Machine learning device, diagnostic system, and equipment

Abstract

To provide a technique capable of more easily making a diagnosis of a diagnosis target including a fluid conveyance machine and a fluid related apparatus.SOLUTION: A machine learning device includes a control section. The control section acquires a first state amount of a fluid conveyance machine included in a diagnosis target and conveying fluid in the diagnosis target, acquires a second state amount related to fluid flowing in a fluid related apparatus included in the diagnosis target and including piping connected to the fluid conveyance machine, acquires a state of the fluid related apparatus, and learns the acquired first state amount of the fluid conveyance machine, the second state amount of the fluid related apparatus and the state of the fluid related apparatus in association with each other.SELECTED DRAWING: Figure 3

Description

Technical Field

[0001] The present disclosure relates to a machine learning device and a diagnostic system and machine to the device related.

Background Art

[0002] For example, there is known a technique for performing a diagnosis regarding an abnormality of a diagnostic target by applying multivariate analysis to a plurality of state quantities of the diagnostic target (see Patent Document 1).

[0003] In Patent Document 1, the Mahalanobis distance of a plurality of state quantities regarding a diagnostic target is calculated, and the presence or absence of an abnormality is diagnosed by comparing the Mahalanobis distance with a threshold value.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, in the method of Patent Document 1, multivariate analysis is performed using a plurality of state quantities, and it is necessary to individually set a threshold value for a reference space corresponding to a normal state of the diagnostic target or an abnormal space corresponding to an abnormal state. For example, when the configuration of the diagnostic target such as a device changes, the state of the diagnostic target changes, so that a technician familiar with the diagnostic target needs to change the threshold values of the reference space and the abnormal space each time, which is a problem that takes time and effort.

[0006] An object of the present disclosure is to provide a technique for more easily diagnosing a diagnostic target including a fluid conveyance machine and fluid-related equipment.

Means for Solving the Problems

[0007] A first aspect of the present disclosure is a machine learning device having a control unit, wherein the control unit acquires a first state quantity of a fluid conveyance machine included in a diagnosis target and conveying a fluid inside the diagnosis target, acquires a second state quantity regarding the fluid flowing inside a fluid-related device including a pipe connected to the fluid conveyance machine and included in the diagnosis target, acquires the state of the fluid-related device, and learns by associating the acquired first state quantity of the fluid conveyance machine, the second state quantity of the fluid-related device, and the state of the fluid-related device.

[0008] According to the first aspect of the present disclosure, it is possible to provide a technique for more easily diagnosing a diagnosis target including a fluid conveyance machine and a fluid-related device.

[0009] A second aspect of the present disclosure is the machine learning device according to the first aspect, wherein the control unit learns the first state quantity of the fluid conveyance machine, the second state quantity of the fluid-related device, and the state of the fluid-related device as teacher data.

[0010] A third aspect of the present disclosure is the machine learning device according to the first aspect, wherein the control unit calculates a reward based on the acquired state of the fluid-related device and the inferred state of the fluid-related device, and learns using the reward.

[0011] A fourth aspect of the present disclosure is the machine learning device according to any one of the first aspect to the third aspect, wherein the control unit infers the state of the fluid-related device from the newly acquired first state quantity of the fluid conveyance machine and the second state quantity of the fluid-related device based on the result of the learning.

[0012] A fifth aspect of the present disclosure is a machine learning device having a control unit, wherein the control unit acquires a first state quantity of a fluid conveyor that is included in a diagnosis target and conveys a fluid inside the diagnosis target, acquires a second state quantity related to the fluid flowing through the inside of a fluid-related device that is included in the diagnosis target and includes a pipe connected to the fluid conveyor, learns to classify the acquired first state quantity of the fluid conveyor and the second state quantity of the fluid-related device, classifies the newly acquired first state quantity of the fluid conveyor and the second state quantity of the fluid-related device based on the result of the learning, and receives an input of information associating the type of the classification with the state of the fluid-related device.

[0013] According to the fifth aspect of the present disclosure, it is possible to provide a technique for more easily diagnosing a diagnosis target including a fluid conveyor and a fluid-related device.

[0014] A sixth aspect of the present disclosure is the machine learning device according to the fifth aspect, wherein the control unit infers the state of the fluid-related device from the type of the classification based on the result of the classification of the newly acquired first state quantity of the fluid conveyor and the second state quantity of the fluid-related device by using the information in which the type of the classification is associated with the state of the fluid-related device.

[0015] A seventh aspect of the present disclosure is a machine learning device having a control unit, wherein the control unit acquires a first state quantity of a fluid conveyor that is included in a diagnosis target and conveys a fluid inside the diagnosis target, acquires a second state quantity related to the fluid flowing through the inside of a fluid-related device that is included in the diagnosis target and includes a pipe connected to the fluid conveyor, acquires a first state of the fluid conveyor, acquires a second state of the fluid-related device, acquires a third state of the fluid conveyor and the fluid-related device, performs a first learning by associating the acquired first state quantity of the fluid conveyor and the first state of the fluid conveyor, performs a second learning by associating the acquired second state quantity of the fluid-related device and the second state of the fluid-related device, and performs a third learning by associating the result of the first learning, the result of the second learning, and the third state of the fluid conveyor and the fluid-related device.

[0016] According to a seventh aspect of the present disclosure, it is possible to provide a technique for more easily diagnosing a diagnostic target including a fluid conveyance machine and fluid-related equipment.

[0017] An eighth aspect of the present disclosure is the machine learning device according to the seventh aspect, wherein the control unit infers the first state of the fluid conveyance machine from the first state quantity of the fluid conveyance machine newly acquired based on the result of the first learning, and infers the second state of the fluid-related equipment from the second state quantity of the fluid-related equipment newly acquired based on the result of the second learning, and infers the third state of the fluid conveyance machine and the fluid-related equipment from the inferred first state of the fluid conveyance machine and the second state of the fluid-related equipment based on the result of the third learning.

[0018] A ninth aspect of the present disclosure is the machine learning device according to any one of the first aspect to the eighth aspect, wherein the fluid conveyance machine is an electric motor and a load driven by the electric motor.

[0019] A tenth aspect of the present disclosure is the machine learning device according to the ninth aspect, wherein the load driven by the electric motor is a compressor, a fan, or a pump.

[0020] An eleventh aspect of the present disclosure is the machine learning device according to any one of the first aspect to the tenth aspect, wherein the diagnostic target is an air conditioner, a refrigeration cycle device, a fluid machine, an oil cooler, or an air handling unit.

[0021] A twelfth aspect of the present disclosure is a diagnostic system for diagnosing the diagnostic target using the machine learning device according to any one of the first aspect to the eleventh aspect.

[0022] A thirteenth aspect of the present disclosure is a device for diagnosing the state of the fluid-related equipment or the third state of the fluid conveyance machine and the fluid-related equipment using the machine learning device according to any one of the first aspect to the eleventh aspect.

[0023] A fourteenth aspect of the present disclosure is a program for causing a control unit included in a machine learning device to acquire a first state quantity of a fluid conveyor that is included in a diagnostic target and conveys a fluid inside the diagnostic target, acquire a second state quantity related to the fluid flowing through the inside of a fluid-related device that is included in the diagnostic target and includes a pipe connected to the fluid conveyor, acquire the state of the fluid-related device, and execute a process of associating and learning the acquired first state quantity of the fluid conveyor, the second state quantity of the fluid-related device, and the state of the fluid-related device.

[0024] According to the fourteenth aspect of the present disclosure, it is possible to provide a technique for more easily diagnosing a diagnostic target including a fluid conveyor and a fluid-related device.

Brief Description of Drawings

[0025]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Mode for Carrying Out the Invention

[0026] Embodiments of the present disclosure will be described in detail.

[0027] [Configuration of Diagnostic System] With reference to FIGS. 1 and 2, the configuration of the diagnostic system 1 according to the present embodiment will be described. FIG. 1 is a diagram showing an example of the configuration of the diagnostic system 1. FIG. 2 is a diagram showing an example of the hardware configuration of the control unit 220. As shown in FIG. 1, the diagnostic system 1 includes an air conditioner 100 and a control unit 220. The diagnostic system 1 performs diagnosis of the diagnosis target in the control unit 220. The air conditioner 100 is an example of the diagnosis target in the diagnostic system 1.

[0028] The air conditioner 100 includes an outdoor unit 110, an indoor unit 120, and refrigerant paths 130 and 140. The air conditioner 100 operates a refrigeration cycle (refrigerant circuit) having a configuration including the outdoor unit 110, the indoor unit 120, and the refrigerant paths 130 and 140, and adjusts the temperature, humidity, etc. inside the room where the indoor unit 120 is installed.

[0029] The outdoor unit 110 is arranged outside the building whose temperature etc. is to be adjusted. The outdoor unit 110 is connected to one end of each of the refrigerant paths 130 and 140, sucks refrigerant from either one of the refrigerant paths 130 and 140, and discharges the refrigerant to the other one. The indoor unit 120 is arranged inside the building whose temperature etc. is to be adjusted. The indoor unit 120 is connected to the other end of each of the refrigerant paths 130 and 140, sucks refrigerant from either one of the refrigerant paths 130 and 140, and discharges the refrigerant to the other one. The refrigerant paths 130 and 140 are constituted by, for example, pipelines, and connect between the outdoor unit 110 and the indoor unit 120 so that the refrigerant can circulate between the outdoor unit 110 and the indoor unit 120.

[0030] The outdoor unit 110 includes refrigerant paths L1 to L6, oil paths L7 and L8, a four-way switching valve 111, an accumulator 112, a compressor 113, an oil separator 114, an outdoor heat exchanger 115, an outdoor expansion valve 116, a drive device 117, a sensor group 118, and a control device 119.

