Learning device, monitoring device, and air conditioning system

Abstract

Operation data of an air conditioning apparatus includes a first data group and a second data group that is not the same as the first data group. A learning device includes: a first calculation unit configured to calculate a first feature amount from the first data group of the air conditioning apparatus during a learning period; and a learning unit configured to generate a first inference model that infers a first normal range of the first feature amount from a second data group by performing supervised learning using the second data group with the first feature amount being set as truth data, the first feature amount being obtained by calculation by the first calculation unit.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to a learning device, a monitoring device, and an air conditioning system.BACKGROUND ART

[0002] Conventionally, a method of detecting leakage of refrigerant in a cooling system based on a value serving as an indicator of an amount of refrigerant introduced therein has been reviewed, and Japanese Patent No. 6791429 discloses an exemplary refrigerant amount determination device configured to facilitate determination of an amount of refrigerant.

[0003] The refrigerant amount determination device includes: a calculation unit configured to calculate a refrigerant amount indicator value from operation data of an air conditioning system; an inference unit configured to infer information about correction of the refrigerant amount indicator value using a correction model and at least one of the operation data and the calculated refrigerant amount indicator value; and a determination unit configured to determine the amount of the refrigerant of the air conditioning system based on the information about correction of the refrigerant amount indicator value.CITATION LISTPatent Literature

[0004] PTL 1: Japanese Patent No. 6791429SUMMARY OF INVENTIONTechnical Problem

[0005] In the refrigerant amount determination device disclosed in Japanese Patent No. 6791429, an influence of a decrease in the amount of the refrigerant over an air conditioning performance cannot be precisely known and a timing for maintenance cannot be known, disadvantageously. For example, when an amount of leakage of the refrigerant is small to be 10%, the air conditioning performance is not decreased so much even though it is known from the refrigerant amount indicator that the refrigerant is being leaked, with the result that it cannot be determined whether the location of the leakage should be checked urgently or the current situation should be kept and observed for a while.

[0006] The present disclosure has been made to solve the above-described problem, and has an object to provide a learning device so as to obtain an inference model useful to improve accuracy of detecting an abnormality of an air conditioning system.Solution to Problem

[0007] The present disclosure relates to a learning device configured to learn a condition of an air conditioning apparatus in which refrigerant circulates. Operation data of the air conditioning apparatus includes a first data group and a second data group that does not contain a same data element as a data element of the first data group. The learning device includes: a first calculation unit configured to calculate a first feature amount from the first data group of the air conditioning apparatus during a learning period; and a learning unit configured to generate a first inference model that infers a first normal range of the first feature amount from the second data group by performing supervised learning using the second data group with the first feature amount being set as truth data.Advantageous Effects of Invention

[0008] According to the learning device of the present disclosure, since the model that infers the normal range of the feature amount from the behavior of the data other than the data that can be directly used for the calculation of the feature amount among the operation data is generated, possibility of detecting an abnormality accurately at an early stage can be increased.BRIEF DESCRIPTION OF DRAWINGS

[0009] FIG. 1 is a block diagram showing a configuration of an air conditioning system according to a first embodiment.

[0010] FIG. 2 is a functional block diagram showing a configuration of an air conditioning apparatus 40 of FIG. 1.

[0011] FIG. 3 is a diagram showing exemplary operation data reflecting a condition of air conditioning apparatus 40 of FIG. 1.

[0012] FIG. 4 is a block diagram showing a configuration of an abnormality detection system 1 of FIG. 1.

[0013] FIG. 5 is a flowchart for illustrating a detail of a learning process performed by abnormality detection system 1.

[0014] FIG. 6 is a diagram showing an exemplary neural network Nw1 included in an inference model M1 constructed in a step S3 of FIG. 5.

[0015] FIG. 7 is a flowchart for illustrating a detail of a poor-condition detection process performed by abnormality detection system 1.

[0016] FIG. 8 is a diagram showing exemplary temporal changes of inferred values and calculated values of a heat exchange performance Qo and a degree of supercooling SC of refrigerant at an outlet of a condenser.

[0017] FIG. 9 is a graph showing a relation between an amount of change of the refrigerant and an amount of change of the degree of supercooling.

[0018] FIG. 10 is a graph showing a relation among decrease of the amount of the refrigerant, a performance decrease operation ratio, and a performance decrease ratio.

[0019] FIG. 11 is a diagram showing an exemplary trend graph displayed to a user.DESCRIPTION OF EMBODIMENTS

[0020] Hereinafter, embodiments of the present disclosure will be described in detail with reference to figures. It should be noted that the same or corresponding portions in the figures are denoted by the same reference characters and will not be described repeatedly in principle.First Embodiment

[0021] FIG. 1 is a block diagram showing a configuration of an air conditioning system according to a first embodiment. An air conditioning system 1000 includes: an abnormality detection system 1; and an air conditioning apparatus 40 having a condition monitored by abnormality detection system 1. As shown in FIG. 1, abnormality detection system 1 is connected to air conditioning apparatus 40 via a network 900.

[0022] Abnormality detection system 1 includes a CPU (Central Processing Unit) 2, a memory 3 (a ROM (Read Only Memory) and a RAM (Random Access Memory)), an input / output buffer (not shown), and the like. CPU 2 loads a program stored in the ROM into the RAM or the like and executes the program. The program stored in the ROM is a program in which a processing procedure of abnormality detection system 1 is written. Abnormality detection system 1 monitors each device in air conditioning apparatus 40 in accordance with such a program. This control is not limited to a process by software, and the process can also be performed by dedicated hardware (electronic circuit). It should be noted that abnormality detection system 1 may be constructed in a server connected to network 900.

[0023] Air conditioning apparatus 40 includes a plurality of indoor units 20, an outdoor unit 10, and a controller 30. Each of the plurality of indoor units 20 is disposed in an indoor space and is connected to outdoor unit 10 by a liquid pipe and a gas pipe through each of which refrigerant passes. Outdoor unit 10 is disposed in a space (outdoor space) outside the indoor space. It should be noted that the number of indoor units 20 included in air conditioning apparatus 40 may be one or two.

[0024] Outdoor unit 10 includes a compressor, an outdoor heat exchanger, and an outdoor fan. Each of the plurality of indoor units 20 includes an expansion valve and an indoor heat exchanger. The refrigerant is supplied from the compressor included in outdoor unit 10 to each of the plurality of indoor units 20. The refrigerant circulates between each of the plurality of indoor units 20 and outdoor unit 10.

[0025] Controller 30 includes a thermostat and integrally controls air conditioning apparatus 40. Controller 30 is connected to an abnormality detection system 1 via network 900. Network 900 includes the Internet and a cloud system. It should be noted that network 900 may be a LAN (local area network).

[0026] Controller 30 includes a CPU 32, a memory 33 (a ROM and a RAM), an input / output buffer (not shown), and the like. CPU 32 loads a program stored in the ROM into the RAM or the like and executes the program. The program stored in the ROM is a program in which a processing procedure of controller 30 is written. Controller 30 performs control of each device in air conditioning apparatus 40 in accordance with such a program. This control is not limited to a process by software, and the process can also be performed by dedicated hardware (electronic circuit).

[0027] FIG. 2 is a functional block diagram showing a configuration of air conditioning apparatus 40 of FIG. 1. As shown in FIG. 2, outdoor unit 10 includes a compressor 11, an outdoor heat exchanger 12, a four-way valve 13, an outdoor fan 14, an accumulator 15, temperature sensors 54 to 56, pressure sensors 61, 63, and a humidity sensor 57.

[0028] Each of the plurality of indoor units 20 includes an expansion valve 21, an indoor heat exchanger 22, an indoor fan 23, and temperature sensors 51 to 53. It should be noted that expansion valve 21 includes, for example, an LEV (linear expansion valve).

[0029] Operation modes of air conditioning apparatus 40 include a heating mode, a cooling mode, and a defrosting mode. In the heating mode, four-way valve 13 connects a discharge port of compressor 11 and indoor heat exchanger 22, and connects outdoor heat exchanger 12 and a refrigerant inlet of accumulator 15. In the heating mode, the refrigerant circulates through compressor 11, four-way valve 13, indoor heat exchanger 22, expansion valve 21, and outdoor heat exchanger 12 in this order. In each of the cooling mode and the defrosting mode, four-way valve 13 connects the discharge port of compressor 11 and outdoor heat exchanger 12, and connects indoor heat exchanger 22 and the refrigerant inlet of accumulator 15. In each of the cooling mode and the defrosting mode, the refrigerant circulates through compressor 11, four-way valve 13, outdoor heat exchanger 12, expansion valve 21, and indoor heat exchanger 22 in this order.

[0030] Temperature sensor 51 measures a temperature (room temperature TH1ic) of air suctioned into indoor heat exchanger 22, and outputs the temperature to controller 30. Temperature sensors 52, 53 respectively measure temperatures (indoor liquid temperature TH2ic and indoor gas temperature TH3ic) of the refrigerant before and after passing through indoor heat exchanger 22, and output the temperatures to controller 30.

[0031] Temperature sensor 54 measures a temperature (discharge temperature TH4) of the refrigerant discharged from compressor 11, and outputs the discharge temperature to controller 30. Temperature sensor 55 measures a temperature (suction temperature TH5) of the refrigerant suctioned into compressor 11 via accumulator 15, and outputs the suction temperature to controller 30. Temperature sensor 56 measures a temperature TH3 of the liquid refrigerant in the pipe that connects outdoor heat exchanger 12 and liquid pipe 41, and outputs the temperature to controller 30.

[0032] Pressure sensor 61 measures a pressure (discharge pressure HS1) of the refrigerant discharged from compressor 11, and outputs discharge pressure HS1 to controller 30. Pressure sensor 63 measures a pressure (suction pressure LS) of the refrigerant suctioned into compressor 11, and outputs suction pressure LS to controller 30.

[0033] Hereinafter, control during cooling will be described as a representative example. Controller 30 controls an amount of the refrigerant discharged by compressor 11 per unit time by controlling an operation frequency fCOMP of compressor 11 so as to allow a suction saturation gas temperature to become a target temperature. Controller 30 controls a degree of opening Li of expansion valve 21 so as to allow a degree of superheat SH (=TH3ic-TH2ic) of the refrigerant at the outlet of indoor heat exchanger 22 to become a target value. Controller 30 switches a circulation direction of the refrigerant by controlling four-way valve 13 to provide a flow path indicated by the solid lines. Controller 30 controls an amount of air sent by outdoor fan 14 per unit time by controlling a rotation frequency fFANo of outdoor fan 14 so as to allow a discharge saturation gas temperature to become a target value. Controller 30 controls a rotation frequency fFANi of indoor fan 23 so as to attain an amount of air set by a user. Controller 30 associates, with a time at the measurement, operation data reflecting the condition of the air conditioning system, and transmits it to abnormality detection system 1.

[0034] FIG. 3 is a diagram showing exemplary operation data reflecting the condition of air conditioning apparatus 40 of FIG. 1. As shown in FIG. 3, the operation data includes, for example, an outdoor air temperature, the discharge temperature (TH4), the evaporation temperature (TH2ic), a condensation temperature, the suction temperature (TH1ic), a send-out temperature, the high pressure (HS1), the low pressure (LS), the operation frequency (fCOMP) of compressor 11, the degree of opening of expansion valve 21, the operation modes, an operation state (operation, halt, or standby), and the rotation speed (fFANo, fFANi)) of each of outdoor fan 14 and indoor fan 23, the temperature (setting temperature) of the indoor space as set by the user, a current value of an inverter of compressor 11, a voltage value of the inverter, a temperature of a heat sink included in outdoor unit 10, and the temperature (liquid pipe temperature TH3) of the liquid pipe that connects outdoor unit 10 and indoor unit 20 (pipe through which the liquid refrigerant flows). It should be noted that the operation frequency of compressor 11, the degree of opening of expansion valve 21, and the rotation speed of outdoor fan 14 are basic operation amounts in VRF (Variable Refrigerant Flow) control.

[0035] An environment in which air conditioning apparatus 40 is operated may have characteristics specific to the environment (for example, a length of the refrigerant pipe, a type of indoor unit 20, the number of indoor units 20, and a difference in height between indoor unit 20 and outdoor unit 10). Therefore, a determination criterion (for example, a threshold value) for detecting an abnormality of air conditioning apparatus 40 can be different for each environment in which air conditioning apparatus 40 is operated. Therefore, when the same determination criterion is used regardless of the environment in which air conditioning apparatus 40 is operated, accuracy of detecting an abnormality of air conditioning apparatus 40 may be decreased.

[0036] Therefore, in abnormality detection system 1, when a normal condition is confirmed during a trial operation after installation of the air conditioning system, the normal condition is regarded as being kept for a subsequent certain period. Then, a trained model is generated in which a relation is learned between operation data of air conditioning apparatus 40 during the certain period and a normal value of a condition indicator value (specific parameter) representing the condition of air conditioning apparatus 40 and corresponding to the operation data. By using the trained model, an abnormality of air conditioning apparatus 40 can be detected in accordance with a determination criterion adapted to the environment in which air conditioning apparatus 40 is operated. As a result, the accuracy of detecting an abnormality of the air conditioning system can be improved.

[0037] FIG. 4 is a block diagram showing a configuration of abnormality detection system 1 of FIG. 1. In abnormality detection system 1, operation data is obtained in each of a plurality of consecutive operation periods (for example, one day, one week, or one month), and trained inference models M1, M2 are constructed using training data including the operation data. The plurality of consecutive operation periods may be the same period (predetermined period) or may be different periods.

