Home appliance system

Abstract

The present invention aims to provide a home appliance system capable of more appropriate fault diagnosis. In one embodiment, the home appliance system includes a diagnostic unit. The diagnostic unit performs fault diagnosis for the home appliance based on a learned model. The learned model is learned to output a fault diagnosis result related to the home appliance when inputted with information based on detection results from one or more sensors included in the home appliance.

Description

Technical Field

[0001] An embodiment of the present invention relates to a home appliance system. Background Art

[0002] Home appliances may experience various malfunctions due to various reasons. However, the causal relationship between the malfunction and the sensor values ​​detected by the appliance may be unclear. This makes proper fault diagnosis difficult.

[0003] Prior art literature

[0004] Patent Literature

[0005] Patent Document 1: Japanese Patent Application Laid-Open No. 2016-173782 Summary of the Invention

[0006] Problems to be solved by the invention

[0007] An object of the present invention is to provide a home appliance system capable of performing more appropriate fault diagnosis.

[0008] Means for solving problems

[0009] A home appliance system according to an embodiment includes a diagnostic unit that performs fault diagnosis of a home appliance based on a learned model that has been learned to output a fault diagnosis result related to the home appliance when information based on detection results of one or more sensors included in the home appliance is input.

[0010] Effects of the Invention

[0011] The above-described home appliance system enables more appropriate fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a diagram showing the configuration of a home appliance system according to the first embodiment.

[0013] Figure 2 It is a diagram showing a part of the configuration of the household appliance according to the first embodiment.

[0014] Figure 3 This is a diagram showing an example of the content of the correspondence table according to the first embodiment.

[0015] Figure 4 This is a diagram showing an example of a first data table according to the first embodiment.

[0016] Figure 5 This is a diagram schematically showing a learned model according to the first embodiment.

[0017] Figure 6 This is a diagram showing an example of the second data table according to the first embodiment.

[0018] Figure 7 This is a diagram showing an example of a first processing flow in the first embodiment.

[0019] Figure 8 This is a diagram showing an example of the second processing flow in the first embodiment. DETAILED DESCRIPTION

[0020] Hereinafter, the household appliance system of the embodiment will be described with reference to the accompanying drawings. In the following description, the same figure marks are marked for the structures with the same or similar functions. The repeated description of the structure is sometimes omitted. "Based on XX" means "at least based on XX", and may include the case of being based on other elements besides XX. "Based on XX" is not limited to the case of directly using XX, and may include the case of being based on the result obtained after calculation and processing relative to XX. "XX or YY" is not limited to the case of one of XX and YY, and may include the case of both XX and YY. This is also the same when the selected elements are three or more. "XX" and "YY" are arbitrary elements (for example, arbitrary information).

[0021] (First embodiment)

[0022] [Composition of home appliance system]

[0023] Figure 1 This diagram shows the configuration of a home appliance system 1 according to the first embodiment. The home appliance system 1 generates a learned model using information accumulated for each home appliance, such as the operating status of the home appliance and physical quantities sensed by sensors. The home appliance system 1 uses the generated learned model to diagnose home appliance faults.

[0024] like Figure 1 As shown, the home appliance system 1 includes a plurality of home appliances (home appliances 10a, 10b, ..., 10c), an interface device (interface unit) 20, a learning device (learning unit) 30, and a fault diagnosis device (fault diagnosis unit) 40. Each of the home appliances 10a, 10b, ..., 10c, the learning device 30, and the fault diagnosis device 40 can communicate with the interface device 2 directly or via a network not shown. For the sake of convenience, the following describes an example in which the interface device 20, the learning device 30, and the fault diagnosis device 40 are independent devices. However, as described later, the interface device 20, the learning device 30, and the fault diagnosis device 40 can also be implemented by one device. In addition, the home appliances 10a, 10b, ..., 10c are collectively referred to as home appliances 10. The home appliances 10 are configured in each home of the user.

[0025] [Composition of home appliances]

[0026] like Figure 1 As shown, the home appliance 10a includes a sensor 102a and a transmitter 101a. The home appliance 10a is, for example, a refrigerator, a washing machine, an air conditioner (air conditioner), a vacuum cleaner, a microwave oven, a television receiver, or a lighting fixture.

