Air conditioning system, air conditioning system abnormality estimation method, air conditioner, and air conditioner abnormality estimation method

Abstract

An air conditioner has a refrigerant circuit configured by connecting an outdoor unit and at least one or more indoor units by refrigerant piping. The air conditioner has a detection unit for detecting a state quantity related to control of the air conditioner, and an acquisition unit for acquiring a detection value of the state quantity detected by the detection unit. Further, the air conditioner has an abnormality estimation unit which, when a state quantity related to an abnormality of the refrigerant circuit is used as a characteristic quantity, estimates occurrence of an abnormality of the refrigerant circuit using the detection value of the characteristic quantity. The abnormality estimation unit groups the outdoor unit and one indoor unit as one set, and estimates occurrence of an abnormality of the refrigerant circuit for each such combination, and in a case where it is estimated that an abnormality occurs in any combination, estimates that an abnormality occurs in the indoor unit of the combination. Further, the abnormality estimation unit, in a case where it is estimated that an abnormality occurs in all combinations, estimates that an abnormality occurs in the outdoor unit. As a result, it is possible to estimate which of the indoor unit or the outdoor unit has an abnormality.

Description

Technical Field

[0001] This invention relates to an air conditioning system, a method for diagnosing anomalies in an air conditioning system, an air conditioner, and a method for diagnosing anomalies in an air conditioner. Background Technology

[0002] For air conditioners, various methods have been proposed for detecting abnormalities or their symptoms related to the refrigerant circuit. For example, Patent Document 1 proposes a method that learns by treating capability values ​​obtained using values ​​detected by various sensors as normal data, and compares capability values ​​obtained using values ​​detected by various sensors during periods different from the learning period with the normal data, thereby detecting abnormalities in the air conditioner.

[0003] Patent Document 1: Japanese Patent Application Publication No. 2020-16358 Summary of the Invention

[0004] However, Patent Document 1 only presumes the occurrence of an air conditioner malfunction. Therefore, for example, in the case of an air conditioner where the outdoor unit is connected to multiple indoor units via refrigerant piping, it is impossible to presume whether the malfunction occurs in the outdoor unit or the indoor unit.

[0005] The present invention is proposed in view of the above-mentioned problems, and its purpose is to provide an air conditioning system capable of presuming which of the indoor or outdoor units is malfunctioning, a method for presuming malfunctions in an air conditioning system, an air conditioner, and a method for presuming malfunctions in an air conditioner.

[0006] An air conditioning system of one form includes an air conditioner and a server. The air conditioner has a refrigerant circuit formed by connecting an outdoor unit to at least one indoor unit via refrigerant piping. The server is communicatively connected to the air conditioner. The air conditioner includes: a detection unit for detecting state quantities related to the control of the air conditioner; an acquisition unit for acquiring detection values ​​of the state quantities detected by the detection unit; and a first communication unit for sending the detection values ​​acquired by the acquisition unit to the server. The server includes: a second communication unit for receiving the detection values ​​from the air conditioner; and an anomaly estimation unit that, when using the state quantities related to an anomaly in the refrigerant circuit as characteristic quantities, uses the detection values ​​of these characteristic quantities to estimate the occurrence of an anomaly in the refrigerant circuit. The anomaly estimation unit groups the outdoor unit and one indoor unit together and estimates the occurrence of an anomaly in the refrigerant circuit for each such combination. If an anomaly is estimated in any combination, an anomaly is estimated in the indoor unit of that combination; and if an anomaly is estimated in all combinations, an anomaly is estimated in the outdoor unit.

[0007] As one aspect, it is possible to determine whether the outdoor unit or the indoor unit is malfunctioning. Attached Figure Description

[0008] Figure 1 This is an explanatory diagram illustrating an example of an air conditioner according to this embodiment.

[0009] Figure 2 This is an illustrative diagram showing an example of an outdoor unit and an indoor unit.

[0010] Figure 3 This is a block diagram illustrating an example of the control circuitry for an outdoor unit.

[0011] Figure 4 This is an explanatory diagram illustrating an example of grouping outdoor units and individual indoor units as a group.

[0012] Figure 5 It is a Morrill diagram that represents the state of refrigerant changes in an air conditioner.

[0013] Figure 6 This is an explanatory diagram illustrating an example of the first characteristic quantity used in the first to third refrigeration estimation models and the second characteristic quantity used in the refrigeration anomaly estimation model.

[0014] Figure 7 This is an explanatory diagram illustrating an example of the first characteristic quantity used in the first to third heating simulation models and the second characteristic quantity used in the heating anomaly simulation model.

[0015] Figure 8A This is an explanatory diagram illustrating an example of a case where interpolation between the estimation results of the first refrigeration estimation model and the estimation results of the second refrigeration estimation model is not performed using a sigmoid curve.

[0016] Figure 8B This is an explanatory diagram illustrating an example of the interpolation between the estimation results of the first refrigeration estimation model and the estimation results of the second refrigeration estimation model using an S-curve.

[0017] Figure 9A This is an explanatory diagram illustrating an example of a case where interpolation between the estimation results of the first heating estimation model and the estimation results of the second heating estimation model is not performed using an S-curve.

[0018] Figure 9B This is an explanatory diagram illustrating an example of the interpolation between the estimation results of the first heating estimation model and the estimation results of the second heating estimation model using an S-curve.

[0019] Figure 10 This is an illustrative diagram illustrating an example of a method for representing the distribution of the detection values ​​of the second feature of an anomaly inference model.

[0020] Figure 11 This is an illustrative diagram illustrating an example of anomaly detection based on outliers.

[0021] Figure 12 This is an explanatory diagram showing an example of the judgment result of the judgment unit.

[0022] Figure 13 This is a flowchart illustrating an example of the processing actions of a control circuit involving presumption processing.

[0023] Figure 14 This is a flowchart illustrating an example of the processing actions of a control circuit involved in estimating residual cooling dose.

[0024] Figure 15 This is an explanatory diagram illustrating an example of the air conditioning system of Embodiment 2. Detailed Implementation

[0025] The following is a detailed description of embodiments of the air conditioning system, the method for estimating anomalies in the air conditioning system, the air conditioner, and the method for estimating anomalies in the air conditioner disclosed in this application, based on the accompanying drawings. However, the disclosed technology is not limited to these embodiments. Furthermore, the various embodiments shown below can be appropriately modified within a reasonable scope.

[0026] Example 1

[0027] Air conditioner structure

[0028] Figure 1 This is an explanatory diagram showing an example of the air conditioner 1 in this embodiment. Figure 1 The air conditioner 1 shown has one outdoor unit 2 and N indoor units 3, where N is a natural number greater than 2. The outdoor unit 2 is connected to each indoor unit 3 in parallel via liquid pipes 4 and gas pipes 5. Furthermore, the outdoor unit 2 and the indoor units 3 are connected via refrigerant piping, such as liquid pipes 4 and gas pipes 5, thereby forming the refrigerant circuit 6 of the air conditioner 1.

[0029] outdoor unit structure

[0030] Figure 2 This is an explanatory diagram showing an example of outdoor unit 2 and N indoor units 3. Outdoor unit 2 includes: compressor 11, four-way valve 12, outdoor heat exchanger 13, outdoor unit expansion valve 14, first shut-off valve 15, second shut-off valve 16, liquid receiver 17, outdoor unit fan 18, and control circuitry 19. Using the compressor 11, four-way valve 12, outdoor heat exchanger 13, outdoor unit expansion valve 14, first shut-off valve 15, second shut-off valve 16, and liquid receiver 17, they are interconnected by refrigerant piping as detailed below, thereby forming an outdoor refrigerant circuit as part of refrigerant circuit 6.

[0031] The compressor 11 is, for example, a high-pressure container type variable capacity compressor capable of changing its operating capacity by being driven by an electric motor (not shown) whose speed is controlled by an inverter. The refrigerant discharge side of the compressor 11 is connected to the first valve port 12A of the four-way valve 12 via a discharge pipe 21. Furthermore, the refrigerant suction side of the compressor 11 is connected to the refrigerant discharge side of the receiver 17 via a suction pipe 22.

[0032] The four-way valve 12 is a valve used to switch the refrigerant flow direction in the refrigerant circuit 6, and it has first to fourth valve ports 12A to 12D. The first valve port 12A is connected to the refrigerant discharge side of the compressor 11 via a discharge pipe 21. The second valve port 12B is connected to one side of the outdoor heat exchanger 13 via an outdoor refrigerant pipe 23. The third valve port 12C is connected to the refrigerant inflow side of the receiver 17 via an outdoor refrigerant pipe 26. Furthermore, the fourth valve port 12D is connected to the second shut-off valve 16 via an outdoor gas pipe 24.

[0033] The outdoor heat exchanger 13 facilitates heat exchange between the refrigerant and the outside air drawn into the outdoor unit 2 by the rotation of the outdoor unit fan 18. One side of the outdoor heat exchanger 13's refrigerant inlet / outlet is connected to the second valve port 12B of the four-way valve 12 via an outdoor refrigerant pipe 23. The other side of the outdoor heat exchanger 13's refrigerant inlet / outlet is connected to the first shut-off valve 15 via an outdoor liquid pipe 25. The outdoor heat exchanger 13 functions as a condenser during cooling operation of the air conditioner 1 and as an evaporator during heating operation.

[0034] The outdoor unit expansion valve 14 is located on the outdoor liquid line 25 and is an electronic expansion valve driven by a pulse motor (not shown). The outdoor unit expansion valve 14 adjusts its opening degree according to the number of pulses supplied to the pulse motor, thereby regulating the amount of refrigerant flowing into or out of the outdoor heat exchanger 13. When the air conditioner 1 is in heating operation, the opening degree of the outdoor unit expansion valve 14 is adjusted to achieve a target suction superheat for the refrigerant on the refrigerant suction side of the compressor 11. Furthermore, the outdoor unit expansion valve 14 is fully open when the air conditioner 1 is in cooling operation.

[0035] The refrigerant inlet side of the receiver 17 is connected to the third valve port 12C of the four-way valve 12 via an outdoor refrigerant pipe 26. Furthermore, the refrigerant outlet side of the receiver 17 is connected to the refrigerant inlet side of the compressor 11 via a suction pipe 22. The receiver 17 separates the refrigerant flowing into it from the outdoor refrigerant pipe 26 into gaseous and liquid refrigerant, thus allowing only gaseous refrigerant to be drawn into the compressor 11.

[0036] The outdoor unit fan 18, made of resin material, is positioned near the outdoor heat exchanger 13. Based on the rotation of a fan motor (not shown), the outdoor unit fan 18 draws outside air into the interior of the outdoor unit 2 through an air intake (not shown) and discharges the outside air, which has exchanged heat with the refrigerant in the outdoor heat exchanger 13, to the outside of the outdoor unit 2 through an exhaust (not shown).

[0037] In addition, the outdoor unit 2 is equipped with multiple sensors. The discharge pipe 21 is equipped with a discharge pressure sensor 31 and a discharge temperature sensor 32. The discharge pressure sensor 31 detects the pressure of the refrigerant discharged from the compressor 11, i.e., the discharge pressure, and the discharge temperature sensor 32 detects the temperature of the refrigerant discharged from the compressor 11, i.e., the discharge temperature. Near the refrigerant inlet of the receiver 17 of the outdoor refrigerant pipe 26, a suction pressure sensor 33 and a suction temperature sensor 34 are installed. The suction pressure sensor 33 detects the pressure of the refrigerant drawn into the compressor 11, i.e., the suction pressure, and the suction temperature sensor 34 detects the temperature of the refrigerant drawn into the compressor 11.

[0038] An outdoor liquid pipe 25 between the outdoor heat exchanger 13 and the outdoor unit expansion valve 14 is equipped with a refrigerant temperature sensor 35, which is used to detect the temperature of the refrigerant flowing into or out of the outdoor heat exchanger 13. Furthermore, near the air intake (not shown) of the outdoor unit 2, an external air temperature sensor 36 is installed, which is used to detect the temperature of the external air flowing into the outdoor unit 2, i.e., the external air temperature.

[0039] The control circuit 19 is used to control the entire air conditioner 1. Figure 3 This is a block diagram illustrating an example of the control circuit 19 for the outdoor unit 2. The control circuit 19 includes: an acquisition unit 41, a communication unit 42, a storage unit 43, a control unit 44, a refrigerant dosage estimation unit 45, and an anomaly estimation unit 46. The acquisition unit 41 acquires sensor values ​​from the various sensors, i.e., the detection units. The communication unit 42 is a communication interface for communicating with the communication units of each indoor unit 3. The storage unit 43 is, for example, a flash memory, used to store: the control program for the outdoor unit 2, operating status quantities such as detection values ​​corresponding to detection signals from various sensors, the drive status of the compressor 11 and the outdoor unit fan 18, operating information sent from each indoor unit 3 (e.g., including operation / stop information, operating modes such as cooling / heating), and the rated capacity of the outdoor unit 2 and the required capacity of each indoor unit 3. Furthermore, the storage unit 43 includes an anomaly record storage unit 43A for storing anomaly records described later.

[0040] The control unit 44 periodically (e.g., every 30 seconds) acquires detection values ​​from various sensors via the communication unit 42. Signals including operating status values ​​sent from each indoor unit 3 are input to the control unit 44 via the communication unit 42. Based on this input information, the control unit 44 adjusts the opening of the outdoor unit expansion valve 14 or controls the drive of the compressor 11.

[0041] The refrigerant dosage estimation unit 45 has a refrigerant dosage estimation model 45A, which uses the detection value of the first characteristic quantity to estimate the refrigerant shortage rate of the refrigerant circuit 6 when the operating state quantity related to the refrigerant dosage of the refrigerant circuit 6 is set as the first characteristic quantity. In this embodiment, the refrigerant dosage remaining in the refrigerant circuit 6 is, for example, a relative refrigerant dosage. Specifically, the refrigerant dosage estimation model 45A is a model for estimating the refrigerant shortage rate of the refrigerant circuit 6 (the amount of reduction relative to the specified amount when the refrigerant is filled to 100%, as will be the case below). The refrigerant dosage estimation model 45A includes: a first refrigerant estimation model 45A1, a second refrigerant estimation model 45A2, a third refrigerant estimation model 45A3, a first heating estimation model 45A4, a second heating estimation model 45A5, and a third heating estimation model 45A6. These refrigerant dosage estimation models 45A will be described in detail below.

