Information Processing System

The information processing system uses dual learning models to enhance failure diagnosis in refrigerators, ensuring accurate detection and prevention of issues, thereby improving user convenience and system reliability.

JP7594369B2Active Publication Date: 2024-12-04MIDEA GROUP CO LTD
View PDF 13 Cites 0 Cited by

Patent Information

Application Number
JP2020088672
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Filing Date
2020-05-21
Publication Date
2024-12-04
Estimated Expiration
2040-05-21

AI Technical Summary

Technical Problem

Existing information processing systems struggle to perform accurate failure diagnosis in devices like refrigerators, particularly when it is difficult to correlate operating states with the presence or absence of failures.

Method used

An information processing system that includes a diagnosis section using a first learning model to detect failures and a second learning model to verify normal operation, with a notification section to alert users and update the learning models based on repair or inspection results, enabling more precise failure detection and prevention.

Benefits of technology

The system enhances failure diagnosis accuracy by detecting potential issues early, allowing for timely repairs and maintaining device functionality, improving user convenience and system reliability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007594369000001
    Figure 0007594369000001
  • Figure 0007594369000002
    Figure 0007594369000002
  • Figure 0007594369000003
    Figure 0007594369000003
Patent Text Reader

Abstract

To provide an information processing system capable of performing a more appropriate failure diagnosis.SOLUTION: An information processing system has a diagnosis unit and a notification unit. The diagnosis unit performs a failure diagnosis of a refrigerator, by using a learning model learned so that a result of the failure diagnosis is outputted when information indicating an operating condition of the refrigerator is inputted. The notification unit outputs the information based on the result of the failure diagnosis by the diagnosis unit from the refrigerator or a terminal device.SELECTED DRAWING: Figure 3
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] FIELD An embodiment of the present invention relates to an information processing system. [Background technology]

[0002] A server that diagnoses equipment using recent data including operation history has been proposed. Meanwhile, as an information processing system for fault diagnosis, an information processing system that can perform more appropriate fault diagnosis is expected. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2019-179300 A Summary of the Invention [Problem to be solved by the invention]

[0004] An object of the present invention is to provide an information processing system capable of performing more appropriate fault diagnosis. [Means for solving the problem]

[0005] An information processing system according to an embodiment includes a diagnosing unit, a notifying unit, a receiving unit, and an updating unit. The diagnosing unit performs fault diagnosis on the refrigerator using a learning model that has been trained to output a result of fault diagnosis when information indicating an operating state of the refrigerator is input. The notifying unit causes the refrigerator or a terminal device to output information based on the result of the fault diagnosis by the diagnosing unit. The receiving unit receives input of information based on a result of repairing or inspecting the refrigerator. The updating unit updates the learning model based on a learning model additionally learned using the information received by the receiving unit. The learning model includes a first learning model and a second learning model. The first learning model is trained to output a result of the fault diagnosis when information indicating an operating state of the refrigerator is input. The first learning model is trained to restore and output information identical to the input information when information indicating a normal operating state of the refrigerator is input as input information. The diagnosis unit performs a fault diagnosis of the refrigerator to detect a fault of a component of the refrigerator or a sign of a fault of the refrigerator by a first fault diagnosis using the first learning model, and when a fault or a sign of a fault is not detected in the refrigerator by the first fault diagnosis, the diagnosis unit further performs a second fault diagnosis using the second learning model. The first learning model is generated based on first data. The second learning model is generated based on the second data for each of second data divided finer than the first data. [Brief description of the drawings]

[0006] [Figure 1] 1 is a diagram showing a fault diagnosis system according to an embodiment; [Diagram 2] FIG. 1 is a block diagram showing a configuration of a refrigerator according to an embodiment. [Diagram 3] FIG. 2 is a block diagram showing the configuration of a server according to the embodiment. [Figure 4] FIG. 2 is a diagram illustrating a first learning model according to the embodiment. [Diagram 5] FIG. 4 is a diagram illustrating a second learning model according to the embodiment. [Figure 6] FIG. 13 is a diagram showing a schematic view of the second learning model according to the embodiment from another perspective. [Figure 7] 10 is a flowchart showing a process flow of a server according to the embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0007] Hereinafter, an information processing system according to an embodiment will be described with reference to the drawings. In the following description, the same reference numerals are used for configurations having the same or similar functions. Then, duplicated descriptions of those configurations may be omitted. "Based on XX" means "based on at least XX" and may include cases based on other elements in addition to XX. "Based on XX" is not limited to cases where XX is directly used, but may also include cases where XX is based on an element that has been subjected to calculation or processing. "XX or YY" is not limited to either one of XX and YY, but may include both XX and YY. This also applies to cases where there are three or more selective elements. "XX" and "YY" are any elements (for example, any information).

[0008] <1. Overall configuration of the fault diagnosis system> 1 is a diagram showing an overall configuration of a fault diagnosis system 1 according to an embodiment. The fault diagnosis system 1 includes, for example, a refrigerator 100, a server 200, and a terminal device 300. The network NW described below may be, for example, the Internet, a cellular network, a Wi-Fi network, a Low Power Wide Area (LPWA), a Wide Area Network (WAN), a Local Area Network (LAN), or other public lines or dedicated lines, depending on the situation.

[0009] The refrigerator 100 is placed in the residence of a user U. The refrigerator 100 has a wireless module 140 (see FIG. 2), which will be described later, and can communicate with the server 200, for example, via a wireless router R and a network NW, which are placed in the same residence. The refrigerator 100 will be described in detail later.

[0010] The server 200 is configured with one or more server devices SD (e.g., cloud servers). The server 200 may be called a "server system." The server 200 may include an information processing unit that performs edge computing or fog computing, such as an information processing unit included in a router in the network NW. The server 200 will be described in detail later.

[0011] The terminal device 300 is a terminal device used by a user U of the refrigerator 100. The terminal device 300 is, for example, a portable terminal device such as a smartphone or a tablet terminal, and includes a display screen capable of displaying the result of the fault diagnosis received from the server 200. However, the terminal device 300 is not limited to the above example, and may be a voice dialogue device (for example, a smart speaker) capable of outputting the result of the fault diagnosis received from the server 200 by voice.