[0027] Sensor 102a is one or more sensors provided in the household appliance 10a. The sensor 102a senses a number of physical quantities corresponding to the number of sensors in the household appliance 10a, etc. The types of physical quantities include temperature, humidity, pressure, current, voltage, the number of rotations of the motor, and the number of times the door is opened and closed, etc. The sensor can also be a photographing device such as a camera. Therefore, the "sensor detection result" can also be image data captured by a camera, etc. In order to simplify the explanation, the case including such image data is referred to as "physical quantity" below. The physical quantity can also be referred to as "diagnostic information". The type of physical quantity sensed by the sensor 102a varies depending on the type of household appliance (for example, refrigerator, washing machine, air conditioner, electric vacuum cleaner, microwave oven, television receiver, lighting fixture, etc.) or the model (model, etc.) of the household appliance.

[0028] The transmitting unit 101a includes a high-frequency circuit and an antenna (not shown), and is capable of communicating with the interface device 20 via, for example, a wireless router and a network within a home where the home appliance 10a is installed. The transmitting unit 101a transmits the physical quantity sensed by the sensor 102a, along with the operating state of the home appliance 10a at the time the physical quantity was sensed and the identifier ID1 of the home appliance 10a, to the interface device 20 at predetermined intervals.

[0029] For example, if the household appliance 10a is a refrigerator, the operating state of the household appliance 10a refers to the operating state determined by whether it is in refrigeration or cooling mode, whether it is in rapid cooling mode, rapid ice making mode, normal freshness mode or rapid freshness mode, or whether it is in defrost mode. The operating state may also include time information from the start of each of the above operating modes. For example, if the household appliance 10a is a washing machine, the operating state of the household appliance 10a refers to the operating state determined by whether it is in standard mode, quick mode, or delicate wash mode, and whether the drying function is used.

[0030] For example, when the household appliance 10a is an air conditioner, the state during operation is determined by whether the heating operation, cooling operation, or dehumidification operation is in progress, whether the air cleaning function is in use, whether the internal automatic cleaning function (such as the mold removal function) is in use, etc. For example, when the household appliance 10a is an electric vacuum cleaner, the state during operation is determined by whether the suction force is strong or weak, etc. For example, when the household appliance 10a is a microwave oven, the state during operation is determined by whether the oven function is in use, whether steam is in use, etc. For example, when the household appliance 10a is a television receiver, the state during operation is determined by whether the recording function is in use, whether the dual-screen display is in use, etc. For example, when the household appliance 10a is a lighting fixture, the state during operation is determined by whether the light intensity is strong, medium, or weak, etc.

[0031] Figure 2 FIG is a diagram showing a portion of the configuration of the home appliance 10a. Figure 2 As shown, sensor 102a includes multiple sensors, including a first sensor S1, a second sensor S2, and a third sensor S3. Sensors S1, S2, and S3 are located in different locations or sense different physical quantities. For example, if home appliance 10a is a refrigerator, sensor S1 detects the current flowing through the compressor. Sensor S2 detects the temperature inside the refrigerator. Sensor S3 is a camera that captures images of the refrigerator interior.

[0032] In this embodiment, the household appliance 10a includes an abnormality determination unit 103. Based on the detection results of the first sensor S1 and a pre-set first threshold value, the abnormality determination unit 103 determines an abnormality in a first component C1 (e.g., a compressor) included in the household appliance 10a. For example, if the magnitude of the current detected by the first sensor S1 is greater than the first threshold value, the abnormality determination unit 103 determines that the compressor has an abnormality. In the household appliance 10a, the detection results of the second sensor S2 and the third sensor S3 are not used to determine an abnormality related to the first component C1.

[0033] Similarly, the abnormality determination unit 103 determines an abnormality in the second component C2 (e.g., a door component) included in the home appliance 10a based on the detection results of the second sensor S2, the operating status of the home appliance 10a, and a pre-set second threshold value. In the home appliance 10a, the detection results of the first sensor S1 and the detection results of the third sensor S3 are not used to determine an abnormality related to the second component C2. On the other hand, the detection results of the third sensor S3 are not used in the home appliance 10a for abnormality determination but are used for, for example, inventory management of food in the refrigerator.