[0042] Figure 4 This is an explanatory diagram illustrating an example of grouping outdoor unit 2 with each indoor unit 3 as a group. Furthermore, for ease of explanation, the case where the outdoor unit 2 of the air conditioner 1 is, for example, one unit, and the indoor units 3 (3A, 3B, 3C, 3D) connected to the outdoor unit 2 are, for example, four units, will be described. In this example, one outdoor unit 2 and one indoor unit 3 are grouped together, and the combination of outdoor unit 2 and indoor unit 3A is designated as P1, the combination of outdoor unit 2 and indoor unit 3B as P2, the combination of outdoor unit 2 and indoor unit 3C as P3, and the combination of outdoor unit 2 and indoor unit 3D as P4.

[0043] The anomaly estimation unit 46 has an anomaly estimation model 46A, which uses the detection value of a second characteristic quantity related to the anomaly of the refrigerant circuit 6 in the operating state quantities to estimate whether the refrigerant circuit 6 of each combination P1 to P4 of outdoor unit 2 and indoor unit 3 is abnormal or normal. The anomaly estimation unit 46 is used to estimate the anomaly of the refrigerant circuit 6 for each combination P1 to P4. If it is estimated that an anomaly of the refrigerant circuit 6 has occurred in any combination, the cause of the anomaly is estimated to be the indoor unit 3 of that combination. Furthermore, if the anomaly estimation unit 46 estimates that an anomaly has occurred in all combinations P1 to P4, the cause of the anomaly is estimated to be the outdoor unit 2.

[0044] The anomaly estimation model 46A includes: a cooling anomaly estimation model 46B used when the air conditioner 1 is in cooling operation, and a heating anomaly estimation model 46C used when the air conditioner 1 is in heating operation. Furthermore, the anomaly estimation unit 46 has a judgment unit 46D, which can determine, based on the anomaly estimation results of each combination P1 to P4, whether the outdoor unit 2 or the indoor unit 3 is the main cause of the anomaly in the refrigerant circuit 6. These anomaly estimation models 46A will be described in detail later.

[0045] Indoor unit structure

[0046] like Figure 2 As shown, the indoor unit 3 includes: an indoor heat exchanger 51, an indoor unit expansion valve 52, a liquid pipe connection 53, a gas pipe connection 54, and an indoor unit fan 55. The indoor heat exchanger 51, the indoor unit expansion valve 52, the liquid pipe connection 53, and the gas pipe connection 54 are connected to each other via refrigerant piping as described later, thereby forming an indoor unit refrigerant circuit as part of the refrigerant circuit 6.

[0047] The indoor heat exchanger 51 facilitates heat exchange between the refrigerant and indoor air drawn into the indoor unit 3 through an air intake (not shown) by the rotation of the indoor unit fan 55. One side of the indoor heat exchanger 51 is connected to the liquid pipe connection 53 via an indoor liquid pipe 56. The other side of the indoor heat exchanger 51 is connected to the gas pipe connection 54 via an indoor gas pipe 57. When the air conditioner 1 is in heating operation, the indoor heat exchanger 51 functions as a condenser. Conversely, when the air conditioner 1 is in cooling operation, the indoor heat exchanger 51 functions as an evaporator.

[0048] The indoor unit expansion valve 52 is an electronic expansion valve located on the indoor liquid line 56. When the indoor heat exchanger 51 functions as an evaporator, i.e., when the indoor unit 3 is in cooling operation, the opening of the indoor unit expansion valve 52 is adjusted to make the refrigerant superheat at the refrigerant outlet (gas pipe connection 54 side) of the indoor heat exchanger 51 the target refrigerant superheat. Furthermore, when the indoor heat exchanger 51 functions as a condenser, i.e., when the indoor unit 3 is in heating operation, the opening of the indoor unit expansion valve 52 is adjusted to make the refrigerant subcooling at the refrigerant outlet (liquid pipe connection 53 side) of the indoor heat exchanger 51 the target refrigerant subcooling. Here, the target refrigerant superheat or target refrigerant subcooling refers to the refrigerant superheat and refrigerant subcooling required for the indoor unit 3 to achieve sufficient cooling or heating capacity.

[0049] The indoor unit fan 55, made of resin material, is positioned near the indoor heat exchanger 51. The indoor unit fan 55 is driven to rotate by a fan motor (not shown), thereby drawing indoor air into the interior of the indoor unit 3 through the air intake (not shown) and discharging the indoor air, which has exchanged heat with the refrigerant in the indoor heat exchanger 51, out of the room through the exhaust (not shown).

[0050] Various sensors are installed in the indoor unit 3. A liquid-side refrigerant temperature sensor 61 is installed in the indoor liquid pipe 56 between the indoor heat exchanger 51 and the indoor unit expansion valve 52. This sensor detects the temperature of the refrigerant flowing into the indoor heat exchanger 51 (the indoor unit-side heat exchange inlet temperature during cooling operation) or the temperature of the refrigerant flowing out of the indoor heat exchanger 51 (the indoor unit-side heat exchange outlet temperature during heating operation). A gas-side temperature sensor 62 is installed in the indoor gas pipe 57. This sensor detects the temperature of the refrigerant flowing out of the indoor heat exchanger 51 (the indoor unit-side heat exchange outlet temperature during cooling operation) or the temperature of the refrigerant flowing into the indoor heat exchanger 51 (the indoor unit-side heat exchange inlet temperature during heating operation). A suction temperature sensor 63 is installed near the suction port (not shown) of the indoor unit 3. This sensor detects the temperature of the indoor air flowing into the interior of the indoor unit 3, i.e., the suction temperature.

[0051] Operation of the refrigerant circuit

[0052] Next, the flow of refrigerant in the refrigerant circuit 6 and the operation of each component during the operation of the air conditioner 1 in this embodiment will be explained. Furthermore, Figure 1 The arrows indicate the direction of refrigerant flow during heating operation.

[0053] When the air conditioner 1 is in heating mode, the four-way valve 12 is switched so that the first valve port 12A is connected to the fourth valve port 12D, and the second valve port 12B is connected to the third valve port 12C. Thus, the refrigerant circuit 6 is formed into a heating cycle where each indoor heat exchanger 51 functions as a condenser and the outdoor heat exchanger 13 functions as an evaporator. Furthermore, for ease of explanation, by... Figure 2 The solid arrows shown indicate the direction of refrigerant flow during heating operation.

[0054] When the refrigerant circuit 6 is in the above-described state, the compressor 11 is driven. The refrigerant discharged from the compressor 11 flows through the discharge pipe 21 and into the four-way valve 12. From the four-way valve 12, it flows through the outdoor gas pipe 24 and then through the second shut-off valve 16 into the gas pipe 5. The refrigerant flowing in the gas pipe 5 is distributed to each indoor unit 3 via the gas pipe connection 54. The refrigerant flowing into each indoor unit 3 flows through each indoor gas pipe 57 and then into each indoor heat exchanger 51. The refrigerant flowing into each indoor heat exchanger 51 exchanges heat with the indoor air drawn into each indoor unit 3 by the rotation of the indoor unit fan 55 and is condensed. That is, each indoor heat exchanger 51 functions as a condenser. The indoor air heated by the refrigerant in each indoor heat exchanger 51 is blown into the room through the exhaust port not shown, thereby heating the room where each indoor unit 3 is installed.

[0055] Refrigerant flowing from each indoor heat exchanger 51 into each indoor liquid line 56 is depressurized as it flows through each indoor unit expansion valve 52. The opening of each indoor unit expansion valve 52 is adjusted to achieve a target refrigerant subcooling at the refrigerant outlet side of each indoor heat exchanger 51. This target refrigerant subcooling is determined based on the required cooling capacity of each indoor unit 3.

[0056] The refrigerant, depressurized in each indoor unit expansion valve 52, flows from each indoor liquid line 56 through each liquid line connection 53 to the liquid line 4. The refrigerant converging in the liquid line 4 flows into the outdoor unit 2 through the first shut-off valve 15. The refrigerant flowing into the first shut-off valve 15 of the outdoor unit 2 flows through the outdoor liquid line 25 and is depressurized through the outdoor unit expansion valve 14. The refrigerant depressurized in the outdoor unit expansion valve 14 flows through the outdoor liquid line 25 and then into the outdoor heat exchanger 13, where it exchanges heat with the outside air flowing in from the air intake (not shown) of the outdoor unit 2 through the rotation of the outdoor unit fan 18, thereby evaporating. The refrigerant flowing out of the outdoor heat exchanger 13 to the outdoor refrigerant line 26 flows sequentially into the four-way valve 12, the outdoor refrigerant line 26, the receiver 17, and the suction pipe 22, and is then drawn into the compressor 11 and compressed again. It then flows out through the first valve port 12A and the fourth valve port 12D of the four-way valve 12 to the outdoor gas line 24.

[0057] Furthermore, when the air conditioner 1 is in cooling operation, the four-way valve 12 switches to connect the first valve port 12A with the second valve port 12B, and the third valve port 12C with the fourth valve port 12D. Thus, the refrigerant circuit 6 forms a refrigeration cycle in which each indoor heat exchanger 51 functions as an evaporator, and the outdoor heat exchanger 13 functions as a condenser. Furthermore, for ease of explanation, [the following is used...] Figure 2 The dashed arrows shown represent the flow of refrigerant during refrigeration operation.

[0058] With the compressor 11 driven in refrigerant circuit 6, the refrigerant discharged from the compressor 11 flows through the discharge pipe 21 and into the four-way valve 12, then through the outdoor refrigerant pipe 26 and into the outdoor heat exchanger 13. The refrigerant flowing into the outdoor heat exchanger 13 exchanges heat with the outdoor air drawn into the outdoor unit 2 by the rotation of the outdoor unit fan 18, thus condensing. In other words, the outdoor heat exchanger 13 functions as a condenser, and the indoor air heated by the refrigerant in the outdoor heat exchanger 13 is blown to the outside through the exhaust port not shown in the diagram.

[0059] Refrigerant flowing from the outdoor heat exchanger 13 into the outdoor liquid pipe 25 is depressurized as it flows through the fully open outdoor unit expansion valve 14. After being depressurized by the outdoor unit expansion valve 14, the refrigerant flows through the liquid pipe 4 via the first shut-off valve 15 and is distributed to each indoor unit 3. The refrigerant flowing into each indoor unit 3 flows through the indoor liquid pipe 56 via the liquid pipe connection 53 and is depressurized by the indoor unit expansion valve 52, wherein the opening of the indoor unit expansion valve 52 is adjusted to achieve the target refrigerant subcooling at the refrigerant outlet of the indoor heat exchanger 51. The refrigerant, depressurized in the indoor unit expansion valve 52, flows through the indoor liquid pipe 56 and into the indoor heat exchanger 51, where it exchanges heat with indoor air flowing in from the indoor unit 3's (not shown) suction port due to the rotation of the indoor unit fan 55, thereby evaporating. In other words, each indoor heat exchanger 51 functions as an evaporator, and the indoor air cooled by the refrigerant in each indoor heat exchanger 51 is blown out of the exhaust port not shown in the figure to cool the room where each indoor unit 3 is installed.

[0060] Refrigerant flowing from indoor heat exchanger 51 into pipe 5 via pipe connection 54 flows through outdoor pipe 24 via second shut-off valve 16 of outdoor unit 2 and into fourth port 12D of four-way valve 12. Refrigerant flowing into fourth port 12D of four-way valve 12 flows into the refrigerant inlet side of receiver 17 via third port 12C. Refrigerant flowing into receiver 17 from the refrigerant inlet side is drawn into compressor 11 via suction pipe 22 and compressed again.

[0061] The acquisition unit 41 within the control circuit 19 acquires sensor values ​​from the discharge pressure sensor 31, discharge temperature sensor 32, suction pressure sensor 33, suction temperature sensor 63, refrigerant temperature sensor 35, and external air temperature sensor 36 within the outdoor unit 2. Furthermore, the acquisition unit 41 acquires sensor values ​​from the liquid-side refrigerant temperature sensor 61, gas-side temperature sensor 62, and suction temperature sensor 63 of each indoor unit 3.

[0062] Figure 5This is a Morrill diagram representing the refrigeration cycle of air conditioner 1. During cooling operation, the outdoor heat exchanger 13 functions as a condenser, and the indoor heat exchanger 51 functions as an evaporator. Furthermore, during heating operation, the outdoor heat exchanger 13 functions as an evaporator, and the indoor heat exchanger 51 functions as a condenser.

[0063] Compressor 11 compresses the low-temperature, low-pressure gaseous refrigerant flowing from the evaporator, while compressing the high-temperature, high-pressure gaseous refrigerant (becoming...) Figure 5 The refrigerant in state B is discharged. Furthermore, the temperature of the gaseous refrigerant discharged by compressor 11 is the discharge temperature, which is detected by discharge temperature sensor 32.

[0064] The condenser condenses the high-temperature, high-pressure gaseous refrigerant from compressor 11 after heat exchange with air. At this point, in the condenser, the gaseous refrigerant has completely transformed into liquid refrigerant through latent heat change, and the temperature of the liquid refrigerant decreases through sensible heat change, becoming subcooled. Figure 5 (The state of point C). Furthermore, the temperature at which the gaseous refrigerant changes into a liquid refrigerant through latent heat change is the high-pressure saturation temperature, and the temperature of the refrigerant in the subcooled state at the condenser outlet is the heat exchange outlet temperature. The high-pressure saturation temperature is related to the pressure value detected by the discharge pressure sensor 31. Figure 5 The temperature equivalent to the pressure value P2 (represented by "HPS"). The heat exchange outlet temperature is the temperature of the refrigerant flowing in the outdoor liquid line 25, detected by the refrigerant temperature sensor 35.

[0065] The expansion valve reduces the pressure of the low-temperature, high-pressure refrigerant flowing from the condenser, turning it into a two-phase refrigerant mixture of gas and liquid. Figure 5 (The state of the refrigerant at point D).

