even though it is difficult to ensure a sufficient distance from the heat sources to the temperature sensor 68. That is to say, the slit 75 according to the present embodiment contributes to downsizing of the control board 50 (i.e., the air-conditioning controller 23).

Second Embodiment

In the above-described first embodiment, the measured temperature by the temperature sensor 67 and the measured temperature by the temperature sensor 68 are used as the training data to generate the trained model 81. The temperature sensors 67 and 68, which are thermistor sensors that change in resistance depending on temperature, perform analog output. The analog output changes in minute amounts. The measured temperatures by the temperature sensors 67 and 68 thus include an unnecessary variable element. In a case where the training data containing such a variable element is used to generate a trained model, the precision of the trained model may decrease due to the variable element. In particular, a combination use of the plurality of temperature sensors 67 and 68 is beneficial in generating a trained model adapted to the influence of the heat sources and the influence of the airflow. However, the inclusion of the variable element in each of the plurality of training data makes a variation in learning large. A second embodiment of the present disclosure has a feature derived from consideration for such circumstances. The feature will be described below with reference to FIG. 11A and FIG. 11B. It should be noted that an air conditioner for model generation (see FIG. 7) is omitted in FIG. 11A.

In the present embodiment, instead of the measured temperatures by the temperature sensors 67 and 68, data given by smoothing each of the measured temperatures using a moving average filter (i.e., a filtering processing or a smoothing processing) of the MCU 61 is to be inputted. That is to say, the smoothened data is training data. A temperature measurement cycle of the temperature sensors 67 and 68 is one second. The moving average filter calculates each average value from the measured temperatures for the previous 60 times.

The room temperature sensor 100 provided in a laboratory is also a thermistor sensor as the temperature sensors 67 and 68. A measured temperature by the room temperature sensor 100 is also not to be inputted as the label data. Data given by smoothing the measured temperature using the moving average filter (i.e., the filtering processing or the smoothing processing) of the MCU 61 is to be inputted. That is to say, the smoothened data is label data. A measurement cycle of the room temperature sensor 100 is one second. The moving average filter calculates an average value from the measured temperatures for the previous 60 times.

As seen from the above, the use of the smoothened data for both the training data and the label data makes it possible to reduce an influence of the above-described variable element on the training of a trained model 81X.

Moreover, in estimating the room temperature using the trained model 81X, the measured temperatures by the temperature sensors 67 and 68 are also not inputted to the trained model 81X. Data given by smoothing each of these measurement values using the moving average filter (i.e., the filtering processing or the smoothing processing) is inputted to the trained model 81X. Specific implementation of the smoothing is similar to that for generating a trained model and description thereof is omitted, accordingly.

The second embodiment described above in detail makes it possible to significantly reduce a differential between an actual room temperature and an estimated temperature as illustrated in FIG. 12 (i.e., an experimental result). Specifically, the differential falls within a range of ±1° F. except at the timing of switching from the non-airflow state to the air-blowing state. Further, as compared with the experimental result in the first embodiment (see FIG. 10), a variation in the differential is almost halved. That is to say, the majority of the differential is within a range of ±0.5° F. Therefore, it is possible to further improve the estimation accuracy of the room temperature using the trained model 81X by smoothing the training data and the label data using the moving average filter.

MODIFICATION EXAMPLES

It should be noted that embodiments of the present disclosure are not limited to the above-described embodiments and may be modified, for example, as follows. It should be noted that the following configurations may be independently applied to the above-described embodiments or all or a part of the configurations may be applied in combination to the above-described embodiments. Moreover, all or a part of the configurations according to the above-described embodiments may be combined as desired. In this case, technical meanings (i.e., effects to be exhibited) of the configurations that are to be combined are ensured. Further, a combination of a part or all of the following configurations may be applied to the combinations of the embodiments.

(1) In the above-described second embodiment, the data given by smoothing each of the measured temperatures by the temperature sensors 67 and 68 using the moving average filter is inputted in estimating the room temperature using the trained model 81. However, an output value from the trained model 81 may be smoothened using the moving average filter as illustrated in FIG. 13. That is to say, the moving average filter may be provided not on the input side but on the output side to output the smoothened data as an estimated temperature. The present modification example makes it possible to reduce the number of configurations for the smoothing as compared with the second embodiment. It should be noted that in a case where an improvement in the estimation accuracy is given priority, the moving average filter may also be provided on the input side.