[0012] <2. Refrigerator> 2 is a block diagram showing the configuration of the refrigerator 100. The refrigerator 100 includes, for example, a refrigerator temperature sensor 101, a vegetable compartment temperature sensor 102, a freezer compartment temperature sensor 103, a refrigerator cooler temperature sensor 104, a freezer cooler temperature sensor 105, an ambient temperature sensor 106, an interior camera 107, a refrigerator door switch 108, a vegetable compartment door switch 109, a freezer door switch 110, a current detection unit 111, a compressor 121, a refrigerant switching valve 122, a refrigerator cooling fan 123, a freezer cooling fan 124, a damper 125, a heater 126, a display device 130, a wireless module (communication module) 140, a memory unit 150, and a control unit 160. In addition to the above-mentioned configuration, the refrigerator 100 includes a general configuration of a refrigerator (such as a housing, a door, and a cooler).

[0013] Refrigerator compartment temperature sensor 101 is provided in the refrigerator compartment of refrigerator 100 and detects the temperature of the refrigerator compartment. Crisper compartment temperature sensor 102 is provided in the vegetable compartment of refrigerator 100 and detects the temperature of the vegetable compartment. Freezer compartment temperature sensor 103 is provided in the freezer compartment of refrigerator 100 and detects the temperature of the freezer compartment. Each of refrigerator compartment temperature sensor 101, vegetable compartment temperature sensor 102, and freezer compartment temperature sensor 103 is an example of an "interior temperature sensor" that detects the interior temperature of refrigerator 100. Hereinafter, refrigerator compartment temperature sensor 101, vegetable compartment temperature sensor 102, and freezer compartment temperature sensor 103 will be collectively referred to as "interior temperature sensor RS."

[0014] The refrigeration cooler temperature sensor 104 is attached to the refrigeration cooler and detects the temperature of the refrigeration cooler. The refrigeration cooler is an evaporator to which refrigerant is supplied from the compressor 121 and which cools the air that cools the refrigerator compartment and the vegetable compartment. The freezer cooler temperature sensor 105 is attached to the freezer cooler and detects the temperature of the freezer cooler. The freezer cooler is an evaporator to which refrigerant is supplied from the compressor 121 and which cools the air that cools the freezer compartment. Hereinafter, the refrigeration cooler temperature sensor 104 and the freezer cooler temperature sensor 105 are collectively referred to as "cooler temperature sensor ES". In addition, the refrigeration cooler and the freezer cooler are collectively referred to as "cooler".

[0015] Ambient temperature sensor 106 detects the temperature around refrigerator 100 (i.e., the room temperature of the room in which refrigerator 100 is installed). Interior camera 107 is a camera that is provided in, for example, the refrigerator compartment, the vegetable compartment, or the freezer compartment, and captures the state inside refrigerator 100. Interior camera 107 is a detection unit that detects the putting in and taking out of food to and from the refrigerator compartment, the vegetable compartment, or the freezer compartment and / or the amount of food stored inside the refrigerator (cooling load).

[0016] Refrigerator door switch 108 detects whether the refrigerator door of refrigerator 100 is open or closed. Vegetable door switch 109 detects whether the vegetable door of refrigerator 100 is open or closed. Freezer door switch 110 detects whether the freezer door of refrigerator 100 is open or closed. Hereinafter, refrigerator door switch 108, vegetable door switch 109, and freezer door switch 110 are collectively referred to as "door switch DS."

[0017] Current detection unit 111 detects the value of a current flowing through an electric component included in refrigerator 100. The electric component is, for example, compressor 121, refrigerator cooling fan 123, freezer cooling fan 124, ice tray motor, water supply motor, heater 126, or the like.

[0018] The compressor 121 compresses the refrigerant supplied to the refrigerating cooler and the freezing cooler. The refrigerant switching valve 122 selectively connects the flow path through which the refrigerant compressed by the compressor 121 flows to a flow path that supplies the refrigerating cooler and a flow path that supplies the freezing cooler. The refrigerating cooling fan 123 circulates the cold air cooled by the refrigerating cooler in the refrigerator compartment and the vegetable compartment. The freezing cooling fan 124 circulates the cold air cooled by the freezing cooler in the freezing compartment. Hereinafter, the refrigerating cooling fan 123 and the freezing cooling fan 124 are collectively referred to as "cooling fan F". The damper 125 is provided in the air flow path in the refrigerator 100 and opens and closes the air flow path. The heater 126 is, for example, a heater for defrosting. The heater 126 is provided in the cooler, and is heated when the amount of frost attached to the cooler increases, melting the frost attached to the cooler.

[0019] The display device 130 is, for example, a liquid crystal display or an organic EL display (Electro Luminescence) display, and includes a display screen capable of displaying the result of the failure detection. The display device 130 is an example of an "alarm". The wireless module 140 includes a high-frequency circuit and an antenna, and is capable of wireless communication with the wireless router R.

[0020] The storage unit 150 is realized by, for example, a random access memory (RAM), a read only memory (ROM), a hard disk drive (HDD), a flash memory, or a combination of a plurality of these. The storage unit 150 stores control amount information CI.

[0021] The control amount information CI is information in which basic control amounts of the compressor 121, the cooling fan F, the heater 126, and the like in each operation mode are registered. The basic control amount is a control amount initially set in the refrigerator 100, and is a control amount when the refrigerator 100 is normal. For example, the control amount information CI registers basic control amounts such as the operation frequency of the compressor 121 and the number of revolutions of the cooling fan F (or the operation frequency of the compressor 121 and the number of revolutions of the cooling fan F according to the difference between the cooling target temperature and the inside temperature) in each of the cases where the strength of the refrigeration operation is "weak", "medium", and "strong", the threshold temperature of the cooler at which defrosting of the cooler is started, and the amount of electricity supplied to the heater 126. The control unit 160 operates each component of the refrigerator 100 based on the control amount of each component registered in the control amount information CI.

[0022] The control unit 160 is realized, for example, by a hardware processor such as a CPU (Central Processing Unit) mounted on the refrigerator 100 executing a program (software). However, all or part of the control unit 160 may be realized by hardware (including circuitry) such as an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), or an FPGA (Field Programmable Gate Array), or may be realized by a combination of software and hardware. The control unit 160 controls the refrigerator 100 and manages the inventory of ingredients based on the detection results of the interior temperature sensor RS, the cooler temperature sensor ES, the ambient temperature sensor 106, the interior camera 107, the door switch DS, and the current detection unit 111.