[0038] As shown in FIG. 4, abnormality detection system 1 includes a learning device 4 and a monitoring device 5. Learning device 4 includes a calculation unit 110A and a learning unit 120. Monitoring device 5 includes a calculation unit 110B, an inference unit 220, a determination unit 310, and a display unit 320.

[0039] Calculation unit 110A and calculation unit 110B basically perform the same calculation and are only different from each other in that they are used respectively at the time of learning and at the time of monitoring. Therefore, calculation unit 110A may also serve as calculation unit 110B.

[0040] Calculation unit 110A calculates a feature amount F1A1 from operation data D1A1 at the time of the normal condition. Learning unit 120 performs machine learning with feature amount F1A1 of an element device and operation data D2A1 of the element device at the time of the normal condition so as to construct inference model M1. Inference unit 220 infers a normal range (upper and lower limit values) F2B1 of the feature amount by inputting, to trained inference model M1, operation data D2B1 at the time of determination. Further, calculation unit 110A calculates feature amount F1A2 from operation data D1A2 at the time of the normal condition. Learning unit 120 performs machine learning with feature amount F1A2 of an element device and operation data D2A2 of the element device at the time of the normal condition so as to construct inference model M2. Inference unit 220 infers a normal range (upper and lower limit values) F2B2 of the feature amount by inputting, to trained inference model M2, operation data D2B2 at the time of determination.

[0041] Calculation unit 110B calculates a feature amount F1B1 from operation data D1B1 at the time of determination. Calculation unit 110B calculates a feature amount F1B2 from operation data D1B2 at the time of determination. Determination unit 310 includes a data processing unit 311 and a data processing unit 312. Data processing unit 311 counts the number of pieces of data (data falling out of the normal range) for which feature amount F1B1 (degree of supercooling SC of the refrigerant at the outlet of the condenser) calculated by calculation unit 110 from operation data D1B1 at the time of determination exceeds or falls below normal range F2B1 inferred by trained inference model M1 from the operation data at the time of poor condition. Data processing unit 312 counts the number of pieces of data for which feature amount F1B2 (heat exchange performance Qo of the heat exchanger) calculated from operation data D1B2 at the time of determination is equal to or less than a predefined performance. Determination unit 310 calculates a ratio by dividing, by the determined number of pieces of operation data, the number of pieces of data for which feature amount F1B1 (degree of supercooling SC) is outside the normal range (falls out of the normal range) and feature amount F1B2 (heat exchange performance Qo) is decreased by a predefined amount (falls out of the normal range) at the same time. This ratio represents a ratio of times in a day at each of which an operation involving a heat exchange performance Qo decreased to be lower than that in the normal condition takes place due to a poor condition of the element device, and is referred to as “performance decrease operation ratio” in the present specification. Determination unit 310 determines whether or not the performance decrease operation ratio is equal to or more than a threshold value.

[0042] Display unit 320 displays trend data indicating a change of the amount of the refrigerant and a change of the performance decrease operation ratio per day for each element device (outdoor heat exchanger 12, indoor heat exchanger 22), and displays that maintenance is necessary when the performance decrease operation ratio is equal to or more than the threshold value.

[0043] CPU 2 shown in FIG. 1 is operated as calculation units 110A, 110B, learning unit 120, inference unit 220, and determination unit 310 in accordance with corresponding programs. Further, memory 3 stores an operation data set of air conditioning apparatus 40 and inference models M1, M2.

[0044] Each of inference models M1, M2 is a regression model that includes a neural network and that infers a normal value of the condition indicator value of air conditioning apparatus 40 from the operation data of air conditioning apparatus 40. Each of inference models M1, M2 may be a classification model that infers a level (classification) of the condition indicator value. The normal value of the condition indicator value may be each of a maximum value and a minimum value of a confidence interval that the condition indicator value can have when air conditioning apparatus 40 is normal. Further, a range having the normal value as a median value (for example, a range of ±10% of the normal value) may be employed as the confidence interval.

[0045] It should be noted that a general AI (Artificial Intelligence) technology can be applied to clustering and weighting of the parameters included in the operation data.

[0046] Learning unit 120 performs machine learning onto inference models M1, M2 so as to construct trained inference models M1, M2. A machine learning algorithm used by learning unit 120 may be a known algorithm such as supervised learning, semi-supervised learning, unsupervised learning, or reinforcement learning. Further, as the machine learning algorithm, deep learning in which extraction of the feature amount itself is learned can be used, or the machine learning may be performed in accordance with another known method such as a neural network, genetic programming, inductive logic programming, or a support vector machine. Hereinafter, a case where the supervised learning is applied to a neural network will be described.

[0047] Calculation unit 110A calculates a condition indicator value (SC) during a first operation period from operation data D1A1 included in operation data DA. Calculation unit 110A calculates a condition indicator value (Qo) during the first operation period from operation data D1A2 included in operation data DA. Learning unit 120 performs the supervised learning onto inference model M1 using training data having the condition indicator value during the first operation period as truth data (teaching data) for the first operation data. Learning unit 120 constructs trained inference models M1, M2, and stores the models into memory 3. It should be noted that since the first operation period is a period immediately after air conditioning apparatus 40 is confirmed to be operated in the trial operation after being installed in the installation location, there is a low possibility that air conditioning apparatus 40 has an abnormality. Since it can be assumed that air conditioning apparatus 40 in the first operation period is normal, the condition indicator value during the first operation period is set as the truth data for the first operation data. In the below-described example, the first operation period is 365 days immediately after the installation.

[0048] FIG. 5 is a flowchart for illustrating a detail of the learning process performed by abnormality detection system 1. When the learning process is started, abnormality detection system 1 shown in FIG. 4 obtains air conditioning operation data DA for 365 days immediately after the installation in step S1. Air conditioning operation data DA includes a first data group D1A1 and a second data group D2A1 that does not contain the same data element as a data element of first data group D1A1. Further, air conditioning operation data DA includes a third data group D1A2 different from first data group D1A1 and a fourth data group D2A2 that does not contain the same data element as a data element of third data group D1A2.

[0049] Then, in a step S2, calculation unit 110A calculates a feature amount F1A1 from first data group D1A1 and calculates feature amount F1A2 from third data group D1A2. Examples of feature amount F1A1 include degree of supercooling SC of the refrigerant at the outlet of the condenser. Examples of feature amount F1A2 include heat exchange performance Qo of outdoor heat exchanger 12.

[0050] For example, degree of supercooling SC at the outlet of the condenser can be calculated by the following formula (1):SC=Tc-TH⁢3(1)

[0051] Here, Tc represents the discharge saturation gas temperature and is a value determined by discharge pressure HS1. TH3 represents a liquid refrigerant temperature.

[0052] Further, for example, heat exchange performance Qo can be calculated by the following formula (2):Qo=Gr×(Hd-Hco)(2)

[0053] Here, Gr represents a refrigerant circulation amount and is a value determined by operation frequency fCOMP of compressor 11, and Hd represents a specific enthalpy of the inlet portion of outdoor heat exchanger 12 and is a value determined by pressure HS1 and temperature TH4. Further, Hco represents a specific enthalpy of the outlet of outdoor heat exchanger 12 and is a value determined by pressure HS1 and temperature TH3.

[0054] The upper part of Table 1 below shows: operation data DA used when training inference model M1 for determining leakage of the refrigerant; operation data D2B1 used as an input to trained inference model M1; and operation data D1B1 used by calculation unit 110B at the time of determination.

[0055] Further, the lower part of Table 1 shows: operation data DA used when training inference model M2 for determining insufficiency of the heat exchange performance of outdoor heat exchanger 12; operation data D2B2 used as an input to trained inference model M2; and operation data D1B2 used by calculation unit 110B at the time of determination.TABLE 1FeatureAmount(F1B1)Used for Learning (DA)Used for Inference (D2B1)Used for Calculation (D1B1)Degree ofDischarge Temperature (TH4)Discharge Temperature (TH4)SupercoolingDischarge Pressure (HS1)Discharge Pressure (HS1)SCLiquid Temperature (TH3)Liquid Temperature (TH3)(Leakage ofIndoor Liquid Temperature (TH2ic)Indoor Liquid Temperature (TH2ic)Refrigerant)Indoor Gas Temperature (TH3ic)Indoor Gas Temperature (TH3ic)Suction Temperature (TH5)Suction Temperature (TH5)Suction Pressure (LS)Suction Pressure (LS)Outdoor Air Relative Humidity (TH7)Outdoor Air Relative Humidity (TH7)Room Temperature (TH1ic)Room Temperature (TH1ic)Frequency of Compressor (fCOMP)Frequency of Compressor (fCOMP)Frequency of Fan (fFANo)Frequency of Fan (fFANo)Degree of Opening of Indoor LEV (Li)Degree of Opening of Indoor LEV (Li)FeatureAmount(F1B2)Used for Learning (DA)Used for Inference (D2B2)Used for Calculation (D1B2)PerformanceDischarge Temperature (TH4)Discharge Temperature (TH4)QoDischarge Pressure (HS1)Discharge Pressure (HS1)Liquid Temperature (TH3)Liquid Temperature (TH3)Indoor Liquid Temperature (TH2ic)Indoor Liquid Temperature (TH2ic)Indoor Gas Temperature (TH3ic)Indoor Gas Temperature (TH3ic)Suction Temperature (TH5)Suction Temperature (TH5)Suction Pressure (LS)Suction Pressure (LS)Outdoor Air Relative Humidity (TH7)Outdoor Air Relative Humidity (TH7)Room Temperature (TH1ic)Room Temperature (TH1ic)Frequency of Compressor (fCOMP)Frequency of Compressor (fCOMP)Frequency of Fan (fFANo)Frequency of Fan (fFANo)Degree of Opening of Indoor LEV (Li)Degree of Opening of Indoor LEV (Li)

[0056] Then, in a step S3 of FIG. 5, learning unit 120 constructs inference model (regression model) M1 of the objective variable (feature amount F1A1) corresponding to the explanatory variable (operation data group D2A1). Similarly, learning unit 120 constructs inference model (regression model) M2 of the objective variable (feature amount F1A2) corresponding to the explanatory variable (operation data group D2A2). On this occasion, learning unit 120 constructs each inference model with the number of feature amounts×3 (upper limit value, median value, and lower limit value). For example, the upper limit value may be 97.5% of the confidence interval and the lower limit value may be 2.5% of the confidence interval (corresponding to 2σ).

[0057] FIG. 6 is a diagram showing an exemplary neural network Nw1 included in inference model M1 constructed in step S3 of FIG. 5. As shown in FIG. 6, neural network Nw1 includes an input layer X10, an intermediate layer (hidden layer) Y10, and an output layer Z10. Input layer X10 includes neurons X11, X12, X13. Intermediate layer Y10 includes neurons Y11, Y12. Output layer Z10 includes a neuron Z11. Input layer X10 and intermediate layer Y10 are fully connected to each other. Intermediate layer Y10 and output layer Z10 are fully connected to each other. Neural network Nw1 may include two or more intermediate layers.

[0058] When a plurality of inputs are respectively input to neurons X11 to X13 of input layer X10, the values are multiplied by weights w11, w12, w13, w14, w15, w16 and are input to neurons Y11, Y12 of intermediate layer Y10. The outputs from neurons Y11, Y12 are multiplied by weights w21, w22 and are output from neuron Z11 of output layer Z10. The output result from output layer Z10 differs depending on the values of weights w11 to w16, w21, w22. The weights and biases of neural network Nw1 are updated by backpropagation of an error between the truth data and the result output from the output layer as a result of inputting the operation data to the input layer such that the result becomes close to the truth data.

[0059] In this way, the training of inference model M1 is completed. It should be noted that since the same applies to inference model M2, the explanation will not be described repeatedly. The following describes a detail of a poor-condition detection process after the completion of the learning.

[0060] FIG. 7 is a flowchart for illustrating the detail of the poor-condition detection process performed by abnormality detection system 1.

[0061] First, in a step S11, abnormality detection system 1 shown in FIG. 4 obtains operation data DB for one day for which determination for a poor condition is to be made.

[0062] Then, in a step S12, calculation unit 110B of FIG. 4 calculates feature amount F1B1 from operation data D1B1 among obtained operation data DB, and calculates feature amount F1B2 from operation data D1B2 among obtained operation data DB. Feature amount F1B1 is, for example, degree of supercooling SC of the refrigerant at the outlet of the condenser, and feature amount F1B2 is heat exchange performance Qo of outdoor heat exchanger 12. Regarding these, the calculation process have already been described with each of the above-described formulas (1) and (2), and therefore will not be described repeatedly. In parallel with this, in step S12, inference unit 220 of FIG. 4 uses trained model M1 to infer a degree of supercooling SCi (upper limit value, lower limit value, median value SCAI) at the outlet of the condenser as inference value F2B1 from operation data D2B1, and inference unit 220 uses trained model M2 to infer a heat exchange performance QoAI (median inference value) of outdoor heat exchanger 12 as inference value F2B2 from operation data D2B2.

[0063] Then, in a step S13, determination unit 310 extracts operation data for which heat exchange performance Qo of outdoor heat exchanger 12 calculated by calculation unit 110B using the operation data obtained in step S11 is decreased by a predefined amount (for example, 20% or more) with respect to median inference value QoAI provided by inference model M2. The extraction of the data is performed by, for example, flagging a set of operation data. It should be noted that the predefined amount at least includes an amount of decrease of heat exchange performance Qo with which cooling is not attained (performance decrease of −20%).