[0031] The refrigerant paths L1 to L6 are constituted as pipelines, for example. The refrigerant path L1 connects between one end of the refrigerant path 130 outside the outdoor unit 110 and the four-way switching valve 111. The refrigerant path L2 connects between the four-way switching valve 111 and the inlet of the compressor 113. The refrigerant path L2 includes refrigerant paths L21 and L22. The refrigerant path L21 connects between the four-way switching valve 111 and the accumulator 112. The refrigerant path L22 connects between the accumulator 112 and the inlet of the compressor 113.

[0032] The refrigerant path L3 connects between the four-way switching valve 111 and the outlet of the compressor 113. The refrigerant path L3 includes refrigerant paths L31 and L32. The refrigerant path L31 connects between the outlet of the compressor 113 and the oil separator 114. The refrigerant path L32 connects between the four-way switching valve 111 and the oil separator 114.

[0033] The refrigerant path L4 connects between the four-way switching valve 111 and the outdoor heat exchanger 115. The refrigerant path L5 connects between the outdoor heat exchanger 115 and the outdoor expansion valve 116. The refrigerant path L6 connects between one end of the refrigerant path 140 outside the outdoor unit 110 and the outdoor expansion valve 116.

[0034] The oil path L7 is configured as a pipeline, for example. The oil path L7 is used to allow the oil separated by the oil separator 114 to flow into the refrigerant path L22 and return to the compressor 113 through the refrigerant path L22. Note that the oil passing through the oil path L7 may contain, for example, liquid-phase refrigerant (hereinafter, "liquid refrigerant") dissolved therein. That is, not only oil but also liquid refrigerant flows through the oil path L7. The oil path L8 is configured as a pipeline, for example, and is used to allow the oil containing the liquid refrigerant separated by the accumulator 112 to flow into the refrigerant path L22 and return to the compressor 113 through the refrigerant path L22.

[0035] The four-way switching valve 111 reverses the flow of refrigerant circulation between the cooling operation and the heating operation of the air conditioner 100. During the cooling operation of the air conditioner 100, the four-way switching valve 111 connects the path of the solid line in FIG. 1. Specifically, during the cooling operation of the air conditioner 100, the four-way switching valve 111 connects between the refrigerant path L1 and the refrigerant path L2, and between the refrigerant path L3 and the refrigerant path L4.

[0036] On the other hand, during the heating operation of the air conditioner 100, the four-way switching valve 111 connects the path of the dotted line in FIG. 1. Specifically, during the heating operation of the air conditioner 100, the four-way switching valve 111 connects between the refrigerant path L4 and the refrigerant path L2, and between the refrigerant path L1 and the refrigerant path L3.

[0037] The accumulator 112 separates the liquid refrigerant contained in the refrigerant inhaled from the refrigerant path L21 and discharges the refrigerant from which part or all of the liquid refrigerant has been removed to the refrigerant path L22. The liquid refrigerant separated by the accumulator 112 contains oil. The accumulator 112 is provided with an oil discharge port connected to the oil path L8, and the oil containing the separated refrigerant flows out to the oil path L8 through the oil discharge port and is returned to the compressor 113 through the oil path L8 and the refrigerant path L22.

[0038] The compressor 113 inhales the refrigerant from the refrigerant path L22, compresses it to a high pressure, and discharges it to the refrigerant path L31. The compressor 113 includes an electric motor 113A and a compression mechanism section 113B. The electric motor 113A rotationally drives the compression mechanism section 113B using predetermined electric power supplied from the drive device 117. The predetermined electric power is, for example, three-phase alternating current power. The compression mechanism section 113B operates with the electric motor 113A as a power source. The compression mechanism section 113B inhales the refrigerant from the inlet of the compressor 113, compresses it to a high pressure, and discharges it to the outside from the outlet of the compressor 113.

[0039] During the cooling operation of the air conditioner 100, the high-temperature and high-pressure refrigerant compressed by the compressor 113 flows into the outdoor heat exchanger 115 through the refrigerant path L3 and the refrigerant path L4.

[0040] On the other hand, during the heating operation of the air conditioner 100, the high-temperature and high-pressure refrigerant compressed by the compressor 113 flows out to the refrigerant path 130 outside the outdoor unit 110 through the refrigerant path L3 and the refrigerant path L1. Then, the high-temperature and high-pressure refrigerant flows into the indoor unit 120 through the refrigerant path 130.

[0041] The oil separator 114 separates oil from the refrigerant flowing in from the refrigerant path L31 and allows the refrigerant after part or all of the oil has been separated and removed to flow out to the refrigerant path L32. The oil separator 114 is also provided with an oil discharge port connected to the oil path L7. The oil separated from the refrigerant flows out to the oil path L7 through the oil discharge port and is returned to the compressor 113 through the oil path L7 and the refrigerant path L22.

[0042] The outdoor heat exchanger 115 performs heat exchange between the outside air and the refrigerant passing through the inside. A fan 115A is provided in parallel with the outdoor heat exchanger 115. The outdoor heat exchanger 115 performs heat exchange between the outside air blown by the fan 115A and the refrigerant flowing through the inside.

[0043] During the cooling operation of the air conditioner 100, the outdoor heat exchanger 115 causes the high-temperature and high-pressure refrigerant compressed by the compressor 113 and flowing in from the refrigerant path L4 to dissipate heat to the outside air, and discharges the condensed and liquefied refrigerant (liquid refrigerant) to the refrigerant path L5.

[0044] Also, during the heating operation of the air conditioner 100, the outdoor heat exchanger 115 causes the low-temperature and low-pressure liquid refrigerant flowing in from the refrigerant path L5 to absorb heat from the outside air, and discharges the evaporated refrigerant to the refrigerant path L4.

[0045] The outdoor expansion valve 116 is closed to a predetermined opening during the heating operation of the air conditioner 100, and reduces the pressure of the refrigerant (liquid refrigerant) flowing in from the refrigerant path L6 to a predetermined pressure. On the other hand, the outdoor expansion valve 116 is fully opened during the cooling operation of the air conditioner 100, and allows the refrigerant (liquid refrigerant) to pass from the refrigerant path L5 to the refrigerant path L6. The outdoor expansion valve 116 is, for example, an electromagnetic valve.

[0046] The indoor unit 120 includes an indoor expansion valve 121, an indoor heat exchanger 122, and a control device 123. The indoor expansion valve 121 is closed to a predetermined opening during the cooling operation of the air conditioner 100, and reduces the pressure of the subcooled liquid refrigerant flowing in from the refrigerant path 140 to a predetermined pressure. On the other hand, the indoor expansion valve 121 is fully opened during the heating operation of the air conditioner 100, and allows the refrigerant (liquid refrigerant) flowing out from the indoor heat exchanger 122 to pass toward the refrigerant path 140. The indoor expansion valve 121 is, for example, an electromagnetic valve.

[0047] The indoor heat exchanger 122 performs heat exchange between the indoor air and the refrigerant passing through its interior. Specifically, due to the action of the fan 122A mounted on the indoor unit 120, the indoor air passes around the indoor heat exchanger 122, and heat exchange is promoted between the indoor air and the refrigerant inside the indoor heat exchanger 122. The indoor air that has undergone heat exchange with the refrigerant inside the indoor heat exchanger 122 is sent outside the indoor unit 120 by the action of the fan 122A, thereby realizing indoor cooling or heating.

[0048] During the cooling operation of the air conditioner 100, the indoor heat exchanger 122 absorbs heat from the indoor air into the low-temperature and low-pressure liquid refrigerant depressurized by the indoor expansion valve 121, thereby lowering the temperature of the indoor air. On the other hand, during the heating operation of the air conditioner 100, the indoor heat exchanger 122 causes the high-temperature and high-pressure refrigerant flowing in from the outdoor unit 110 through the refrigerant path 130 to release heat to the indoor air, thereby raising the temperature of the indoor air. The control device 123 controls the operation of the indoor unit 120 by directly controlling, for example, the indoor expansion valve 121 or the fan 122A.

[0049] The drive device 117 drives the electric motor 113A using electric power supplied from a predetermined power source. For example, the drive device 117 is an inverter device that drives the electric motor 113A by generating three-phase alternating current electric power with a predetermined frequency and a predetermined voltage using the electric power supplied from a predetermined power source and supplying it to the electric motor 113A.

[0050] The sensor group 118 is an example of a plurality of sensors mounted on the outdoor unit 110 and used for controlling the air conditioner 100 and the like. The output of the sensor group 118 is taken into the control device 119 and the control unit 220. The sensor group 118 in FIG. 1 includes a current sensor 118A, a temperature sensor 118B, and a pressure sensor 118C. The current sensor 118A acquires information regarding the current of the motor 113A and detects a signal representing the information. For example, the temperature sensor 118B may directly measure the temperature of the refrigerant at the outlet of the compressor 113 in the pipe, or may measure the temperature of the pipe through which the refrigerant flowing out from the outlet of the compressor 113 flows from the outside. The temperature sensor 118B detects the temperature of the refrigerant at the outlet of the compressor 113 (temperature of the high-pressure refrigerant) or the temperature of the refrigerant at the inlet (temperature of the low-pressure refrigerant), etc. The temperature sensor 118B is, for example, a thermistor. The pressure sensor 118C acquires information regarding the pressure of the refrigerant at the outlet or the inlet of the compressor 113 and outputs a signal representing the information.

[0051] Based on the output of the sensor group 118, the control device 119 controls the operation of the outdoor unit 110, taking, for example, the four-way switching valve 111, the compressor 113 (drive device 117), the outdoor expansion valve 116, etc. as direct control targets. For example, the control device 119 may be provided separately from the drive device 117 or may be integrated with the drive device 117.

[0052] The control unit 220 diagnoses the air conditioner 100 using machine learning. The diagnosis of the air conditioner 100 performed by the control unit 220 includes the diagnosis of the state of the diagnosis target. The diagnosis of the state of the diagnosis target includes, for example, the diagnosis of abnormalities of the air conditioner 100.