(2) In the above-described embodiments, the temperature sensor 67 disposed in the first region 91 and the temperature sensor 68 disposed in the second region 92 are used to generate the trained model 81 and estimate the room temperature. The heat sources are gathered in the first region 91, which increases the temperature of the entire region. However, depending on a distance from the heat sources, a difference in temperature may arise and heat may be transferred in different manners. Accordingly, a plurality of temperature sensors 67 may be disposed in the first region 91 at respective positions where a difference in the influence of the heat sources is to arise (i.e., positions having different temperatures when heat is generated). For example, in an example illustrated in FIG. 14A, a temperature sensor 67A is disposed at a position with a relatively high temperature in the first region 91 and a temperature sensor 67B is disposed at a position with a relatively low temperature (specifically, a position near the second region 92 in the first region 91) in the first region 91. Using the temperature sensors 67A, 67B, and 68 to generate a trained model is contributable to further enhancement of the properness of the trained model. It should be noted that the temperature data inputted to the trained model in estimating the room temperature includes temperature data according to the temperature sensor 67A, temperature data according to the temperature sensor 67B, and temperature data according to the temperature sensor 68.

Moreover, a difference in temperature also arises between the corner 72 and any other portion in the second region 92. Accordingly, as illustrated in FIG. 14B, a temperature sensor 69 may be added in the portion other than the corner 72 in the second region 92 so that a trained model is generated on the basis of training data from the three temperature sensors 67 to 69.

It should be noted that in the modification examples illustrated in FIG. 14A and FIG. 14B, a trained model is to be generated on the basis of the training data from the three temperature sensors. However, the number of the temperature sensors may be four or may be five or more.

(3) In the above-described embodiments, the training data according to the temperature sensor 67 located in the first region 91 and the training data according to the temperature sensor 68 are used for the generation of the trained model 81. However, the temperature sensors may be located as desired as long as a difference in temperature useful in training a trained model is ensured for at least both the influence of the heat sources and the influence of the airflow. For example, as illustrated in FIG. 15A, the temperature sensor 69 may be located in the second region 92 (except the corner 72) in place of the temperature sensor 67 in the first region 91 so that training data according to the temperature sensor 68 and training data according to the temperature sensor 69 are used to generate a trained model. In particular, in a case where the slit 75 formed to surround the corner 72 reduces the transfer of heat to the corner 72, a difference in temperature between the corner 72 and any other portion in the second region 92 is large. Consequently, even in a case where the temperature sensors 68 and 69 are located in the second region 92, a combination use of the slit 75 (i.e., the configuration that reduces the transfer of heat) makes it possible to ensure the above-described difference in temperature. This is also favorable for improving the flexibility in locating the temperature sensors.

(4) In the above-described embodiments, the temperature sensor 68 is disposed at a position (i.e., the corner 72) near the vents 34 in the second region 92 of the control board 50. Since a difference in temperature arises between the second region 92 and the first region 91 as described above, the temperature sensor 68 may be located in the portion other than the corner 72 of the second region 92. For example, as illustrated in FIG. 15B, the temperature sensor 68 may be located in the control board 50 at a corner 73 distant from the first region 91 (i.e., a position less affected by the heat sources).

(5) In the above-described embodiments, the slit 75 serves to reduce the transfer of heat to the corner 72. However, any configuration for reducing the transfer of heat is usable. For example, a groove may be provided in place of the slit 75. A heat-releasing member having better heat-releasing performance than the control board 50 may be provided in place of the slit 75.

(6) In the above-described embodiments, in generating the trained model 81, the control parameters such as the airflow direction and the airflow volume are changed during the running of the air-conditioning apparatus for experiment. However, a control parameter to be changed is not necessarily limited to the above. For example, the control parameter to be changed may be only the airflow direction or only the airflow volume. Moreover, a control parameter to be changed may be a set temperature in addition to the airflow direction and the airflow volume.

(7) In the above-described embodiments, the model generator 61b of the air-conditioning controller 23 generates the trained model 81 for room temperature estimation. However, the generation of the trained model 81 is not limited to the above. The trained model 81 may be generated by an apparatus (for example, a machine learning apparatus) other than the air-conditioning controller 23 and the generated trained model 81 may be stored in the air-conditioning controller 23.

Features Derived from Above-Described Embodiments

In the following, the features of configurations derived from the above-described embodiments are described and effects and the like are presented, if necessary. It should be noted that in the following, a corresponding component in the above-described embodiments is written in parentheses as appropriate to facilitate understanding, but the specific component in parentheses is not limiting.

Feature 1

An air-conditioning controller (the air-conditioning controller 23) including: a control board (the control board 50) that controls an air-conditioning apparatus (the air-conditioning apparatus 21); an enclosure (the enclosure 30) housing the control board; and a plurality of temperature sensors (the temperature sensors 67 and 68), in which

• • at least one heat-generating component (the MCU 61, the P/S 55, and the LDO 56) are mounted on the control board, the at least one heat-generating component being a heat source that raises a temperature in the enclosure when powered,
  • the plurality of temperature sensors are located at a plurality of respective positions in the enclosure and heat from the at least one heat-generating component causes a difference in temperature between the plurality of positions, and
  • the control board includes a trained model for room temperature estimation (the trained model 81) generated by machine learning and estimates a temperature of a room where the air-conditioning controller is installed using temperature data from each of the plurality of temperature sensors and the trained model.