[0066] The evaporator allows the incoming gas-liquid two-phase refrigerant to exchange heat with the air, causing it to evaporate. At this point, the gas-liquid two-phase refrigerant in the evaporator has completely transformed into a gaseous refrigerant through latent heat change, and then the temperature of the gaseous refrigerant rises through sensible heat change, reaching a superheated state. Figure 5 (at point A), and then drawn into compressor 11. Furthermore, the temperature at which the liquid refrigerant changes to a gaseous refrigerant through latent heat change is the low-pressure saturation temperature. The low-pressure saturation temperature is related to the pressure value detected by the suction pressure sensor 33 ( Figure 5 The temperature corresponding to the pressure value P1 (represented by "LPS" in the text) is also the suction temperature. Additionally, the temperature of the refrigerant that is superheated in the evaporator and then drawn into the compressor 11 is the suction temperature. The suction temperature is detected by the suction temperature sensor 34.

[0067] Furthermore, the refrigerant subcooling when the refrigerant flows out of the condenser can be calculated by subtracting the refrigerant temperature at the outlet of the heat exchanger (mentioned above) from the high-pressure saturation temperature. Similarly, the suction superheat of the refrigerant when it flows out of the evaporator can be calculated by subtracting the suction temperature from the low-pressure saturation temperature.

[0068] First characteristic quantity

[0069] Figure 6 This is an explanatory diagram illustrating an example of the first characteristic quantity used in the first to third refrigeration estimation models 45A1, 45A2, and 45A3, and the second characteristic quantity used in the refrigeration anomaly estimation model 46B. As an operating state quantity used in the refrigerant dosage estimation model 45A, there exists a first characteristic quantity. Examples of the first characteristic quantities used in the first to third refrigeration estimation models 45A1, 45A2, and 45A3 include: compressor 11 speed, high-pressure saturation temperature, suction temperature, low-pressure refrigerant temperature, refrigerant subcooling (outdoor heat exchange subcooling), and outside air temperature. The compressor 11 speed is detected by a speed sensor (not shown) of compressor 11. The high-pressure saturation temperature is a value obtained by converting the pressure value detected by discharge pressure sensor 31 to temperature. The suction temperature is detected by suction temperature sensor 34. The low-pressure refrigerant temperature is the temperature of the refrigerant that is superheated in the evaporator and then drawn into compressor 11. The refrigerant subcooling is, for example, a value calculated by (high-pressure saturation temperature - outdoor heat exchange outlet temperature). The outside air temperature is detected by outside air temperature sensor 36. Furthermore, the outdoor heat exchange outlet temperature is detected by the refrigerant temperature sensor 35. For example, the operating status quantities, including the first characteristic quantity, used in the first to third refrigeration estimation models 45A1, 45A2, and 45A3 are periodically detected by detection units such as a speed sensor, discharge pressure sensor 31, suction temperature sensor 34, external air temperature sensor 36, and refrigerant temperature sensor 35. Furthermore, while the air conditioner 1 is operating, the control unit 44 instructs the detection unit to periodically (e.g., every 10 minutes) acquire the operating status quantities. The instructed detection unit detects the operating status quantities from various sensors installed on the air conditioner 1. The periodically acquired operating status quantities are also marked with the acquisition time information.

[0070] Figure 7This is an explanatory diagram illustrating an example of the first characteristic quantity used in the first to third heating estimation models 45A4, 45A5, and 45A6, and the second characteristic quantity used in the heating anomaly estimation model 46C. Examples of the first characteristic quantities used in the first to third heating estimation models 45A4, 45A5, and 45A6 include: the opening degree of the outdoor unit expansion valve 14, the speed of the compressor 11, the suction superheat, and the outside air temperature. The opening degree of the outdoor unit expansion valve 14 is the number of pulses applied by the control unit 44 to the outdoor unit expansion valve 14 by a stepper motor (not shown). The speed of the compressor 11 is detected by a speed sensor (not shown) of the compressor 11. The suction superheat is, for example, a value calculated by (suction temperature - low-pressure saturation temperature). The outside air temperature is detected by the outside air temperature sensor 36. The suction temperature is a value detected by the suction temperature sensor 34, and the low-pressure saturation temperature is a value obtained by converting the pressure value detected by the suction pressure sensor 33 to temperature. In addition, for example, the operating state quantities containing the first characteristic quantity used in the first to third heating estimation models 45A4, 45A5 and 45A6 are periodically detected by detection units such as speed sensor, intake temperature sensor 34 and external temperature sensor 36.

[0071] Second characteristic quantity

[0072] As an operational state quantity used in the anomaly prediction model 46A, there is a second characteristic quantity related to the anomaly of refrigerant circuit 6. The second characteristic quantity used in generating the anomaly prediction model 46A is, for example, a value obtained when reproducing refrigerant circuit 6 on a computer and performing numerical analysis (hereinafter, numerical analysis will also be referred to as simulation), under the condition that refrigerant circuit 6 operates normally and only the residual refrigerant charge is changed. Furthermore, the second characteristic quantity used in generating the anomaly prediction model 46A is represented as a simulated value (sometimes simply referred to as a "value"). The second characteristic quantity includes at least one operational state quantity included in the first characteristic quantity, and at least one operational state quantity not included in the first characteristic quantity.

[0073] like Figure 6As shown, the second characteristic quantities used in the refrigeration anomaly prediction model 46B include, for example, the compressor 11 speed, high-pressure saturation temperature, suction temperature, low-pressure refrigerant temperature, outside air temperature, high-pressure sensor (HPS), and heat exchange outlet temperature. The compressor 11 speed is detected by a speed sensor (not shown) of the compressor 11. The high-pressure saturation temperature is the value obtained by converting the pressure value detected by the discharge pressure sensor 31 to a temperature value. The suction temperature is detected by the suction temperature sensor 34. The low-pressure refrigerant temperature is the temperature of the refrigerant that is superheated in the evaporator and then drawn into the compressor 11. The outside air temperature is detected by the outside air temperature sensor 36. The high-pressure sensor is the pressure value detected by the discharge pressure sensor 31. The heat exchange outlet temperature is detected by the refrigerant temperature sensor 35. Furthermore, for example, the operating state quantities including the second characteristic quantities used in the refrigeration anomaly prediction model 46B are periodically detected by detection units such as the speed sensor, discharge pressure sensor 31, suction temperature sensor 34, outside air temperature sensor 36, and refrigerant temperature sensor 35.

[0074] In addition, such as Figure 7 As shown, the second characteristic quantities used in the heating anomaly prediction model 46C include, for example, the opening degree of the outdoor unit expansion valve 14, the compressor speed 11, the outside air temperature, the discharge temperature, the suction temperature, the low-pressure saturation temperature, and the low-pressure sensor (LPS). The opening degree of the outdoor unit expansion valve 14 is detected by a sensor (not shown). The compressor speed 11 is detected by a speed sensor of the compressor 11 (not shown). The outside air temperature is detected by the outside air temperature sensor 36. The discharge temperature is detected by the discharge temperature sensor 32. The suction temperature is detected by the suction temperature sensor 34. The low-pressure saturation temperature is the value obtained by converting the pressure value detected by the suction pressure sensor 33 to temperature. The low-pressure sensor is the pressure value detected by the suction pressure sensor 33. Furthermore, for example, the operating state quantities including the second characteristic quantities used in the heating anomaly prediction model 46C are periodically detected by detection units such as the speed sensor, the suction temperature sensor 34, the outside air temperature sensor 36, and the suction pressure sensor 33.

[0075] As a second characteristic quantity commonly used in the anomaly estimation model 46B during cooling and the anomaly estimation model 46C during heating, the compressor speed and suction temperature can be used as operating state quantities on both sides of the outdoor unit.

[0076] Furthermore, as a second characteristic quantity common to both the cooling anomaly estimation model 46B and the heating anomaly estimation model 46C, there may be operating state quantities on the indoor unit 3 side, such as: the indoor unit side heat exchange inlet temperature (detected by the liquid-side refrigerant temperature sensor 61 during cooling operation / detected by the gas-side temperature sensor 62 during heating operation), the indoor unit side heat exchange outlet temperature (detected by the gas-side temperature sensor 62 during cooling operation / detected by the liquid-side refrigerant temperature sensor 61 during heating operation), and the opening degree of the indoor unit expansion valve 52. Moreover, examples of the second characteristic quantities on the indoor unit 3 side, such as the indoor unit side heat exchange inlet temperature, the indoor unit side heat exchange outlet temperature, and the opening degree of the indoor unit expansion valve 52, are characteristic quantities that can be commonly obtained regardless of whether the indoor unit 3 is a ducted type, ceiling-mounted type, etc.

[0077] Structure of the Cooling Dosage Estimation Model

[0078] The refrigerant dosage estimation model 45A is generated using the detection value of the first characteristic quantity. The refrigerant dosage estimation unit 45 applies the detection value of the first characteristic quantity, which is acquired at a time different from the time the refrigerant dosage estimation model 45A is generated, to the refrigerant dosage estimation model 45A, thereby estimating the refrigerant shortage rate of the refrigerant circuit 6.

[0079] The refrigerant dosage estimation model 45A is generated using any one of multiple operating state variables (the detected value of the first characteristic variable) through a regression analysis method, namely, multiple regression analysis. Multiple regression analysis involves selecting the regression equation with the smallest P-value (a preset weighting parameter representing the influence of the operating state variable on the accuracy of the generated estimation model) and the largest possible correction value R2 (representing the accuracy of the generated refrigerant dosage estimation model 45A) between 0.9 and 1.0 from among the regression equations obtained based on multiple simulation results (numerical calculations to reproduce refrigerant loop 6 and to calculate the values ​​of the operating state variables relative to the remaining refrigerant dosage) to generate the refrigerant dosage estimation model 45A. Specifically, the P-value and correction value R2 are values ​​related to the accuracy of the refrigerant dosage estimation model 45A when it is generated using multiple regression analysis; the smaller the P-value and the closer the correction value R2 is to 1.0, the higher the accuracy of the generated refrigerant dosage estimation model 45A. As a result, when the refrigerant deficit rate is 0% to 30% during cooling, operating state parameters such as refrigerant subcooling, ambient temperature, high-pressure saturation temperature, and compressor speed are used as the first characteristic parameters. When the refrigerant deficit rate is 40% to 70% during cooling, operating state parameters such as suction temperature, ambient temperature, and compressor speed are used as the first characteristic parameters. When the refrigerant deficit rate is 0% to 20% during heating, the opening degree of outdoor unit expansion valve 14 is used as the characteristic parameter. Furthermore, when the refrigerant deficit rate is 30% to 70% during heating, operating state parameters such as suction superheat (suction temperature - low-pressure saturation temperature), ambient temperature, compressor speed, and outdoor unit expansion valve 14 are used as the first characteristic parameters.

[0080] As described above, the refrigerant dosage estimation model 45A includes: a first refrigerant estimation model 45A1, a second refrigerant estimation model 45A2, a third refrigerant estimation model 45A3, a first heating estimation model 45A4, a second heating estimation model 45A5, and a third heating estimation model 45A6. In this embodiment, each of the above estimation models is generated using the simulation results described later and is pre-stored in the refrigerant dosage estimation unit 45 within the control circuit 19 of the air conditioner 1.

[0081] The first refrigeration estimation model 45A1 is a refrigerant dosage estimation model 45A effective when the refrigerant shortage rate is 0% to 30% (first range). It is a first regression equation that can estimate the refrigerant shortage rate with high accuracy. The first regression equation is, for example, (α1 × refrigerant subcooling) + (α2 × outside air temperature) + (α3 × high pressure saturation temperature) + (α4 × compressor speed 11) + α5. The coefficients α1 to α5 are determined when generating the estimation model. The refrigerant dosage estimation unit 45 calculates the current refrigerant shortage rate of the refrigerant circuit 6 by substituting the current refrigerant subcooling, outside air temperature, high pressure saturation temperature, and compressor speed 11 acquired by the acquisition unit 41 into the first regression equation. Furthermore, the reason for substituting the refrigerant subcooling, outside air temperature, high pressure saturation temperature, and compressor speed 11 is to use the first characteristic quantity used when generating the first refrigeration estimation model 45A1. The refrigerant subcooling can be calculated, for example, by (high pressure saturation temperature - heat exchange outlet temperature). The outside air temperature is detected by the outside air temperature sensor 36. The high-pressure saturation temperature is a value obtained by converting the pressure value detected by the discharge pressure sensor 31 into a temperature value. The rotational speed of the compressor 11 is detected by a rotational speed sensor of the compressor 11 (not shown).

[0082] The second refrigeration estimation model 45A2 is an effective refrigerant dosage estimation model 45A when the refrigerant shortage rate is 40% to 70% (second range). It is a second regression equation that can estimate the refrigerant shortage rate with high accuracy. The second regression equation is, for example, (α11 × suction temperature) + (α12 × outside temperature) + (α13 × compressor speed 11) + α14. The coefficients α11 to α14 are determined when generating the estimation model. The refrigerant dosage estimation unit 45 calculates the current refrigerant shortage rate of the refrigerant circuit 6 by substituting the current suction temperature, outside temperature, and compressor speed 11 obtained by the acquisition unit 41 into the second regression equation. Furthermore, the reason for substituting the suction temperature, outside temperature, and compressor speed 11 is to use the characteristic quantities used when generating the second refrigeration estimation model 45A2. The suction temperature is detected by the suction temperature sensor 34. The outside temperature is detected by the outside temperature sensor 36. The compressor speed 11 is detected by a speed sensor of the compressor 11 (not shown).

[0083] On the other hand, as mentioned above, the refrigerant deficit rate can be calculated from 0% to 30% using the first regression equation, and from 40% to 70% using the second regression equation. In this case, when the refrigerant deficit rate is 30% to 40%, the first regression equation calculates a deficit rate of 30%, while the second regression equation calculates a deficit rate of 40%. That is, when the refrigerant deficit rate is 30% to 40%, either the refrigerant subcooling, which contributes significantly to the deficit rate below 30%, or the suction temperature, which contributes significantly to the deficit rate above 40%, changes little, thus failing to generate an effective estimation model. Therefore, if either the first or second regression equation is used, then... Figure 8A As shown, the refrigerant shortage rate varies greatly depending on which model is used.