[0053] The control unit 220 may be included in the air conditioner 100 to be diagnosed or may be provided outside the air conditioner 100. For example, the control unit 220 may be realized as a function of a control device that controls the operation of the air conditioner 100. In this case, the control device that controls the operation of the air conditioner 100 is used for both the function of controlling the operation of the air conditioner 100 and the function of diagnosing the air conditioner 100.

[0054] Further, the control unit 220 may be provided separately from the control device that controls the operation of the air conditioner 100. In this case, the control unit 220 may be dedicated to the function of diagnosing the air conditioner 100, or may be shared with other functions. For example, the control unit 220 may be realized as one function of the drive device 117 of the air conditioner 100.

[0055] When provided outside the air conditioner 100 to be diagnosed, the control unit 220 may be realized, for example, as one function of a terminal device for managing the air conditioner 100, an edge controller, or an edge server. The control unit 220 may be realized as one function of an on-premises server or a cloud server. Further, the control unit 220 may be a portable terminal device (mobile terminal). The mobile terminal may be, for example, a dedicated mobile terminal such as a remote controller of the air conditioner 100, or a smartphone, a tablet terminal, or a PC (Personal Computer), etc. The control unit 220 may be configured to diagnose a plurality of air conditioners 100.

[0056] The function of the control unit 220 is realized by the cooperation of arbitrary hardware and software. For example, as shown in FIG. 2, the control unit 220 is configured to include an external interface 201, an auxiliary storage device 202, a memory device 203, a CPU 204, a high-speed arithmetic device 205, a communication interface 206, an input device 207, and an output device 208.

[0057] Further, the external interface 201, the auxiliary storage device 202, the memory device 203, the CPU 204, the high-speed arithmetic device 205, the communication interface 206, the input device 207, and the output device 208 are connected by a bus BS2.

[0058] The external interface 201 functions as an interface for reading data from the recording medium 201A and writing data to the recording medium 201A. The recording medium 201A includes, for example, general-purpose recording media such as flexible disks, CDs (Compact Discs), DVDs (Digital Versatile Discs), BDs (Blu-ray (registered trademark) Discs), SD memory cards, and USB memories. Further, the recording medium 201A may be, for example, a dedicated recording medium used in a manufacturing factory or a repair facility of the air conditioner 100, etc.

[0059] Thereby, the control unit 220 can read various data used in the processing through the recording medium 201A, store it in the auxiliary storage device 202, or install a program for realizing various functions. The control unit 220 may acquire various data and programs used in the processing from an external device through the communication interface 206.

[0060] The auxiliary storage device 202 stores installed various programs and also stores files, data, etc. necessary for various processes. The auxiliary storage device 202 includes, for example, HDDs (Hard Disc Drives), SSDs (Solid State Discs), flash memories, etc.

[0061] When there is an instruction to start a program, the memory device 203 reads and stores the program from the auxiliary storage device 202. The memory device 203 includes, for example, DRAMs (Dynamic Random Access Memories) and SRAMs (Static Random Access Memories).

[0062] The CPU 204 executes various programs loaded from the auxiliary storage device 202 to the memory device 203, and realizes various functions related to the control unit 220 according to the programs. The high-speed arithmetic unit 205 operates in conjunction with the CPU 204 and performs arithmetic processing at a higher speed than the CPU 204. The high-speed arithmetic unit 205 includes, for example, a GPU (Graphics Processing Unit), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), and the like.

[0063] Note that the high-speed arithmetic unit 205 may be omitted according to the speed of the required arithmetic processing. The communication interface 206 is used as an interface for communicably connecting to an external device. Thereby, the control unit 220 can acquire, for example, the output data of the sensor group 118 through the communication interface 206. Further, the communication interface 206 may have a plurality of types of communication interfaces depending on the communication method and the like with the connected device.

[0064] When the control unit 220 is included in the air conditioner 100, the communication interface 206 is an interface that performs only communication with other devices included in the air conditioner 100, for example. When the control unit 220 is included in the air conditioner 100, the communication interface 206 may include both an interface for communicating with other devices included in the air conditioner 100 and an interface for communicating with devices external to the air conditioner 100.

[0065] The input device 207 receives an operation input from the user. The input device 207 includes, for example, buttons, toggles, levers, keyboards, mice, touch panels, touch pads, and the like. Further, the input device 207 may be capable of receiving a voice input from the user. The input device 207 capable of receiving a voice input from the user is a microphone or the like capable of collecting the user's voice.

[0066] The input device 207 may be capable of receiving a gesture input from the user. The input device 207 capable of receiving a gesture input from the user is, for example, a camera or the like that can image the state of the user's gesture. The input device 207 may be capable of receiving a biometric input from the user. The input device 207 capable of receiving a biometric input from the user includes a camera capable of acquiring image data containing information regarding the user's fingerprint or iris.

[0067] The output device 208 outputs information to the user by the control unit 220. The output device 208 is, for example, an illumination device or a display device that outputs information visually. The illumination device is, for example, an indicator lamp or the like. The display device is, for example, a liquid crystal display, an organic EL display, or the like. Further, the output device 208 may be a buzzer, an alarm, a speaker, or the like that outputs auditory information.

[0068] Note that when the control unit 220 is included in the air conditioner 100, the input device 207 and the output device 208 may be omitted. This is because various input devices and output devices mounted on the air conditioner 100 may be also used for the purpose related to the control unit 220.

[0069] [Processing] Next, the diagnostic process of the air conditioner 100 using machine learning will be described. The control unit 220 executes a diagnostic process of the air conditioner 100 including a learning mode process for training a machine learning model (hereinafter simply referred to as a model) and an inference mode process for performing inference using the trained model. The learning mode process may include, in addition to the initial learning of the model, re-learning for improving the accuracy of the trained model. The diagnostic process of the air conditioner 100 executed by the control unit 220 is realized, for example, by loading a program stored in the auxiliary storage device 202 into the memory device 203 and executing it by the CPU 204. Hereinafter, examples of using supervised learning, reinforcement learning, and unsupervised learning as machine learning methods will be described.

[0070] [Supervised Learning] As shown in FIGS. 3 and 4, the control unit 220 according to this embodiment executes diagnostic processing of the air conditioner 100 including learning mode processing and inference mode processing of machine learning using supervised learning. FIG. 3 is an explanatory diagram of an example of learning mode processing of machine learning using supervised learning. FIG. 4 is an explanatory diagram of an example of inference mode processing of machine learning using supervised learning.

[0071] The control unit 220 in FIG. 3 executes learning mode processing of machine learning including the processing of the first state quantity acquisition unit 300 of the fluid conveyor, the second state quantity acquisition unit 302 of the fluid-related device, the state acquisition unit 304 of the fluid-related device, the teacher data storage unit 306, the learning unit 308, and the inference unit 310. The learning mode processing of machine learning is realized, for example, by loading a program stored in the auxiliary storage device 202 into the memory device 203 and executing it with the CPU 204.

[0072] The fluid conveyor is an electric motor 113A and a load driven by the electric motor 113A. The load driven by the electric motor 113A is a compressor 113, a fan 115A, or a pump. The pump operates, for example, in a hydraulic device or a device related to water. The fluid conveyor is included in the air conditioner 100 and conveys a fluid such as refrigerant inside the air conditioner 100. Also, a pump, which is an example of the fluid conveyor, conveys a fluid such as machine oil (lubricating oil, hydraulic oil, or cutting oil) or water.

[0073] The fluid-related device includes a pipe connected to the fluid conveyor and is a device connected to the fluid conveyor. The fluid-related device is included in the air conditioner 100 and fluid flows inside. The fluid-related device is a pipe, a solenoid valve, an evaporator, a condenser, a sensor, etc.

[0074] For example, the fluid-related device includes refrigerant paths 130, 140, and refrigerant paths L1 to L6 as pipes. Also, for example, the refrigerant-related device includes solenoid valves such as the outdoor expansion valve 116 installed in the pipe. Also, for example, the refrigerant-related device includes sensors such as the temperature sensor 118B and the pressure sensor 118C.

[0075] The first state quantity acquisition unit 300 of the fluid conveyor acquires the first state quantity of the fluid conveyor included in the air conditioner 100. For example, the first state quantity of the fluid conveyor is a state quantity based on the current detection value of the electric motor 113A. The current detection value of the electric motor 113A is acquired based on the output of the current sensor 118A.

[0076] For example, the first state quantity of the fluid conveyor is the first harmonic component (hereinafter referred to as the current vector primary component) of the magnitude (amplitude) of the current vector of the electric motor 113A. When an abnormality occurs in which the oil seal performance of the compression chamber of the compressor 113 deteriorates, the airtightness of the compression chamber is insufficient, and the refrigerant gas leaks, so that the current vector primary component becomes relatively smaller compared to the normal state where the oil seal performance of the compression chamber is sufficient. In addition, when the liquid-phase refrigerant (hereinafter referred to as the liquid refrigerant) enters the compression chamber of the compressor 113 and an abnormal state (hereinafter referred to as liquid compression) occurs in which the liquid refrigerant is compressed, due to the abnormal pressure increase in the compression chamber, the current vector primary component becomes relatively larger compared to the normal state where liquid compression does not occur.

[0077] The first state quantity of the fluid conveyor may be a state quantity different from the current vector primary component as a physical quantity correlated with the current of the compressor 113. For example, instead of or in addition to the current vector primary component, the first state quantity of the fluid conveyor may use the harmonic component of an integer order of the magnitude of the current vector (that is, the frequency component that is an integer multiple of the rotational frequency at the mechanical angle). In addition, instead of or in addition to the physical quantity correlated with the current, the first state quantity of the fluid conveyor may use physical quantities correlated with the voltage, power, vibration, sound, ultrasonic wave, magnetic flux, temperature, torque, etc. of the compressor 113. In addition, instead of the frequency component, the state quantity in the time domain may be used as the first state quantity of the fluid conveyor. The state quantity in the time domain is, for example, a sinusoidal time waveform for one cycle of the phase current of the electric motor 113A at the mechanical angle or the electrical angle.