[0064] FIG. 8 is a diagram showing exemplary temporal changes of inferred values and calculated values of heat exchange performance Qo and degree of supercooling SC of the refrigerant at the outlet of the condenser. Decrease of the amount of the refrigerant is less than −10% during a period from a time t0 to a time t1, the decrease of the amount of the refrigerant is −10% during a subsequent period TP1, and the decrease of the amount of refrigerant is −20% during a period TP2.

[0065] As the refrigerant decrease amount shown in the upper part of FIG. 8 is larger, the number of pieces of extracted data indicated by white circles are also larger.

[0066] A step S14 is performed after step S13 in FIG. 7. In step S14, determination unit 310 extracts, from the operation data obtained in step S11, data for which degree of supercooling SC of the refrigerant at the outlet of the condenser falls below the lower limit value. The lower limit value is the value inferred by inference model M1.

[0067] Then, in a step S15, determination unit 310 calculates a ratio (performance decrease operation ratio R) of pieces of data for which heat exchange performance Qo is decreased to be lower than the lower limit value and degree of supercooling SC of the refrigerant at the outlet of the condenser is decreased to be lower than the lower limit value. Ratio R is a value obtained by dividing, by the total number of the pieces of the operation data, the number of the pieces of the data extracted in steps S13 and S14. That is, performance decrease operation ratio R is as follows: (the number of times of observing the pieces of the data flagged in both S13 and S14) / (the total number of times of making observations).

[0068] Thus, since the condition for the extraction is an AND condition for the data for which heat exchange performance Qo is decreased and the data for which each feature amount falls out of the appropriate range, the performance decrease due to the poor condition of each element can be specified. With this, for example, decreased performance caused by failure of heat transfer of outdoor heat exchanger 12 due to rust, adhesion of dust, or the like is not detected. The data extracted in this way is indicated by circles in FIG. 8.

[0069] For example, since heat exchange performance Qo of outdoor heat exchanger 12 is smaller than the lower limit value of −20% from the median value and is equal to or smaller than the inferred lower limit value of degree of supercooling SC of the refrigerant at the outlet of the condenser at a time t2, data is extracted in step S15 and is used for the calculation of the performance decrease operation ratio. On the other hand, at a time t7, degree of supercooling SC of the refrigerant at the outlet of the condenser is equal to or less than the inferred lower limit value but heat exchange performance Qo is not equal to or less than the lower limit value of −20% from the median value, and no data is therefore extracted in step S15.

[0070] The number of the pieces of the data thus extracted from the operation data for one day is divided by the total number of the pieces of the operation data for one day, thereby calculating performance decrease operation ratio R.

[0071] Then, in a step S16, display unit 320 displays a trend graph indicating that the feature amount is changed as the time elapses. When the feature amount becomes equal to or more than the threshold value, the user (maintenance worker) is notified of the poor condition.

[0072] FIG. 9 is a graph showing a relation between an amount of change of the refrigerant and an amount of change of the degree of supercooling. As shown in the lower part of FIG. 8, a difference ΔSC between inferred median value SCAI and degree of supercooling SC, which is a calculated feature amount, tends to be larger as the amount of the refrigerant is decreased, thereby finding out a relation in which the amount of change (amount of decrease) of the refrigerant and ΔSC are uniquely determined as shown in FIG. 9. For example, when the correlation between ΔSC and the liquid refrigerant amount as shown in FIG. 9 is set for each apparatus model, the amount of change of the refrigerant can be estimated. ΔSC may be calculated by the following formula. It should be noted that an index “ave” refers to an average in a predetermined period.Δ⁢SC=SCAI_ave-SC_ave

[0073] FIG. 10 is a graph showing a relation among the decrease of the amount of the refrigerant, the performance decrease operation ratio, and the performance decrease ratio. In FIG. 10, the horizontal axis represents the performance decrease ratio (Qo / QoAI), and the vertical axis represents the performance decrease operation ratio.

[0074] FIG. 11 is a diagram showing an exemplary trend graph displayed to the user. FIG. 11 shows a temporal change of the performance decrease operation ratio when the performance decrease ratio in FIG. 10 is −10%. For example, in the case of slow leakage of the refrigerant, the maintenance worker can confirm that the times of occurrence of performance decrease due to the leakage of the refrigerant are increased day after day, and therefore can determine a timing for maintenance.

[0075] When the leakage of the refrigerant is progressed with passage of time (−10%→−30%), the trend data of the performance decrease ratio is increased to exceed a threshold value R1. When the data exceeds threshold value R1, display unit 320 displays the poor condition on the screen for the sake of notification thereof.

[0076] Further, when an amount of insufficiency of the refrigerant corresponding to ΔSC derived from the operation data is also shown based on the relation shown in FIG. 9, the maintenance worker can predict and prepare an amount of refrigerant to be added and replenished.

[0077] According to monitoring device 5 of the present embodiment as described above, the ratio (performance decrease operation ratio) of the times of occurrence of the operation in a day in which degree of supercooling SC, which is a feature amount representing the leakage of the refrigerant, falls out of the normal range and heat exchange performance Qo is decreased to be lower than that in the normal condition is sorted as the trend data per day, and when the poor condition determination threshold value is exceeded, the maintenance worker is notified of a timing for maintenance. Thus, the maintenance worker can know the timing to perform the maintenance work.

[0078] Further, since the trend data is displayed rather than an instantaneous value, the poor condition can be correctly determined even when the threshold value is instantaneously exceeded due to an internal or external disturbance.

[0079] For example, even when the refrigerant is insufficient, the same performance Qo may be exhibited under the same environment condition as that in the normal condition. This is because performance Qo is balanced with an air conditioning load by control. However, since the operation condition of air conditioning apparatus 40 is changed when the refrigerant is insufficient, the median inference value of performance Qo inferred from the operation condition is increased to cause the calculated value of performance Qo to fall below the inferred normal range, with the result that the decrease of the performance can be detected. Therefore, the insufficiency of the refrigerant can be detected in an intermediate season such as spring, autumn, or the like, with the result that insufficiency of the performance can be prevented in advance before summer during which full operation is performed for cooling.Various Modifications

[0080] It should be noted that the leakage of the refrigerant during the cooling has been described with reference to the flowcharts of FIGS. 5 and 7 and has been explained using the formula (1); however, also in the case of failure of heat transfer of outdoor heat exchanger 12, failure of heat transfer of indoor heat exchanger 22, failure of the indoor expansion valve, and failure of the indoor expansion valve during the cooling, the feature amounts can be changed as shown in the below-described Table 2 and the poor condition can be determined by the same process.TABLE 2Type of PoorCondition(During Cooling)Feature AmountLeakage of RefrigerantOC_SC = Tc − TH3Failure of HeatOC_GaCpε = Gr × (hd − hco) / (Tc − TH7)Transfer of OutdoorGr = ρs × fCOMP × St ×ηv × 10−6Heat ExchangerFailure of HeatIC_GaCpε = Gr_lev × (hs − hco) / (TH1ic −Transfer of IndoorTH2ic)ExchangerGr_lev = A × Cv ×√{square root over (ρco × (HS1 − LS))}Failure of IndoorIC_LEV = LiExpansion Valve

[0081] In Table 2, details of the variables are as follows. It should be noted that OC is a sign representing the outdoor unit, and IC_ is a sign representing the indoor unit. Further, it is indicated that f( ) below is defined as a function for performing calculation using a value obtained from a refrigerant physical property table with respect to a numerical value that is in the parentheses and that is detected by a sensor.

[0082] Tc: compressor discharge saturated gas temperature f(HS1) “° C.”

[0083] hd: compressor discharge enthalpy f(HS1, TH4) [kJ / kg]

[0084] hco: enthalpy of the refrigerant at the outlet of outdoor heat exchanger 12 f(HS1, TH3) [kJ / kg]

[0085] ρs: compressor suction density f(LS, TH5) [kg / m3]

[0086] fCOMP: compressor frequency [Hz]

[0087] St: excluded volume [cc]

[0088] ηv: volume efficiency 0.95 [-]

[0089] hs: compressor suction enthalpy f(LS, TH5) [kJ / kg]

[0090] A: coefficient (value corresponding to the unit of pressure)

[0091] Cv: flow rate coefficient (value corresponding to the specification and degree of opening of the LEV)

[0092] ρco: density of the refrigerant at the outlet of outdoor heat exchanger 12 f(HS1, TH3) [kg / m3]

[0093] Further, the tendencies in each of which the feature amount falls out of the normal range during the cooling are as follows.

[0094] Leakage of refrigerant: falling below the normal range

[0095] Failure of heat transfer of outdoor heat exchanger 12: falling below the normal range

[0096] Failure of heat transfer of indoor heat exchanger 22: falling below the normal range

[0097] Failure of indoor expansion valve: exceeding and falling below the normal range

[0098] Further, the poor condition of outdoor heat exchanger 12 with regard to the heat exchange performance during the cooling has been shown in the formula (2); however, the poor condition of indoor heat exchanger 22 with regard to the heat exchange performance can also be determined by the same process with the feature amount being changed as shown in Table 3 below.TABLE 3Type of Poor Condition (DuringCooling)Feature AmountOutdoor Heat ExchangerOC_Q = Gr × (hd − hco)Heat Exchange PerformanceGr = ρs × fCOMP × St ×ηv × 10−6Indoor Heat ExchangerIC_Q = Gr_lev × (hs − hco)Heat Exchange PerformanceGr_lev = A ×Cv ×√{square root over (ρco × (HS1 − LS))}

[0099] In Table 3, details of the variables are as follows. It should be noted that OC_ is a sign representing the outdoor unit, and IC_ is a sign representing the indoor unit. Further, it is indicated that f( ) below is defined as a function for performing calculation using a value obtained from the refrigerant physical property table with respect to a numerical value that is in the parentheses and that is detected by a sensor.

[0100] hd: compressor discharge enthalpy f(HS1, TH4) [kJ / kg]

[0101] hco: enthalpy of the refrigerant at the outlet of outdoor heat exchanger 12

[0102] f(HS1, TH3) [kJ / kg]

[0103] ρs: compressor suction density f(LS, TH5) [kg / m3]

[0104] fCOMP: compressor frequency [Hz]

[0105] St: excluded volume [cc]

[0106] ηv: volume efficiency 0.95 [-]

[0107] hs: compressor suction enthalpy f(LS, TH5) [kJ / kg]

[0108] A: coefficient (value corresponding to the unit of pressure)

[0109] Cv: flow rate coefficient (value corresponding to the specification and degree of opening of the LEV)

[0110] ρco: density of the refrigerant at the outlet of outdoor heat exchanger 12 f(HS1, TH3) [kg / m3]

[0111] Further, also in the case of the leakage of the refrigerant, the failure of heat transfer of outdoor heat exchanger 12, the failure of the heat transfer of indoor heat exchanger 22, and the failure of the indoor expansion valve during the heating, the feature amounts can be changed as shown in Table 4 below and the poor condition can be determined by the same process.TABLE 4Type of PoorCondition(During Heating)Feature AmountLeakage of RefrigerantOC_SH = TH5 − TeFailure of HeatOC_Gaε = Gr × (hs − hco_ave) / (Ia7 − Ie)Transfer of OutdoorGr = ρs × fCOMP × St x ηv × 10−6Heat ExchangerFailure of HeatIC_GaCpε = Gr_lev × (hs − hco) / (TH1ic −Transfer of IndoorTH2ic)Heat ExchangerGr_lev = A × Cv ×√{square root over (ρco × (HS1 − LS))}Failure of IndoorIC_LEV = LiExpansion Valve

[0112] In Table 4, details of the variables are as follows. It should be noted that OC is a sign representing the outdoor unit, and IC_ is a sign representing the indoor unit. Further, it is indicated that f( ) below is defined as a function for performing calculation using a value obtained from the refrigerant physical property table with respect to a numerical value that is in the parentheses and that is detected by a sensor.

[0113] Te: compressor suction saturated gas temperature f(LS) [C]

[0114] hs: compressor suction enthalpy f(LS, TH5) [kJ / kg]

[0115] hco_ave: enthalpy of the refrigerant at the outlet of indoor heat exchanger 22

[0116] f(HS1, TH2ic_ave) [kJ / kg]

[0117] TH2ic_ave is an average value of outlet temperatures Th2ic of all the indoor heat exchangers 22.

[0118] Ia7: outdoor air enthalpy f(TH7, relative humidity) [kJ / kg]

[0119] Ie: surface enthalpy of outdoor heat exchanger 12 f(Te, relative humidity 100%) [kJ / kg]

[0120] ρs: compressor suction density f(LS, TH5) [kg / m3]

[0121] fCOMP: compressor frequency [Hz]

[0122] St: excluded volume [cc]

[0123] ηv: volume efficiency 0.95 [-]

[0124] hco: enthalpy of the refrigerant at the outlet of indoor heat exchanger 22 f(HS1, TH2ic) [kJ / kg]

[0125] A: coefficient (value corresponding to the unit of pressure)

[0126] Cv: flow rate coefficient (value corresponding to specification and degree of opening of the LEV)

[0127] ρco: density of the refrigerant at the outlet of indoor heat exchanger 22 f(HS1, TH2ic) [kg / m3]

[0128] Further, the tendencies in each of which the feature amount falls out of the normal range during the heating are as follows.