[0084] The third refrigeration estimation model 45A3 is a formula for calculating the refrigerant shortage rate during refrigeration, covering a range of 0% to 70%. Its coverage includes the range where the refrigerant shortage rate cannot be estimated using either the first or second regression equation described above. Figure 8B As shown, the refrigerant shortage rate calculation formula during cooling is a formula used to continuously connect the refrigerant shortage rate estimated as the result of the first regression equation and the refrigerant shortage rate estimated as the result of the second regression equation using an S-curve with an S-shaped coefficient. Specifically, the refrigerant shortage rate calculation formula during cooling is: (refrigerant shortage rate obtained by S-shaped coefficient × first regression equation) + (refrigerant shortage rate obtained by (1 - S-shaped coefficient) × second regression equation). The refrigerant dosage estimation unit 45 substitutes the current operating status quantity obtained by the acquisition unit 41 into the first regression equation and the second regression equation to calculate the refrigerant shortage rate respectively, and substitutes the refrigerant shortage rate into the refrigerant shortage rate calculation formula during cooling to calculate the current refrigerant shortage rate of the refrigerant circuit 6.

[0085] The S-curve coefficient is calculated using any of the operating state variables. In this embodiment, considering that the result based on the first regression equation remains almost unchanged when the subcooling is 0°C, the calculation formula is set to 0.5 for the S-curve coefficient when the subcooling is 5°C.

[0086] p = 1 / (1 + exp(-(sc-5)))

[0087] p: S-type coefficient

[0088] sc: supercooling value

[0089] Therefore, by determining the S-type coefficient and applying it to the third refrigeration estimation model 45A3, when the refrigerant shortage rate is 0% to 30%, that is, when the refrigerant shortage rate is in the first range, the estimated value of the first refrigeration estimation model 45A1 dominates among the estimated values ​​based on the third refrigeration estimation model 45A3. Furthermore, when the refrigerant shortage rate is 40% to 70%, that is, when the refrigerant shortage rate is in the second range, the estimated value of the second refrigeration estimation model 45A2 dominates among the estimated values ​​based on the third refrigeration estimation model 45A3.

[0090] Furthermore, the calculation of the S-type coefficient is not limited to the above method. The S-type coefficient can be determined as follows: when the actual refrigerant shortage rate is above 30%, that is, when the actual refrigerant shortage rate is not within the first range, the estimated value of the second refrigerant estimation model 45A2 among the estimated values ​​based on the third refrigerant estimation model 45A3 becomes dominant; and when the actual refrigerant shortage rate is below 40%, that is, when the actual refrigerant shortage rate is not within the second range, the estimated value of the first refrigerant estimation model 45A1 among the estimated values ​​based on the third refrigerant estimation model 45A3 becomes dominant.

[0091] The first heating estimation model 45A4 is a refrigerant dosage estimation model 45A that is effective when the refrigerant shortage rate is 0% to 20% (third range). It is a fourth regression equation that can estimate the refrigerant shortage rate with high accuracy. The fourth regression equation is, for example, (α31 × opening degree of outdoor unit expansion valve 14) + α32. The refrigerant dosage estimation unit 45 calculates the refrigerant shortage rate by substituting the current opening degree of outdoor unit expansion valve 14 acquired by the acquisition unit 41 into the fourth regression equation. Furthermore, the reason for substituting the opening degree of outdoor unit expansion valve 14 is to use the characteristic quantity used when generating the first heating estimation model 45A4.

[0092] The second heating estimation model 45A5 is an effective refrigerant dosage estimation model 45A when the refrigerant shortage rate is 30% to 70% (fourth range). It is a fifth regression equation that can estimate the refrigerant shortage rate with high accuracy. The fifth regression equation is, for example, (α41 × suction superheat) + (α42 × outside temperature) + (α43 × compressor speed 11) + (α44 × outdoor unit expansion valve 14 opening) + α45. The coefficients α41 to α45 are determined when generating the estimation model. The refrigerant dosage estimation unit 45 calculates the current refrigerant shortage rate of the refrigerant circuit 6 by substituting the current suction superheat, outside temperature, compressor speed 11, and main-side expansion valve opening obtained by the acquisition unit 41 into the fifth regression equation. Furthermore, the reason for substituting the suction superheat, outside temperature, compressor speed 11, and outdoor unit expansion valve opening is to use the characteristic quantities used when generating the second heating estimation model 45A5. The suction superheat can be calculated, for example, by (suction temperature - low-pressure saturation temperature). The outside air temperature is detected by the outside air temperature sensor 36. The compressor speed 11 is detected by a speed sensor of compressor 11 (not shown). The opening degree of the outdoor unit expansion valve 14 is detected by a sensor (not shown).

[0093] Furthermore, as mentioned above, the refrigerant shortage rate can be calculated from 0% to 20% using the fourth regression equation, and from 30% to 70% using the fifth regression equation. In this case, when the refrigerant shortage rate is 20% to 30%, the fourth regression equation calculates a refrigerant shortage rate of 20%, while the fifth regression equation calculates a refrigerant shortage rate of 30%. That is, when the refrigerant shortage rate is 20% to 30%, either the opening of the outdoor unit expansion valve 14, which contributes significantly to the refrigerant shortage rate below 20%, or the suction superheat, which contributes significantly to the refrigerant shortage rate above 30%, changes little, thus failing to generate an effective estimation model. Therefore, if the fourth or fifth regression equation is used, then... Figure 9A As shown, the refrigerant shortage rate varies greatly depending on which model is used.

[0094] The third model for heating, 45A6, is a formula for calculating refrigerant shortage rate during heating, covering a range of 0% to 70%. Its coverage includes ranges where neither the fourth nor fifth regression equations described above can accurately estimate the refrigerant shortage rate. Figure 9BAs shown, the refrigerant shortage rate calculation formula during heating is a formula used to continuously connect the refrigerant shortage rate estimated as the result of the fourth regression equation and the refrigerant shortage rate estimated as the result of the fifth regression equation using an S-curve with an S-shaped coefficient. Specifically, the refrigerant shortage rate calculation formula during heating is: (refrigerant shortage rate obtained by S-shaped coefficient × fifth regression equation) + (refrigerant shortage rate obtained by (1 - S-shaped coefficient) × fourth regression equation). The refrigerant dosage estimation unit 45 substitutes the current operating status quantity obtained by the acquisition unit 41 into the fourth regression equation and the fifth regression equation to calculate the refrigerant shortage rate respectively, and substitutes the refrigerant shortage rate into the refrigerant shortage rate calculation formula during heating to calculate the current refrigerant shortage rate of the refrigerant circuit 6.

[0095] Here, as with cooling operation, the S-coefficient is calculated using any of the operating state variables. In this embodiment, considering that when the outdoor unit expansion valve 14 is fully closed (0) / fully open (100), if the outdoor unit expansion valve 14 is fully open, the result based on the fourth regression equation remains almost unchanged. Therefore, the calculation formula is set to 0.5 for the S-coefficient when the outdoor unit expansion valve 14 is 90°.

[0096] p = 1 / (1 + exp(-(D / 10-45)))

[0097] p: S-type coefficient

[0098] D: Opening degree of outdoor unit expansion valve 14

[0099] Therefore, by determining the S-type coefficient and applying it to the third heating estimation model 45A6, when the refrigerant shortage rate is 0% to 20%, that is, when the refrigerant shortage rate is in the third range, the estimated value of the first heating estimation model 45A4 dominates among the estimated values ​​based on the third heating estimation model 45A6. Furthermore, when the refrigerant shortage rate is 30% to 70%, that is, when the refrigerant shortage rate is in the fourth range, the estimated value of the second heating estimation model 45A5 dominates among the estimated values ​​based on the third heating estimation model 45A6.

[0100] Furthermore, the calculation of the S-type coefficient is not limited to the above method. The S-type coefficient can be determined as follows: when the actual refrigerant shortage rate is above 20%, that is, when the actual refrigerant shortage rate is not within the third range, the estimated value of the second heating estimation model 45A5 among the estimated values ​​based on the third heating estimation model 45A6 becomes dominant; and when the actual refrigerant shortage rate is below 30%, that is, when the actual refrigerant shortage rate is not within the fourth range, the estimated value of the first heating estimation model 45A4 among the estimated values ​​based on the third heating estimation model 45A6 becomes dominant.

[0101] As described above, during refrigeration operation, the refrigerant deficiency rate is estimated using the first regression equation, the second regression equation, and the formula for calculating the refrigerant deficiency rate during refrigeration. The refrigerant subcooling during refrigeration is greater than the first threshold (…). Figure 8A and Figure 8B When considering the value of Tv1, choosing the first regression equation provides a more accurate estimate of the refrigerant under-refrigerant rate than choosing the second regression equation. Furthermore, when the refrigerant subcooling during cooling is less than a first threshold, choosing the second regression equation provides a more accurate estimate of the refrigerant under-refrigerant rate than choosing the first regression equation. Moreover, when the refrigerant subcooling during cooling is near the first threshold, the estimated refrigerant under-refrigerant rate changes significantly depending on which regression equation is used. Therefore, during cooling, a refrigerant under-refrigerant rate calculation formula that includes both the first and second regression equations is selected. This allows for a highly accurate estimate of the refrigerant under-refrigerant rate during cooling.

[0102] Furthermore, during heating operation, the refrigerant shortage rate is estimated using the fourth regression equation, the fifth regression equation, and the formula for calculating the refrigerant shortage rate during heating. During heating, the opening degree of the outdoor unit expansion valve 14 is less than the second threshold (…). Figure 9A and Figure 9B In the case of TV2), choosing the fourth regression equation provides a more accurate estimate of the refrigerant shortage rate than choosing the fifth regression equation. Furthermore, when the opening degree of the outdoor unit expansion valve 14 during heating is above the second threshold, choosing the fifth regression equation provides a more accurate estimate of the refrigerant shortage rate than choosing the fourth regression equation. Moreover, when the opening degree of the outdoor unit expansion valve 14 during heating is near the second threshold, the estimated value of the refrigerant shortage rate changes significantly depending on which regression equation is used. Therefore, during heating, the formula for calculating the refrigerant shortage rate during heating, which includes both the fourth and fifth regression equations, is selected. This allows for a highly accurate estimate of the refrigerant shortage rate during heating.

[0103] Structure of anomaly inference model

[0104] The anomaly estimation model 46A is generated using the value of the second characteristic quantity, i.e., the simulated value. The value of the second characteristic quantity is derived from a simulation of the operation of the refrigerant circuit 6 when its operation is normal and only the residual refrigerant charge is changed. The anomaly estimation unit 46 applies the detection values ​​of the second characteristic quantity of each combination P1 to P4 obtained from the operating air conditioner 1 to the anomaly estimation model 46A to estimate whether the detection value of the second characteristic quantity of each combination P1 to P4 is abnormal or normal. That is, when the detection value of the second characteristic quantity of each combination P1 to P4 is abnormal, the anomaly estimation unit 46 estimates that the refrigerant circuit 6 of each combination P1 to P4 is abnormal. When the detection value of the second characteristic quantity of each combination P1 to P4 is normal, the anomaly estimation unit 46 estimates that the refrigerant circuit 6 of each combination P1 to P4 is normal.

[0105] In the generation of the anomaly inference model 46A, kernel density estimation is employed, for example. Kernel density estimation is a method that infers the overall distribution based on a finite number of sample points. The anomaly inference model 46A calculates the degree of deviation (hereinafter also referred to as outlier) from the density function of the overall distribution inferred from the finite number of sample points, relative to the maximum value of the density function (the center of a cluster). Furthermore, after the data to be judged is input, the anomaly inference model 46A calculates the outlier of that data and determines whether the outlier is within a specified range (whether the data to be judged is contained in a cluster).

[0106] Figure 10 This is an illustrative diagram illustrating an example of the distribution method of the detection values ​​of the second feature of the anomaly inference model 46A. (See diagram for example.) Figure 10As shown, the anomaly prediction model 46A clusters and classifies the sets of second characteristic quantity values ​​(hereinafter also referred to as "simulated values ​​of the second characteristic quantity") obtained through simulation of the refrigerant circuit 6 under normal, stable, and refrigerant leakage conditions as normal. The detected value of the second characteristic quantity under stable conditions is the detected value of the second characteristic quantity after simulating the operation of the refrigerant circuit 6 under normal conditions. The simulation conditions are that the refrigerant circuit 6 is under normal, stable, or in a state of reduced refrigerant charge (refrigerant leakage state). The simulated value of the second characteristic quantity under normal conditions is the value of the second characteristic quantity obtained under the assumption that the simulation is performed under the condition that all components constituting the air conditioner 1 (refrigerant circuit 6, compressor, expansion valve, etc.) are working normally. In addition, the simulated value of the second characteristic quantity under the condition of refrigerant leakage is the value of the second characteristic quantity obtained under the assumption that the simulation is performed under the condition that all components constituting the air conditioner (refrigerant circuit 6, compressor, expansion valve, etc.) are working normally, and only the amount of refrigerant remaining in the refrigerant circuit 6 changes (decreases). Furthermore, if the input is a detection value for a second feature that does not belong to a cluster classified as normal by the anomaly inference model 46A, the detection value is classified as an anomaly. Additionally, as... Figure 10 As shown, when plotting the detection values ​​on a graph, the detection values ​​classified as abnormal are those that deviate from the clusters classified as normal. Furthermore, abnormality indicates a higher probability of a malfunction in the device constituting refrigerant circuit 6.

[0107] The anomaly prediction model 46A converts the difference between the values ​​of the second characteristic quantities obtained through simulation of the refrigerant circuit 6 under normal stable and refrigerant leakage conditions and the detection values ​​of the second characteristic quantities of each combination P1 to P4 obtained from the operating air conditioner 1 into numerical values, and calculates outliers. Specifically, the anomaly prediction model 46A uses the values ​​of the second characteristic quantities used when generating the anomaly prediction model 46A as normal sample values ​​(clusters classified as normal), and calculates outliers representing the degree of deviation from the normal sample values ​​for the detection values ​​of the second characteristic quantities of each group obtained by the acquisition unit 41 of the operating air conditioner 1. The outlier value is a numerical value converted from the distance of deviation from the boundary of the cluster classified as normal; the larger the absolute value of the value, the greater the degree of deviation. As the degree of deviation increases, the probability of the detection value of the second characteristic quantity being abnormal increases.