In a case where the heat sources generate heat, differences in measured temperature and the amount of change of the measured temperature may arise between the plurality of temperature sensors built into the air-conditioning controller due to a difference between the influences of the heat sources received by the plurality of temperature sensors. Moreover, in a case where a difference in temperature arises between the plurality of temperature sensors depending on the respective positions thereof and the temperature in the enclosure decreases due to the airflow from the air-conditioning device blowing on the air-conditioning controller, differences in measured temperature and the amount of change of the measured temperature may also arise. That is to say, the influence of the heat sources and the influence of the airflow are shown in the measured temperatures by the temperature sensors. Here, since there are a variety of possible manners in which the airflow blows on the air-conditioning controller, the temperatures of the heat sources are not always stable. That is to say, the influence of the heat sources and the influence of the airflow may change depending on a variety of factors in an environment where the air-conditioning controller is actually used. In this feature, temperature data from the plurality of temperature sensors is inputted to the trained model for room temperature estimation and the room temperature is estimated. This makes it possible to accurately estimate the room temperature even though the heat source and the airflow affect the measured temperatures.

Feature 2

The air-conditioning controller according to Feature 1, in which the trained model is generated by the machine learning based on training data based on the temperature data from each of the plurality of temperature sensors and label data indicating a room temperature.

As described in relation to Feature 1, the influence of the heat source and the influence of the airflow are shown in the temperature detected by each of the temperature sensors. Accordingly, in the present feature, the trained model is generated on the basis of the training data indicating the temperature detected by each of the temperature sensors and the label data indicating the room temperature. This makes it possible to accurately estimate the room temperature irrespective of the influences of the heat source and the airflow.

Feature 3

The air-conditioning controller according to Feature 1 or Feature 2, in which

• • the plurality of heat-generating components are mounted on the control board,
  • the plurality of temperature sensors include a first temperature sensor (the temperature sensor 67) and a second temperature sensor (the temperature sensor 68), and
  • a distance from each of the plurality of heat-generating components to the first temperature sensor is shorter than a distance from each of the plurality of heat-generating components to the second temperature sensor.

For the generation of the trained model, it is favorable that the influence of the heat source and the influence of the airflow be considerably different between the temperature sensors. Accordingly, in a case where a plurality of heat-generating components are mounted, the temperature sensors are located in such a way that a distance from each of the plurality of heat-generating components to the first temperature sensor is shorter than a distance from each of the plurality of heat-generating components to the second temperature sensor. This makes it possible to suitably improve the accuracy of the trained model generated using the training data indicating the temperature detected by each of the temperature sensors.

Feature 4

The air-conditioning controller according to Feature 3, in which

• • the plurality of heat-generating components include a component (the P/S 55, the LDO 56) included in a power supply unit (the power supply unit 51) of the control board and a component (the MCU 61) that controls the air-conditioning apparatus.

In the air-conditioning controller, the component included in the power supply unit and the component that controls the air-conditioning apparatus may generate a large amount of heat. Accordingly, it is possible to implement a favorable configuration for practical use by locating each of the temperature sensors with reference to the positions of these heat-generating components.

Feature 5

The air-conditioning controller according to Feature 1 or Feature 2, in which

• • the plurality of heat-generating components are mounted on the control board, and
  • the plurality of temperature sensors include a first temperature sensor (the temperature sensor 67) located in a predetermined region (the first region 91) in which the plurality of heat-generating components are gathered and a second temperature sensor (the temperature sensor 68) located in a region (the second region 92) other than the predetermined region.

Gathering the plurality of heat-generating components in the predetermined region makes it possible to not only increase a temperature of the predetermined region but also reduce a variation in the temperature. The first temperature sensor is located in such a region and the second temperature sensor is located in another region, which makes the influence of the heat source and the influence of the airflow likely to be shown in the temperature detected by each of the temperature sensors.

In addition to the heat-generating components, other electronic components are also mounted on the control board. Further, wiring patterns connecting these components is also formed. Consequently, in a case where the first temperature sensor is located near a specific heat-generating component, the first temperature sensor may be difficult to coexist with the other components and the wiring patterns. In the present feature, gathering the plurality of heat-generating components in the predetermined region makes it possible to increase a temperature of the entire predetermined region. The flexibility in locating the temperature sensor within the predetermined region is thus increased. As a result, it is possible to reduce an increase in restriction on the locations of the other components and the wiring patterns.