[0129] Leakage of refrigerant: exceeding the normal range

[0130] Failure of heat transfer of outdoor heat exchanger 12: falling below the normal range

[0131] Failure of heat transfer of indoor heat exchanger 22: falling below the normal range

[0132] Failure of indoor expansion valve: exceeding and falling below the normal range

[0133] Further, each of the poor condition of outdoor heat exchanger 12 with regard to the heat exchange performance and the poor condition of indoor heat exchanger 22 with regard to the heat exchange performance during the heating can be determined by the same process with the feature amount being changed as shown in Table 5 below.TABLE 5Type of Poor Condition (DuringHeating)Feature AmountOutdoor Heat ExchangerOC_Q = Gr × (hs − hco_ave)Heat Exchange PerformanceGr = ρs × fCOMP × St ×ηv × 10−6Indoor Heat ExchangerIC_Q = Gr_lev × (hs − hco)Heat Exchange PerformanceGr_lev = A ×Cv ×√{square root over (ρco × (HS1 − LS))}

[0134] In Table 5, details of the variables are as follows. It should be noted that OC_ is a sign representing the outdoor unit, and IC_ is a sign representing the indoor unit. Further, it is indicated that f( ) below is defined as a function for performing calculation using a value obtained from the refrigerant physical property table with respect to a numerical value that is in the parentheses and that is detected by a sensor.

[0135] hs: compressor suction enthalpy f(LS, TH5) [kJ / kg]

[0136] hco_ave: enthalpy of the refrigerant at the outlet of indoor heat exchanger 22 f(HS1, TH2ic_ave) [kJ / kg]

[0137] TH2ic_ave is an average value of refrigerant temperatures Th2ic at the outlets of all the indoor heat exchangers 22.

[0138] ρs: compressor suction density f(LS, TH5) [kg / m3]

[0139] fCOMP: compressor frequency [Hz]

[0140] St: excluded volume [cc]

[0141] ηv: volume efficiency 0.95 [-]

[0142] hco: enthalpy of the refrigerant at the outlet of indoor heat exchanger 22 f(HS1, TH2ic) [kJ / kg]

[0143] A: coefficient (value corresponding to the unit of pressure)

[0144] Cv: flow rate coefficient (value corresponding to the degree of opening determined by the specification of the expansion valve)

[0145] ρco: density of the refrigerant at the outlet of indoor heat exchanger 22 f(HS1, TH2ic) [kg / m3]

[0146] As described above, the insufficiency of the refrigerant can be determined when the falling-out of the feature amount with regard to the leakage of the refrigerant from the normal range and the decrease of heat exchange performance OC_Q from the normal range as described in Table 2 and Table 4 are simultaneously detected.

[0147] Further, the failure of the heat transfer of outdoor heat exchanger 12 can be determined when the falling-out of the feature amount with regard to the failure of the heat transfer of outdoor heat exchanger 12 from the normal range as described in Tables 2 and 4 and the decrease of heat exchange performance OC_Q from the normal range as described in Tables 3 and 5 are simultaneously detected.

[0148] Further, the failure of the heat transfer of indoor heat exchanger 22 can be determined when the falling-out of the feature amount with regard to the failure of the heat transfer of the indoor heat exchanger from the normal range as described in Tables 2 and 4 and the decrease of heat exchange performance IC_Q from the normal range as described in Tables 3 and 5 are simultaneously detected.

[0149] Further, the failure of the indoor expansion valve can be determined when the falling-out of feature amount IC_LEV with regard to the indoor expansion valve and the decrease of heat exchange performance IC_Q from the normal range are simultaneously detected.CONCLUSION

[0150] The learning device, the monitoring device, and the air conditioning system according to the present disclosure will be summarized with reference to the figures again.

[0151] (Item 1) The present disclosure relates to a learning device 4 configured to learn a condition of an air conditioning apparatus 40 in which refrigerant circulates. As shown in FIG. 4, operation data DA of air conditioning apparatus 40 includes a first data group D1A1 and a second data group D2A1 that does not include a same data element as a data element of first data group D1A1. Learning device 4 includes: a first calculation unit 110A configured to calculate a first feature amount F1A1 from first data group D1A1 of the air conditioning apparatus during a learning period; and a learning unit 120 configured to generate a first inference model M1 that infers a first normal range of first feature amount F1A1 from second data group D2A1 by performing supervised learning using second data group D2A1 with first feature amount F1A1 being set as truth data.

[0152] (Item 2) In the learning device according to item 1, as shown in FIG. 4, operation data DA includes a third data group D1A2 different from first data group D1A1, and a fourth data group D2A2 that does not contain a same data element as a data element of third data group D1A2. First calculation unit 110A is configured to calculate a second feature amount F1A2 from third data group D1A2 during the learning period. Learning unit 120 is configured to generate a second inference model M2 that infers a second normal range of second feature amount F1A2 from fourth data group D2A2 by performing supervised learning using fourth data group D2A2 with second feature amount F1A2 being set as truth data, second feature amount F1A2 being obtained by the calculation by first calculation unit 110A.

[0153] (Item 3) In another aspect, the present disclosure relates to a monitoring device 5 for an air conditioning apparatus 40, monitoring device 5 using first inference model M1 and second inference model M2 each generated by learning device 4 according to item 2, each of first inference model M1 and second inference model M2 being trained. As shown in FIG. 4, monitoring device 5 includes: a second calculation unit 110B configured to calculate first feature amount F1B1 from first data group D1B1 and second feature amount F1B2 from third data group D1B2 among operation data DB of air conditioning apparatus 40 during a monitoring period; an inference unit 220 configured to output, from second data group D2B1 among operation data DB of the air conditioning apparatus during the monitoring period, first normal range F2B1 using first inference model M1 trained, and to output second normal range F2B2 from fourth data group D2B2 among the operation data of the air conditioning apparatus during the monitoring period; and a determination unit 310 configured to determine whether or not the condition of the air conditioning apparatus is normal based on a frequency at which a first poor condition and a second poor condition are simultaneously detected, the first poor condition being a condition in which first feature amount F1B1 calculated by second calculation unit 110B falls out of first normal range F2B1 inferred by inference unit 220, the second poor condition being a condition in which second feature amount F1B2 calculated by second calculation unit 110B falls out of second normal range F2B2 inferred by inference unit 220.

[0154] (Item 4) In monitoring device 5 according to item 3, air conditioning apparatus 40 includes a refrigerant circuit C1 in which the refrigerant circulates. As shown in FIG. 2, refrigerant circuit C1 includes a compressor 11, a condenser (outdoor heat exchanger 12), an expansion valve 21, and an evaporator (indoor heat exchanger 22). The first feature amount is a degree of supercooling SC of the refrigerant at an outlet of the condenser, and the second feature amount is a heat exchange performance Qo of the condenser.

[0155] (Item 5) In monitoring device 5 according to item 4, as shown in Table 1, first data group D1B1 at least includes a pressure HS1 of the refrigerant discharged from the compressor and a temperature TH3 of the refrigerant at the outlet of the condenser, and second data group D2B1 at least includes a temperature TH1ic of an air conditioning target space, an operation frequency fCOMP of compressor 11, and a degree of opening Li of expansion valve 21. Third data group D1B2 at least includes pressure HS1 of the refrigerant discharged from compressor 11, refrigerant temperature TH3 at the outlet of the condenser, and operation frequency fCOMP of compressor 11, and fourth data group D2B2 at least includes temperature TH1ic of the air conditioning target space and degree of opening Li of expansion valve 21.

[0156] (Item 6) Monitoring device 5 according to item 3 further includes a display unit 320 configured to display a change in the frequency per certain period elapsed and configured to display that maintenance is necessary when the frequency exceeds a determination threshold value.

[0157] (Item 7) In still another aspect, the present disclosure relates to a monitoring device 5 for an air conditioning apparatus 40, monitoring device 5 using first inference model M1 generated by learning device 4 according to item 1, first inference model M1 being trained. As shown in FIG. 4, monitoring device 5 includes: a second calculation unit 110B configured to calculate first feature amount F1B1 from first data group D1B1 among operation data DB of air conditioning apparatus 40 during a monitoring period; an inference unit 220 configured to output, using first inference model M1 trained, first normal range F2B1 from second data group D2B1 among operation data DB of air conditioning apparatus 40 during the monitoring period; and a determination unit 310 configured to determine whether or not the condition of air conditioning apparatus 40 is normal by comparing first feature amount F1B1 calculated by second calculation unit 110B with first normal range F2B1 inferred by inference unit 220.

[0158] (Item 8) In monitoring device 5 according to item 7, air conditioning apparatus 40 includes a refrigerant circuit C1 in which the refrigerant circulates. As shown in FIG. 2, refrigerant circuit C1 includes a compressor 11, a condenser (outdoor heat exchanger 12), an expansion valve 21, and an evaporator (indoor heat exchanger 22). First feature amount F1B1 is a degree of supercooling SC of the refrigerant at the outlet of the condenser.

[0159] (Item 9) In the monitoring device according to item 8, as shown in Table 1, first data group D1B1 includes a pressure HS1 of the refrigerant discharged from compressor 11 and a temperature TH3 of the refrigerant at the outlet of the condenser, and second data group D2B1 at least includes a temperature TH1ic of an air conditioning target space and an operation frequency fCOMP of compressor 11.

[0160] (Item 10) In yet another aspect, the present disclosure relates to an air conditioning system 1000 shown in FIG. 1, air conditioning system 1000 including: monitoring device 5 according to any one of items 3 to 9; and air conditioning apparatus 40.

[0161] It is expected to implement the embodiments disclosed herein in appropriate combinations unless contradicted. The embodiments disclosed herein are illustrative and non-restrictive in any respect. The scope of the present disclosure is defined by the terms of the claims, rather than the embodiments described above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.REFERENCE SIGNS LIST1: abnormality detection system; 2, 32: CPU; 3, 33: memory; 4: learning device; 5: monitoring device; 10: outdoor unit; 11: compressor; 12, 22 heat exchanger; 13: four-way valve; 14: outdoor fan; 15: accumulator; 20: indoor unit; 21: expansion valve; 23: indoor fan; 30: controller; 40: air conditioning apparatus; 51, 52, 53, 54, 55, 56: temperature sensor; 57: humidity sensor; 61, 63: pressure sensor; 110, 110A, 110B: calculation unit; 120: learning unit; 220: inference unit; 310: determination unit; 311, 312: data processing unit; 320: display unit; 900: network; 1000: air conditioning system; C1: refrigerant circuit; M1, M2: inference model; Nw1: neural network.

Examples

first embodiment

[0021]FIG. 1 is a block diagram showing a configuration of an air conditioning system according to a first embodiment. An air conditioning system 1000 includes: an abnormality detection system 1; and an air conditioning apparatus 40 having a condition monitored by abnormality detection system 1. As shown in FIG. 1, abnormality detection system 1 is connected to air conditioning apparatus 40 via a network 900.

[0022]Abnormality detection system 1 includes a CPU (Central Processing Unit) 2, a memory 3 (a ROM (Read Only Memory) and a RAM (Random Access Memory)), an input / output buffer (not shown), and the like. CPU 2 loads a program stored in the ROM into the RAM or the like and executes the program. The program stored in the ROM is a program in which a processing procedure of abnormality detection system 1 is written. Abnormality detection system 1 monitors each device in air conditioning apparatus 40 in accordance with such a program. This control is not limited to a process by softwa...

TECHNICAL FIELD

[0001] The present disclosure relates to a learning device, a monitoring device, and an air conditioning system.BACKGROUND ART

[0002] Conventionally, a method of detecting leakage of refrigerant in a cooling system based on a value serving as an indicator of an amount of refrigerant introduced therein has been reviewed, and Japanese Patent No. 6791429 discloses an exemplary refrigerant amount determination device configured to facilitate determination of an amount of refrigerant.

[0003] The refrigerant amount determination device includes: a calculation unit configured to calculate a refrigerant amount indicator value from operation data of an air conditioning system; an inference unit configured to infer information about correction of the refrigerant amount indicator value using a correction model and at least one of the operation data and the calculated refrigerant amount indicator value; and a determination unit configured to determine the amount of the refrigerant of the air conditioning system based on the information about correction of the refrigerant amount indicator value.CITATION LISTPatent Literature

[0004] PTL 1: Japanese Patent No. 6791429SUMMARY OF INVENTIONTechnical Problem

[0005] In the refrigerant amount determination device disclosed in Japanese Patent No. 6791429, an influence of a decrease in the amount of the refrigerant over an air conditioning performance cannot be precisely known and a timing for maintenance cannot be known, disadvantageously. For example, when an amount of leakage of the refrigerant is small to be 10%, the air conditioning performance is not decreased so much even though it is known from the refrigerant amount indicator that the refrigerant is being leaked, with the result that it cannot be determined whether the location of the leakage should be checked urgently or the current situation should be kept and observed for a while.