[0108] Figure 11This is an explanatory diagram illustrating an example of anomaly detection based on outliers. The anomaly estimation unit 46 classifies the detected value of the second feature quantity as normal if the absolute value of the outlier is, for example, less than -150, and classifies it as abnormal if the absolute value of the outlier is, for example, greater than or equal to -150. Furthermore, the deviation threshold X is a value determined based on the results of actually verifying values ​​judged as abnormal after collecting the fault history records of the air conditioner 1; it represents the degree to which normal data will not be mistakenly classified as abnormal. The anomaly estimation unit 46 classifies the detected value of the second feature quantity as abnormal if the calculated absolute value of the outlier is greater than or equal to the absolute value of the deviation threshold X.

[0109] If the detection value of the second characteristic quantity is classified as abnormal, the anomaly estimation unit 46 does not perform the refrigerant shortage rate estimation operation performed by the refrigerant dosage estimation unit 45, which uses the detection value of the first characteristic quantity acquired simultaneously with the detection value of the second characteristic quantity. Furthermore, the anomaly estimation unit 46 stores the detection value of the second characteristic quantity classified as abnormal as an anomaly in the anomaly record storage unit 43A.

[0110] If the absolute value of the estimated outlier is less than the absolute value of the deviation threshold X, the anomaly estimation unit 46 classifies the detected value of the second characteristic quantity as normal. In this case, the anomaly estimation unit 46 uses the detected value of the first characteristic quantity, which is acquired simultaneously with the detected value of the second characteristic quantity, to perform the refrigerant shortage rate estimation operation performed by the refrigerant dosage estimation unit 45. Furthermore, the anomaly estimation unit 46 classifies the detected value of the second characteristic quantity as normal even if only the refrigerant leakage state changes.

[0111] Furthermore, for ease of explanation, a case is shown where the deviation threshold X is set to, for example, "-150", but it can also be appropriately adjusted based on the results of actually verifying the values ​​determined to be abnormal after collecting fault history records.

[0112] Figure 12This is an explanatory diagram illustrating an example of the judgment result of the judgment unit 46D. The anomaly estimation unit 46 outputs an estimation result classifying the detected values ​​of the second characteristic quantities of each combination P1 to P4 as abnormal or normal. The judgment unit 46D stores the estimation results of the detected values ​​of the second characteristic quantities of each combination P1 to P4. The judgment unit 46D determines whether the estimation results of the detected values ​​of the second characteristic quantities of the combinations P1 to P4 are abnormal. If the detected values ​​of the second characteristic quantities of each combination P1 to P4 are abnormal, for example, if the detected values ​​of the second characteristic quantities of all combinations P1 to P4 are abnormal, the judgment unit 46D determines that the anomaly of the refrigerant circuit 6 is caused by the outdoor unit 2 common to all combinations P1 to P4. If the detected values ​​of the second characteristic quantities of each combination P1 to P4 are abnormal, and only the detected values ​​of the second characteristic quantities of some combinations are abnormal, the judgment unit 46D determines that the anomaly of the refrigerant circuit 6 is caused by the indoor unit 3 of the abnormal combination.

[0113] exist Figure 12 In, for example, if the detection values ​​of the second characteristic quantity of combinations P1, P2, and P4 are normal, but the detection value of the second characteristic quantity of combination P3 is abnormal, the judgment unit 46D determines that the abnormality of the refrigerant circuit 6 is caused by the indoor unit 3C of combination P3. Furthermore, although in Figure 12 The diagram is not shown. For example, if the detection values ​​of the second characteristic quantities of combination P1 and combination P2 are presumed to be normal, while the detection values ​​of the second characteristic quantities of combination 3 and combination 4 are abnormal, the judgment unit 46D determines that the abnormality of the refrigerant circuit 6 is caused by the indoor unit 3C of combination P3 and the indoor unit 3D of combination P4.

[0114] Presumed treatment actions

[0115] Figure 13This is a flowchart illustrating an example of the processing operation of the control circuit 19 involved in the estimation process. Furthermore, the refrigerant dosage estimation unit 45 within the control circuit 19 holds: a pre-generated first refrigerant estimation model 45A1, a second refrigerant estimation model 45A2, a third refrigerant estimation model 45A3, a first heating estimation model 45A4, a second heating estimation model 45A5, and a third heating estimation model 45A6. Further, the anomaly estimation unit 46 within the control circuit 19 holds a pre-generated cooling anomaly estimation model 46B and a heating anomaly estimation model 46C. For example, during a preset time period once a day (e.g., nighttime), estimation processing is periodically performed on the operating status quantities detected sequentially by the detection unit every 10 minutes, for a total of 24 hours. Furthermore, nighttime is shown as an example of a preset time period; for example, the operating status quantities for one day can also be obtained after the air conditioner 1 stops operating during the nighttime period when the air conditioner 1 operates less frequently. Furthermore, the preset time period can be determined not at night, but rather based on, for example, the operating status of the air conditioner 1 over a month.

[0116] exist Figure 13 In this process, the control unit 44 within the control circuit 19 collects operating state quantities as operating data through the acquisition unit 41 (step S11). The control unit 44 performs data filtering processing, that is, extracts arbitrary operating state quantities from the collected operating data (step S12). The control unit 44 performs data cleaning processing (step S13). Further, the anomaly estimation unit 46 performs anomaly estimation processing for each combination P1 to P4, that is, using the anomaly estimation model 46A, classifies the detection value of the second feature quantity after data cleaning processing as normal or abnormal (step S14). In the anomaly estimation processing, the anomaly estimation model 46A is used to estimate the classification result of each combination P1 to P4 as abnormal or normal.

[0117] The control unit 44 determines whether there are any abnormalities in the detection values ​​of the second characteristic quantities of each combination P1 to P4 (step S15). If there are no abnormalities in the detection values ​​of the second characteristic quantities of each combination P1 to P4 (step S15: No), the anomaly estimation unit 46 performs residual refrigerant dosage estimation processing, that is, it applies the detection values ​​of the first characteristic quantities obtained simultaneously with the detection values ​​of the second characteristic quantities of the combinations classified as normal to each refrigerant dosage estimation model (step S16). Then, the refrigerant dosage estimation unit 45 calculates the refrigerant shortage rate of the refrigerant circuit 6 (step S17) and ends the process. Figure 13 The processing actions shown.

[0118] Furthermore, if the detection value of the second characteristic quantity of each combination P1 to P4 is abnormal (step S15: Yes), the judgment unit 46D within the anomaly estimation unit 46 determines that the refrigerant circuit 6 is abnormal, and judges whether the detection values ​​of the second characteristic quantities of all combinations P1 to P4 are abnormal (step S18). If the detection values ​​of the second characteristic quantities of all combinations P1 to P4 are all abnormal (step S18: Yes), the judgment unit 46D determines that the cause of the abnormality in the refrigerant circuit 6 is an abnormality in the outdoor unit 2 (step S19). Then, the anomaly estimation unit 46 performs anomaly output processing (step S20) and ends. Figure 13 The processing actions shown are as follows. As a result, the anomaly estimation unit 46 is able to determine that the cause of the anomaly in the refrigerant circuit 6 is an anomaly in the outdoor unit 2.

[0119] If the detection values ​​of the second characteristic quantity of all combinations P1 to P4 are not all abnormal (step S18: No), the determination unit 46D determines that the detection values ​​of the second characteristic quantity of only some combinations are abnormal (step S21). Furthermore, if the determination unit 46D determines that the detection values ​​of the second characteristic quantity of only some combinations are abnormal, as described above, it can identify the combination where the abnormality occurs. Further, if the determination unit 46D determines that the cause of the abnormality in the refrigerant circuit 6 is an abnormality in the indoor unit 3 of the combination identified as abnormal (step S22), it returns to step S20 to perform abnormality output processing. As a result, the abnormality estimation unit 46 is able to identify the indoor unit 3 from among the multiple indoor units 3 that is the cause of the abnormality in the refrigerant circuit 6.

[0120] The data filtering process, based on preset filtering conditions, extracts only a portion of the operating state quantities (the detection values ​​of the first and second characteristic quantities) required for anomaly estimation and refrigerant shortage rate calculation from multiple operating state quantities, rather than using all of them. By substituting the detection values ​​of the first and second characteristic quantities (after the data filtering process described later, which removes outliers or escape values) into the generated refrigerant charge estimation model 45A or anomaly estimation model 46A, it is possible to more accurately estimate anomalies using the second characteristic quantity or estimate the refrigerant shortage rate using the first characteristic quantity.

[0121] The preset filtering conditions include a first filtering condition, a second filtering condition, and a third filtering condition. The first filtering condition, for example, is a filtering condition for data extracted under all operating modes of the air conditioner 1. The second filtering condition is a filtering condition for data extracted during cooling operation. The third filtering condition is a filtering condition for data extracted during heating operation.

[0122] The first filtering conditions include, for example, the driving state of compressor 11, identification of operating modes, exclusion of special operations, exclusion of missing values ​​in the acquired values, and selection of values ​​with smaller changes for operating state quantities that have a significant impact on the generation of various regression equations. The driving state of compressor 11 is a condition that needs to be judged because unless the compressor operates stably and refrigerant circulates in the refrigerant circuit 6, it is impossible to deduce the refrigerant shortage rate. Furthermore, the driving state of compressor 11 is a filtering condition set to remove operating state quantities detected during the transition period, such as when compressor 11 starts up.

[0123] Operating mode identification is a filter condition used to extract only the operating status quantities acquired during cooling and heating operations. Therefore, operating status quantities acquired during dehumidification or air supply operations are removed. Special operation exclusion is a filter condition used to remove operating status quantities acquired during special operations, such as refrigerant oil recovery or defrosting operations, where the state of refrigerant circuit 6 differs significantly from that during cooling or heating operations. Defect value exclusion refers to a filter condition that removes operating status quantities containing defect values ​​from the operating status quantities used to determine refrigerant deficiency rate, as using these quantities to generate regression equations might decrease accuracy.

[0124] Selecting values ​​with small changes for the operating state variables substituted into each regression equation or refrigerant deficiency rate calculation formula, and only extracting the operating state variables when the air conditioner 1 is in a stable state, is a necessary condition to improve the estimation accuracy based on each regression equation and refrigerant deficiency rate calculation formula. Furthermore, operating state variables with significant impact include: the refrigerant subcooling used when the refrigerant deficiency rate is 0-30% during cooling operation, the suction temperature used when the refrigerant deficiency rate is 40-70% during cooling operation, and the suction superheat during heating operation.

[0125] Secondary filtration conditions include, for example, the elimination of heat exchange outlet temperature, abnormal subcooling, and abnormal discharge temperature.

[0126] The exclusion of heat exchange outlet temperature is a filtering condition that takes into account the fact that the external air temperature sensor 36 and the refrigerant temperature sensor 35 are located close to each other, so that the heat exchange outlet temperature detected by the refrigerant temperature sensor 35 will not be lower than the external air temperature detected by the external air temperature sensor 36 during cooling operation. It is a filtering condition used to remove heat exchange outlet temperatures that are lower than the external air temperature.

[0127] Subcooling anomaly is a filter condition that removes a refrigerant subcooling that is detected due to an extremely high or low refrigeration load. Discharge temperature anomaly is a filter condition that removes a discharge temperature detected under what is commonly referred to as a gas shortage condition, which is a condition in which the amount of refrigerant drawn into the compressor 11 is reduced due to a low refrigeration load.

[0128] The third filtering condition is, for example, an abnormal discharge temperature. If the discharge temperature becomes high due to a large heating load during heating operation, and discharge temperature protection control is executed, the discharge temperature is reduced, for example, by reducing the speed of compressor 11, thus removing the filtering condition of the detected discharge temperature.

[0129] Data cleaning is used to remove detection values ​​of the first characteristic quantity that pose a risk of erroneous inferences, rather than using all the acquired detection values ​​of the first characteristic quantity to infer the refrigerant shortage rate. Furthermore, data cleaning is used to remove detection values ​​of the second characteristic quantity that pose a risk of erroneous inferences, rather than using all the acquired detection values ​​of the second characteristic quantity for abnormal inference processing. Specifically, this includes smoothing the acquired operating status quantities to suppress noise and limiting the amount of data. Smoothing the data to suppress noise means calculating the average value of the interval to derive moving averages for, for example, refrigerant subcooling, suction temperature, and suction superheat in each model, thereby suppressing noise. Limiting the amount of data means, for example, removing data with a small number of data points because of their lower reliability. For example, if filtering the input data for a day leaves more than X data points, these are used for inferring the refrigerant shortage rate or for abnormal inference processing of the second characteristic quantity; if there are fewer than X data points, all data from that day are not used. In other words, in the data cleaning process, by substituting the operating status quantities after removing outliers and escape values ​​into the refrigerant dosage estimation model 45A, the refrigerant shortage rate can be estimated more accurately; by substituting the operating status quantities after removing outliers and escape values ​​into the anomaly estimation model 46A, anomaly estimation can be performed more accurately.

[0130] Anomaly inference processing is a process that calculates the degree of deviation (outlier value) of the overall distribution density function estimated from the simulated value of the second feature quantity, relative to the maximum value (center of the cluster) of the density function, and judges whether the outlier value is within a specified range (whether the data to be judged is included in the cluster). The second feature quantities of each combination P1 to P4 obtained from the operating air conditioner 1 are applied to this anomaly inference model 46A to calculate the outlier value. In the anomaly inference processing, the value of the second feature quantity used when generating the anomaly inference model 46A is assumed to be a normal sample value, thereby calculating the outlier value of the detected value of the second feature quantity of each combination P1 to P4 obtained by the acquisition unit 41 at different times relative to the normal sample value. Further, in the anomaly inference processing, when the absolute value of the calculated outlier value is above the absolute value of the deviation threshold X, the detected value of the second feature quantity of that combination is classified as anomaly. Further, in the anomaly inference processing, when the absolute value of the calculated outlier value is less than the absolute value of the deviation threshold X, the detected value of the second feature quantity of that combination is classified as normal.