[0023] In this embodiment, the control unit 160 transmits the detection results of the interior temperature sensor RS, the cooler temperature sensor ES, the ambient temperature sensor 106, the interior camera 107, the door switch DS, and the current detection unit 111 to the server 200 as information indicating the operation state of the refrigerator 100 (hereinafter referred to as "operation state information"). The control unit 160 also transmits information indicating the operation mode of the refrigerator 100 at that time to the server 200 as a part of the operation state information. The control unit 160 transmits the operation state information to the server 200 via the wireless module 140 at a predetermined cycle (for example, every 10 minutes). The control unit 160 also controls the display device 130 to display the result of the fault diagnosis received from a notification unit 250 of the server 200 (described later) on the display device 130.

[0024] <3. Server> 3 is a block diagram showing the configuration of the server 200. The server 200 has, for example, an information acquisition unit 210, an information accumulation unit 220, a learning unit (update unit) 230, a diagnosis unit 240, a notification unit 250, a reception unit 260, and a control change unit 270. These functional units are realized, for example, by a hardware processor such as a CPU mounted on the server 200 executing a program (software). However, all or a part of these functional units may be realized by hardware (including a circuit unit; circuitry) such as an ASIC, PLD, or FPGA, or may be realized by cooperation between software and hardware.

[0025] Furthermore, the server 200 has a storage unit 280. The storage unit 280 is realized by, for example, a RAM, a ROM, a HDD, a flash memory, or a combination of a plurality of these. The storage unit 280 stores past data PD, a first learning model M1, and a second learning model M2. These will be described in detail later.

[0026] In this embodiment, server 200 is an example of an "information processing system." Some or all of information acquisition unit 210, information accumulation unit 220, learning unit 230, diagnosis unit 240, notification unit 250, reception unit 260, control change unit 270, and storage unit 280 described above may be provided in refrigerator 100 instead of server 200. In these cases, server 200 and refrigerator 100 may cooperate to configure an example of an "information processing system," or refrigerator 100 may independently correspond to an example of an "information processing system." An example in which the above-mentioned functional units are provided in server 200 will be described below.

[0027] The information acquiring unit 210 communicates via a communication module (not shown) and acquires (receives) operating state information transmitted from the refrigerator 100. That is, the information acquiring unit 210 acquires detection results of the inside temperature sensor RS, the cooler temperature sensor ES, the ambient temperature sensor 106, the inside camera 107, the door switch DS, and the current detecting unit 111 at a predetermined period. During the execution stage of the fault diagnosis, the information acquiring unit 210 acquires operating state information from the refrigerator 100 that is the target of the fault diagnosis. Meanwhile, during the learning stage of the learning model described below, the information acquiring unit 210 collects operating state information from a large number of refrigerators 100 as learning data.

[0028] When new operating state information is acquired by the information acquisition unit 210, the information accumulation unit 220 associates the newly acquired operating state information with date and time information, and adds and accumulates the information to the past data PD in the storage unit 280. In this embodiment, the information accumulation unit 220 adds and accumulates operating state information acquired from the refrigerator 100 at a predetermined cycle (for example, every 10 minutes) to the past data PD as needed. In this way, the information accumulation unit 220 accumulates a history of changes in the operating state of the refrigerator 100.

[0029] The learning unit 230 generates a first learning model M1 and a second learning model M2. The first learning model M1 and the second learning model M2 are trained models trained by machine learning, and are configured by a support vector machine or a neural network, etc.

[0030] FIG. 4 is a diagram showing a schematic diagram of the first learning model M1. For example, the first learning model M1 is configured by a neural network including an input layer, a hidden layer (intermediate layer), and an output layer. The first learning model M1 learns by adjusting weighting coefficients between nodes n included in the input layer, hidden layer, and output layer by deep learning using teacher data. Note that FIG. 4 shows only information related to some of the components as input information and output information. This is the same in FIG. 5 and FIG. 6.

[0031] The first learning model M1 is a learning model that is trained to output a result of a fault diagnosis when the operating state information of the refrigerator 100 is input. For example, when the operating state information of the refrigerator 100 is input as input information, the first learning model M1 outputs, as output information, the possibility that a fault has occurred in the refrigerator 100 and the possibility that a symptom of a fault has occurred in the refrigerator 100.

[0032] The operating state information, which is the input information, may be the most recently acquired single piece of operating state information (a single piece of operating state information acquired at a predetermined cycle), or may be operating state information accumulated by the information accumulation unit 220 for a certain period of time. The "operating state information accumulated for a certain period of time" may be, for example, operating state information for one day, one week, one month, or other period of time. In this embodiment, the operating state information, which is the input information, is operating state information for a first period (for example, one month). The operating state information, which is the input information, includes the detection results of the inside temperature sensor RS, the cooler temperature sensor ES, the ambient temperature sensor 106, the inside camera 107, the door switch DS, and the current detection unit 111.

[0033] In this embodiment, the output information of the first learning model M1 includes the possibility of failure and possible signs of failure for each of the multiple components (compressor 121, cooling fan F, cooler, heater 126, door, housing, etc.) included in the refrigerator 100. In this embodiment, "failure" refers to a state in which the normal functioning of a component is impaired. On the other hand, in this embodiment, "sign" refers to a state in which a small change occurs prior to a failure, such as a state in which a component operates but the operation may become unstable or may be accompanied by abnormal operation.

[0034] With respect to the compressor 121, a "failure" is when the compression ratio decreases due to a refrigerant leak or when an overcurrent flows, and a "sign" is when the vibration of the compressor 121 increases or when an abnormal sound is generated from the compressor 121. With respect to the cooling fan F, a "failure" is when the rotation of the cooling fan F stops, and a "sign" is when the vibration of the cooling fan F increases or when an abnormal sound is generated from the cooling fan F. With respect to the cooler, a "failure" is when the surface coating peels off due to aging deterioration, a lot of frost adheres, and the cooler cannot maintain its function, and a "sign" is when frost becomes more likely to adhere, and the cooling of the refrigerator compartment or freezer compartment (such as the rate of temperature reduction during cooling or the minimum temperature) becomes worse depending on conditions such as the operating mode and the ambient temperature of the refrigerator 100.