Feature 6

The air-conditioning controller according to Feature 5, in which

• • a shape of the control board is a rectangle,
  • the predetermined region is provided along one of a pair of sides of the control board, and
  • the second temperature sensor is located along the other one of the pair of sides in the other region.

The present feature makes it possible to increase a difference between the temperature detected by the first temperature sensor and the temperature detected by the second temperature sensor.

Feature 7

The air-conditioning controller according to Feature 5 or Feature 6, in which

• • the plurality of heat-generating components disposed in the predetermined region include a component (the P/S 55, the LDO 56) included in a power supply unit (the power supply unit 51) of the control board and a component (the MCU 61) that controls the air-conditioning apparatus.

In the air-conditioning controller, the component included in the power supply unit and the component that controls the air-conditioning apparatus may generate a large amount of heat. Accordingly, it is possible to implement a favorable configuration for practical use by locating each of the temperature sensors with reference to the positions of these heat-generating components.

Feature 8

The air-conditioning controller according to any one of Feature 3 to Feature 7, in which

• • the first temperature sensor is located at a position to which heat is to be transferred from the at least one heat-generating component through the control board, and
  • the second temperature sensor is located on the control board and partially separated from the at least one heat-generating component on the control board.

The first temperature sensor is located at the position to which heat is to be transferred from the plurality of heat-generating components through the control board and the second temperature sensor is located at the position to which heat is unlikely to be transferred, which makes it possible to distinguish a difference between the temperature detected by the first temperature sensor and the temperature detected by the second temperature sensor.

Feature 9: Three Temperature Sensors

The air-conditioning controller according to any one of Feature 3 to Feature 8, in which

• • the first temperature sensor includes a plurality of temperature sensors, and
  • the plurality of temperature sensors are located at a plurality of respective positions different in influence of heat generation of the at least one heat-generating component.

The first temperature sensor includes the plurality of temperature sensors and the plurality of temperature sensors are located at the plurality of respective positions different in influence of heat generation. This makes it possible to contribute to further improving the accuracy of estimation using the trained model. It should be noted that in generating the trained model, a temperature detected by each of the plurality of temperature sensors included in the first temperature sensor is inputted as the training data.

Feature 10: Shape of Board

The air-conditioning controller according to any one of Feature 3 to Feature 8, in which

• • an inhibitor (the slit 75) is provided in the control board at a portion between the at least one heat-generating component and the second temperature sensor, the inhibitor preventing transfer of the heat from the at least one heat-generating component toward the second temperature sensor.

In a case where the control board is large, it is easy to ensure a distance from the heat source to the second temperature sensor. However, in a case where the control board is small in size, a distance from the heat source to the second temperature sensor is short and thus it may be difficult to ensure a difference in temperature between the temperature sensors. In the present feature, the inhibitor serves to prevent the transfer of the heat to the second temperature sensor. This makes it possible to ensure a difference in temperature between the temperature sensors even though it is difficult to ensure a distance from the heat source to the second temperature sensor. That is to say, the configuration according to the present feature contributes to achieving downsizing of the control board (i.e., the air-conditioning controller).

Feature 11: Moving Average

The air-conditioning controller according to Feature 1, in which

• • the trained model is generated by the machine learning based on training data based on the temperature data from each of the plurality of temperature sensors and label data based on room temperature data, and
  • the training data is moving average data regarding the temperature data from each of the plurality of temperature sensors.

In a case where a temperature that is detected by the temperature sensor and that includes an unnecessary variable element is inputted as the training data, the variable element may become a factor hampering enhancement of the properness of the trained model. In particular, in a case where the training data is provided from the plurality of temperature sensors, proper training may become difficult due to the training data containing the variable element. Accordingly, in the present feature, data smoothened using the moving average filter is used as the training data, which makes it possible to overcome the above-described difficulty to contribute to a further improvement in the estimation accuracy of the room temperature.

(Feature 12: Estimation using Moving Average Data)

The air-conditioning controller according to Feature 11, in which

• • data inputted to the trained model in estimating the temperature of the room is moving average data of the temperature data from each of the plurality of temperature sensors.

The moving average data of the temperature data (i.e., data given by smoothing the temperature data) is inputted to the trained model in estimating the room temperature, which makes it possible to estimate the room temperature using the trained model.

(Feature 13: Moving Average of Output Value) The air-conditioning controller according to Feature 11, in which

• • the label data is moving average data of the room temperature data.

The moving average data of the room temperature data is used as the label data, so that data in a form of moving average data is outputted from the trained model. The control board can acquire the output data in the form of moving average data without processing the output data from the trained model with the moving average filter.