[0131] Based on the classification results of each combination P1 to P4, the judgment unit 46D can determine whether the indoor unit 3 or the outdoor unit 2 is the main cause of the abnormality in the refrigerant circuit 6. If the detection values ​​of the second characteristic quantity of all combinations P1 to P4 are abnormal, the judgment unit 46D determines that the main cause of the abnormality in the refrigerant circuit 6 is the abnormality of the outdoor unit 2. Furthermore, if the detection values ​​of the second characteristic quantity of some combinations are abnormal, the judgment unit 46D determines that the main cause of the abnormality in the refrigerant circuit 6 is the abnormality of the indoor unit 3, which is classified as an abnormal combination.

[0132] Figure 14 This is a flowchart illustrating an example of the processing actions of the control circuit 19 involved in the residual refrigerant dosage estimation process. The estimation of the residual refrigerant dosage is performed as follows: for example, by substituting the detection value of the first characteristic quantity, obtained simultaneously with the detection value of the second characteristic quantity classified as normal by the anomaly estimation process, from the current operating status quantity (sensor value) after data filtering and cleaning, into the regression equations or refrigerant shortage rate calculation formulas of the refrigerant dosage estimation model 45A, the current refrigerant shortage rate of the refrigerant loop 6 is calculated. Figure 14 In the process, the refrigerant dosage estimation unit 45 within the control circuit 19 determines whether the acquired first characteristic quantity is a characteristic quantity acquired during the refrigeration operation (step S31). If the acquired first characteristic quantity is a characteristic quantity acquired during the refrigeration operation (step S31: Yes), the refrigerant dosage estimation unit 45 applies the first characteristic quantity to the first refrigeration estimation model 45A1 to the third refrigeration estimation model 45A3 respectively (step S32).

[0133] If the first characteristic quantity acquired by the refrigerant dosage estimation unit 45 is not acquired during refrigeration operation (step S31: No), that is, if the first characteristic quantity acquired is acquired during heating operation, the unit applies the first characteristic quantity to the first heating estimation model 45A4 to the third heating estimation model 45A6 respectively (step S33). Then, the refrigerant dosage estimation unit 45 integrates the results obtained by applying the first characteristic quantity to the first refrigeration estimation model 45A1 to the third refrigeration estimation model 45A3 respectively, and the results obtained by applying the first characteristic quantity to the first heating estimation model 45A4 to the third heating estimation model 45A6 respectively, thereby calculating the current refrigerant shortage rate (step S34), and then ends the process. Figure 14 The processing actions shown.

[0134] The anomaly output processing stores the detection value of the second feature quantity, which is classified as an anomaly in the anomaly induction processing, as an anomaly record in the anomaly record storage unit 43A and outputs an alarm. As a result, it is able to store the detection value of the second feature quantity of the anomaly.

[0135] Effects of Example 1

[0136] In the air conditioner 1 of Example 1, the value of the second feature quantity used when generating the anomaly estimation model 46A is set as a normal sample value, thereby calculating the outlier values ​​of the detected values ​​of the second feature quantities of each combination P1 to P4 obtained at different times relative to the normal sample values. Further, in the air conditioner 1, when the absolute value of the calculated outlier value is above the absolute value of the deviation threshold X, the detected value of the second feature quantity of that combination is classified as an anomaly, and an anomaly in the refrigerant circuit 6 is inferred. Further, the air conditioner 1 does not use the detected value of the first feature quantity obtained simultaneously with the detected value of the second feature quantity of the combination classified as an anomaly for the refrigerant dosage estimation model 45A. As a result, the refrigerant shortage rate of the refrigerant circuit 6 can be accurately estimated.

[0137] Based on the classification results of each combination P1 to P4, air conditioner 1 can determine whether the indoor unit 3 or outdoor unit 2 is the primary cause of the abnormality in refrigerant circuit 6. If the detection values ​​of the second characteristic quantity for all combinations P1 to P4 are abnormal, air conditioner 1 determines that the primary cause of the abnormality in refrigerant circuit 6 is the abnormality in outdoor unit 2. Furthermore, if the detection values ​​of the second characteristic quantity for some combinations are abnormal, air conditioner 1 determines that the primary cause of the abnormality in refrigerant circuit 6 is the abnormality in indoor unit 3, which is classified as an abnormal combination. As a result, even if an abnormality other than a change in residual refrigerant charge is presumed, it is possible to presume which of the outdoor unit 2 and indoor unit 3 is experiencing the abnormality.

[0138] For example, when estimating the refrigerant shortage rate using the refrigerant dosage estimation model 45A generated by linear analysis of multiple regression analysis, a malfunction other than refrigerant leakage may occur simultaneously, causing a change in the first characteristic quantity. In this case, depending on the degree of change in each characteristic quantity, it is possible that a situation that should have resulted in a larger refrigerant shortage rate (=abnormality) may be estimated as a smaller refrigerant shortage rate. For example, it is possible that the compressor speed and suction temperature change due to a malfunction other than refrigerant leakage, and the result of these changes canceling each other out leads to the refrigerant shortage rate being estimated as a smaller value (=normal). However, in the air conditioner 1 of this embodiment, the detection value of the first characteristic quantity, which is obtained simultaneously with the detection value of the second characteristic quantity classified as abnormal by the anomaly estimation model 46A, is not used in the refrigerant dosage estimation model 45A, where the anomaly estimation model 46A is generated by nonlinear analysis such as kernel density estimation. As a result, it is possible to prevent the estimation of an incorrect refrigerant shortage rate.

[0139] Furthermore, it is possible that, when using the refrigerant dosage estimation model 45A generated by linear analysis, a value that is actually a small refrigerant shortage rate (=normal) may be estimated as a large refrigerant shortage rate (=abnormal). For example, a change in compressor speed due to a fault other than refrigerant leakage may result in an estimated large refrigerant shortage rate. However, in the air conditioner 1 of this embodiment 1, the detection value of the first feature quantity, which is obtained simultaneously with the detection value of the second feature quantity classified as abnormal by the abnormality estimation model 46A, is not used in the refrigerant dosage estimation model 45A, whereby the abnormality estimation model 46A is generated through nonlinear analysis. As a result, it is possible to prevent the estimation of an incorrect refrigerant shortage rate.

[0140] In the anomaly prediction model 46A of air conditioner 1, when the absolute value of the calculated outlier is less than the absolute value of the deviation threshold X, the detection value of the second characteristic quantity of the combination is classified as normal. Furthermore, in air conditioner 1, by performing multivariate regression analysis on the detection values ​​of the first characteristic quantity obtained simultaneously with the detection values ​​of the second characteristic quantity of the combination classified as normal, the refrigerant shortage rate of refrigerant circuit 6 is calculated. As a result, the refrigerant shortage rate of refrigerant circuit 6 can be accurately estimated.

[0141] The anomaly estimation model 46A, installed in the air conditioner 1, is generated using a portion of the detected values ​​of the first characteristic quantity used in the refrigerant dosage estimation model 45A, along with the values ​​of the second characteristic quantity, which includes operating state quantities that significantly affect the refrigeration cycle operation, through nonlinear analysis such as kernel density estimation. In the anomaly estimation model 46A, the detected values ​​of the second characteristic quantities for each combination P1 to P4 are classified as normal or abnormal. Furthermore, for the refrigerant dosage estimation model 45A, not all operating state quantities are used; instead, the detected values ​​of the first characteristic quantity, obtained simultaneously with the detected values ​​of the second characteristic quantity classified as normal, are used to generate the refrigerant dosage estimation model 45A. As a result, a high-precision refrigerant dosage estimation model 45A can be generated.

[0142] In this embodiment, the generation of each regression equation in the refrigerant dosage estimation model 45A uses the detection value of the first characteristic quantity obtained through simulation. This first characteristic quantity does not contain abnormal values ​​or values ​​that are significantly larger or smaller than others. The detection values ​​of the operating status quantity, after undergoing data filtering and cleaning to remove outliers and escape values, are substituted into each regression equation or refrigerant shortage rate calculation formula of the refrigerant dosage estimation model 45A generated using this simulation-obtained characteristic quantity. At this point, by substituting only the detection value of the first characteristic quantity obtained simultaneously with the detection value of the second characteristic quantity classified as normal by the anomaly estimation model 46A, the refrigerant shortage rate can be estimated more accurately.

[0143] The anomaly prediction model 46A is generated using feature quantities obtained through simulation, which do not contain outlier values ​​or values ​​that are significantly large or small compared to other values. By applying the detection value of the second feature quantity, which has undergone data filtering and cleaning to remove outliers and escape values, to this anomaly prediction model 46A generated using feature quantities that do not contain outliers or escape values, the detection value of the second feature quantity can be accurately judged. Furthermore, in the control circuit 19, by performing data filtering and data cleaning, the amount of data used when calculating outliers through the anomaly prediction model 46A can be reduced. Therefore, the time spent calculating outliers through the anomaly prediction model 46A can be shortened, and the load on the control circuit 19 can be reduced.

[0144] Furthermore, in Embodiment 1 described above, the following situation was shown: during the design phase of the air conditioner 1, simulation results of various operating state quantities were obtained; by having an information processing device such as a server with learning capabilities learn the simulation results, a refrigerant dosage estimation model 45A and an anomaly estimation model 46A were obtained and stored in the control circuit 19. Alternatively, a server 120 may exist, connected to the air conditioner 1 via a communication network 110. This server 120 generates the refrigerant dosage estimation model 45A and the anomaly estimation model 46A, and sends the estimation results of the refrigerant dosage estimation model 45A and the anomaly estimation model 46A to the air conditioner 1. This embodiment will be described below.

[0145] Example 2

[0146] Structure of air conditioning system

[0147] Figure 15 This is an explanatory diagram illustrating an example of the air conditioning system 100 of Embodiment 2. Furthermore, the same symbols are used to denote structures identical to those in the air conditioner 1 of Embodiment 1, thus omitting descriptions of repetitive structures and operations. Figure 15 The air conditioning system 100 shown includes: an air conditioner 1, a communication network 110, and a server 120. The air conditioner 1 includes an outdoor unit 2, an indoor unit 3, and a control circuit 19A. The outdoor unit 2 includes a compressor 11, an outdoor heat exchanger 13, and an outdoor unit expansion valve 14. The indoor unit 3 includes an indoor heat exchanger 51. The air conditioner 1 has a refrigerant circuit 6 formed by connecting the outdoor unit 2 and the indoor unit 3 through refrigerant piping such as liquid pipes 4 and gas pipes 5. This refrigerant circuit 6 is filled with a preset amount of refrigerant. The control circuit 19A includes: an acquisition unit 41, a communication unit 42, a storage unit 43, and a control unit 44. However, the control circuit 19A does not include a refrigerant dosage estimation unit 45, an anomaly estimation unit 46, or an anomaly record storage unit 43A.

[0148] Server 120 includes a generation unit 121, a communication unit 121A, a refrigerant dosage estimation unit 122, an anomaly estimation unit 123, and a storage unit 124. Storage unit 124 includes an anomaly record storage unit 124A. Generation unit 121 uses the detected value or simulated value of a first characteristic quantity related to the estimation of the refrigerant shortage rate of the refrigerant filled in the refrigerant circuit 6, and generates a refrigerant dosage estimation model 45A using a multiple regression analysis method. Furthermore, the refrigerant dosage estimation model 45A includes, for example, the following models described in the first embodiment: a first refrigerant estimation model 45A1, a second refrigerant estimation model 45A2, a third refrigerant estimation model 45A3, a first heating estimation model 45A4, a second heating estimation model 45A5, and a third heating estimation model 45A6. Refrigerant dosage estimation unit 122 stores the refrigerant dosage estimation model 45A generated by generation unit 121. Furthermore, the generation unit 121 uses the detection values ​​of the second characteristic quantity of all combinations P1 to P4 obtained through simulation under steady state and refrigerant leakage state to generate an anomaly estimation model 46A by kernel density estimation. In addition, the anomaly estimation model 46A includes, for example, the anomaly estimation model 46B during cooling and the anomaly estimation model 46C during heating described in Example 1.

[0149] The anomaly estimation unit 123 stores the anomaly estimation model 46A generated by the generation unit 121. The anomaly estimation unit 123 uses the anomaly estimation model 46A to classify the detected value of the second characteristic quantity as normal or abnormal. If the detected value of the second characteristic quantity is classified as abnormal, the anomaly estimation unit 123 stores the detected value of the classified second characteristic quantity as an anomaly record in the anomaly record storage unit 124A. Further, the judgment unit 46D within the anomaly estimation unit 123 determines the indoor unit 3 or outdoor unit 2 as the main cause of the anomaly in the refrigerant circuit 6 based on the classification results of the anomaly estimation unit 123, i.e., the classification results of each combination P1 to P4. The communication unit 121A sends the determination result of the judgment unit 46D regarding the indoor unit 3 or outdoor unit 2 as the main cause of the anomaly in the refrigerant circuit 6 to the air conditioner 1 via the communication network 110. The control circuit 19A of the air conditioner 1 can determine the main cause of the abnormality in the refrigerant circuit 6 based on the determination result of the indoor unit 3 or the outdoor unit 2, which is the main cause of the abnormality in the refrigerant circuit 6, received from the server 120.

[0150] Furthermore, the refrigerant dosage estimation unit 122 uses the detection value of the first characteristic quantity, which is obtained simultaneously with the detection value of the normal second characteristic quantity classified by the anomaly estimation model 46A, and the received refrigerant dosage estimation model 45A, to calculate the refrigerant shortage rate in the refrigerant circuit 6 of the air conditioner 1. The communication unit 121A sends the refrigerant shortage rate calculated by the refrigerant dosage estimation unit 122 to the air conditioner 1 via the communication network 110. The control circuit 19A of the air conditioner 1 can determine the refrigerant shortage rate of the refrigerant circuit 6 based on the refrigerant shortage rate received from the server 120.

[0151] The generation unit 121 uses the values ​​of the second characteristic quantities obtained through simulation of the stable state of all combinations P1 to P4 under normal refrigerant circuit 6 during refrigeration and the value of the second characteristic quantity under refrigerant leakage state to generate or update the refrigeration anomaly estimation model 46B.