[0006] The present disclosure has been made to solve the above-described problem, and has an object to provide a learning device so as to obtain an inference model useful to improve accuracy of detecting an abnormality of an air conditioning system.Solution to Problem

[0007] The present disclosure relates to a learning device configured to learn a condition of an air conditioning apparatus in which refrigerant circulates. Operation data of the air conditioning apparatus includes a first data group and a second data group that does not contain a same data element as a data element of the first data group. The learning device includes: a first calculation unit configured to calculate a first feature amount from the first data group of the air conditioning apparatus during a learning period; and a learning unit configured to generate a first inference model that infers a first normal range of the first feature amount from the second data group by performing supervised learning using the second data group with the first feature amount being set as truth data.Advantageous Effects of Invention

[0008] According to the learning device of the present disclosure, since the model that infers the normal range of the feature amount from the behavior of the data other than the data that can be directly used for the calculation of the feature amount among the operation data is generated, possibility of detecting an abnormality accurately at an early stage can be increased.BRIEF DESCRIPTION OF DRAWINGS

[0009] FIG. 1 is a block diagram showing a configuration of an air conditioning system according to a first embodiment.

[0010] FIG. 2 is a functional block diagram showing a configuration of an air conditioning apparatus 40 of FIG. 1.

[0011] FIG. 3 is a diagram showing exemplary operation data reflecting a condition of air conditioning apparatus 40 of FIG. 1.

[0012] FIG. 4 is a block diagram showing a configuration of an abnormality detection system 1 of FIG. 1.

[0013] FIG. 5 is a flowchart for illustrating a detail of a learning process performed by abnormality detection system 1.

[0014] FIG. 6 is a diagram showing an exemplary neural network Nw1 included in an inference model M1 constructed in a step S3 of FIG. 5.

[0015] FIG. 7 is a flowchart for illustrating a detail of a poor-condition detection process performed by abnormality detection system 1.

[0016] FIG. 8 is a diagram showing exemplary temporal changes of inferred values and calculated values of a heat exchange performance Qo and a degree of supercooling SC of refrigerant at an outlet of a condenser.

[0017] FIG. 9 is a graph showing a relation between an amount of change of the refrigerant and an amount of change of the degree of supercooling.

[0018] FIG. 10 is a graph showing a relation among decrease of the amount of the refrigerant, a performance decrease operation ratio, and a performance decrease ratio.

[0019] FIG. 11 is a diagram showing an exemplary trend graph displayed to a user.DESCRIPTION OF EMBODIMENTS

[0020] Hereinafter, embodiments of the present disclosure will be described in detail with reference to figures. It should be noted that the same or corresponding portions in the figures are denoted by the same reference characters and will not be described repeatedly in principle.First Embodiment

[0021] FIG. 1 is a block diagram showing a configuration of an air conditioning system according to a first embodiment. An air conditioning system 1000 includes: an abnormality detection system 1; and an air conditioning apparatus 40 having a condition monitored by abnormality detection system 1. As shown in FIG. 1, abnormality detection system 1 is connected to air conditioning apparatus 40 via a network 900.

[0022] Abnormality detection system 1 includes a CPU (Central Processing Unit) 2, a memory 3 (a ROM (Read Only Memory) and a RAM (Random Access Memory)), an input / output buffer (not shown), and the like. CPU 2 loads a program stored in the ROM into the RAM or the like and executes the program. The program stored in the ROM is a program in which a processing procedure of abnormality detection system 1 is written. Abnormality detection system 1 monitors each device in air conditioning apparatus 40 in accordance with such a program. This control is not limited to a process by software, and the process can also be performed by dedicated hardware (electronic circuit). It should be noted that abnormality detection system 1 may be constructed in a server connected to network 900.

[0023] Air conditioning apparatus 40 includes a plurality of indoor units 20, an outdoor unit 10, and a controller 30. Each of the plurality of indoor units 20 is disposed in an indoor space and is connected to outdoor unit 10 by a liquid pipe and a gas pipe through each of which refrigerant passes. Outdoor unit 10 is disposed in a space (outdoor space) outside the indoor space. It should be noted that the number of indoor units 20 included in air conditioning apparatus 40 may be one or two.

[0024] Outdoor unit 10 includes a compressor, an outdoor heat exchanger, and an outdoor fan. Each of the plurality of indoor units 20 includes an expansion valve and an indoor heat exchanger. The refrigerant is supplied from the compressor included in outdoor unit 10 to each of the plurality of indoor units 20. The refrigerant circulates between each of the plurality of indoor units 20 and outdoor unit 10.

[0025] Controller 30 includes a thermostat and integrally controls air conditioning apparatus 40. Controller 30 is connected to an abnormality detection system 1 via network 900. Network 900 includes the Internet and a cloud system. It should be noted that network 900 may be a LAN (local area network).

[0026] Controller 30 includes a CPU 32, a memory 33 (a ROM and a RAM), an input / output buffer (not shown), and the like. CPU 32 loads a program stored in the ROM into the RAM or the like and executes the program. The program stored in the ROM is a program in which a processing procedure of controller 30 is written. Controller 30 performs control of each device in air conditioning apparatus 40 in accordance with such a program. This control is not limited to a process by software, and the process can also be performed by dedicated hardware (electronic circuit).

[0027] FIG. 2 is a functional block diagram showing a configuration of air conditioning apparatus 40 of FIG. 1. As shown in FIG. 2, outdoor unit 10 includes a compressor 11, an outdoor heat exchanger 12, a four-way valve 13, an outdoor fan 14, an accumulator 15, temperature sensors 54 to 56, pressure sensors 61, 63, and a humidity sensor 57.

[0028] Each of the plurality of indoor units 20 includes an expansion valve 21, an indoor heat exchanger 22, an indoor fan 23, and temperature sensors 51 to 53. It should be noted that expansion valve 21 includes, for example, an LEV (linear expansion valve).

[0029] Operation modes of air conditioning apparatus 40 include a heating mode, a cooling mode, and a defrosting mode. In the heating mode, four-way valve 13 connects a discharge port of compressor 11 and indoor heat exchanger 22, and connects outdoor heat exchanger 12 and a refrigerant inlet of accumulator 15. In the heating mode, the refrigerant circulates through compressor 11, four-way valve 13, indoor heat exchanger 22, expansion valve 21, and outdoor heat exchanger 12 in this order. In each of the cooling mode and the defrosting mode, four-way valve 13 connects the discharge port of compressor 11 and outdoor heat exchanger 12, and connects indoor heat exchanger 22 and the refrigerant inlet of accumulator 15. In each of the cooling mode and the defrosting mode, the refrigerant circulates through compressor 11, four-way valve 13, outdoor heat exchanger 12, expansion valve 21, and indoor heat exchanger 22 in this order.

[0030] Temperature sensor 51 measures a temperature (room temperature TH1ic) of air suctioned into indoor heat exchanger 22, and outputs the temperature to controller 30. Temperature sensors 52, 53 respectively measure temperatures (indoor liquid temperature TH2ic and indoor gas temperature TH3ic) of the refrigerant before and after passing through indoor heat exchanger 22, and output the temperatures to controller 30.

[0031] Temperature sensor 54 measures a temperature (discharge temperature TH4) of the refrigerant discharged from compressor 11, and outputs the discharge temperature to controller 30. Temperature sensor 55 measures a temperature (suction temperature TH5) of the refrigerant suctioned into compressor 11 via accumulator 15, and outputs the suction temperature to controller 30. Temperature sensor 56 measures a temperature TH3 of the liquid refrigerant in the pipe that connects outdoor heat exchanger 12 and liquid pipe 41, and outputs the temperature to controller 30.

[0032] Pressure sensor 61 measures a pressure (discharge pressure HS1) of the refrigerant discharged from compressor 11, and outputs discharge pressure HS1 to controller 30. Pressure sensor 63 measures a pressure (suction pressure LS) of the refrigerant suctioned into compressor 11, and outputs suction pressure LS to controller 30.

[0033] Hereinafter, control during cooling will be described as a representative example. Controller 30 controls an amount of the refrigerant discharged by compressor 11 per unit time by controlling an operation frequency fCOMP of compressor 11 so as to allow a suction saturation gas temperature to become a target temperature. Controller 30 controls a degree of opening Li of expansion valve 21 so as to allow a degree of superheat SH (=TH3ic-TH2ic) of the refrigerant at the outlet of indoor heat exchanger 22 to become a target value. Controller 30 switches a circulation direction of the refrigerant by controlling four-way valve 13 to provide a flow path indicated by the solid lines. Controller 30 controls an amount of air sent by outdoor fan 14 per unit time by controlling a rotation frequency fFANo of outdoor fan 14 so as to allow a discharge saturation gas temperature to become a target value. Controller 30 controls a rotation frequency fFANi of indoor fan 23 so as to attain an amount of air set by a user. Controller 30 associates, with a time at the measurement, operation data reflecting the condition of the air conditioning system, and transmits it to abnormality detection system 1.

[0034] FIG. 3 is a diagram showing exemplary operation data reflecting the condition of air conditioning apparatus 40 of FIG. 1. As shown in FIG. 3, the operation data includes, for example, an outdoor air temperature, the discharge temperature (TH4), the evaporation temperature (TH2ic), a condensation temperature, the suction temperature (TH1ic), a send-out temperature, the high pressure (HS1), the low pressure (LS), the operation frequency (fCOMP) of compressor 11, the degree of opening of expansion valve 21, the operation modes, an operation state (operation, halt, or standby), and the rotation speed (fFANo, fFANi)) of each of outdoor fan 14 and indoor fan 23, the temperature (setting temperature) of the indoor space as set by the user, a current value of an inverter of compressor 11, a voltage value of the inverter, a temperature of a heat sink included in outdoor unit 10, and the temperature (liquid pipe temperature TH3) of the liquid pipe that connects outdoor unit 10 and indoor unit 20 (pipe through which the liquid refrigerant flows). It should be noted that the operation frequency of compressor 11, the degree of opening of expansion valve 21, and the rotation speed of outdoor fan 14 are basic operation amounts in VRF (Variable Refrigerant Flow) control.

[0035] An environment in which air conditioning apparatus 40 is operated may have characteristics specific to the environment (for example, a length of the refrigerant pipe, a type of indoor unit 20, the number of indoor units 20, and a difference in height between indoor unit 20 and outdoor unit 10). Therefore, a determination criterion (for example, a threshold value) for detecting an abnormality of air conditioning apparatus 40 can be different for each environment in which air conditioning apparatus 40 is operated. Therefore, when the same determination criterion is used regardless of the environment in which air conditioning apparatus 40 is operated, accuracy of detecting an abnormality of air conditioning apparatus 40 may be decreased.

[0036] Therefore, in abnormality detection system 1, when a normal condition is confirmed during a trial operation after installation of the air conditioning system, the normal condition is regarded as being kept for a subsequent certain period. Then, a trained model is generated in which a relation is learned between operation data of air conditioning apparatus 40 during the certain period and a normal value of a condition indicator value (specific parameter) representing the condition of air conditioning apparatus 40 and corresponding to the operation data. By using the trained model, an abnormality of air conditioning apparatus 40 can be detected in accordance with a determination criterion adapted to the environment in which air conditioning apparatus 40 is operated. As a result, the accuracy of detecting an abnormality of the air conditioning system can be improved.

[0037] FIG. 4 is a block diagram showing a configuration of abnormality detection system 1 of FIG. 1. In abnormality detection system 1, operation data is obtained in each of a plurality of consecutive operation periods (for example, one day, one week, or one month), and trained inference models M1, M2 are constructed using training data including the operation data. The plurality of consecutive operation periods may be the same period (predetermined period) or may be different periods.

[0038] As shown in FIG. 4, abnormality detection system 1 includes a learning device 4 and a monitoring device 5. Learning device 4 includes a calculation unit 110A and a learning unit 120. Monitoring device 5 includes a calculation unit 110B, an inference unit 220, a determination unit 310, and a display unit 320.

[0039] Calculation unit 110A and calculation unit 110B basically perform the same calculation and are only different from each other in that they are used respectively at the time of learning and at the time of monitoring. Therefore, calculation unit 110A may also serve as calculation unit 110B.

[0040] Calculation unit 110A calculates a feature amount F1A1 from operation data D1A1 at the time of the normal condition. Learning unit 120 performs machine learning with feature amount F1A1 of an element device and operation data D2A1 of the element device at the time of the normal condition so as to construct inference model M1. Inference unit 220 infers a normal range (upper and lower limit values) F2B1 of the feature amount by inputting, to trained inference model M1, operation data D2B1 at the time of determination. Further, calculation unit 110A calculates feature amount F1A2 from operation data D1A2 at the time of the normal condition. Learning unit 120 performs machine learning with feature amount F1A2 of an element device and operation data D2A2 of the element device at the time of the normal condition so as to construct inference model M2. Inference unit 220 infers a normal range (upper and lower limit values) F2B2 of the feature amount by inputting, to trained inference model M2, operation data D2B2 at the time of determination.

[0041] Calculation unit 110B calculates a feature amount F1B1 from operation data D1B1 at the time of determination. Calculation unit 110B calculates a feature amount F1B2 from operation data D1B2 at the time of determination. Determination unit 310 includes a data processing unit 311 and a data processing unit 312. Data processing unit 311 counts the number of pieces of data (data falling out of the normal range) for which feature amount F1B1 (degree of supercooling SC of the refrigerant at the outlet of the condenser) calculated by calculation unit 110 from operation data D1B1 at the time of determination exceeds or falls below normal range F2B1 inferred by trained inference model M1 from the operation data at the time of poor condition. Data processing unit 312 counts the number of pieces of data for which feature amount F1B2 (heat exchange performance Qo of the heat exchanger) calculated from operation data D1B2 at the time of determination is equal to or less than a predefined performance. Determination unit 310 calculates a ratio by dividing, by the determined number of pieces of operation data, the number of pieces of data for which feature amount F1B1 (degree of supercooling SC) is outside the normal range (falls out of the normal range) and feature amount F1B2 (heat exchange performance Qo) is decreased by a predefined amount (falls out of the normal range) at the same time. This ratio represents a ratio of times in a day at each of which an operation involving a heat exchange performance Qo decreased to be lower than that in the normal condition takes place due to a poor condition of the element device, and is referred to as “performance decrease operation ratio” in the present specification. Determination unit 310 determines whether or not the performance decrease operation ratio is equal to or more than a threshold value.