[0078] The magnitude of the current vector of the electric motor 113A is represented by the square root of the sum of the squared values of the three-phase current detection values of the electric motor 113A. The current vector primary component represents the frequency component of the rotational frequency at the mechanical angle of the compressor 113 among the frequency components of the magnitude of the current vector.

[0079] Further, the second state quantity acquisition unit 302 of the fluid-related device acquires the second state quantity of the fluid-related device included in the air conditioner 100. For example, the second state quantity of the fluid-related device is a state quantity related to the fluid flowing inside the fluid-related device, and is pressure, temperature, saturation temperature, flow velocity, or flow rate. The second state quantity of the fluid-related device may include, for example, the number of pulses representing the opening degree of the outdoor expansion valve 116 or the like.

[0080] For example, the second state quantity acquisition unit 302 of the fluid-related device acquires the temperature and pressure of the high-pressure refrigerant at the outlet of the compressor 113 and the temperature and pressure of the low-pressure refrigerant at the inlet of the compressor 113 as the second state quantity of the fluid-related device. The second state quantity of the fluid-related device may include the superheat degree (superheat) of the high-pressure refrigerant at the outlet of the compressor 113. The superheat degree of the high-pressure refrigerant at the outlet of the compressor 113 is the difference between the temperature of the high-pressure refrigerant at the outlet of the compressor 113 and the saturation temperature (evaporation temperature). The saturation temperature of the high-pressure refrigerant at the outlet of the compressor 113 is determined from the pressure of the high-pressure refrigerant at the outlet of the compressor 113. The second state quantity of the fluid-related device may include the subcooling degree (subcool) of the low-pressure refrigerant at the inlet of the compressor 113. The subcooling degree of the low-pressure refrigerant at the inlet of the compressor 113 is the difference between the saturation temperature (condensation temperature) and the temperature of the low-pressure refrigerant at the inlet of the compressor 113. The saturation temperature of the low-pressure refrigerant at the inlet of the compressor 113 is determined from the pressure of the low-pressure refrigerant at the inlet of the compressor 113.

[0081] Further, the state acquisition unit 304 of the fluid-related device acquires the state of the fluid-related device included in the air conditioner 100. Note that the state acquisition unit 304 of the fluid-related device may receive an input of the state of the fluid-related device from the user, or may determine the state of the fluid-related device from the information obtained from the air conditioner 100. The state of the fluid-related device acquired by the state acquisition unit 304 of the fluid-related device becomes a label required for the teacher data.

[0082] For example, the state of the fluid-related equipment refers to a state indicating an abnormality in the fluid-related equipment. For example, it includes states such as clogging of the oil circuit, failure of the parts causing wet operation, failure of the temperature sensor 118B, refrigerant leakage, and failure of heat exchangers (condensers or evaporators) such as the outdoor heat exchanger 115. Also, the state of the fluid-related equipment may include states such as clogging of the refrigerant circuit and failure of the fan (for example, deformation, breakage, or looseness of the rotating shaft of the fan). Further, when a strainer for dust collection or a drier for refrigerant drying is provided in the middle of the piping of the refrigerant circuit, the state of clogging of the strainer for dust collection or the drier for refrigerant drying may be included in the state of the fluid-related equipment.

[0083] Note that the parts causing wet operation are specific parts that cause wet operation in the refrigerant-related equipment, and are the solenoid valves in the refrigerant circuit. The solenoid valves in the refrigerant circuit are the outdoor expansion valve 116 or the indoor expansion valve 121, etc.

[0084] The teacher data storage unit 306 associates the first state quantity of the fluid conveyance machine acquired by the first state quantity acquisition unit 300 of the fluid conveyance machine, the second state quantity of the fluid-related equipment acquired by the second state quantity acquisition unit 302 of the fluid-related equipment, and the state of the fluid-related equipment acquired by the state acquisition unit 304 of the fluid-related equipment, and stores them as teacher data. The teacher data may be data collected in the market, data obtained from experiments in the laboratory, data obtained from simulations, alone or in combination.

[0085] The learning unit 308 adjusts the parameters of the model by performing supervised learning on the teacher data stored in the teacher data storage unit 306. Specifically, the learning unit 308 learns the correspondence between the first state quantity of the fluid conveyance machine and the second state quantity of the fluid-related equipment in the teacher data and the state of the fluid-related equipment. The learned model has its parameters adjusted such that when the first state quantity of the fluid conveyance machine and the second state quantity of the fluid-related equipment in the teacher data are input, the state of the fluid-related equipment corresponding to the first state quantity of the fluid conveyance machine and the second state quantity of the fluid-related equipment is output. The learning unit 308 provides the learned model to the inference unit 310.

[0086] For example, when the mechanical angular frequency first component of the current vector of the electric motor 113A, which is an example of the first state quantity of the fluid conveyance machine, and the saturation temperature of the high-pressure refrigerant and the temperature of the high-pressure refrigerant, which are examples of the second state quantity of the fluid-related equipment, are input, the learning unit 308 adjusts the parameters of the model so that the state of clogging (normal or abnormal) of the oil circuit, which is an example of the state of the fluid-related equipment, is output as a diagnosis result.

[0087] Also, for example, when the mechanical angular frequency first component of the current vector of the electric motor 113A, which is an example of the first state quantity of the fluid conveyance machine, and the saturation temperature of the high-pressure refrigerant and the temperature of the high-pressure refrigerant, which are examples of the second state quantity of the fluid-related equipment, are input, the learning unit 308 adjusts the parameters of the model so that the state of failure (normal or abnormal) of the temperature sensor 118B, which is an example of the state of the fluid-related equipment, is output as a diagnosis result.

[0088] Also, for example, in the case of refrigerant leakage, when the following first state quantity of the fluid conveyance machine and the second state quantity of the fluid-related equipment are input, the learning unit 308 adjusts the parameters of the model so that the state of the fluid-related equipment is output as a diagnosis result. FIG. 14 is an example diagram showing the relationship between the first state of the fluid conveyance machine and the second state of the fluid-related equipment and the abnormal location of the air conditioner 100. For example, as shown in FIG. 14, for the combination of the first state of the fluid conveyance machine and the second state of the fluid-related equipment, the presence or absence of abnormalities and the abnormal location of the fluid conveyance machine and the fluid-related equipment are uniquely determined. For example, the first state of the fluid conveyance machine is the result of the abnormality diagnosis of the compressor 113 based on the effective value of the current. Also, the second state of the fluid-related equipment is the result of the temperature range deviation diagnosis by the subcooling of the refrigerant.

[0089] Specifically, when the first state of the fluid conveyor represents a normal state and the second state of the fluid-related device represents an abnormal state, the control unit 220 may diagnose it as a failure of the temperature sensor. When the first state of the fluid conveyor represents an abnormal state and the second state of the fluid-related device represents a normal state, the control unit 220 may diagnose it as an abnormality of the fluid conveyor such as bearing wear. Further, when the first state of the fluid conveyor represents an abnormal state and the second state of the fluid-related device represents an abnormal state, the control unit 220 may diagnose it as a simultaneous occurrence of refrigerant leakage, or a failure of the temperature sensor and an abnormality of the fluid conveyor.

[0090] The first state quantity acquisition unit 300 of the fluid conveyor acquires the first state quantity of the fluid conveyor included in the air conditioner 100. For example, the first state quantity of the fluid conveyor is a state quantity based on the current detection value of the electric motor 113A. The current detection value of the electric motor 113A is acquired based on the output of the current sensor 118A.

[0091] As a physical quantity correlated with the current of the compressor 113, the first state quantity of the fluid conveyor may use physical quantities correlated with the voltage, power, vibration, sound, ultrasonic wave, magnetic flux, temperature, torque, etc. of the compressor 113.

[0092] In addition, the second state quantity acquisition unit 302 of the fluid-related device acquires the second state quantity of the fluid-related device included in the air conditioner 100. For example, the second state quantity of the fluid-related device is a state quantity related to the fluid flowing inside the fluid-related device, and is pressure, temperature, or saturation temperature.

[0093] For example, the second state quantity acquisition unit 302 of the fluid-related device acquires the temperature of the low-pressure refrigerant at the inlet of the compressor 113, the pressure of the low-pressure refrigerant at the inlet of the compressor 113, and the saturation temperature of the low-pressure refrigerant at the inlet of the compressor 113 as the second state quantity of the fluid-related device.

[0094] In addition, the state acquisition unit 304 of the fluid-related device acquires the state of the fluid-related device included in the air conditioner 100. Note that the state acquisition unit 304 of the fluid-related device may receive an input of the state of the fluid-related device from the user, or may determine the state of the fluid-related device from the information obtained from the air conditioner 100. The state of the fluid-related device acquired by the state acquisition unit 304 of the fluid-related device becomes the label required for the teacher data.

[0095] For example, the state of the fluid-related device includes states such as normal, refrigerant leakage, or failure of the temperature sensor 118B. The teacher data storage unit 306 associates the first state quantity of the fluid transfer device acquired by the first state quantity acquisition unit 300 of the fluid transfer device, the second state quantity of the fluid-related device acquired by the second state quantity acquisition unit 302 of the fluid-related device, and the state of the fluid-related device acquired by the state acquisition unit 304 of the fluid-related device, and stores them as teacher data. The teacher data may be data collected in the market, data obtained from experiments in the laboratory, data obtained from simulations, or may be used alone or in combination.