[0035] With regard to heater 126, a "failure" is when the cooler cannot be heated or when an overcurrent flows, and a "sign" is when the cooler cannot be heated sufficiently (cannot reach the target temperature within a specified time) depending on conditions such as the operation mode or the ambient temperature of refrigerator 100. With regard to the housing or door, a "failure" is when the heat retention function drops below a specified level, and a "sign" is when a slow leak occurs in the vacuum insulation material contained inside the housing or door, causing the cooling of the refrigerator or freezer to gradually deteriorate.

[0036] The learning unit 230 performs supervised learning based on the teacher data generated based on the past data PD stored in the storage unit 280, and generates a first learning model M1. The teacher data includes a first data set in which information on the normal operation state of the refrigerator 100 is associated with information indicating that each component of the refrigerator 100 is normal (correct answer data), a second data set in which information on the operation state in a state in which one or more components of the refrigerator 100 have actually had a failure is associated with information indicating that the one or more components have had a failure (correct answer data), and a third data set in which information on the operation state in a state in which one or more components of the refrigerator 100 have actually had a failure prediction is associated with information indicating that the one or more components have had a failure prediction (correct answer data). These correct answer data are registered based on the contents confirmed in a support service such as an actual repair or maintenance of the refrigerator 100, for example.

[0037] The first learning model M1 is generated with the first division as a group (one unit). The first division is a division divided by model of refrigerator 100, or a division divided by multiple models having the same basic design. In the following, an example in which the first division is set by model of refrigerator 100 will be described. That is, the first learning model M1 is trained based on operating state information and repair and inspection results collected from multiple refrigerators 100 of the same model installed in multiple homes.

[0038] 5 is a diagram showing a schematic diagram of the second learning model M2. For example, the second learning model M2 is configured by a neural network including an input layer, a hidden layer (intermediate layer), and an output layer. The second learning model M2 is learned by adjusting weighting coefficients between nodes n included in the input layer, hidden layer, and output layer by deep learning using teacher data.

[0039] In the second learning model M2, the number of nodes in the input layer is the same as the number of nodes in the output layer. In the second learning model M2, the number of nodes in at least one hidden layer is less than the number of nodes in the input layer and is also less than the number of nodes in the output layer. The second learning model M2 is a model that, when certain information is input to the input layer, compresses the input information in the hidden layer, restores the compressed information, and outputs the restored information from the output layer. In this embodiment, the second learning model M2 is trained to restore and output information identical to the input information when normal operating state information of the refrigerator 100 is input as input information. Information identical to the input information means information whose difference with respect to the input information is less than a predetermined threshold value set in advance.

[0040] In this embodiment, the driving state information, which is the input information input to the second learning model M2, may be, for example, the most recently acquired one piece of driving state information (one piece of driving state information acquired at a predetermined cycle), or may be driving state information accumulated by the information accumulation unit 220 for a certain period. The "driving state information accumulated for a certain period" may be, for example, one day's worth of driving state information, one week's worth of driving state information, one month's worth of driving state information, or other period's worth of driving state information. In this embodiment, the driving state information, which is the input information input to the second learning model M2, is accumulated driving state information for a second period (for example, one day) shorter than the first period. The driving state information, which is the input information, includes the detection results of the inside temperature sensor RS, the cooler temperature sensor ES, the ambient temperature sensor 106, the inside camera 107, the door switch DS, and the current detection unit 111. In this embodiment, the output information is the same as the input information, that is, the detection results of the inside temperature sensor RS, the cooler temperature sensor ES, the ambient temperature sensor 106, the inside camera 107, the door switch DS, and the current detection unit 111, etc.

[0041] The learning unit 230 performs supervised learning based on the teacher data generated based on the past data PD stored in the storage unit 280, and generates the second learning model M2. The teacher data is a data set in which the same operating state information (correct answer data) is associated as output information with the input information, which is the operating state information of the refrigerator 100 in a normal state. In other words, the teacher data for the second learning model M2 only needs to be the operating state information of the refrigerator 100 in a normal state, and does not need to be the operating state information when the refrigerator 100 has a malfunction or when a symptom of a malfunction has occurred in the refrigerator 100.

[0042] 6 is a diagram showing the second learning model M2 from another perspective. As a result of the above-described learning, when the operating state information when an abnormality occurs in the refrigerator 100 is input as input information, the output information output from the second learning model M2 does not match the input information. In other words, the difference between the input information and the output information is equal to or greater than the above-mentioned predetermined threshold.

[0043] The second learning model M2 is generated by treating the second division, which is different from the first division described above, as a group (one unit). The second division is, for example, a division that is more finely divided than the first division. The second division is a division for each user U of the refrigerator 100, a division for each user U with the same attribute (single person / nuclear family / large family, or young people / child-rearing generation / elderly, etc.), or a division for each model of the refrigerator 100. In the following, an example in which the second division is set for each user U of the refrigerator 100 will be described. That is, the second learning model M2 is learned for each user U based on normal operating state information collected from the refrigerator 100 of a certain user U.

[0044] The diagnosing unit 240 performs a fault diagnosis on the refrigerator 100 based on information indicating the operating state of the refrigerator 100. For example, the diagnosing unit 240 includes a first diagnosing unit 241 and a second diagnosing unit 242.

[0045] The first diagnosis unit 241 uses the first learning model M1 to perform a first fault diagnosis of the refrigerator 100. That is, the first diagnosis unit 241 inputs operating state information (for example, operating state information for the above-mentioned first period) acquired from the refrigerator 100 to be diagnosed for faults to the first learning model M1, and causes the first diagnosis unit 241 to output the possibility that each component of the refrigerator 100 to be diagnosed for faults has a fault and the possibility that each component of the refrigerator 100 has a symptom of a fault.

[0046] As a result of the first fault diagnosis of refrigerator 100, first diagnosis unit 241 outputs to notification unit 250 information indicating the possibility that a fault has occurred in each component of refrigerator 100 (for example, "There is a XX% possibility that the compressor has a fault") and information indicating the possibility that a symptom of a fault has occurred in each component of refrigerator 100 (for example, "There is a XX% possibility that a symptom of a fault has occurred in the compressor").