[0042] Display unit 320 displays trend data indicating a change of the amount of the refrigerant and a change of the performance decrease operation ratio per day for each element device (outdoor heat exchanger 12, indoor heat exchanger 22), and displays that maintenance is necessary when the performance decrease operation ratio is equal to or more than the threshold value.

[0043] CPU 2 shown in FIG. 1 is operated as calculation units 110A, 110B, learning unit 120, inference unit 220, and determination unit 310 in accordance with corresponding programs. Further, memory 3 stores an operation data set of air conditioning apparatus 40 and inference models M1, M2.

[0044] Each of inference models M1, M2 is a regression model that includes a neural network and that infers a normal value of the condition indicator value of air conditioning apparatus 40 from the operation data of air conditioning apparatus 40. Each of inference models M1, M2 may be a classification model that infers a level (classification) of the condition indicator value. The normal value of the condition indicator value may be each of a maximum value and a minimum value of a confidence interval that the condition indicator value can have when air conditioning apparatus 40 is normal. Further, a range having the normal value as a median value (for example, a range of ±10% of the normal value) may be employed as the confidence interval.

[0045] It should be noted that a general AI (Artificial Intelligence) technology can be applied to clustering and weighting of the parameters included in the operation data.

[0046] Learning unit 120 performs machine learning onto inference models M1, M2 so as to construct trained inference models M1, M2. A machine learning algorithm used by learning unit 120 may be a known algorithm such as supervised learning, semi-supervised learning, unsupervised learning, or reinforcement learning. Further, as the machine learning algorithm, deep learning in which extraction of the feature amount itself is learned can be used, or the machine learning may be performed in accordance with another known method such as a neural network, genetic programming, inductive logic programming, or a support vector machine. Hereinafter, a case where the supervised learning is applied to a neural network will be described.

[0047] Calculation unit 110A calculates a condition indicator value (SC) during a first operation period from operation data D1A1 included in operation data DA. Calculation unit 110A calculates a condition indicator value (Qo) during the first operation period from operation data D1A2 included in operation data DA. Learning unit 120 performs the supervised learning onto inference model M1 using training data having the condition indicator value during the first operation period as truth data (teaching data) for the first operation data. Learning unit 120 constructs trained inference models M1, M2, and stores the models into memory 3. It should be noted that since the first operation period is a period immediately after air conditioning apparatus 40 is confirmed to be operated in the trial operation after being installed in the installation location, there is a low possibility that air conditioning apparatus 40 has an abnormality. Since it can be assumed that air conditioning apparatus 40 in the first operation period is normal, the condition indicator value during the first operation period is set as the truth data for the first operation data. In the below-described example, the first operation period is 365 days immediately after the installation.

[0048] FIG. 5 is a flowchart for illustrating a detail of the learning process performed by abnormality detection system 1. When the learning process is started, abnormality detection system 1 shown in FIG. 4 obtains air conditioning operation data DA for 365 days immediately after the installation in step S1. Air conditioning operation data DA includes a first data group D1A1 and a second data group D2A1 that does not contain the same data element as a data element of first data group D1A1. Further, air conditioning operation data DA includes a third data group D1A2 different from first data group D1A1 and a fourth data group D2A2 that does not contain the same data element as a data element of third data group D1A2.

[0049] Then, in a step S2, calculation unit 110A calculates a feature amount F1A1 from first data group D1A1 and calculates feature amount F1A2 from third data group D1A2. Examples of feature amount F1A1 include degree of supercooling SC of the refrigerant at the outlet of the condenser. Examples of feature amount F1A2 include heat exchange performance Qo of outdoor heat exchanger 12.

[0050] For example, degree of supercooling SC at the outlet of the condenser can be calculated by the following formula (1):SC=Tc-TH⁢3(1)

[0051] Here, Tc represents the discharge saturation gas temperature and is a value determined by discharge pressure HS1. TH3 represents a liquid refrigerant temperature.

[0052] Further, for example, heat exchange performance Qo can be calculated by the following formula (2):Qo=Gr×(Hd-Hco)(2)

[0053] Here, Gr represents a refrigerant circulation amount and is a value determined by operation frequency fCOMP of compressor 11, and Hd represents a specific enthalpy of the inlet portion of outdoor heat exchanger 12 and is a value determined by pressure HS1 and temperature TH4. Further, Hco represents a specific enthalpy of the outlet of outdoor heat exchanger 12 and is a value determined by pressure HS1 and temperature TH3.

[0054] The upper part of Table 1 below shows: operation data DA used when training inference model M1 for determining leakage of the refrigerant; operation data D2B1 used as an input to trained inference model M1; and operation data D1B1 used by calculation unit 110B at the time of determination.

[0055] Further, the lower part of Table 1 shows: operation data DA used when training inference model M2 for determining insufficiency of the heat exchange performance of outdoor heat exchanger 12; operation data D2B2 used as an input to trained inference model M2; and operation data D1B2 used by calculation unit 110B at the time of determination.TABLE 1FeatureAmount(F1B1)Used for Learning (DA)Used for Inference (D2B1)Used for Calculation (D1B1)Degree ofDischarge Temperature (TH4)Discharge Temperature (TH4)SupercoolingDischarge Pressure (HS1)Discharge Pressure (HS1)SCLiquid Temperature (TH3)Liquid Temperature (TH3)(Leakage ofIndoor Liquid Temperature (TH2ic)Indoor Liquid Temperature (TH2ic)Refrigerant)Indoor Gas Temperature (TH3ic)Indoor Gas Temperature (TH3ic)Suction Temperature (TH5)Suction Temperature (TH5)Suction Pressure (LS)Suction Pressure (LS)Outdoor Air Relative Humidity (TH7)Outdoor Air Relative Humidity (TH7)Room Temperature (TH1ic)Room Temperature (TH1ic)Frequency of Compressor (fCOMP)Frequency of Compressor (fCOMP)Frequency of Fan (fFANo)Frequency of Fan (fFANo)Degree of Opening of Indoor LEV (Li)Degree of Opening of Indoor LEV (Li)FeatureAmount(F1B2)Used for Learning (DA)Used for Inference (D2B2)Used for Calculation (D1B2)PerformanceDischarge Temperature (TH4)Discharge Temperature (TH4)QoDischarge Pressure (HS1)Discharge Pressure (HS1)Liquid Temperature (TH3)Liquid Temperature (TH3)Indoor Liquid Temperature (TH2ic)Indoor Liquid Temperature (TH2ic)Indoor Gas Temperature (TH3ic)Indoor Gas Temperature (TH3ic)Suction Temperature (TH5)Suction Temperature (TH5)Suction Pressure (LS)Suction Pressure (LS)Outdoor Air Relative Humidity (TH7)Outdoor Air Relative Humidity (TH7)Room Temperature (TH1ic)Room Temperature (TH1ic)Frequency of Compressor (fCOMP)Frequency of Compressor (fCOMP)Frequency of Fan (fFANo)Frequency of Fan (fFANo)Degree of Opening of Indoor LEV (Li)Degree of Opening of Indoor LEV (Li)

[0056] Then, in a step S3 of FIG. 5, learning unit 120 constructs inference model (regression model) M1 of the objective variable (feature amount F1A1) corresponding to the explanatory variable (operation data group D2A1). Similarly, learning unit 120 constructs inference model (regression model) M2 of the objective variable (feature amount F1A2) corresponding to the explanatory variable (operation data group D2A2). On this occasion, learning unit 120 constructs each inference model with the number of feature amounts×3 (upper limit value, median value, and lower limit value). For example, the upper limit value may be 97.5% of the confidence interval and the lower limit value may be 2.5% of the confidence interval (corresponding to 2σ).

[0057] FIG. 6 is a diagram showing an exemplary neural network Nw1 included in inference model M1 constructed in step S3 of FIG. 5. As shown in FIG. 6, neural network Nw1 includes an input layer X10, an intermediate layer (hidden layer) Y10, and an output layer Z10. Input layer X10 includes neurons X11, X12, X13. Intermediate layer Y10 includes neurons Y11, Y12. Output layer Z10 includes a neuron Z11. Input layer X10 and intermediate layer Y10 are fully connected to each other. Intermediate layer Y10 and output layer Z10 are fully connected to each other. Neural network Nw1 may include two or more intermediate layers.

[0058] When a plurality of inputs are respectively input to neurons X11 to X13 of input layer X10, the values are multiplied by weights w11, w12, w13, w14, w15, w16 and are input to neurons Y11, Y12 of intermediate layer Y10. The outputs from neurons Y11, Y12 are multiplied by weights w21, w22 and are output from neuron Z11 of output layer Z10. The output result from output layer Z10 differs depending on the values of weights w11 to w16, w21, w22. The weights and biases of neural network Nw1 are updated by backpropagation of an error between the truth data and the result output from the output layer as a result of inputting the operation data to the input layer such that the result becomes close to the truth data.

[0059] In this way, the training of inference model M1 is completed. It should be noted that since the same applies to inference model M2, the explanation will not be described repeatedly. The following describes a detail of a poor-condition detection process after the completion of the learning.

[0060] FIG. 7 is a flowchart for illustrating the detail of the poor-condition detection process performed by abnormality detection system 1.

[0061] First, in a step S11, abnormality detection system 1 shown in FIG. 4 obtains operation data DB for one day for which determination for a poor condition is to be made.

[0062] Then, in a step S12, calculation unit 110B of FIG. 4 calculates feature amount F1B1 from operation data D1B1 among obtained operation data DB, and calculates feature amount F1B2 from operation data D1B2 among obtained operation data DB. Feature amount F1B1 is, for example, degree of supercooling SC of the refrigerant at the outlet of the condenser, and feature amount F1B2 is heat exchange performance Qo of outdoor heat exchanger 12. Regarding these, the calculation process have already been described with each of the above-described formulas (1) and (2), and therefore will not be described repeatedly. In parallel with this, in step S12, inference unit 220 of FIG. 4 uses trained model M1 to infer a degree of supercooling SCi (upper limit value, lower limit value, median value SCAI) at the outlet of the condenser as inference value F2B1 from operation data D2B1, and inference unit 220 uses trained model M2 to infer a heat exchange performance QoAI (median inference value) of outdoor heat exchanger 12 as inference value F2B2 from operation data D2B2.

[0063] Then, in a step S13, determination unit 310 extracts operation data for which heat exchange performance Qo of outdoor heat exchanger 12 calculated by calculation unit 110B using the operation data obtained in step S11 is decreased by a predefined amount (for example, 20% or more) with respect to median inference value QoAI provided by inference model M2. The extraction of the data is performed by, for example, flagging a set of operation data. It should be noted that the predefined amount at least includes an amount of decrease of heat exchange performance Qo with which cooling is not attained (performance decrease of −20%).

[0064] FIG. 8 is a diagram showing exemplary temporal changes of inferred values and calculated values of heat exchange performance Qo and degree of supercooling SC of the refrigerant at the outlet of the condenser. Decrease of the amount of the refrigerant is less than −10% during a period from a time t0 to a time t1, the decrease of the amount of the refrigerant is −10% during a subsequent period TP1, and the decrease of the amount of refrigerant is −20% during a period TP2.

[0065] As the refrigerant decrease amount shown in the upper part of FIG. 8 is larger, the number of pieces of extracted data indicated by white circles are also larger.

[0066] A step S14 is performed after step S13 in FIG. 7. In step S14, determination unit 310 extracts, from the operation data obtained in step S11, data for which degree of supercooling SC of the refrigerant at the outlet of the condenser falls below the lower limit value. The lower limit value is the value inferred by inference model M1.

[0067] Then, in a step S15, determination unit 310 calculates a ratio (performance decrease operation ratio R) of pieces of data for which heat exchange performance Qo is decreased to be lower than the lower limit value and degree of supercooling SC of the refrigerant at the outlet of the condenser is decreased to be lower than the lower limit value. Ratio R is a value obtained by dividing, by the total number of the pieces of the operation data, the number of the pieces of the data extracted in steps S13 and S14. That is, performance decrease operation ratio R is as follows: (the number of times of observing the pieces of the data flagged in both S13 and S14) / (the total number of times of making observations).

[0068] Thus, since the condition for the extraction is an AND condition for the data for which heat exchange performance Qo is decreased and the data for which each feature amount falls out of the appropriate range, the performance decrease due to the poor condition of each element can be specified. With this, for example, decreased performance caused by failure of heat transfer of outdoor heat exchanger 12 due to rust, adhesion of dust, or the like is not detected. The data extracted in this way is indicated by circles in FIG. 8.

[0069] For example, since heat exchange performance Qo of outdoor heat exchanger 12 is smaller than the lower limit value of −20% from the median value and is equal to or smaller than the inferred lower limit value of degree of supercooling SC of the refrigerant at the outlet of the condenser at a time t2, data is extracted in step S15 and is used for the calculation of the performance decrease operation ratio. On the other hand, at a time t7, degree of supercooling SC of the refrigerant at the outlet of the condenser is equal to or less than the inferred lower limit value but heat exchange performance Qo is not equal to or less than the lower limit value of −20% from the median value, and no data is therefore extracted in step S15.