[0096] The learning unit 308 adjusts the parameters of the model by performing supervised learning on the teacher data stored in the teacher data storage unit 306. Specifically, the learning unit 308 learns the correspondence between the first state quantity of the fluid transfer device in the teacher data, the second state quantity of the fluid-related device, and the state of the fluid-related device. The learned model has its parameters adjusted such that when the first state quantity of the fluid transfer device and the second state quantity of the fluid-related device in the teacher data are input, the state of the fluid-related device corresponding to the first state quantity of the fluid transfer device and the second state quantity of the fluid-related device is output. The learning unit 308 provides the learned model to the inference unit 310.

[0097] For example, when the first state quantity of the fluid transfer device and the second state quantity of the fluid-related device are input, the learning unit 308 may adjust the parameters of the model so as to output normal operation, failure of the temperature sensor 118B (an example of the state of the fluid-related device), or refrigerant leakage as the diagnosis result.

[0098] Also, for example, when the learning unit 308 receives the first state quantity of the fluid conveyor and the second state quantity of the fluid-related device as described below in the case of a fan failure, it adjusts the parameters of the model so that the state of the fluid-related device is output as a diagnosis result. FIG. 15 is a diagram showing an example of the relationship between the first state of the fluid conveyor and the second state of the fluid-related device and the abnormal part of the air conditioner 100. For example, as shown in FIG. 15, for the combination of the first state of the fluid conveyor and the second state of the fluid-related device, the presence or absence of abnormalities and the abnormal part of the fluid conveyor and the fluid-related device are uniquely determined. For example, the first state of the fluid conveyor is the result of disturbance diagnosis based on the frequency component of the mechanical angular frequency × the number of propeller fan blades of the current vector of the fan motor current of the heat exchanger and fan failure diagnosis based on the mechanical angular frequency component. Also, the second state of the fluid-related device is the result of condenser (evaporator) temperature range deviation diagnosis based on the temperature of the heat exchanger.

[0099] Specifically, when the first state of the fluid conveyor represents a normal state and the second state of the fluid-related device represents an abnormal state, the control unit 220 may diagnose refrigerant leakage or a temperature sensor failure. When the first state of the fluid conveyor represents a disturbed state and the second state of the fluid-related device represents a normal state, the control unit 220 may diagnose that the heat exchanger is normal and the disturbance is caused by the reverse wind to the fan. Also, when the first state of the fluid conveyor represents a disturbed and fan failure state and the second state of the fluid-related device represents a normal state, the control unit 220 may diagnose fan imbalance such as foreign matter attachment or frosting. Also, when the first state of the fluid conveyor represents a disturbed state and the second state of the fluid-related device represents an abnormal state, the control unit 220 may diagnose an abnormality of the heat exchanger such as blockage. Also, when the first state of the fluid conveyor represents a disturbed and fan failure state and the second state of the fluid-related device represents an abnormal state, the control unit 220 may diagnose a fan failure such as breakage or bearing wear.

[0100] The first state quantity acquisition unit 300 of the fluid conveyor acquires the first state quantity of the fluid conveyor included in the air conditioner 100. For example, the first state quantity of the fluid conveyor is a state quantity based on the detected current value of the fan 115A. The detected current value of the fan 115A is acquired based on the output of the current sensor 118A.

[0101] As a physical quantity correlated with the current of the fan 115A, the first state quantity of the fluid conveyor may use physical quantities correlated with the voltage, power, vibration, sound, ultrasonic wave, magnetic flux, temperature, torque, etc. of the fan 115A.

[0102] Further, the second state quantity acquisition unit 302 of the fluid-related device acquires the second state quantity of the fluid-related device included in the air conditioner 100. For example, the second state quantity of the fluid-related device is a state quantity regarding the fluid flowing inside the fluid-related device, and is the temperature of a heat exchanger (condenser or evaporator) such as the outdoor heat exchanger 115.

[0103] For example, the second state quantity acquisition unit 302 of the fluid-related device acquires the temperature of a heat exchanger (condenser or evaporator) such as the outdoor heat exchanger 115 as the second state quantity of the fluid-related device.

[0104] Moreover, the state acquisition unit 304 of the fluid-related device acquires the state of the fluid-related device included in the air conditioner 100. Note that the state acquisition unit 304 of the fluid-related device may receive an input of the state of the fluid-related device from the user, or may determine the state of the fluid-related device from the information obtained from the air conditioner 100. The state of the fluid-related device acquired by the state acquisition unit 304 of the fluid-related device becomes a label necessary for the teacher data.

[0105] For example, the states of fluid-related equipment include states such as refrigerant leakage, abnormalities in heat exchangers (condensers or evaporators), or normal conditions. The teacher data storage unit 306 associates the first state quantity of the fluid conveyor obtained by the first state quantity acquisition unit 300 of the fluid conveyor, the second state quantity of the fluid-related equipment obtained by the second state quantity acquisition unit 302 of the fluid-related equipment, and the state of the fluid-related equipment obtained by the state acquisition unit 304 of the fluid-related equipment, and stores them as teacher data. The teacher data can be data collected in the market, data obtained from experiments in the laboratory, data obtained through simulation, alone or in combination.

[0106] The learning unit 308 adjusts the model parameters by performing supervised learning on the teacher data stored in the teacher data storage unit 306. Specifically, the learning unit 308 learns the correspondence between the first state quantity of the fluid conveyor in the teacher data, the second state quantity of the fluid-related equipment, and the state of the fluid-related equipment. The trained model has its parameters adjusted such that when the first state quantity of the fluid conveyor and the second state quantity of the fluid-related equipment in the teacher data are input, the state of the fluid-related equipment corresponding to the first state quantity of the fluid conveyor and the second state quantity of the fluid-related equipment is output. The learning unit 308 provides the trained model to the inference unit 310.

[0107] For example, the learning unit 308 may adjust the model parameters so that when the first state quantity of the fluid conveyor and the second state quantity of the fluid-related equipment are input, normal operation, an abnormality in the heat exchanger (condenser or evaporator), or refrigerant leakage is output as the diagnosis result.

[0108] Further, for example, when the learning unit 308 receives the first state quantity of the fluid transfer machine and the second state quantity of the fluid-related device as described below in the event of a pump failure, it adjusts the parameters of the model so that the state of the fluid-related device is output as a diagnosis result. FIG. 16 is a diagram showing an example of the relationship between the first state of the fluid transfer machine and the second state of the fluid-related device and the abnormal location of the air conditioner 100. For example, as shown in FIG. 16, for a combination of the first state of the fluid transfer machine and the second state of the fluid-related device, the presence or absence of an abnormality and the abnormal location of the fluid transfer machine and the fluid-related device are uniquely determined. For example, the first state of the fluid transfer machine is the result of abnormal diagnosis of the pump based on the mechanical angular frequency component of the pump vibration. Also, the second state of the fluid-related device is the result of deviation diagnosis from the normal range based on the flow velocity or flow rate of the fluid circulated by the pump.

[0109] Specifically, when the first state of the fluid transfer machine represents a normal state and the second state of the fluid-related device represents an abnormal state, the control unit 220 may diagnose a blockage in the flow path of the fluid-related device outside the pump. When the first state of the fluid transfer machine represents an abnormal state and the second state of the fluid-related device represents a normal state, the control unit 220 may diagnose an abnormality in the pump such as bearing wear, shaft eccentricity, or shaft misalignment. Also, when the first state of the fluid transfer machine represents an abnormal state and the second state of the fluid-related device represents an abnormal state, the control unit 220 may diagnose a blockage in the flow path inside the pump.

[0110] The first state quantity acquisition unit 300 of the fluid transfer machine acquires the first state quantity of the fluid transfer machine included in the air conditioner 100. For example, the first state quantity of the fluid transfer machine is a state quantity based on the detected current value of the pump. The detected current value of the pump is acquired based on the output of the current sensor 118A.

[0111] As a physical quantity correlated with the current of the pump, the first state quantity of the fluid transfer machine may use a physical quantity correlated with the voltage, power, vibration, sound, ultrasonic wave, magnetic flux, temperature, torque, etc. of the pump.

[0112] Further, the second state quantity acquisition unit 302 of the fluid-related device acquires the second state quantity of the fluid-related device included in the air conditioner 100. For example, the second state quantity of the fluid-related device is a state quantity related to the fluid flowing inside the fluid-related device, and is the flow velocity or flow rate of the fluid.

[0113] For example, the second state quantity acquisition unit 302 of the fluid-related device acquires the flow velocity or flow rate of the fluid flowing inside the fluid-related device as the second state quantity of the fluid-related device.

[0114] In addition, the state acquisition unit 304 of the fluid-related device acquires the state of the fluid-related device included in the air conditioner 100. Note that the state acquisition unit 304 of the fluid-related device may receive an input of the state of the fluid-related device from the user, or may determine the state of the fluid-related device from the information obtained from the air conditioner 100. The state of the fluid-related device acquired by the state acquisition unit 304 of the fluid-related device becomes the label required for the teacher data.

[0115] For example, the state of the fluid-related device includes states such as clogging of the flow path of the fluid-related device outside the pump. The teacher data storage unit 306 associates the first state quantity of the fluid conveyor acquired by the first state quantity acquisition unit 300 of the fluid conveyor, the second state quantity of the fluid-related device acquired by the second state quantity acquisition unit 302 of the fluid-related device, and the state of the fluid-related device acquired by the state acquisition unit 304 of the fluid-related device, and stores them as teacher data. The teacher data may be data collected in the market, data obtained from experiments in the laboratory, data obtained from simulations, alone, or in combination.

[0116] The learning unit 308 adjusts the parameters of the model by performing supervised learning on the teacher data stored in the teacher data storage unit 306. Specifically, the learning unit 308 learns the correspondence between the first state quantity of the fluid conveyor in the teacher data, the second state quantity of the fluid-related equipment, and the state of the fluid-related equipment. When the first state quantity of the fluid conveyor in the teacher data and the second state quantity of the fluid-related equipment are input, the parameters of the learned model are adjusted so that the state of the fluid-related equipment corresponding to the first state quantity of the fluid conveyor and the second state quantity of the fluid-related equipment is output. The learning unit 308 provides the learned model to the inference unit 310.