[0034] Then, return to Figure 1 Next, household appliances 10b and 10c will be described. Household appliance 10b includes a sensor 102b and a transmitter 101b. Sensors 102b are one or more sensors installed in household appliance 10b. Sensors 102b sense a number of physical quantities corresponding to the number of sensors in household appliance 10b. The types of physical quantities sensed by sensors 102b are the same as those sensed by sensor 102a, but vary depending on the type of household appliance, the model within that type, and other factors.

[0035] At predetermined intervals, the transmitter 101b transmits the physical quantity sensed by the sensor 102b, the operating state of the home appliance 10b at the time of sensing, and the identifier ID2 of the home appliance 10b to the interface device 20. The operating state of the home appliance 10b is the same as that of the home appliance 10a.

[0036] Home appliance 10c includes sensors 102c and a transmitter 101c. Sensors 102c are one or more sensors installed in home appliance 10c. Sensors 102c sense a number of physical quantities corresponding to the number of sensors in home appliance 10c. The types of physical quantities sensed by sensors 102c are the same as those sensed by sensors 102a, but vary depending on the type of home appliance 10 and the model within that type.

[0037] The transmitting unit 101c transmits the physical quantity sensed by the sensor 102b, the operating state of the home appliance 10c at the time the physical quantity is sensed, and the identifier ID3 of the home appliance 10c to the interface device 20 at every predetermined time. The operating state of the home appliance 10c is the same as that of the home appliance 10a. Figure 2 The configuration of the household appliance 10a described above is also the same for the household appliances 10b and 10c. Hereinafter, the transmitters 101a, 101b, and 101c are collectively referred to as the transmitter 101. Hereinafter, the sensors 102a, 102b, and 102c are collectively referred to as the sensor 102.

[0038] [Composition of interface device]

[0039] The interface device 20 mediates communication between the home appliance 10 and the learning device 30. The interface device 20 also mediates communication between the home appliance 10 and the fault diagnosis device 40. The interface device 20 also mediates communication between the learning device 30 and the fault diagnosis device 40. The interface device 20 includes an intermediary unit 201. For example, the interface device 20 is included in a server connected to a network.

[0040] The intermediary unit 201 receives the physical quantity sensed by the sensor 102, the operating state of the home appliance 10, and the identifier of the home appliance 10 from the home appliance 10. Specifically, the intermediary unit 201 receives the physical quantity sensed by the sensor 102a of the home appliance 10a, the operating state of the home appliance 10a, and the identifier ID1 of the home appliance 10a from the home appliance 10a. The intermediary unit 201 receives the physical quantity sensed by the sensor 102b of the home appliance 10b, the operating state of the home appliance 10b, and the identifier ID2 of the home appliance 10b from the home appliance 10b. The intermediary unit 201 receives the physical quantity sensed by the sensor 102c of the home appliance 10c, the operating state of the home appliance 10c, and the identifier ID3 of the home appliance 10c from the home appliance 10c.

[0041] The intermediary unit 201 distributes the data received from the home appliance 10 to one or both of the learning device 30 and the fault diagnosis device 40. Specifically, the intermediary unit 201 determines the destination for sending the physical quantity sensed by the sensor 102, the operating state of the home appliance 10, and the identifier of the home appliance 10 based on the identifier of the home appliance 10 received from the home appliance 10 and the identifier received from the learning device 30 along with the learned model.

[0042] Specifically, the intermediary unit 201 stores a correspondence table CT that registers the correspondence between the identifier of the home appliance 10 and the identifier of the learned model that can be used to diagnose faults in the home appliance 10. For example, the intermediary unit 201 stores the identifier received from the learning device 30 along with the learned model in the table CT. Then, if the intermediary unit 201 determines that the learned model identifier corresponding to the identifier of the home appliance 10 received from the home appliance 10 is not present in the table CT, the intermediary unit 201 transmits the physical quantity sensed by the sensor 102, the operating state of the home appliance 10, and the identifier of the home appliance 10 to the learning device 30. On the other hand, if the intermediary unit 201 determines that the learned model identifier corresponding to the identifier of the home appliance 10 received from the home appliance 10 is present in the table CT, the intermediary unit 201 transmits the physical quantity sensed by the sensor 102, the operating state of the home appliance 10, and the identifier of the home appliance 10 to the fault diagnosis device 40.