[0152] The generation unit 121 periodically collects operating status quantities during refrigeration operation from a standard unit of the air conditioner 1 (located in the manufacturer's laboratory, etc.) that can actually measure the stable state and refrigerant leakage state of the refrigerant circuit 6 under normal refrigeration conditions. It then uses the refrigeration anomaly estimation model 46B to compare the classification results of normal or abnormal conditions with the actual measured classification results, along with the collected operating status quantities, to generate or update the refrigeration anomaly estimation model 46B. As a result, a more accurate refrigeration anomaly estimation model 46B can be generated.

[0153] The generation unit 121 periodically collects operating status quantities during refrigeration operation from a standard unit of the air conditioner 1 (located in the manufacturer's laboratory, etc.) capable of actually measuring the refrigerant deficiency rate in the refrigerant circuit 6. Using the comparison results between the refrigerant deficiency rate estimated by each refrigerant dosage estimation model 45A and the actually measured refrigerant deficiency rate, along with the collected operating status quantities, it generates or updates the first refrigeration estimation model 45A1, the second refrigeration estimation model 45A2, and the third refrigeration estimation model 45A3. Alternatively, as in Embodiment 1, the operating status quantities used to generate each refrigerant dosage estimation model 45A can be obtained through simulation, and the generation unit 121 uses the operating status quantities obtained through simulation to generate each refrigerant dosage estimation model 45A.

[0154] The generation unit 121 uses the values ​​of the second characteristic quantities obtained through simulation of the stable state of all combinations P1 to P4 under normal refrigerant circuit 6 during heating and the refrigerant leakage state to generate or update the heating anomaly estimation model 46C.

[0155] The generation unit 121 periodically collects operating status data during heating operation from a standard unit of the air conditioner 1 (located in the manufacturer's laboratory, etc.) that can actually measure the stable state of the refrigerant circuit 6 under normal conditions and the refrigerant leakage state. It then uses the heating anomaly estimation model 46C to compare the classification results of normal or abnormal conditions with the actual measured classification results, along with the collected operating status data, to generate or update the heating anomaly estimation model 46C. As a result, a more accurate heating anomaly estimation model 46C can be generated.

[0156] The generation unit 121 periodically collects operating status data during heating operation from the standard unit of the aforementioned air conditioner 1, and uses the comparison results between the refrigerant shortage rate estimated by each refrigerant dosage estimation model 45A and the actual measured refrigerant shortage rate, as well as the collected operating status data, to generate a first heating estimation model 45A4, a second heating estimation model 45A5, and a third heating estimation model 45A6. Alternatively, as in Embodiment 1, operating status data for generating each refrigerant dosage estimation model 45A can be obtained through simulation, and the generation unit 121 uses the operating status data obtained through simulation to generate each refrigerant dosage estimation model 45A.

[0157] The anomaly prediction model 46A generated by the generation unit 121 uses feature quantities obtained through simulation. The values ​​of these feature quantities do not contain outlier values ​​or values ​​that are significantly large or small compared to other values. By applying the detection values ​​of the second feature quantities, which have undergone data filtering and cleaning to remove outliers and escape values, to this anomaly prediction model 46A generated using feature quantities that do not contain outliers or escape values, a more accurate judgment of the detection values ​​of the second feature quantities can be achieved. Furthermore, by performing the data filtering and data cleaning processes for the second feature quantities described in the first embodiment in the generation unit 121, the amount of data used when calculating outliers based on the anomaly prediction model 46A can be reduced. This shortens the time spent calculating outliers based on the anomaly prediction model 46A and reduces the utilization rate of the server 120, thus suppressing the cost of calculating outliers under a pay-as-you-go system, such as charging based on server 120 usage.

[0158] Effects of Example 2

[0159] In Example 2, the server 120 uses the values ​​of the second characteristic quantities of all combinations P1 to P4 under normal, stable, and refrigerant leakage states obtained from simulation to generate an anomaly estimation model 46A, and stores the generated anomaly estimation model 46A in the anomaly estimation unit 123. The anomaly estimation unit 123 within the server 120 can use the stored anomaly estimation model 46A to classify the detection values ​​of the second characteristic quantities acquired at different times for each combination P1 to P4 as normal or abnormal. Furthermore, the air conditioner 1 estimates whether the refrigerant circuit 6 of each combination P1 to P6 is abnormal or normal based on the classification results of the detection values ​​of the second characteristic quantities of each combination. Based on the estimation results of the anomalies in the refrigerant circuit 6 of each combination, the anomaly estimation unit 123 determines whether the outdoor unit 2 or indoor unit 3 is the main cause of the anomaly in the refrigerant circuit 6. The communication unit 121A sends the determination result of whether the outdoor unit 2 or indoor unit 3 is the main cause of the anomaly in the refrigerant circuit 6 to the air conditioner 1. As a result, air conditioner 1 is able to identify the outdoor unit 2 or indoor unit 3 as the main cause of the abnormality in refrigerant circuit 6.

[0160] Server 120 uses the value of the first characteristic quantity obtained from air conditioner 1 to generate a refrigerant dosage estimation model 45A, and stores the generated refrigerant dosage estimation model 45A in the refrigerant dosage estimation unit 122. Server 120 uses the stored refrigerant dosage estimation model 45A to estimate the refrigerant shortage rate, and sends the estimation result to air conditioner 1 via communication network 110. As a result, air conditioner 1 is able to identify the refrigerant shortage rate of refrigerant circuit 6.

[0161] Furthermore, in the air conditioner 1 of embodiments 1 and 2, a case is shown in which one outdoor unit 2 is connected to four indoor units 3, but it is not limited to four indoor units 3. As long as there are multiple indoor units 3, appropriate changes can be made.

[0162] Furthermore, this embodiment describes the case where a relative refrigerant charge is estimated to represent the refrigerant charge remaining in the refrigerant circuit 6. Specifically, it describes the case where a refrigerant shortage rate is estimated and provided, which is the ratio of the refrigerant charge leaking from the refrigerant circuit 6 to the amount of refrigerant added when filling the refrigerant circuit 6 (initial value). However, the present invention is not limited to this; the estimated refrigerant shortage rate can also be multiplied by the initial value to provide the refrigerant charge leaking from the refrigerant circuit 6 to the outside. In addition, an estimation model for estimating the absolute refrigerant charge leaking from the refrigerant circuit 6 to the outside or the absolute refrigerant charge remaining in the refrigerant circuit 6 can be generated, and the estimation result based on the estimation model can be provided. When generating an estimation model for estimating the absolute refrigerant charge leaking from the refrigerant circuit 6 to the outside or the absolute refrigerant charge remaining in the refrigerant circuit 6, in addition to the various operating state quantities described so far, the volumes of the outdoor heat exchanger 13 and each indoor heat exchanger 51, as well as the volume of the liquid pipe 4, need not be taken into account.

[0163] Variations

[0164] Furthermore, in this embodiment, for example, an interpolation is shown between the estimation result of the first refrigeration estimation model 45A1 and the estimation result of the second refrigeration estimation model 45A2 using an S-shaped coefficient. However, it is not limited to the S-shaped coefficient. For example, interpolation methods such as linear interpolation can also be used, and appropriate changes can be made.

[0165] In this embodiment, not all simulation results from multiple simulations are used; instead, only a portion of the simulation results are used. For example, separate models are generated, such as a first refrigeration estimation model 45A1 used when the refrigerant shortage rate is 0-30% during refrigeration operation, a second refrigeration estimation model 45A2 used when the refrigerant shortage rate is 40-70%, and a third refrigeration estimation model 45A3 used when the refrigerant shortage rate is 30-40%. Therefore, since the operating state quantities are prepared through simulation, the required amount of operating state quantities can be easily collected compared to the case where the air conditioner 1 is operated to collect operating state quantities.

[0166] In this embodiment, the case where the refrigerant dosage estimation model 45A and the anomaly estimation model 46A are generated via server 120 or control circuit 19 is shown. However, the user can also calculate the refrigerant dosage estimation model 45A and the anomaly estimation model 46A based on simulation results. Furthermore, this embodiment illustrates the use of multiple regression analysis to generate each estimation model. However, machine learning algorithms capable of performing common regression analysis, such as SVR (Support Vector Regression) or NN (Neural Network), can also be used to generate the estimation models. In this case, instead of the P-value and correction value R used in multiple regression analysis, common methods for selecting features (such as forward feature selection or backward feature elimination) can be used to improve the accuracy of the estimation model.

[0167] The diagram illustrates the scenario where the anomaly prediction model 46A is generated using the values ​​of the second characteristic quantity under both the stable state and refrigerant leakage states of all combinations in a normal refrigerant circuit 6, derived through simulation. It shows setting the values ​​of the second characteristic quantity for all combinations as normal sample values ​​and converting the distance between the detected value of the second characteristic quantity for each combination and the normal sample value into a numerical value to calculate outliers. However, it can also be generated using the values ​​of the second characteristic quantity for each combination in a normal refrigerant circuit 6, derived through simulation, with the second characteristic quantity for each combination used in the generation as a normal sample value, and converting the distance between the detected value of the second characteristic quantity for the same combination and the normal sample value for the same combination into a numerical value to calculate outliers, allowing for appropriate modifications.

[0168] Furthermore, it is shown that the anomaly estimation model 46A is generated using the values ​​of the second characteristic quantity obtained by simulation in the stable state and the refrigerant leakage state of the refrigerant circuit 6 under normal conditions. However, it is also possible to generate the model using only the value of the second characteristic quantity in the stable state instead of the value of the second characteristic quantity obtained by simulation in the refrigerant leakage state of the refrigerant circuit 6 under normal conditions.

[0169] Furthermore, this embodiment shows the use of kernel density estimation to generate anomaly inference model 46A, but it is not limited to kernel density estimation method. Any nonlinear analysis method can be used and appropriate modifications can be made.

[0170] Furthermore, this embodiment shows an example of an air conditioner 1 in which one outdoor unit 2 is connected to one or more indoor units 3, but it can also be applied to an air conditioner 1 in which two or more outdoor units 2 are connected to one or more indoor units 3.

[0171] Example 1 illustrates a scenario where, during the design phase of the air conditioner 1, simulation results of various operating state variables are obtained. These simulation results are then learned by an information processing device such as a server with learning capabilities, resulting in a refrigerant dosage estimation model 45A and an anomaly estimation model 46A, which are then stored in the control circuit 19. Alternatively, a server connected to the air conditioner 1 via a communication network can be used to generate the refrigerant dosage estimation model 45A and the anomaly estimation model 46A and send them to the air conditioner 1. Furthermore, the air conditioner 1 can also store the refrigerant dosage estimation model 45A and the anomaly estimation model 46A received from the server in the control circuit 19.

[0172] In refrigerant circuit 6, at least one indoor unit 3 is connected to at least one outdoor unit 2 via refrigerant piping. Therefore, the refrigerant charge estimation model 45A is a model capable of estimating the refrigerant shortage rate using the detection values ​​of a first characteristic quantity of a representative outdoor unit 2 from the at least one outdoor unit 2 and a representative indoor unit 3 from the at least one indoor unit 3. Furthermore, the representative outdoor unit 2 can be selected from the at least one operating outdoor unit 2 using any rule, and the representative indoor unit 3 can also be selected from the at least one operating indoor unit 3 using any rule. An arbitrary rule could be, for example, the ascending order of the identification numbers labeled on each machine.

[0173] Furthermore, each structural element of the various parts shown in the attached figures does not necessarily need to be physically constructed as shown in the figures. That is, the specific form of the distribution / combination of the various parts is not limited to that shown in the attached figures, and can be functionally or physically distributed or combined in any unit according to various loads or usage conditions.

[0174] Furthermore, the various processing functions performed by each device can also be executed, in whole or in part, on a microcomputer such as a CPU (Central Processing Unit) (or MPU (Micro Processing Unit), MCU (Micro Controller Unit), etc.). In addition, it is obvious that the various processing functions can also be executed, in whole or in part, on a program analyzed and executed by the CPU (or MPU, MCU, etc.), or on hardware using wiring logic.

[0175] Furthermore, in the embodiments described above, the refrigerant shortage rate is defined as the amount of refrigerant reduction calculated from the specified amount when the specified amount of refrigerant is filled to 100%. Alternatively, after the specified amount of refrigerant has just been filled into the refrigerant circuit 6, the refrigerant shortage rate can be estimated using the method described in this embodiment, and this estimated result can be taken as 100%. For example, if the estimated refrigerant shortage rate is 90% after the specified amount of refrigerant has just been filled into the refrigerant circuit 6, that is, if the amount of refrigerant currently filled into the refrigerant circuit 6 is estimated to be 10% less than the specified amount, this 10% less refrigerant amount can also be defined as 100%. By combining the refrigerant amount defined as 100% with the estimated result, the subsequent refrigerant shortage rate can be estimated more accurately.

[0176] Symbol Explanation

[0177] 1 air conditioner

[0178] 2 outdoor units

[0179] 3 indoor units

[0180] 41 Acquisition Department

[0181] 44 Control Department

[0182] 45. Refrigerant Dosage Estimation Section

[0183] 45A Refrigerant Estimation Model

[0184] 46. ​​Anomaly Prediction Department

[0185] 46A Anomaly Prediction Model

[0186] 46B Anomaly Prediction Model During Cooling

[0187] 46°C Heating Anomaly Prediction Model

[0188] 46D Judgment Department

[0189] 100 air conditioning system

[0190] 120 server

[0191] 121 Production Department

[0192] 121A Communications Department

[0193] 122 Cooling Dosage Estimation Section

[0194] 123 Abnormal Presumption Section.