[0117] For example, when the first state quantity of the fluid conveyor and the second state quantity of the fluid-related equipment are input, the learning unit 308 may adjust the parameters of the model so as to output normal operation or clogging of the flow path of the fluid-related equipment other than the pump as a diagnosis result.

[0118] The control unit 220 in FIG. 4 executes the inference mode process of machine learning, including the processes of the first state quantity acquisition unit 300 of the fluid conveyor, the second state quantity acquisition unit 302 of the fluid-related equipment, and the inference unit 310. The inference mode process of machine learning is realized, for example, by loading a program stored in the auxiliary storage device 202 into the memory device 203 and executing it by the CPU 204.

[0119] The first state quantity acquisition unit 300 of the fluid conveyor newly acquires the first state quantity of the fluid conveyor included in the air conditioner 100. Further, the second state quantity acquisition unit 302 of the fluid-related equipment newly acquires the second state quantity of the fluid-related equipment included in the air conditioner 100.

[0120] The inference unit 310 can infer the state of the fluid-related equipment output from the learned model by inputting the first state quantity of the fluid conveyor newly acquired by the first state quantity acquisition unit 300 of the fluid conveyor and the second state quantity of the fluid-related equipment newly acquired by the second state quantity acquisition unit 302 of the fluid-related equipment into the learned model.

[0121] <Reinforcement learning> As shown in FIGS. 5 and 6, the control unit 220 according to the present embodiment executes diagnostic processing of the air conditioner 100 including learning mode processing and inference mode processing of machine learning using reinforcement learning. Note that descriptions of parts similar to supervised learning will be omitted as appropriate. FIG. 5 is an explanatory diagram of an example of learning mode processing of machine learning using reinforcement learning. FIG. 6 is an explanatory diagram of an example of inference mode processing of machine learning using reinforcement learning.

[0122] The control unit 220 in FIG. 5 executes learning mode processing of machine learning including the processing of the first state quantity acquisition unit 300 of the fluid conveyance machine, the second state quantity acquisition unit 302 of the fluid-related device, the state acquisition unit 304 of the fluid-related device, the learning unit 308, the inference unit 310, and the reward calculation unit 312. The learning mode processing of machine learning is realized, for example, by loading a program stored in the auxiliary storage device 202 into the memory device 203 and executing it by the CPU 204.

[0123] The first state quantity acquisition unit 300 of the fluid conveyance machine acquires the first state quantity of the fluid conveyance machine included in the air conditioner 100. Further, the second state quantity acquisition unit 302 of the fluid-related device acquires the second state quantity of the fluid-related device included in the air conditioner 100. Further, the state acquisition unit 304 of the fluid-related device acquires the state of the fluid-related device included in the air conditioner 100 and provides it to the reward calculation unit 312.

[0124] The inference unit 310 inputs the first state quantity of the fluid conveyance machine acquired by the first state quantity acquisition unit 300 of the fluid conveyance machine and the second state quantity of the fluid-related device acquired by the second state quantity acquisition unit 302 of the fluid-related device into the model, and obtains the state of the fluid-related device output from the model. The inference unit 310 provides the inferred state of the fluid-related device to the reward calculation unit 312.

[0125] The reward calculation unit 312 calculates a reward based on the state of the fluid-related device provided by the state acquisition unit 304 of the fluid-related device and the state of the fluid-related device inferred by the inference unit 310, and outputs it to the learning unit 308. The reward calculation unit 312 calculates a higher reward as the state of the fluid-related device provided by the state acquisition unit 304 of the fluid-related device is closer to the state of the fluid-related device inferred by the inference unit 310.

[0126] The learning unit 308 repeatedly adjusts the parameters of the model used by the inference unit 310 so as to maximize the reward calculated by the reward calculation unit 312. Specifically, the learning unit 308 learns the correspondence between the first state quantity of the fluid conveyor and the second state quantity of the fluid-related device, and the state of the fluid-related device. When the first state quantity of the fluid conveyor and the second state quantity of the fluid-related device are input, the learning unit 308 adjusts the parameters of the model so that the state of the fluid-related device corresponding to the first state quantity of the fluid conveyor and the second state quantity of the fluid-related device is output.

[0127] The control unit 220 in FIG. 6 executes the inference mode process of machine learning, including the process of the first state quantity acquisition unit 300 of the fluid conveyor, the second state quantity acquisition unit 302 of the fluid-related device, and the inference unit 310. The inference mode process of machine learning is realized, for example, by loading a program stored in the auxiliary storage device 202 into the memory device 203 and executing it by the CPU 204.

[0128] The first state quantity acquisition unit 300 of the fluid conveyor newly acquires the first state quantity of the fluid conveyor included in the air conditioner 100. Also, the second state quantity acquisition unit 302 of the fluid-related device newly acquires the second state quantity of the fluid-related device included in the air conditioner 100.

[0129] The inference unit 310 can infer the state of the fluid-related device output from the learned model by inputting the first state quantity of the fluid conveyor newly acquired by the first state quantity acquisition unit 300 of the fluid conveyor and the second state quantity of the fluid-related device newly acquired by the second state quantity acquisition unit 302 of the fluid-related device into the learned model.

[0130] <Unsupervised learning> As shown in FIGS. 7 and 8, the control unit 220 according to the present embodiment executes diagnostic processing of the air conditioner 100 including learning mode processing and inference mode processing of machine learning using unsupervised learning. Note that descriptions of parts similar to supervised learning will be omitted as appropriate. FIG. 7 is an explanatory diagram of an example of learning mode processing of machine learning using unsupervised learning. FIG. 8 is an explanatory diagram of an example of inference mode processing of machine learning using unsupervised learning.

[0131] The control unit 220 in FIG. 7 executes learning mode processing of machine learning including the processing of the first state quantity acquisition unit 300 of the fluid conveyance machine, the second state quantity acquisition unit 302 of the fluid-related device, the learning unit 308, the inference unit 310, the data storage unit 320, and the state input reception unit 322 of the fluid-related device. The learning mode processing of machine learning is realized, for example, by loading a program stored in the auxiliary storage device 202 into the memory device 203 and executing it by the CPU 204.

[0132] The first state quantity acquisition unit 300 of the fluid conveyance machine acquires the first state quantity of the fluid conveyance machine included in the air conditioner 100. Further, the second state quantity acquisition unit 302 of the fluid-related device acquires the second state quantity of the fluid-related device included in the air conditioner 100.

[0133] The data storage unit 320 stores the first state quantity of the fluid conveyance machine acquired by the first state quantity acquisition unit 300 of the fluid conveyance machine and the second state quantity of the fluid-related device acquired by the second state quantity acquisition unit 302 of the fluid-related device in association with each other.

[0134] The learning unit 308 clusters the data stored in the teacher data storage unit 306 by performing unsupervised learning. Clustering is to group (classify) from the characteristics of the data. Specifically, the learning unit 308 learns the classification of the first state quantity of the fluid conveyance machine and the second state quantity of the fluid-related device. The data stored in the teacher data storage unit 306 may be data collected in the market, data obtained from experiments in the laboratory, data obtained from simulations, or alone or in combination.

[0135] The state input reception unit 322 of the fluid-related device receives, from the user, an input of information (state of the fluid-related device) associated with the type of classification clustered by the learning unit 308. The learning unit 308 adjusts the parameters of the model and clusters the first state quantity of the fluid transfer machine and the second state quantity of the fluid-related device when the first state quantity of the fluid transfer machine and the second state quantity of the fluid-related device are input. Further, the learning unit 308 uses the learning result of clustering the first state quantity of the fluid transfer machine and the second state quantity of the fluid-related device, and the information in which the input clustered classification type and the state of the fluid-related device are associated, to perform learning so that the state of the fluid-related device based on the classification type of the newly acquired first state quantity of the fluid transfer machine and the second state quantity of the fluid-related device is output. The learning unit 308 provides the inference unit 310 with the learning result including the learned model.

[0136] The control unit 220 in FIG. 8 executes the inference mode process of machine learning, including the processes of the first state quantity acquisition unit 300 of the fluid transfer machine, the second state quantity acquisition unit 302 of the fluid-related device, and the inference unit 310. The inference mode process of machine learning is realized, for example, by loading a program stored in the auxiliary storage device 202 into the memory device 203 and executing it by the CPU 204.

[0137] The first state quantity acquisition unit 300 of the fluid transfer machine newly acquires the first state quantity of the fluid transfer machine included in the air conditioner 100. Further, the second state quantity acquisition unit 302 of the fluid-related device newly acquires the second state quantity of the fluid-related device included in the air conditioner 100.

[0138] The inference unit 310 clusters the first state quantity of the fluid conveyance device newly acquired by the first state quantity acquisition unit 300 of the fluid conveyance device and the second state quantity of the fluid-related device newly acquired by the second state quantity acquisition unit 302 of the fluid-related device by inputting them into the learned model. Based on the types of the classified clusters, the inference unit 310 can infer the state of the fluid-related device associated with the types of the classification.

[0139] <Other diagnostic processes using machine learning> As shown in FIGS. 9 to 12, for example, the control unit 220 according to the present embodiment may execute a diagnostic process of the air conditioner 100 including a learning mode process and an inference mode process of machine learning using supervised learning. FIGS. 9 to 11 are explanatory diagrams of an example of the learning mode process of machine learning using supervised learning. FIG. 12 is an explanatory diagram of an example of the inference mode process of machine learning using supervised learning. Note that descriptions of the same parts as those described above are omitted as appropriate.

[0140] The control unit 220 in FIG. 9 executes a learning mode process of machine learning including the processes of the first state quantity acquisition unit 300 of the fluid conveyance device, the teacher data storage unit 306, the first state acquisition unit 330, the first learning unit 332, and the first inference unit 334. The learning mode process of machine learning is realized, for example, by loading a program stored in the auxiliary storage device 202 into the memory device 203 and executing it by the CPU 204.