[0043] Figure 3: is a diagram showing an example of the contents of the above-mentioned table CT. In this embodiment, for the same type of household electrical appliance 10, a first learned model and a second learned model are included, wherein the first learned model is generated for each large group classified based on a first benchmark, and the second learned model is generated for each small group by subdividing the large group based on a second benchmark different from the above-mentioned first benchmark. Large groups and small groups will be described in detail later. In this embodiment, the corresponding table CT registers the correspondence between the identifier of the household electrical appliance 10, the identifier of the first learned model that can be used for fault diagnosis of the household electrical appliance 10 (the identifier of the learned model classified by large group), and the identifier of the second learned model that can be used for fault diagnosis of the household electrical appliance 10 (the identifier of the learned model classified by small group).

[0044] return Figure 1 The explanation continues. The intermediary unit 201 receives data of a certain period specified in each group of data from the fault diagnosis device 40 as teacher data, and the data of each group consists of the actual fault condition generated by the household appliance 10 and the physical quantity and operation status at each moment associated with the fault condition recorded by the accumulation unit 403 of the fault diagnosis device 40 described later. Then, the intermediary unit 201 sends the received teacher data to the learning device 30. In addition, the content of the actual fault condition is determined based on the abnormal value of the physical quantity sensed by the sensor 102 (for example, the abnormality determined by the abnormality determination unit 103), the content of the fault condition for which service support is actually performed (repair content), etc. In other words, in this embodiment, the following teacher data is used: the detection results of one or more sensors provided in the household appliance 10 and the operation status of the household appliance 10 are set as input information, and the content of the fault condition corresponding to the repair actually performed, such as component exchange, is set as correct answer data of the output information.

[0045] Due to the presence of the interface device 20, each home appliance 10 does not need to communicate separately with the learning device 30 and the fault diagnosis device 40. In other words, each home appliance 10 only needs to communicate with the interface device 20 at all times without changing the setting of the communication destination.

[0046] [Composition of the learning device]

[0047] The learning device 30 generates a learned model such as a neural network according to the type and model of the home appliance 10 that is determined to be capable of using the same learned model. Figure 1 As shown, the learning device 30 includes a receiving unit 301, an accumulating unit 302, and a learning unit 303. The accumulating unit 302 is an example of a "second information accumulating unit."

[0048] The receiving unit 301 receives the physical quantity sensed by the sensor 102a, the operating state of the household appliance 10a, and the identifier ID1 of the household appliance 10a from the interface device 20. Furthermore, the receiving unit 301 receives the physical quantity sensed by the sensor 102b, the operating state of the household appliance 10b, and the identifier ID2 of the household appliance 10b from the interface device 20. Furthermore, the receiving unit 301 receives the physical quantity sensed by the sensor 102c, the operating state of the household appliance 10c, and the identifier ID3 of the household appliance 10c from the interface device 20.

[0049] The accumulation unit 302 records the physical quantity sensed by sensor 102a and the operating state of household appliance 10a, as received by the receiving unit 301, in association with the learning model determined based on the identifier ID1 of household appliance 10a. Furthermore, the accumulation unit 302 records the physical quantity sensed by sensor 102b and the operating state of household appliance 10b, as received by the receiving unit 301, in association with the learning model determined based on the identifier ID2 of household appliance 10b. Furthermore, the accumulation unit 302 records the physical quantity sensed by sensor 102c and the operating state of household appliance 10c, as received by the receiving unit 301, in association with the learning model determined based on the identifier ID3 of household appliance 10c. Furthermore, if a malfunction occurs in household appliance 10, the accumulation unit 302 records the details of the malfunction in the household appliance 10 in association with the physical quantity and operating state accumulated before the malfunction occurred. As described above, the details of the fault condition of the home appliance 10 are determined based not only on abnormal values ​​of physical quantities sensed by the sensor 102 (e.g., abnormalities determined by the abnormality determination unit 103), but also on the details of the fault condition (repair details) for which actual service support, such as repairs, has been performed. For example, if a fault condition is detected in a first component C1 of the home appliance 10 during actual service support, the fault condition associated with the first component C1 is registered as the fault condition details. Furthermore, if a fault condition is detected in a second component C2 of the home appliance 10 during actual service support, the fault condition associated with the second component C2 is registered as the fault condition details.