[0070] The number of the pieces of the data thus extracted from the operation data for one day is divided by the total number of the pieces of the operation data for one day, thereby calculating performance decrease operation ratio R.

[0071] Then, in a step S16, display unit 320 displays a trend graph indicating that the feature amount is changed as the time elapses. When the feature amount becomes equal to or more than the threshold value, the user (maintenance worker) is notified of the poor condition.

[0072] FIG. 9 is a graph showing a relation between an amount of change of the refrigerant and an amount of change of the degree of supercooling. As shown in the lower part of FIG. 8, a difference ΔSC between inferred median value SCAI and degree of supercooling SC, which is a calculated feature amount, tends to be larger as the amount of the refrigerant is decreased, thereby finding out a relation in which the amount of change (amount of decrease) of the refrigerant and ΔSC are uniquely determined as shown in FIG. 9. For example, when the correlation between ΔSC and the liquid refrigerant amount as shown in FIG. 9 is set for each apparatus model, the amount of change of the refrigerant can be estimated. ΔSC may be calculated by the following formula. It should be noted that an index “ave” refers to an average in a predetermined period.Δ⁢SC=SCAI_ave-SC_ave

[0073] FIG. 10 is a graph showing a relation among the decrease of the amount of the refrigerant, the performance decrease operation ratio, and the performance decrease ratio. In FIG. 10, the horizontal axis represents the performance decrease ratio (Qo / QoAI), and the vertical axis represents the performance decrease operation ratio.

[0074] FIG. 11 is a diagram showing an exemplary trend graph displayed to the user. FIG. 11 shows a temporal change of the performance decrease operation ratio when the performance decrease ratio in FIG. 10 is −10%. For example, in the case of slow leakage of the refrigerant, the maintenance worker can confirm that the times of occurrence of performance decrease due to the leakage of the refrigerant are increased day after day, and therefore can determine a timing for maintenance.

[0075] When the leakage of the refrigerant is progressed with passage of time (−10%→−30%), the trend data of the performance decrease ratio is increased to exceed a threshold value R1. When the data exceeds threshold value R1, display unit 320 displays the poor condition on the screen for the sake of notification thereof.

[0076] Further, when an amount of insufficiency of the refrigerant corresponding to ΔSC derived from the operation data is also shown based on the relation shown in FIG. 9, the maintenance worker can predict and prepare an amount of refrigerant to be added and replenished.

[0077] According to monitoring device 5 of the present embodiment as described above, the ratio (performance decrease operation ratio) of the times of occurrence of the operation in a day in which degree of supercooling SC, which is a feature amount representing the leakage of the refrigerant, falls out of the normal range and heat exchange performance Qo is decreased to be lower than that in the normal condition is sorted as the trend data per day, and when the poor condition determination threshold value is exceeded, the maintenance worker is notified of a timing for maintenance. Thus, the maintenance worker can know the timing to perform the maintenance work.

[0078] Further, since the trend data is displayed rather than an instantaneous value, the poor condition can be correctly determined even when the threshold value is instantaneously exceeded due to an internal or external disturbance.

[0079] For example, even when the refrigerant is insufficient, the same performance Qo may be exhibited under the same environment condition as that in the normal condition. This is because performance Qo is balanced with an air conditioning load by control. However, since the operation condition of air conditioning apparatus 40 is changed when the refrigerant is insufficient, the median inference value of performance Qo inferred from the operation condition is increased to cause the calculated value of performance Qo to fall below the inferred normal range, with the result that the decrease of the performance can be detected. Therefore, the insufficiency of the refrigerant can be detected in an intermediate season such as spring, autumn, or the like, with the result that insufficiency of the performance can be prevented in advance before summer during which full operation is performed for cooling.Various Modifications

[0080] It should be noted that the leakage of the refrigerant during the cooling has been described with reference to the flowcharts of FIGS. 5 and 7 and has been explained using the formula (1); however, also in the case of failure of heat transfer of outdoor heat exchanger 12, failure of heat transfer of indoor heat exchanger 22, failure of the indoor expansion valve, and failure of the indoor expansion valve during the cooling, the feature amounts can be changed as shown in the below-described Table 2 and the poor condition can be determined by the same process.TABLE 2Type of PoorCondition(During Cooling)Feature AmountLeakage of RefrigerantOC_SC = Tc − TH3Failure of HeatOC_GaCpε = Gr × (hd − hco) / (Tc − TH7)Transfer of OutdoorGr = ρs × fCOMP × St ×ηv × 10−6Heat ExchangerFailure of HeatIC_GaCpε = Gr_lev × (hs − hco) / (TH1ic −Transfer of IndoorTH2ic)ExchangerGr_lev = A × Cv ×√{square root over (ρco × (HS1 − LS))}Failure of IndoorIC_LEV = LiExpansion Valve

[0081] In Table 2, details of the variables are as follows. It should be noted that OC is a sign representing the outdoor unit, and IC_ is a sign representing the indoor unit. Further, it is indicated that f( ) below is defined as a function for performing calculation using a value obtained from a refrigerant physical property table with respect to a numerical value that is in the parentheses and that is detected by a sensor.

[0082] Tc: compressor discharge saturated gas temperature f(HS1) “° C.”

[0083] hd: compressor discharge enthalpy f(HS1, TH4) [kJ / kg]

[0084] hco: enthalpy of the refrigerant at the outlet of outdoor heat exchanger 12 f(HS1, TH3) [kJ / kg]

[0085] ρs: compressor suction density f(LS, TH5) [kg / m3]

[0086] fCOMP: compressor frequency [Hz]

[0087] St: excluded volume [cc]

[0088] ηv: volume efficiency 0.95 [-]

[0089] hs: compressor suction enthalpy f(LS, TH5) [kJ / kg]

[0090] A: coefficient (value corresponding to the unit of pressure)

[0091] Cv: flow rate coefficient (value corresponding to the specification and degree of opening of the LEV)

[0092] ρco: density of the refrigerant at the outlet of outdoor heat exchanger 12 f(HS1, TH3) [kg / m3]

[0093] Further, the tendencies in each of which the feature amount falls out of the normal range during the cooling are as follows.

[0094] Leakage of refrigerant: falling below the normal range

[0095] Failure of heat transfer of outdoor heat exchanger 12: falling below the normal range

[0096] Failure of heat transfer of indoor heat exchanger 22: falling below the normal range

[0097] Failure of indoor expansion valve: exceeding and falling below the normal range

[0098] Further, the poor condition of outdoor heat exchanger 12 with regard to the heat exchange performance during the cooling has been shown in the formula (2); however, the poor condition of indoor heat exchanger 22 with regard to the heat exchange performance can also be determined by the same process with the feature amount being changed as shown in Table 3 below.TABLE 3Type of Poor Condition (DuringCooling)Feature AmountOutdoor Heat ExchangerOC_Q = Gr × (hd − hco)Heat Exchange PerformanceGr = ρs × fCOMP × St ×ηv × 10−6Indoor Heat ExchangerIC_Q = Gr_lev × (hs − hco)Heat Exchange PerformanceGr_lev = A ×Cv ×√{square root over (ρco × (HS1 − LS))}

[0099] In Table 3, details of the variables are as follows. It should be noted that OC_ is a sign representing the outdoor unit, and IC_ is a sign representing the indoor unit. Further, it is indicated that f( ) below is defined as a function for performing calculation using a value obtained from the refrigerant physical property table with respect to a numerical value that is in the parentheses and that is detected by a sensor.

[0100] hd: compressor discharge enthalpy f(HS1, TH4) [kJ / kg]

[0101] hco: enthalpy of the refrigerant at the outlet of outdoor heat exchanger 12

[0102] f(HS1, TH3) [kJ / kg]

[0103] ρs: compressor suction density f(LS, TH5) [kg / m3]

[0104] fCOMP: compressor frequency [Hz]

[0105] St: excluded volume [cc]

[0106] ηv: volume efficiency 0.95 [-]

[0107] hs: compressor suction enthalpy f(LS, TH5) [kJ / kg]

[0108] A: coefficient (value corresponding to the unit of pressure)

[0109] Cv: flow rate coefficient (value corresponding to the specification and degree of opening of the LEV)

[0110] ρco: density of the refrigerant at the outlet of outdoor heat exchanger 12 f(HS1, TH3) [kg / m3]

[0111] Further, also in the case of the leakage of the refrigerant, the failure of heat transfer of outdoor heat exchanger 12, the failure of the heat transfer of indoor heat exchanger 22, and the failure of the indoor expansion valve during the heating, the feature amounts can be changed as shown in Table 4 below and the poor condition can be determined by the same process.TABLE 4Type of PoorCondition(During Heating)Feature AmountLeakage of RefrigerantOC_SH = TH5 − TeFailure of HeatOC_Gaε = Gr × (hs − hco_ave) / (Ia7 − Ie)Transfer of OutdoorGr = ρs × fCOMP × St x ηv × 10−6Heat ExchangerFailure of HeatIC_GaCpε = Gr_lev × (hs − hco) / (TH1ic −Transfer of IndoorTH2ic)Heat ExchangerGr_lev = A × Cv ×√{square root over (ρco × (HS1 − LS))}Failure of IndoorIC_LEV = LiExpansion Valve

[0112] In Table 4, details of the variables are as follows. It should be noted that OC is a sign representing the outdoor unit, and IC_ is a sign representing the indoor unit. Further, it is indicated that f( ) below is defined as a function for performing calculation using a value obtained from the refrigerant physical property table with respect to a numerical value that is in the parentheses and that is detected by a sensor.

[0113] Te: compressor suction saturated gas temperature f(LS) [C]

[0114] hs: compressor suction enthalpy f(LS, TH5) [kJ / kg]

[0115] hco_ave: enthalpy of the refrigerant at the outlet of indoor heat exchanger 22

[0116] f(HS1, TH2ic_ave) [kJ / kg]

[0117] TH2ic_ave is an average value of outlet temperatures Th2ic of all the indoor heat exchangers 22.

[0118] Ia7: outdoor air enthalpy f(TH7, relative humidity) [kJ / kg]

[0119] Ie: surface enthalpy of outdoor heat exchanger 12 f(Te, relative humidity 100%) [kJ / kg]

[0120] ρs: compressor suction density f(LS, TH5) [kg / m3]

[0121] fCOMP: compressor frequency [Hz]

[0122] St: excluded volume [cc]

[0123] ηv: volume efficiency 0.95 [-]

[0124] hco: enthalpy of the refrigerant at the outlet of indoor heat exchanger 22 f(HS1, TH2ic) [kJ / kg]

[0125] A: coefficient (value corresponding to the unit of pressure)

[0126] Cv: flow rate coefficient (value corresponding to specification and degree of opening of the LEV)

[0127] ρco: density of the refrigerant at the outlet of indoor heat exchanger 22 f(HS1, TH2ic) [kg / m3]

[0128] Further, the tendencies in each of which the feature amount falls out of the normal range during the heating are as follows.

[0129] Leakage of refrigerant: exceeding the normal range

[0130] Failure of heat transfer of outdoor heat exchanger 12: falling below the normal range

[0131] Failure of heat transfer of indoor heat exchanger 22: falling below the normal range

[0132] Failure of indoor expansion valve: exceeding and falling below the normal range

[0133] Further, each of the poor condition of outdoor heat exchanger 12 with regard to the heat exchange performance and the poor condition of indoor heat exchanger 22 with regard to the heat exchange performance during the heating can be determined by the same process with the feature amount being changed as shown in Table 5 below.TABLE 5Type of Poor Condition (DuringHeating)Feature AmountOutdoor Heat ExchangerOC_Q = Gr × (hs − hco_ave)Heat Exchange PerformanceGr = ρs × fCOMP × St ×ηv × 10−6Indoor Heat ExchangerIC_Q = Gr_lev × (hs − hco)Heat Exchange PerformanceGr_lev = A ×Cv ×√{square root over (ρco × (HS1 − LS))}

[0134] In Table 5, details of the variables are as follows. It should be noted that OC_ is a sign representing the outdoor unit, and IC_ is a sign representing the indoor unit. Further, it is indicated that f( ) below is defined as a function for performing calculation using a value obtained from the refrigerant physical property table with respect to a numerical value that is in the parentheses and that is detected by a sensor.

[0135] hs: compressor suction enthalpy f(LS, TH5) [kJ / kg]

[0136] hco_ave: enthalpy of the refrigerant at the outlet of indoor heat exchanger 22 f(HS1, TH2ic_ave) [kJ / kg]

[0137] TH2ic_ave is an average value of refrigerant temperatures Th2ic at the outlets of all the indoor heat exchangers 22.

[0138] ρs: compressor suction density f(LS, TH5) [kg / m3]

[0139] fCOMP: compressor frequency [Hz]

[0140] St: excluded volume [cc]

[0141] ηv: volume efficiency 0.95 [-]

[0142] hco: enthalpy of the refrigerant at the outlet of indoor heat exchanger 22 f(HS1, TH2ic) [kJ / kg]

[0143] A: coefficient (value corresponding to the unit of pressure)

[0144] Cv: flow rate coefficient (value corresponding to the degree of opening determined by the specification of the expansion valve)

[0145] ρco: density of the refrigerant at the outlet of indoor heat exchanger 22 f(HS1, TH2ic) [kg / m3]

[0146] As described above, the insufficiency of the refrigerant can be determined when the falling-out of the feature amount with regard to the leakage of the refrigerant from the normal range and the decrease of heat exchange performance OC_Q from the normal range as described in Table 2 and Table 4 are simultaneously detected.