[0047] The second diagnosis unit 242 performs a second fault diagnosis of the refrigerator 100 using the second learning model M2. The second diagnosis unit 242 inputs operating state information (e.g., operating state information for the above-mentioned second period) acquired from the refrigerator 100 to be diagnosed as a fault, to the second learning model M2. The second diagnosis unit 242 compares the operating state information of each component of the refrigerator 100 input to the second learning model M2 with the operating state information of each component of the refrigerator 100 output from the second learning model M2. The operating state information output from the second learning model M2 is an example of a "predetermined comparison target."

[0048] Then, when the difference between the operating state information of the component of the refrigerator 100 input to the second learning model M2 and the operating state information of the component output from the second learning model M2 is equal to or greater than the above-mentioned predetermined threshold, the second diagnosis unit 242 determines that an abnormality exists related to the component. On the other hand, when the difference between the operating state information of all the components input to the second learning model M2 and the operating state information of all the components output from the second learning model M2 is less than the above-mentioned predetermined threshold, the second diagnosis unit 242 determines that no abnormality exists in any of the components of the refrigerator 100. As a result of the second fault diagnosis of the refrigerator 100, the second diagnosis unit 242 outputs information indicating the presence or absence of an abnormality for each component of the refrigerator 100 to the notification unit 250. Note that the second fault diagnosis may only determine the presence or absence of an abnormality for the entire refrigerator 100 instead of determining the presence or absence of an abnormality for each component of the refrigerator 100.

[0049] Notification unit 250 determines whether the possibility of a failure output as a result of the first failure diagnosis for each component included in refrigerator 100 is equal to or greater than a first threshold. When the possibility of a failure for each component is equal to or greater than the first threshold, notification unit 250 transmits a predetermined control command to refrigerator 100 or terminal device 300 to output a predetermined first notification from display device 130 of refrigerator 100 or terminal device 300. "Outputting a notification" means, for example, displaying a predetermined message on the display screen or outputting an audio message from a speaker.

[0050] The first notification includes information indicating the result of the first fault diagnosis and a notification prompting the user U to perform repairs. The information indicating the result of the first fault diagnosis included in the first notification is an example of "information based on the result of the fault diagnosis". The information indicating the result of the first fault diagnosis included in the first notification may include information indicating which component included in the refrigerator 100 may have a fault. The first notification is, for example, a message such as "The compressor may have a fault. Please contact the support center for repairs". On the other hand, when the possibility that all the components of the refrigerator 100 have a fault is less than the first threshold, the notification unit 250 does not output the first notification.

[0051] Furthermore, the notification unit 250 judges whether or not the possibility of occurrence of a failure sign outputted as a result of the first failure diagnosis for each component included in the refrigerator 100 is equal to or greater than a second threshold. When the possibility of occurrence of a failure for each component is less than the first threshold but the possibility of occurrence of a failure sign for the component is equal to or greater than the second threshold, the notification unit 250 transmits a predetermined control command to the refrigerator 100 or the terminal device 300 to cause the display device 130 of the refrigerator 100 or the terminal device 300 to output a predetermined second notification. The second threshold may be the same value as the first threshold, may be a value higher than the first threshold, or may be a value lower than the first threshold. For example, the second threshold is set lower than the first threshold. This allows the user U to be notified of a failure sign at an early stage before a failure actually occurs.

[0052] The second notification includes information indicating the result of the first fault diagnosis and a notification prompting the user U to perform an inspection. The information indicating the result of the first fault diagnosis included in the second notification is an example of "information based on the result of the fault diagnosis". The information indicating the result of the first fault diagnosis included in the second notification may include information indicating which component included in the refrigerator 100 may have a sign of a fault. The second notification is, for example, a message such as "There may be a sign of a fault in the compressor. Please check whether there is any abnormal noise coming from the compressor". On the other hand, when the possibility that a sign of a fault has occurred in all the components of the refrigerator 100 is less than the second threshold, the notification unit 250 does not output the second notification.

[0053] Furthermore, when it is determined as a result of the second fault diagnosis that an abnormality exists in at least one component included in refrigerator 100, notification unit 250 transmits a predetermined control command to refrigerator 100 or terminal device 300, thereby causing display device 130 of refrigerator 100 or terminal device 300 to output a predetermined third notification.

[0054] The third notification includes information indicating the result of the second fault diagnosis and a notification prompting the user U to check the condition of the refrigerator 100. The information indicating the result of the second fault diagnosis included in the third notification is an example of "information based on the result of the fault diagnosis". The information indicating the result of the second fault diagnosis included in the third notification may include information indicating which component of the refrigerator 100 has an abnormality. The third notification is, for example, a message such as "There may be an abnormality in the compressor. Please check whether there is any abnormality such as an abnormal sound coming from the compressor". On the other hand, when it is determined that no abnormality has occurred in any of the components of the refrigerator 100, the notification unit 250 does not output the third notification.

[0055] Furthermore, when the user's input of the confirmation result for the third notification is accepted by the accepting unit 260 described below and the accepted user's input suggests that there may be a malfunction in the refrigerator 100, the notifying unit 250 may transmit a predetermined control command to the refrigerator 100 or the terminal device 300 to output a predetermined fourth notification from the display device 130 of the refrigerator 100 or the terminal device 300. The fourth notification is a notification that informs the user that the abnormality that the user has confirmed is a malfunction or a sign of a malfunction in a component of the refrigerator 100.

[0056] The reception unit 260 receives input of information based on the results of the repair or inspection. The information based on the results of the repair includes, for example, information indicating whether or not a failure actually occurred when the repair or maintenance was performed in response to the first notification, and, if a failure occurred, which component actually had the failure. The information based on the results of the repair is input, for example, by a worker at the support center using an input device (not shown). The reception unit 260 receives the information input using the input device. The reception unit 260 adds the information received by the reception unit 260 (correct answer data) to a part of the teacher data in association with the operating state information used in the first fault diagnosis.

[0057] On the other hand, the information based on the result of the inspection includes, for example, information indicating whether or not a sign of failure actually occurred when the user U performed an inspection in response to the second notification, and, if a sign of failure occurred, information indicating which component actually had the sign of failure. The information based on the result of the inspection is input by the user U of the refrigerator 100 using, for example, the terminal device 300 or the display device 130 of the refrigerator 100. The reception unit 260 receives the information input using, for example, the terminal device 300 or the display device 130 of the refrigerator 100. The reception unit 260 adds the information received by the reception unit 260 (correct answer data) to a part of the teacher data in association with the operating state information used in the first fault diagnosis.