[0141] The first state quantity acquisition unit 300 of the fluid conveyance device acquires the first state quantity of the fluid conveyance device included in the air conditioner 100. Also, the first state acquisition unit 330 acquires the first state of the fluid conveyance device that conveys fluid inside the air conditioner 100.

[0142] For example, the first state of the fluid conveyance machine is the state inside the compressor 113 or the fan 115A. The state inside the compressor 113 includes an abnormal state in which the oil sealing property inside the compressor 113 has deteriorated, an abnormal state in which liquid compression is occurring inside the compressor 113, and a normal state in which these abnormalities do not occur inside the compressor 113.

[0143] Note that the first state acquisition unit 330 may receive an input of the first state of the fluid conveyance machine from the user, or may determine the first state of the fluid conveyance machine from the information obtained from the air conditioner 100. The first state of the fluid conveyance machine acquired by the first state acquisition unit 330 serves as the label required for the teacher data.

[0144] The teacher data storage unit 306 associates the first state quantity of the fluid conveyance machine acquired by the first state quantity acquisition unit 300 of the fluid conveyance machine and the first state of the fluid conveyance machine acquired by the first state acquisition unit 330, and stores them as teacher data. The teacher data may be data collected in the market, data obtained from experiments in the laboratory, data obtained from simulations, or may be used alone or in combination.

[0145] The first learning unit 332 adjusts the parameters of the model by performing supervised learning on the teacher data stored in the teacher data storage unit 306. Specifically, the first learning unit 332 learns the correspondence between the first state quantity of the fluid conveyance machine in the teacher data and the first state of the fluid conveyance machine. In the learned model, the parameters are adjusted such that when the first state quantity of the fluid conveyance machine in the teacher data is input, the first state of the fluid conveyance machine corresponding to the first state quantity of the fluid conveyance machine is output. The first learning unit 332 provides the learned model to the inference unit 310.

[0146] In addition, the control unit 220 in FIG. 10 executes a learning mode process of machine learning including the processes of the second state quantity acquisition unit 302, the teacher data storage unit 306, the second state acquisition unit 336, the second learning unit 338, and the second inference unit 340 of the fluid-related device.

[0147] The second state quantity acquisition unit 302 of the fluid-related device acquires the second state quantity of the fluid-related device included in the air conditioner 100. Further, the second state acquisition unit 336 acquires the second state of the fluid-related device included in the air conditioner 100. Note that the second state acquisition unit 336 may receive an input of the second state of the fluid-related device from the user, or may determine the second state of the fluid-related device from information obtained from the air conditioner 100. The second state of the fluid-related device acquired by the second state acquisition unit 336 serves as a label necessary for the training data.

[0148] The second state of the fluid-related device is the operating state of the refrigerant-related device including the refrigerant circuit. The operating state of the refrigerant-related device is, for example, the operating state of the refrigerant flowing through the inside of the refrigerant-related device conveyed by the compressor 113. The operating state of the refrigerant-related device includes a wet operation state and a superheat operation state.

[0149] The wet operation state means an abnormal operation state in which the high-pressure refrigerant at the outlet of the compressor 113 contains liquid refrigerant exceeding a predetermined standard assumed from operating conditions and the like. Whether the refrigerant at the outlet of the compressor 113 is in the wet operation state may be determined based on the superheat of the refrigerant at the outlet (discharge side) of the compressor 113. The superheat operation state means a normal operation state that is not in the wet operation state. The second state of the fluid-related device may mean an operation state in which the temperature of a heat exchanger (condenser or evaporator) such as the outdoor heat exchanger 115 has increased or decreased.

[0150] The operating state of the refrigerant-related device is, for example, the operating state of the refrigerant flowing through the inside of the refrigerant-related device conveyed by the compressor 113. The operating state of the refrigerant-related device includes a predetermined wet operation state and a superheat operation state. The predetermined wet operation state means an operation state in which the superheat (superheat degree) of the refrigerant at the outlet of the compressor 113 is relatively small with respect to a predetermined standard assumed from operating conditions and the like, and the superheat operation state means an operation state that is not in the predetermined wet operation state.

[0151] The teacher data storage unit 306 stores, as teacher data, the second state quantity of the fluid-related equipment acquired by the second state quantity acquisition unit 302 of the fluid-related equipment and the second state of the fluid-related equipment acquired by the second state acquisition unit 336 in association with each other. The teacher data may be data collected in the market, data obtained from experiments in a laboratory, data obtained from simulations, or may be used alone or in combination.

[0152] The second learning unit 338 adjusts the model parameters by performing teacher-based learning on the teacher data stored in the teacher data storage unit 306. Specifically, the second learning unit 338 learns the correspondence between the second state quantity of the fluid-related equipment in the teacher data and the second state of the fluid-related equipment. The parameters of the learned model are adjusted such that when the second state quantity of the fluid-related equipment in the teacher data is input, the second state of the fluid-related equipment corresponding to the second state quantity of the fluid-related equipment is output. The second learning unit 338 provides the learned model to the inference unit 310.

[0153] Also, the control unit 220 in FIG. 11 executes a learning mode process of machine learning including the processes of the first inference unit 334, the second inference unit 340, the third state acquisition unit 342, the third learning unit 344, and the third inference unit 346.

[0154] The third state acquisition unit 342 acquires the third state of the fluid conveyor and the fluid-related equipment. Note that the third state acquisition unit 342 may receive an input of the third state of the fluid conveyor and the fluid-related equipment from the user, or may determine the third state of the fluid conveyor and the fluid-related equipment from the information obtained from the air conditioner 100. The third state of the fluid conveyor and the fluid-related equipment acquired by the third state acquisition unit 342 becomes the label required for the teacher data.

[0155] For example, the third state of the fluid conveyance machine and fluid-related equipment includes states such as clogging of the oil circuit, failure of the wet operation cause parts, failure of the temperature sensor 118B, refrigerant leakage, failure of heat exchangers (condensers or evaporators) such as the outdoor heat exchanger 115, and normality. Further, the third state of the fluid conveyance machine and fluid-related equipment may include states such as clogging of the refrigerant circuit, failure of the pump (bearing wear, shaft eccentricity, shaft misalignment), and failure of the fan.

[0156] The third state of the fluid conveyance machine and fluid-related equipment may be related failure events of the compressor 113, such as refrigerant liquid backflow to the compressor 113, oil depletion or deterioration inside the compressor 113, abnormal bearings of the compressor 113, abnormal wobbling of the compressor 113, abnormal motor of the compressor 113, damage to the compression chamber components such as valves, scrolls, and rotors of the compressor 113, and problems related to the compressor 113 such as tooth contact of the compressor.

[0157] Further, the third state of the fluid conveyance machine and fluid-related equipment may include states such as deterioration or liquid backflow due to the lifespan of the compressor 113, fouling or damage of the condenser or evaporator, deterioration or failure of the condenser blower or evaporator blower, clogging of the strainer for dust collection or drier for refrigerant drying provided in the middle of the refrigerant circuit piping, and deterioration of the oil used in the compressor 113 (detected by clogging of the piping, poor lubrication of the compressor 113, change in the heat transfer amount, etc.).

[0158] The third learning unit 344 adjusts the parameters of the model by performing supervised learning using, as teacher data, the first state of the fluid transfer machine inferred by the first inference unit 334, the second state of the fluid-related equipment inferred by the second inference unit 340, and the third state of the fluid transfer machine and the fluid-related equipment acquired by the third state acquisition unit 342. Specifically, the third learning unit 344 learns the correspondence between the first state of the fluid transfer machine and the second state of the fluid-related equipment in the teacher data and the third state of the fluid transfer machine and the fluid-related equipment. The learned model has its parameters adjusted such that when the first state of the fluid transfer machine and the second state of the fluid-related equipment in the teacher data are input, the third state of the fluid transfer machine and the fluid-related equipment corresponding to the first state of the fluid transfer machine and the second state of the fluid-related equipment are output. The third learning unit 344 provides the learned model to the third inference unit 346.

[0159] The control unit 220 in FIG. 12 executes inference mode processing of machine learning, including the processing of the first state quantity acquisition unit 300 of the fluid transfer machine, the second state quantity acquisition unit 302 of the fluid-related equipment, the first inference unit 334, the second inference unit 340, and the third inference unit 346. The inference mode processing of machine learning in FIG. 12 is realized, for example, by loading a program stored in the auxiliary storage device 202 into the memory device 203 and executing it with the CPU 204.

[0160] The first state quantity acquisition unit 300 of the fluid transfer machine newly acquires the first state quantity of the fluid transfer machine included in the air conditioner 100. Further, the second state quantity acquisition unit 302 of the fluid-related equipment newly acquires the second state quantity of the fluid-related equipment included in the air conditioner 100.

[0161] The first inference unit 334 infers the first state of the fluid transfer machine output from the learned model by inputting the first state quantity of the fluid transfer machine newly acquired by the first state quantity acquisition unit 300 of the fluid transfer machine into the learned model.

[0162] The second inference unit 340 infers the second state of the fluid-related equipment output from the learned model by inputting the second state quantity of the fluid-related equipment newly acquired by the second state quantity acquisition unit 302 of the fluid-related equipment into the learned model. The third inference unit 346 can infer the third state of the fluid conveyance device and the fluid-related equipment output from the learned model by inputting the first state of the fluid conveyance device inferred by the first inference unit 334 and the second state of the fluid-related equipment inferred by the second inference unit 340 into the learned model.

[0163] Further, the third state of the fluid conveyance device and the fluid-related equipment inferred by the third inference unit 346 may, for example, estimate an abnormal location of the air conditioner 100. FIG. 13 is a diagram showing an example of the relationship between the first state of the fluid conveyance device and the second state of the fluid-related equipment and the abnormal location of the air conditioner 100.