[0147] Further, the failure of the heat transfer of outdoor heat exchanger 12 can be determined when the falling-out of the feature amount with regard to the failure of the heat transfer of outdoor heat exchanger 12 from the normal range as described in Tables 2 and 4 and the decrease of heat exchange performance OC_Q from the normal range as described in Tables 3 and 5 are simultaneously detected.

[0148] Further, the failure of the heat transfer of indoor heat exchanger 22 can be determined when the falling-out of the feature amount with regard to the failure of the heat transfer of the indoor heat exchanger from the normal range as described in Tables 2 and 4 and the decrease of heat exchange performance IC_Q from the normal range as described in Tables 3 and 5 are simultaneously detected.

[0149] Further, the failure of the indoor expansion valve can be determined when the falling-out of feature amount IC_LEV with regard to the indoor expansion valve and the decrease of heat exchange performance IC_Q from the normal range are simultaneously detected.CONCLUSION

[0150] The learning device, the monitoring device, and the air conditioning system according to the present disclosure will be summarized with reference to the figures again.

[0151] (Item 1) The present disclosure relates to a learning device 4 configured to learn a condition of an air conditioning apparatus 40 in which refrigerant circulates. As shown in FIG. 4, operation data DA of air conditioning apparatus 40 includes a first data group D1A1 and a second data group D2A1 that does not include a same data element as a data element of first data group D1A1. Learning device 4 includes: a first calculation unit 110A configured to calculate a first feature amount F1A1 from first data group D1A1 of the air conditioning apparatus during a learning period; and a learning unit 120 configured to generate a first inference model M1 that infers a first normal range of first feature amount F1A1 from second data group D2A1 by performing supervised learning using second data group D2A1 with first feature amount F1A1 being set as truth data.

[0152] (Item 2) In the learning device according to item 1, as shown in FIG. 4, operation data DA includes a third data group D1A2 different from first data group D1A1, and a fourth data group D2A2 that does not contain a same data element as a data element of third data group D1A2. First calculation unit 110A is configured to calculate a second feature amount F1A2 from third data group D1A2 during the learning period. Learning unit 120 is configured to generate a second inference model M2 that infers a second normal range of second feature amount F1A2 from fourth data group D2A2 by performing supervised learning using fourth data group D2A2 with second feature amount F1A2 being set as truth data, second feature amount F1A2 being obtained by the calculation by first calculation unit 110A.

[0153] (Item 3) In another aspect, the present disclosure relates to a monitoring device 5 for an air conditioning apparatus 40, monitoring device 5 using first inference model M1 and second inference model M2 each generated by learning device 4 according to item 2, each of first inference model M1 and second inference model M2 being trained. As shown in FIG. 4, monitoring device 5 includes: a second calculation unit 110B configured to calculate first feature amount F1B1 from first data group D1B1 and second feature amount F1B2 from third data group D1B2 among operation data DB of air conditioning apparatus 40 during a monitoring period; an inference unit 220 configured to output, from second data group D2B1 among operation data DB of the air conditioning apparatus during the monitoring period, first normal range F2B1 using first inference model M1 trained, and to output second normal range F2B2 from fourth data group D2B2 among the operation data of the air conditioning apparatus during the monitoring period; and a determination unit 310 configured to determine whether or not the condition of the air conditioning apparatus is normal based on a frequency at which a first poor condition and a second poor condition are simultaneously detected, the first poor condition being a condition in which first feature amount F1B1 calculated by second calculation unit 110B falls out of first normal range F2B1 inferred by inference unit 220, the second poor condition being a condition in which second feature amount F1B2 calculated by second calculation unit 110B falls out of second normal range F2B2 inferred by inference unit 220.

[0154] (Item 4) In monitoring device 5 according to item 3, air conditioning apparatus 40 includes a refrigerant circuit C1 in which the refrigerant circulates. As shown in FIG. 2, refrigerant circuit C1 includes a compressor 11, a condenser (outdoor heat exchanger 12), an expansion valve 21, and an evaporator (indoor heat exchanger 22). The first feature amount is a degree of supercooling SC of the refrigerant at an outlet of the condenser, and the second feature amount is a heat exchange performance Qo of the condenser.

[0155] (Item 5) In monitoring device 5 according to item 4, as shown in Table 1, first data group D1B1 at least includes a pressure HS1 of the refrigerant discharged from the compressor and a temperature TH3 of the refrigerant at the outlet of the condenser, and second data group D2B1 at least includes a temperature TH1ic of an air conditioning target space, an operation frequency fCOMP of compressor 11, and a degree of opening Li of expansion valve 21. Third data group D1B2 at least includes pressure HS1 of the refrigerant discharged from compressor 11, refrigerant temperature TH3 at the outlet of the condenser, and operation frequency fCOMP of compressor 11, and fourth data group D2B2 at least includes temperature TH1ic of the air conditioning target space and degree of opening Li of expansion valve 21.

[0156] (Item 6) Monitoring device 5 according to item 3 further includes a display unit 320 configured to display a change in the frequency per certain period elapsed and configured to display that maintenance is necessary when the frequency exceeds a determination threshold value.

[0157] (Item 7) In still another aspect, the present disclosure relates to a monitoring device 5 for an air conditioning apparatus 40, monitoring device 5 using first inference model M1 generated by learning device 4 according to item 1, first inference model M1 being trained. As shown in FIG. 4, monitoring device 5 includes: a second calculation unit 110B configured to calculate first feature amount F1B1 from first data group D1B1 among operation data DB of air conditioning apparatus 40 during a monitoring period; an inference unit 220 configured to output, using first inference model M1 trained, first normal range F2B1 from second data group D2B1 among operation data DB of air conditioning apparatus 40 during the monitoring period; and a determination unit 310 configured to determine whether or not the condition of air conditioning apparatus 40 is normal by comparing first feature amount F1B1 calculated by second calculation unit 110B with first normal range F2B1 inferred by inference unit 220.

[0158] (Item 8) In monitoring device 5 according to item 7, air conditioning apparatus 40 includes a refrigerant circuit C1 in which the refrigerant circulates. As shown in FIG. 2, refrigerant circuit C1 includes a compressor 11, a condenser (outdoor heat exchanger 12), an expansion valve 21, and an evaporator (indoor heat exchanger 22). First feature amount F1B1 is a degree of supercooling SC of the refrigerant at the outlet of the condenser.

[0159] (Item 9) In the monitoring device according to item 8, as shown in Table 1, first data group D1B1 includes a pressure HS1 of the refrigerant discharged from compressor 11 and a temperature TH3 of the refrigerant at the outlet of the condenser, and second data group D2B1 at least includes a temperature TH1ic of an air conditioning target space and an operation frequency fCOMP of compressor 11.

[0160] (Item 10) In yet another aspect, the present disclosure relates to an air conditioning system 1000 shown in FIG. 1, air conditioning system 1000 including: monitoring device 5 according to any one of items 3 to 9; and air conditioning apparatus 40.

[0161] It is expected to implement the embodiments disclosed herein in appropriate combinations unless contradicted. The embodiments disclosed herein are illustrative and non-restrictive in any respect. The scope of the present disclosure is defined by the terms of the claims, rather than the embodiments described above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.REFERENCE SIGNS LIST1: abnormality detection system; 2, 32: CPU; 3, 33: memory; 4: learning device; 5: monitoring device; 10: outdoor unit; 11: compressor; 12, 22 heat exchanger; 13: four-way valve; 14: outdoor fan; 15: accumulator; 20: indoor unit; 21: expansion valve; 23: indoor fan; 30: controller; 40: air conditioning apparatus; 51, 52, 53, 54, 55, 56: temperature sensor; 57: humidity sensor; 61, 63: pressure sensor; 110, 110A, 110B: calculation unit; 120: learning unit; 220: inference unit; 310: determination unit; 311, 312: data processing unit; 320: display unit; 900: network; 1000: air conditioning system; C1: refrigerant circuit; M1, M2: inference model; Nw1: neural network.

Examples

first embodiment

[0021]FIG. 1 is a block diagram showing a configuration of an air conditioning system according to a first embodiment. An air conditioning system 1000 includes: an abnormality detection system 1; and an air conditioning apparatus 40 having a condition monitored by abnormality detection system 1. As shown in FIG. 1, abnormality detection system 1 is connected to air conditioning apparatus 40 via a network 900.

[0022]Abnormality detection system 1 includes a CPU (Central Processing Unit) 2, a memory 3 (a ROM (Read Only Memory) and a RAM (Random Access Memory)), an input / output buffer (not shown), and the like. CPU 2 loads a program stored in the ROM into the RAM or the like and executes the program. The program stored in the ROM is a program in which a processing procedure of abnormality detection system 1 is written. Abnormality detection system 1 monitors each device in air conditioning apparatus 40 in accordance with such a program. This control is not limited to a process by softwa...

Claims

1. A learning device configured to learn a condition of an air conditioning apparatus in which refrigerant circulates, wherein operation data of the air conditioning apparatus includes a first data group and a second data group that does not contain a same data element as a data element of the first data group,the learning device comprising:a first calculation unit configured to calculate a first feature amount from the first data group of the air conditioning apparatus during a learning period; anda learning unit configured to generate a first inference model that infers a first normal range of the first feature amount from the second data group by performing supervised learning using the second data group with the first feature amount being set as truth data.

2. The learning device according to claim 1, whereinthe operation data includes a third data group different from the first data group and a fourth data group that does not contain a same data element as a data element of the third data group,the first calculation unit is configured to calculate a second feature amount from the third data group during the learning period, andthe learning unit is configured to generate a second inference model that infers a second normal range of the second feature amount from the fourth data group by performing supervised learning using the fourth data group with the second feature amount being set as truth data, the second feature amount being obtained by the calculation by the first calculation unit.

3. A monitoring device for an air conditioning apparatus, the monitoring device using the first inference model and the second inference model each generated by the learning device according to claim 2, each of the first inference model and the second inference model being trained,the monitoring device comprising:a second calculation unit configured to calculate the first feature amount from the first data group and the second feature amount from the third data group among the operation data of the air conditioning apparatus during a monitoring period;an inference unit configured to output, from the second data group among the operation data of the air conditioning apparatus during the monitoring period, the first normal range using the first inference model trained, and to output the second normal range from the fourth data group among the operation data of the air conditioning apparatus during the monitoring period; anda determination unit configured to determine whether or not the condition of the air conditioning apparatus is normal based on a frequency at which a first poor condition and a second poor condition are simultaneously detected, the first poor condition being a condition in which the first feature amount calculated by the second calculation unit falls out of the first normal range inferred by the inference unit, the second poor condition being a condition in which the second feature amount calculated by the second calculation unit falls out of the second normal range inferred by the inference unit.

4. The monitoring device according to claim 3, whereinthe air conditioning apparatus includes a refrigerant circuit in which the refrigerant circulates,the refrigerant circuit includes a compressor, a condenser, an expansion valve, and an evaporator,the first feature amount is a degree of supercooling of the refrigerant at an outlet of the condenser, andthe second feature amount is a heat exchange performance of the condenser.

5. The monitoring device according to claim 4, whereinthe first data group at least includes a pressure of the refrigerant discharged from the compressor and a temperature of the refrigerant at the outlet of the condenser,the second data group at least includes a temperature of an air conditioning target space, an operation frequency of the compressor, and a degree of opening of the expansion valve,the third data group at least includes the pressure of the refrigerant discharged from the compressor, the temperature of the refrigerant at the outlet of the condenser, and the operation frequency of the compressor, andthe fourth data group at least includes the temperature of the air conditioning target space and the degree of opening of the expansion valve.

6. The monitoring device according to claim 3, further comprising a display unit configured to display a change in the frequency per certain period elapsed and configured to display that maintenance is necessary when the frequency exceeds a determination threshold value.

7. A monitoring device for an air conditioning apparatus, the monitoring device using the first inference model generated by the learning device according to claim 1, the first inference model being trained,the monitoring device comprising:a second calculation unit configured to calculate the first feature amount from the first data group among the operation data of the air conditioning apparatus during a monitoring period;an inference unit configured to output, using the first inference model trained, the first normal range from the second data group among the operating data of the air conditioning apparatus during the monitoring period; anda determination unit configured to determine whether or not the condition of the air conditioning apparatus is normal by comparing the first feature amount calculated by the second calculation unit with the first normal range inferred by the inference unit.

8. The monitoring device according to claim 7, whereinthe air conditioning apparatus includes a refrigerant circuit in which the refrigerant circulates,the refrigerant circuit includes a compressor, a condenser, an expansion valve, and an evaporator, andthe first feature amount is a degree of supercooling of the refrigerant at an outlet of the condenser.

9. The monitoring device according to claim 8, whereinthe first data group includes a pressure of the refrigerant discharged from the compressor and a temperature of the refrigerant at the outlet of the condenser, andthe second data group at least includes a temperature of an air conditioning target space and an operation frequency of the compressor.

10. An air conditioning system comprising:the monitoring device according to claim 3; andthe air conditioning apparatus.

11. An air conditioning system comprising:the monitoring device according to claim 7; andthe air conditioning apparatus.

Images