[0058] In this embodiment, the learning unit 230 additionally learns the first learning model M1 using the information accepted by the accepting unit 260. Then, the learning unit 230 updates the old first learning model M1 using the first learning model M1 newly generated by the additional learning. The learning unit 230 may additionally learn the first learning model M1 every time new information is accepted by the accepting unit 260, or may additionally learn the first learning model M1 when the information accepted by the accepting unit 260 is accumulated to a predetermined amount or more.

[0059] When it is determined that the possibility of a malfunction is equal to or greater than the third threshold as a result of the first malfunction diagnosis, the control change unit 270 changes the control of the refrigerator 100. For example, the control change unit 270 changes the control amount information CI stored in the storage unit 150 of the refrigerator 100 by transmitting a predetermined control command to the refrigerator 100. For example, the control change unit 270 changes the basic control amount included in the control amount information CI to the first corrected control amount. As a result, the control unit 160 of the refrigerator 100 operates the refrigerator 100 based on the first corrected control amount. The third threshold may be the same as the first threshold, or may be higher than the first threshold, or may be lower than the first threshold. For example, when the third threshold is set lower than the first threshold, the control of the refrigerator 100 can be changed so as to prevent malfunction at an earlier stage. Changing the basic control amount to the first corrected control amount is an example of "changing in the first manner".

[0060] In this embodiment, when it is determined that the possibility of a malfunction occurring in the compressor 121 is equal to or greater than the third threshold, the control change unit 270 changes the control of the refrigerator 100 so as to reduce the drive amount of the compressor 121 for the same operating conditions (e.g., the same operating mode) before and after the change. Similarly, when it is determined that the possibility of a malfunction occurring in the cooling fan F is equal to or greater than the third threshold, the control change unit 270 changes the control of the refrigerator 100 so as to reduce the drive amount of the cooling fan F for the same operating conditions (e.g., the same operating mode) before and after the change.

[0061] In this embodiment, when it is determined that the possibility of the cooler malfunctioning is equal to or greater than the third threshold, the control change unit 270 changes the control of the refrigerator 100 so that the amount of heat generated by the heater 126 provided in the cooler increases for the same operating conditions (e.g., the same operating mode) before and after the change. In this embodiment, when it is determined that the possibility of the heater 126 malfunctioning is equal to or greater than the third threshold, the control change unit 270 changes the control of the refrigerator 100 so that the amount of heat generated by the heater 126 decreases for the same operating conditions (e.g., the same operating mode) before and after the change. In this embodiment, when it is determined that the possibility of the housing or the door malfunctioning is equal to or greater than the third threshold, the control change unit 270 changes the control of the refrigerator 100 so that the amount of drive of the compressor 121 and / or the cooling fan F increases for the same operating conditions (e.g., the same operating mode) before and after the change.

[0062] On the other hand, when it is determined that the possibility of a failure occurring as a result of the first failure diagnosis is less than the third threshold value, but the possibility of a failure occurring is equal to or greater than the fourth threshold value, the control change unit 270 changes the control of the refrigerator 100. For example, the control change unit 270 changes the control amount information CI stored in the storage unit 150 of the refrigerator 100 by transmitting a predetermined control command to the refrigerator 100. For example, the control change unit 270 changes the basic control amount included in the control amount information CI to the second corrected control amount. As a result, the control unit 160 of the refrigerator 100 operates the refrigerator 100 based on the second corrected control amount. For example, the second corrected control amount is a control amount in which the correction amount for the basic control amount is smaller than that of the first corrected control amount. The fourth threshold value may be the same as the second threshold value, or may be higher or lower than the second threshold value. For example, when the fourth threshold value is set lower than the second threshold value, the control of the refrigerator 100 can be changed so as to prevent a failure at an earlier stage. Changing the basic control amount to the second adjusted control amount is an example of "changing in the second manner."

[0063] In this embodiment, the control change unit 270 changes the control amount in basically the same direction (for example, in the direction of decreasing or increasing the driving amount) when there is a possibility that a failure has occurred in each component included in the refrigerator 100 and when there is a possibility that a symptom of a failure has occurred in the component. For this reason, in the control example when there is a possibility that a failure symptom has occurred, "when it is determined that the possibility of a failure occurring is equal to or greater than the first threshold value" in the above-mentioned explanation of each control example regarding the compressor 121, the cooling fan F, the cooler, the heater 126, the housing, and the door can be read as "when it is determined that the possibility of a failure occurring is equal to or greater than the second threshold value."

[0064] <4. Processing flow> 7 is a flowchart showing a process flow of the server 200 according to the embodiment. First, the information acquiring unit 210 acquires operation state information from the refrigerator 100 to be diagnosed (S101). The information storing unit 220 adds the information acquired by the information acquiring unit 210 to the past data PD and stores it (S102).

[0065] The diagnosis unit 240 inputs the latest operating state information accumulated by the information acquisition unit 210 over the first period as input information to the first learning model M1, and performs a first fault diagnosis (S103). That is, as the first fault diagnosis, the diagnosis unit 240 determines whether all components of the refrigerator 100 are normal (that is, whether the possibility that a fault has occurred in each component is less than a first threshold value and the possibility that a symptom of a fault has occurred in each component is less than a second threshold value).

[0066] When it is determined in the first fault diagnosis that at least one component of the refrigerator 100 is not normal (S103: fault / sign of fault), the notification unit 250 determines whether or not the possibility of fault occurring in the component is equal to or greater than a first threshold value (S104). When it is determined that the possibility of fault occurring in at least one component is equal to or greater than the first threshold value (S104: fault), the notification unit 250 performs the first notification described above as it is highly likely that a fault has occurred (S105).

[0067] On the other hand, when the possibility that a component included in refrigerator 100 has a malfunction is less than the first threshold value but the possibility that a symptom of a malfunction has occurred in the component is equal to or greater than the second threshold value (S104: malfunction indication), notification unit 250 determines that there is a high possibility that a symptom of a malfunction has occurred and issues the above-mentioned second notification (S106).