[0164] For example, as shown in FIG. 13, for the combination of the first state of the fluid conveyance device and the second state of the fluid-related equipment, the presence or absence of an abnormality and the abnormal location of the fluid conveyance device and the fluid-related equipment are uniquely determined.

[0165] When both the first state of the fluid conveyance device and the second state of the fluid-related equipment represent normal states, the control unit 220 diagnoses that the fluid conveyance device and the fluid-related equipment are normal.

[0166] On the other hand, when at least one of the first state of the fluid conveyance device and the second state of the fluid-related equipment represents an abnormal state, the control unit 220 diagnoses that the fluid conveyance device and the fluid-related equipment are abnormal.

[0167] Specifically, when the first state of the fluid conveyance device represents a normal state and the second state of the fluid-related equipment represents an abnormal wet operation state, the control unit 220 may diagnose that the component causing the wet operation has failed. The component causing the wet operation is a specific component that causes the wet operation, for example, a solenoid valve in the refrigerant circuit. The solenoid valve as a specific component causing the wet operation includes, for example, the outdoor expansion valve 116 and the indoor expansion valve 121.

[0168] Further, when the first state of the fluid conveyor represents an abnormal state of a decrease in the oil sealing property of the compressor 113 and the second state of the fluid-related device represents a state of normal superheat operation, the control unit 220 may diagnose that there is a blockage in the oil circuit (oil path L7 or oil path L8).

[0169] Further, when the first state of the fluid conveyor represents an abnormal state of a decrease in the oil sealing property of the compressor 113 and the second state of the fluid-related device represents an abnormal wet operation state, the control unit 220 may diagnose that the wet operation causative component is malfunctioning.

[0170] Further, when the first state of the fluid conveyor represents an abnormal state of liquid compression of the compressor 113 and the second state of the fluid-related device represents a state of normal superheat operation, the control unit 220 may diagnose that the temperature sensor 118B is malfunctioning.

[0171] Further, when the first state of the fluid conveyor represents an abnormal state of liquid compression of the compressor 113 and the second state of the fluid-related device represents an abnormal wet operation state, the control unit 220 may diagnose that the wet operation causative component is malfunctioning.

[0172] The control unit 220 is an example of a machine learning device. Further, the air conditioner 100 having the control unit 220, a terminal device, a mobile terminal, an edge controller, an edge server, an on-premises server, or a cloud server that functions as the control unit 220 is also an example of a machine learning device.

[0173] For example, the control unit 220 included in the drive device 117 of the air conditioner 100 may execute diagnostic processing of the air conditioner 100 including learning mode processing and inference mode processing. Note that the learning mode processing includes initial learning of the model. The learning mode processing may include re-learning for improving the accuracy of the learned model.

[0174] Further, for example, the control unit 220 included in the control device that controls the operation of the air conditioner 100 may execute diagnostic processing of the air conditioner 100 including learning mode processing and inference mode processing. Note that the learning mode processing includes initial learning of the model. The learning mode processing may include relearning for improving the accuracy of the learned model.

[0175] Further, for example, the control unit 220 included in the terminal device, mobile terminal, edge controller, edge server, or on-premises server of the air conditioner 100 may execute diagnostic processing of the air conditioner 100 including learning mode processing and inference mode processing. Note that the learning mode processing includes initial learning of the model. The learning mode processing may include relearning for improving the accuracy of the learned model.

[0176] Further, for example, the control unit 220 included in the cloud server may execute diagnostic processing of the air conditioner 100 including learning mode processing and inference mode processing. Note that the learning mode processing includes initial learning of the model. The learning mode processing may include relearning for improving the accuracy of the learned model.

[0177] Further, for example, the control unit 220 included in the terminal device, mobile terminal, edge controller, edge server, or on-premises server of the air conditioner 100 may be configured to execute learning mode processing. In this case, for example, the control unit 220 included in the control device that controls the operation of the air conditioner 100 executes inference mode processing. Relearning for improving the accuracy of the learned model may or may not be included.

[0178] Further, for example, the control unit 220 included in the terminal device, mobile terminal, edge controller, edge server, or on-premises server of the air conditioner 100 may be configured to execute learning mode processing. In this case, for example, the control unit 220 included in the drive device 117 of the air conditioner 100 executes inference mode processing. Relearning for improving the accuracy of the learned model may or may not be included.

[0179] Also, for example, the control unit 220 included in the cloud server may execute learning mode processing. In this case, for example, the control unit 220 included in the control device that controls the operation of the air conditioner 100 executes inference mode processing. Relearning for improving the accuracy of the learned model may or may not be included.

[0180] Also, for example, the control unit 220 included in the cloud server may execute learning mode processing. In this case, for example, the control unit 220 included in the drive device 117 of the air conditioner 100 executes inference mode processing. Relearning for improving the accuracy of the learned model may or may not be included.

[0181] The control unit 220 records and stores the diagnostic results in, for example, the auxiliary storage device or the recording medium 201A, etc. Thereby, the user of the diagnostic system 1 can take out the data stored from the auxiliary storage device 202 or the recording medium 201A of the control unit 220 and confirm the diagnostic results of the air conditioner 100 in time series.

[0182] Also, the control unit 220 may output the diagnostic results to the user through the output device 208. Thereby, the user of the diagnostic system 1 can confirm the diagnostic results of the air conditioner 100 based on the visual or auditory information from the output device 208.

[0183] Also, the control unit 220 may transmit the diagnostic results to an external device through the communication interface 206. Thereby, the external device can store the diagnostic results received from the control unit 220. The external device is, for example, a terminal device used by the user of the diagnostic system 1.

[0184] As a result, the user of the diagnostic system 1 can check the diagnostic results of the air conditioner 100 in chronological order through the terminal device used by the user himself / herself. When the control unit 220 is included in the air conditioner 100, the external device may be a terminal device for managing the air conditioner 100, an edge controller, an edge server, an on-premises server, a cloud server, or the like.

[0185] [Other Embodiments] The above-described embodiments may be appropriately modified or changed.

[0186] For example, in the above-described embodiments, the diagnostic target may be a device or system having a refrigeration cycle different from that of the air conditioner 100. The diagnostic target includes a refrigeration cycle device, a fluid machine, an oil cooler, an air handling unit, and the like.

[0187] Also, the diagnostic target may be a device or system different from a device having a refrigeration cycle. Specifically, the diagnostic target may be a device or system including a fluid conveyance machine that conveys a fluid flowing through the inside and a fluid-related device through which the fluid conveyed by the fluid conveyance machine flows through the inside. Thereby, the device or system to be diagnosed can be diagnosed by the same method as described above.

[0188] Also, the teacher data may be accumulated and used in a plurality of diagnostic targets, or an edge controller, an edge server, an on-premises server, a cloud server, etc. that collect various data from the microcomputer of the diagnostic target.

[0189] As described above, although this embodiment has been described, it will be understood that various changes in form and details are possible without departing from the spirit and scope of the claims. Note that in this embodiment, "A or B" may include both the meanings of "only A, only B" and "A and B".

Description of Reference Numerals

[0190] 1 Diagnostic system 100 Air conditioner 110 Outdoor unit 113 Compressor 120 Indoor unit 130, 140, L1 - L6 Refrigerant paths 220 Control unit L7 - L8 Oil paths

Claims

1. A machine learning device having a control unit, wherein the control unit acquires a first state quantity of a fluid conveyor that is included in a diagnosis target and conveys a fluid inside the diagnosis target, acquires a second state quantity regarding the fluid that flows inside a fluid-related device that is included in the diagnosis target and includes a pipe connected to the fluid conveyor, acquires a first normal or abnormal state of the fluid conveyor, acquires a second normal or abnormal state of the fluid-related device, acquires a third normal or abnormal state of the fluid conveyor and the fluid-related device, performs first learning by associating the acquired first state quantity of the fluid conveyor and the first normal or abnormal state of the fluid conveyor, performs second learning by associating the acquired second state quantity of the fluid-related device and the second normal or abnormal state of the fluid-related device, performs third learning by associating the result of the first learning, the result of the second learning, and the third normal or abnormal state of the fluid conveyor and the fluid-related device, the third abnormal state of the fluid conveyor and the fluid-related device is a blockage of an oil circuit, a failure of an outdoor expansion valve, a failure of an indoor expansion valve, a failure of a temperature sensor, a failure of a heat exchanger, a blockage of a refrigerant circuit, a failure of a fan, a blockage of a strainer for dust collection, a blockage of a drier for refrigerant drying, or a blockage of a flow path of a fluid-related device outside the pump Machine learning device.

2. The control unit infers the first normal or abnormal state of the fluid conveyor from the newly acquired first state quantity of the fluid conveyor based on the result of the first learning, infers the second normal or abnormal state of the fluid-related device from the newly acquired second state quantity of the fluid-related device based on the result of the second learning, infers the third normal or abnormal state of the fluid conveyor and the fluid-related device from the inferred first normal or abnormal state of the fluid conveyor and the inferred second normal or abnormal state of the fluid-related device based on the result of the third learning The machine learning device according to claim 1.

3. The fluid conveyor is an electric motor and a load driven by the electric motor The machine learning device according to claim 1 or 2.

4. The load driven by the electric motor is a compressor, a fan, or a pump The machine learning device according to claim 3.

5. The diagnosis target is an air conditioner, a refrigeration cycle device, a fluid machine, an oil cooler, or an air handling unit The machine learning device according to claim 1 or 2.

6. A diagnostic system that diagnoses the object to be diagnosed using the machine learning device according to claim 1 or 2.

7. A device that diagnoses the normal or abnormal state of the fluid-related equipment or the third normal or abnormal state of the fluid conveyance machine and the fluid-related equipment using the machine learning device according to claim 1 or 2.

Images