Claims

1. An air conditioning system comprising an air conditioner and a server, the air conditioner having a refrigerant circuit connecting an outdoor unit to at least one indoor unit via refrigerant piping, the server being communicatively connected to the air conditioner, the air conditioning system being characterized in that... The air conditioner has the following features: The detection unit is used to detect state quantities related to the control of the air conditioner; The acquisition unit is used to acquire the detection value of the state quantity detected by the detection unit; and The first communication unit sends the detection value obtained by the acquisition unit to the server. The server has: A second communication unit, configured to receive the detected value from the air conditioner; and The anomaly estimation unit, when using the state quantity related to the anomaly in the refrigerant circuit as a characteristic quantity, uses the detection value of this characteristic quantity to estimate the occurrence of the anomaly in the refrigerant circuit. The anomaly estimation unit groups the outdoor unit and one indoor unit together, and estimates the occurrence of an anomaly in the refrigerant circuit for each such combination. If an anomaly is estimated in any combination, it is estimated that an anomaly has occurred in the indoor unit of that combination. Furthermore, if it is assumed that an anomaly has occurred in all combinations, it is assumed that an anomaly has occurred in the outdoor unit. The server has a cooling dose estimation unit that, when using the state quantity related to the cooling dose of the refrigerant circuit as a first characteristic quantity, uses the detection value of the first characteristic quantity to estimate the residual cooling dose remaining in the refrigerant circuit. When a state quantity that includes at least one state quantity contained in the first characteristic quantity and at least one state quantity not contained in the first characteristic quantity is used as a second characteristic quantity, the anomaly estimation unit uses the detection value of the second characteristic quantity to estimate the occurrence of an anomaly in the refrigerant circuit.

2. The air conditioning system according to claim 1, characterized in that, The refrigerant circuit is filled with a specified amount of refrigerant. The anomaly estimation unit estimates that a change in only the residual refrigerant amount in the refrigerant circuit is normal.

3. The air conditioning system according to claim 1, characterized in that, The cooling dose estimation unit has a cooling dose estimation model generated using the first characteristic quantity. The refrigerant dosage estimation unit applies the detected value of the first characteristic quantity to the refrigerant dosage estimation model to estimate the residual refrigerant dosage in the refrigerant circuit. The anomaly estimation unit has an anomaly estimation model generated using the second feature quantity. The anomaly estimation unit applies the detection value of the second characteristic quantity to the anomaly estimation model to estimate the occurrence of an anomaly in the refrigerant circuit.

4. The air conditioning system according to claim 3, characterized in that, The anomaly inference model uses the second feature value used in its generation as a normal sample value, and calculates an outlier value representing the degree of deviation of the detected value of the second feature value obtained by the acquisition unit from the normal sample value. If the absolute value of the outlier calculated by the anomaly estimation model is above a predetermined threshold, the anomaly estimation unit estimates that an anomaly has occurred in the refrigerant circuit. Furthermore, if the absolute value of the outlier calculated by the anomaly prediction model is less than a specified threshold, the refrigerant circuit is presumed to be normal.

5. The air conditioning system according to claim 4, characterized in that, The refrigerant dosage estimation unit estimates the residual refrigerant dosage of the refrigerant circuit only when the abnormality estimation unit estimates the refrigerant circuit to be normal, using the detection value of the first characteristic quantity obtained simultaneously with the detection value of the second characteristic quantity when the refrigerant circuit is estimated to be normal.

6. The air conditioning system according to claim 5, characterized in that, Before the remaining refrigerant dose is estimated by the refrigerant dose estimation unit, the occurrence of an anomaly in the refrigerant circuit is estimated by the anomaly estimation unit.

7. The air conditioning system according to claim 1, characterized in that, The second characteristic quantity is a state quantity obtained based on the simulation of the operation of the refrigerant circuit, which is the operation when the refrigerant circuit is operating normally and only the residual refrigerant charge is changed.

8. The air conditioning system according to any one of claims 3 to 6, characterized in that, The second characteristic quantity is a state quantity obtained based on the simulation of the operation of the refrigerant circuit, which is the operation when the refrigerant circuit is operating normally and only the residual refrigerant charge is changed.

9. The air conditioning system according to claim 3, characterized in that, The estimated cooling dose model was generated using linear analysis. The anomaly prediction model was generated using nonlinear analysis.

10. An anomaly estimation method for an air conditioning system, wherein the anomaly estimation method is executed by the air conditioning system, the air conditioning system having an air conditioner and a server, the air conditioner having a refrigerant circuit forming an outdoor unit connected to at least one indoor unit via refrigerant piping, the server being communicatively connected to the air conditioner, the anomaly estimation method for the air conditioning system being characterized in that... The air conditioner performs the following steps: The detection department detects the status variables related to the control of the air conditioner. The detection value of the detected state quantity is acquired by the acquisition unit; and The first communication unit sends the acquired detection value to the server. The server performs the following steps: The detection value is received from the air conditioner by the second communication unit; and The anomaly estimation step involves using the state quantity related to the anomaly in the refrigerant circuit as a characteristic quantity, and then using the detected value of this characteristic quantity to group the outdoor unit and one indoor unit together. For each such combination, the occurrence of an anomaly in the refrigerant circuit is estimated. If an anomaly is estimated in any combination, an anomaly is estimated in the indoor unit of that combination. If an anomaly is estimated in all combinations, an anomaly is estimated in the outdoor unit. The server performs a refrigerant dose estimation step, that is, when using the state quantity related to the refrigerant dose in the refrigerant circuit as a first characteristic quantity, it uses the detection value of the first characteristic quantity to estimate the residual refrigerant dose remaining in the refrigerant circuit. In the anomaly estimation step, when a state quantity that includes at least one state quantity contained in the first characteristic quantity and at least one state quantity not contained in the first characteristic quantity is used as a second characteristic quantity, the detection value of the second characteristic quantity is used to estimate the occurrence of an anomaly in the refrigerant circuit.

11. The method for presuming anomalies in an air conditioning system according to claim 10, characterized in that, The refrigerant circuit is filled with a specified amount of refrigerant. In the abnormality estimation step, the situation where only the residual refrigerant amount in the refrigerant circuit changes is estimated to be normal.

12. The method for presuming anomalies in an air conditioning system according to claim 10, characterized in that, It has a cooling dose estimation model generated using the first feature quantity. In the refrigerant dosage estimation step, the detected value of the first characteristic quantity is applied to the refrigerant dosage estimation model to estimate the residual refrigerant dosage in the refrigerant circuit. Having an anomaly estimation model generated using the second feature quantity, In the anomaly estimation step, the detection value of the second characteristic quantity is applied to the anomaly estimation model to estimate the occurrence of an anomaly in the refrigerant circuit.

13. The method for presuming anomalies in an air conditioning system according to claim 12, characterized in that, The anomaly inference model uses the second feature value used in its generation as a normal sample value, and calculates an outlier value representing the degree of deviation of the detected value of the second feature value relative to the normal sample value. In the anomaly estimation step, If the absolute value of the outlier calculated by the anomaly prediction model is above a specified threshold, an anomaly is presumed to have occurred in the refrigerant circuit. Furthermore, if the absolute value of the outlier calculated by the anomaly prediction model is less than a specified threshold, the refrigerant circuit is presumed to be normal.

14. The method for presuming anomalies in an air conditioning system according to claim 13, characterized in that, In the step of estimating the cooling dosage, Only when the refrigerant circuit is presumed to be normal in the anomaly estimation step, the residual refrigerant charge of the refrigerant circuit is estimated using the detection value of the first characteristic quantity obtained simultaneously with the detection value of the second characteristic quantity when the refrigerant circuit is presumed to be normal.

15. The method for presuming anomalies in an air conditioning system according to claim 14, characterized in that, The anomaly estimation step is performed before the cooling dosage estimation step.

16. An air conditioner having a refrigerant circuit connecting an outdoor unit to at least one indoor unit via refrigerant piping, characterized in that it has: The detection unit is used to detect state quantities related to the control of the air conditioner; The acquisition unit is used to acquire the detection value of the state quantity detected by the detection unit; and The anomaly estimation unit, when using the state quantity related to the anomaly in the refrigerant circuit as a characteristic quantity, uses the detection value of this characteristic quantity to estimate the occurrence of the anomaly in the refrigerant circuit. The anomaly estimation unit groups the outdoor unit and one indoor unit together, and estimates the occurrence of an anomaly in the refrigerant circuit for each such combination. If an anomaly is estimated in any combination, it is estimated that an anomaly has occurred in the indoor unit of that combination. Furthermore, if it is assumed that an anomaly has occurred in all combinations, it is assumed that an anomaly has occurred in the outdoor unit. The air conditioner further includes a refrigerant dosage estimation unit, which, when using the state quantity related to the refrigerant dosage in the refrigerant circuit as a first characteristic quantity, uses the detection value of the first characteristic quantity to estimate the residual refrigerant dosage remaining in the refrigerant circuit. When a state quantity that includes at least one state quantity contained in the first characteristic quantity and at least one state quantity not contained in the first characteristic quantity is used as a second characteristic quantity, the anomaly estimation unit uses the detection value of the second characteristic quantity to estimate the occurrence of an anomaly in the refrigerant circuit.

17. The air conditioner according to claim 16, characterized in that, The refrigerant circuit is filled with a specified amount of refrigerant. The anomaly estimation unit estimates that a change in only the residual refrigerant amount in the refrigerant circuit is normal.

18. The air conditioner according to claim 16, characterized in that, The cooling dose estimation unit has a cooling dose estimation model generated using the first characteristic quantity. The refrigerant dosage estimation unit applies the detected value of the first characteristic quantity to the refrigerant dosage estimation model to estimate the residual refrigerant dosage in the refrigerant circuit. The anomaly estimation unit has an anomaly estimation model generated using the second feature quantity. The anomaly estimation unit applies the detection value of the second characteristic quantity to the anomaly estimation model to estimate the occurrence of an anomaly in the refrigerant circuit.

19. The air conditioner according to claim 18, characterized in that, The anomaly inference model uses the second feature value used in its generation as a normal sample value, and calculates an outlier value representing the degree of deviation of the detected value of the second feature value obtained by the acquisition unit from the normal sample value. If the absolute value of the outlier calculated by the anomaly estimation model is above a predetermined threshold, the anomaly estimation unit estimates that an anomaly has occurred in the refrigerant circuit. Furthermore, if the absolute value of the outlier calculated by the anomaly prediction model is less than a specified threshold, the refrigerant circuit is presumed to be normal.

20. The air conditioner according to claim 19, characterized in that, The refrigerant dosage estimation unit estimates the residual refrigerant dosage of the refrigerant circuit only when the abnormality estimation unit estimates the refrigerant circuit to be normal, using the detection value of the first characteristic quantity obtained simultaneously with the detection value of the second characteristic quantity when the refrigerant circuit is estimated to be normal.

21. The air conditioner according to claim 20, characterized in that, Before the remaining refrigerant dose is estimated by the refrigerant dose estimation unit, the occurrence of an anomaly in the refrigerant circuit is estimated by the anomaly estimation unit.

22. The air conditioner according to any one of claims 16 to 21, characterized in that, The second characteristic quantity is a state quantity obtained based on the simulation of the operation of the refrigerant circuit, and the operation of the refrigerant circuit is the operation when the refrigerant circuit is operating normally and only the residual refrigerant charge is changed.

23. The air conditioner according to claim 18, characterized in that, The estimated cooling dose model was generated using linear analysis. The anomaly prediction model was generated using nonlinear analysis.

24. A method for diagnosing anomalies in an air conditioner, the air conditioner having a refrigerant circuit formed by connecting an outdoor unit to at least one indoor unit via refrigerant piping, the method for diagnosing anomalies in the air conditioner being characterized by comprising the following steps: The state variables related to the control of the air conditioner are detected; Obtain the detection value of the detected state quantity; When the state quantity related to the anomaly of the refrigerant circuit is used as a characteristic quantity, the detection value of this characteristic quantity is used to estimate the occurrence of the anomaly in the refrigerant circuit; and The outdoor unit and one indoor unit are grouped together. For each such combination, the occurrence of an abnormality in the refrigerant circuit is presumed. If an abnormality is presumed in any combination, an abnormality is presumed in the indoor unit of that combination. If an abnormality is presumed in all combinations, an abnormality is presumed in the outdoor unit. In the method for estimating anomalies in the air conditioner, a refrigerant dosage estimation step is performed, that is, when the state quantity related to the refrigerant dosage in the refrigerant circuit is used as a first characteristic quantity, the detection value of the first characteristic quantity is used to estimate the residual refrigerant dosage in the refrigerant circuit. In the anomaly estimation step for estimating the occurrence of an anomaly in the refrigerant circuit, when a state quantity that includes at least one state quantity contained in the first characteristic quantity and at least one state quantity not contained in the first characteristic quantity is used as a second characteristic quantity, the detection value of the second characteristic quantity is used to estimate the occurrence of an anomaly in the refrigerant circuit.

25. The method for diagnosing anomalies in an air conditioner according to claim 24, characterized in that, The refrigerant circuit is filled with a specified amount of refrigerant. In the anomaly estimation step used to estimate the occurrence of an anomaly in the refrigerant circuit, a situation where only the residual amount of refrigerant in the refrigerant circuit changes is estimated to be normal.

26. The method for presuming an anomaly in an air conditioner according to claim 24, characterized in that, It has a cooling dose estimation model generated using the first feature quantity. In the refrigerant dosage estimation step, the detected value of the first characteristic quantity is applied to the refrigerant dosage estimation model to estimate the residual refrigerant dosage in the refrigerant circuit. Having an anomaly estimation model generated using the second feature quantity, In the anomaly estimation step, the detection value of the second characteristic quantity is applied to the anomaly estimation model to estimate the occurrence of an anomaly in the refrigerant circuit.

27. The method for presuming an anomaly in an air conditioner according to claim 26, characterized in that, The anomaly inference model uses the second feature value used in its generation as a normal sample value, and calculates an outlier value representing the degree of deviation of the detected value of the second feature value relative to the normal sample value. In the anomaly estimation step, If the absolute value of the outlier calculated by the anomaly prediction model is above a specified threshold, an anomaly is presumed to have occurred in the refrigerant circuit. Furthermore, if the absolute value of the outlier calculated by the anomaly prediction model is less than a specified threshold, the refrigerant circuit is presumed to be normal.

28. The method for diagnosing anomalies in an air conditioner according to claim 27, characterized in that, In the refrigerant dosage estimation step, only when the refrigerant circuit is estimated to be normal in the abnormality estimation step, the residual refrigerant dosage of the refrigerant circuit is estimated by using the detection value of the first characteristic quantity obtained simultaneously with the detection value of the second characteristic quantity when the refrigerant circuit is estimated to be normal.

29. The method for diagnosing anomalies in an air conditioner according to claim 28, characterized in that, The anomaly estimation step is performed before the cooling dosage estimation step.

Images