[0068] Next, the learning unit 230 determines whether or not the input of information for additional learning based on the results of the repair or inspection has been accepted by the accepting unit 260 (S111). If the input of information for additional learning has been accepted by the accepting unit 260 (S111: YES), the learning unit 230 additionally learns the first learning model M1 based on the information accepted by the accepting unit 260, and updates the first learning model M1 (S112). On the other hand, if the input of information for additional learning has not been accepted by the accepting unit 260 (S111: NO), the process returns to S101, and the above-mentioned process is repeated.

[0069] Next, when it is determined in the first fault diagnosis that at least one component part of refrigerator 100 is not normal, control change unit 270 changes control amount information CI of refrigerator 100 based on the first corrected control amount or the second corrected control amount. In this way, control change unit 270 changes the control of refrigerator 100 (S113). Then, the process returns to S101, and the above-mentioned processes are repeated.

[0070] On the other hand, when it is determined in the first fault diagnosis that all components of the refrigerator 100 are normal (S103: normal), the information accumulation unit 220 determines whether or not sufficient operating state information (normal data) of the refrigerator 100 in a normal state has been accumulated to generate the second learning model M2 (S121). When sufficient operating state information (normal data) of the refrigerator 100 in a normal state has not been accumulated (S121: NO), the information accumulation unit 220 accumulates the most recent acquired operating state information as part of the normal data (S122).

[0071] On the other hand, if sufficient operating state information (normal data) of the refrigerator 100 under normal conditions has been accumulated and the second learning model M2 has already been learned and generated (S121: YES), the diagnosis unit 240 inputs the latest operating state information accumulated over the second period by the information acquisition unit 210 as input information to the second learning model M2, and performs a second fault diagnosis (S123). That is, the diagnosis unit 240 determines whether or not there is an abnormality in each component part included in the refrigerator 100, as the second fault diagnosis.

[0072] Then, when each component included in refrigerator 100 is judged to be normal (S123: normal), information accumulation unit 220 accumulates the operating state information used in the second fault diagnosis as part of the normal data (S122). On the other hand, when at least one component included in refrigerator 100 is judged to be abnormal (S123: abnormal), notification unit 250 issues a third notification to prompt the user to check the condition of refrigerator 100 (S124).

[0073] Next, the reception unit 260 receives an input of a user's confirmation result for the third notification, and judges whether or not the received user's input is a confirmation result suggesting that the refrigerator 100 may have a malfunction (S125). When the reception unit 260 receives an input from the user indicating that the refrigerator 100 may not have a malfunction (S125: NO), the information storage unit 220 stores the operating state information used in the second malfunction diagnosis as a part of the normal data (S122). On the other hand, when the reception unit 260 receives an input from the user indicating that the refrigerator 100 may have a malfunction (S125: YES), the information storage unit 220 performs the fourth notification described above (S126). In this case, the control change unit 270 changes the control of the refrigerator 100 (S113). After that, the process returns to S101, and the above-mentioned process is repeated.

[0074] <5. Effect> As a comparative example, a fault diagnosis system that performs fault diagnosis by setting thresholds for current values ​​and sensor values ​​will be considered. In such a fault diagnosis system, in the case of a fault in which it is difficult to associate the operating state of the refrigerator with the presence or absence of a fault, it may be difficult to improve the accuracy of the fault diagnosis.

[0075] In the present embodiment, the information processing system includes a diagnosing unit 240 that performs a first fault diagnosis on the refrigerator 100 using a first learning model M1 that has been trained to output a result of the first fault diagnosis when information indicating an operating state of the refrigerator is input, and a notifying unit 250 that causes the refrigerator 100 or the terminal device 300 to output information based on the result of the first fault diagnosis by the diagnosing unit 240. With this configuration, even in the case of a fault where it is difficult to associate the operating state of the refrigerator with the presence or absence of the fault, a more appropriate fault diagnosis can be performed.

[0076] In this embodiment, when it is determined that the possibility of a malfunction is equal to or greater than the first threshold as a result of the first malfunction diagnosis, the notification unit 250 outputs a notification urging the user U to perform repairs. With this configuration, the user U can take action at the stage when there is a possibility of a malfunction, and can perform repairs or maintenance before the functions of the refrigerator 100 stop. This improves convenience for the user U.

[0077] In this embodiment, when it is determined that the first fault diagnosis indicates that the possibility of a fault is less than a first threshold value but that the possibility of a fault symptom is equal to or greater than a second threshold value, the notification unit 250 outputs a notification urging the user U to perform an inspection. With this configuration, the user U can take action at an earlier stage, and can more reliably perform repairs or maintenance before the refrigerator 100 stops functioning. This further improves the convenience for the user U.

[0078] In this embodiment, the information processing system includes a receiving unit 260 that receives input of information based on the results of repair or inspection, and a learning unit 230 that updates the first learning model M1 based on a learning model additionally learned using the information received by the receiving unit 260. With this configuration, the first learning model M1 can be updated using information based on the results of repair or inspection. This can further improve the accuracy of fault diagnosis.

[0079] In this embodiment, the diagnosis unit 240 performs a first fault diagnosis using a first learning model M1 that has been trained to output a fault diagnosis result when information indicating the operating state of the refrigerator 100 is input, and if the first fault diagnosis determines that the refrigerator 100 is normal, performs a second fault diagnosis to additionally determine the presence or absence of an abnormality based on the information indicating the operating state of the refrigerator 100. With this configuration, it is possible to detect the presence or absence of an abnormality even for faults that occur infrequently and for which it is difficult to collect sufficient teacher data. This allows for more appropriate fault diagnosis.

[0080] In the present embodiment, the diagnosis unit 240 performs the second fault diagnosis by comparing the operating state of the refrigerator 100 with a predetermined comparison target. With such a configuration, the presence or absence of an abnormality can be detected by a relatively simple method.

[0081] In this embodiment, when information indicating the normal operating state of the refrigerator 100 is input as input information, the diagnosing unit 240 uses the second learning model M2 that has been trained to restore and output information identical to the input information, and performs the second fault diagnosis based on a comparison between information indicating the operating state of the refrigerator 100 input to the second learning model M2 and output information output from the second learning model M2. With this configuration, fault diagnosis can be performed with higher accuracy even for faults that occur infrequently and for which it is difficult to collect sufficient teacher data. For example, it is possible to handle anomaly judgment that reflects the usage of the user U and cases in which it is difficult to associate the operating state with the presence or absence of a fault.

[0082] In this embodiment, the information processing system further includes a control change unit 270 that changes control of the refrigerator 100 when it is determined that the possibility of a malfunction is equal to or greater than a third threshold as a result of the malfunction diagnosis. With this configuration, it is possible to prevent a malfunction when a malfunction has not actually occurred yet.

[0083] In this embodiment, when the failure diagnosis result indicates that the possibility of a failure is less than the third threshold value, but that the possibility of a failure symptom is equal to or greater than the fourth threshold value, the control change unit 270 changes the control of the refrigerator 100. With this configuration, it is possible to prevent failures at an earlier stage.

[0084] <6. Variations> In the above-described embodiment, an example has been described in which the second fault diagnosis is performed using the second learning model M2. However, the second fault diagnosis is not limited to the above example, and may be performed based on a comparison between a most recent operating state of the refrigerator 100 and a past operating state of the refrigerator 100. The most recent operating state of the refrigerator 100 is, for example, an operating state in a most recent second period. The past operating state of the refrigerator 100 is an operating state in a past second period accumulated in the past data PD. The past operating state of the refrigerator 100 is an example of a "predetermined comparison target."

[0085] In this modification, in the process of S123 shown in Fig. 7, a second fault diagnosis is performed based on a comparison between the latest operating state of the refrigerator 100 and the past operating state of the refrigerator 100. Other configurations and processes are similar to those of the above embodiment.

[0086] According to at least one of the embodiments described above, the information processing system includes a diagnosis unit that performs a fault diagnosis on the refrigerator using a learning model that is trained to output a fault diagnosis result when information indicating an operating state of the refrigerator is input, and a notification unit that causes the refrigerator or a terminal device to output information based on the fault diagnosis result by the diagnosis unit. With this configuration, it is possible to perform a more appropriate fault diagnosis.

[0087] Although some embodiments of the present invention have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their modifications are included in the scope of the invention and its equivalents described in the claims, as well as in the scope and spirit of the invention. [Explanation of symbols]

[0088] 1...fault diagnosis system, 100...refrigerator, 200...server, 210...information acquisition unit, 220...information storage unit, 230...learning unit (update unit), 240...diagnosis unit, 250...notification unit, 260...reception unit, 270...control change unit, 300...terminal device.

Claims

1. a diagnosis unit that performs a fault diagnosis on the refrigerator by using a learning model that is trained to output a result of a fault diagnosis when information indicating an operation state of the refrigerator is input; a notification unit that causes the refrigerator or the terminal device to output information based on a result of the fault diagnosis by the diagnosis unit; and a reception unit that receives an input of information based on a result of repairing the refrigerator or inspecting the refrigerator; and an update unit that updates the learning model based on a learning model additionally learned using the information accepted by the acceptance unit; Equipped with The learning model is a first learning model that is trained to output a result of the fault diagnosis when information indicating an operation state of the refrigerator is input; a second learning model that is trained to restore and output information identical to the input information when information indicating a normal operating state of the refrigerator is input as input information, The diagnosis unit includes: performing a fault diagnosis of the refrigerator by a first fault diagnosis using the first learning model, the first fault diagnosis detecting a fault of a component of the refrigerator or a sign of a fault of the refrigerator, and when a fault or a sign of a fault is not detected in the refrigerator by the first fault diagnosis, further performing a second fault diagnosis using the second learning model; The first learning model is generated based on first data; The second learning model is generated based on each of second data that is divided finer than the first data. Information processing system.

2. the notification unit outputs a notification to a user to prompt the user to perform the repair when it is determined that a possibility of occurrence of a malfunction is equal to or greater than a first threshold value as a result of the malfunction diagnosis. The information processing system according to claim 1 .

3. When it is determined that the possibility of occurrence of a failure is less than the first threshold value as a result of the failure diagnosis, but the possibility of occurrence of a sign of a failure is equal to or greater than a second threshold value, the notification unit outputs a notification urging the user to perform the inspection. The information processing system according to claim 2 .

4. The first data is data from a plurality of refrigerators, The second data is data divided into smaller parts than the first data according to a category associated with each user.

3. The information processing system according to claim 1 or 2.

5. The diagnosis unit performs the second fault diagnosis by comparing an operation state of the refrigerator with a predetermined comparison target. The information processing system according to claim 1 .

6. The diagnosis unit performs the second fault diagnosis based on a comparison between a most recent operation state of the refrigerator and a past operation state of the refrigerator. The information processing system according to any one of claims 1 to 5.

7. and a control change unit that changes control of the refrigerator to a first mode when it is determined as a result of the failure diagnosis that a possibility of a failure is equal to or greater than a third threshold. The information processing system according to any one of claims 1 to 6.

8. when it is determined that a possibility of occurrence of a failure is less than the third threshold value as a result of the failure diagnosis but a possibility of occurrence of a sign of a failure is equal to or greater than a fourth threshold value, the control change unit changes the control of the refrigerator to a second mode different from the first mode. The information processing system according to claim 7.

9. A diagnostic unit which performs a first fault diagnosis using a first learning model trained to output the result of a fault diagnosis when information indicating the operating state of the refrigerator is input, and if no fault or signs of fault are detected in the refrigerator in the first fault diagnosis, uses a second learning model trained to restore and output information identical to the input information when information indicating the normal operating state of the refrigerator is input as input information, and performs a second fault diagnosis of the refrigerator based on a comparison between the information indicating the operating state of the refrigerator which is input to the second learning model and the output information output from the second learning model. a notification unit that causes the refrigerator or the terminal device to output information based on a result of the fault diagnosis by the diagnosis unit; and a reception unit that receives an input of information based on a result of repairing the refrigerator or inspecting the refrigerator; and an update unit that updates the learning model based on a learning model additionally learned using the information accepted by the acceptance unit; Equipped with The first learning model is generated based on first data; The second learning model is generated based on each of second data that is divided finer than the first data. Information processing system.

Citation Information

Patent Citations

  • A fault prediction method of refrigeration equipment based on neural network

    CN109308519A

  • Electric appliance fault determination method and device

    CN110954758A

  • Broadcast master control fault trend pre-judgment method based on machine learning

    CN110995384A

  • Kitchen electric equipment predictive maintenance system and method based on side-cloud collaboration

    CN111160616A

  • Operating state managing apparatus for equipment

    JP1998238920A