[0050] Figure 4 FIG. 1 is a diagram showing an example of the first data table TBL1 of the first embodiment. Figure 4 As shown, the accumulation unit 302 records the physical quantity (for example, Figure 2 The physical quantities sensed by the first sensor S1, the second sensor S2, ...) and the action state. Figure 4The first data table TBL1 shown is an example of data recorded by the accumulation unit 302 when the learning model determined based on the identifier ID1 of the home appliance 10a, the learning model determined based on the identifier ID2 of the home appliance 10b, and the learning model determined based on the identifier ID3 of the home appliance 10c are the same. That is, Figure 4 The first data table TBL1 shown is an example of data recorded by the accumulation unit 302 when household appliances 10a, 10b, and 10c can predict the occurrence of a problem using the same learning model. The situation in which household appliances 10a, 10b, and 10c can predict the occurrence of a problem using the same learning model refers to, for example, a situation in which household appliances 10a, 10b, and 10c are of the same type and model. Furthermore, the situation in which household appliances 10a, 10b, and 10c can predict the occurrence of a problem using the same learning model refers to a situation in which, for example, household appliances 10a, 10b, and 10c are of the same type but different models, but have, for example, the same main component configuration, and the difference in model does not substantially affect the prediction of the occurrence of a problem.

[0051] The learning unit 303 uses the teacher data of the correct answer data with physical quantities and action states as input information and the content of the bad condition as output information to determine the coefficients in the learning model (such as the weight coefficients between nodes) according to the learning model, thereby generating a completed learning model.

[0052] For example, the learning model is a neural network learning model consisting of an input layer, an intermediate layer, and an output layer. For the home appliance 10, it is assumed that Figure 4 The first data table TBL1 shown is recorded in the accumulation unit 302. That is, it is assumed that household appliances 10a, 10b, and 10c can predict the occurrence of a fault condition using the same learning model #1. Furthermore, it is assumed that household appliance 10a experiences fault condition 1 at time t100 and undergoes processing to eliminate fault condition 1. The physical quantities and operating states from time t1 to t100 are associated with fault condition 1. Furthermore, it is assumed that household appliance 10a experiences fault condition 2 at time t200 and undergoes processing to eliminate fault condition 2. The physical quantities and operating states from time t150 to t200 are associated with fault condition 2. Furthermore, it is assumed that household appliance 10b experiences fault condition 2 at time t140 and undergoes processing to eliminate fault condition 2. The physical quantities and operating states from time t1 to t140 are associated with fault condition 2.

[0053] In this case, fault condition 1 and the physical quantities and operating states at each time from t1 to t100 are set as the first set of data. Furthermore, fault condition 2 and the physical quantities and operating states at each time from t150 to t100 are set as the second set of data. Furthermore, fault condition 2 and the physical quantities and operating states at each time from t1 to t140 are set as the third set of data. Furthermore, in addition to the three sets of data described above, fault conditions occurring at other times in home appliances 10a and 10b, and fault conditions occurring in home appliances 10c other than home appliances 10a and 10b whose occurrence can be predicted using learning #1, are also set as separate sets of data, similar to those for home appliances 10a and 10b. Next, the learning unit 303 sets the data for a predetermined period of time within each set of data consisting of the fault condition and the physical quantities and operating states at each time associated with the fault condition as the teacher data. The predetermined period is a predetermined period (e.g., three weeks) from the time when the malfunction actually occurred in the home appliance 10, looking back to the time point after the predetermined period (e.g., one week). In this embodiment, the predetermined period is 24 days, from 30 days to 7 days past the time when each malfunction occurred. The learning unit 303 inputs the malfunction, the physical quantities during the predetermined period, and the operating state in each data set as training data into the learning model, and determines the coefficients in the learning model. Thus, a learned model is generated.

[0054] Specifically, assuming Figure 4 The times t1 to t200 in the first data table TBL1 shown are in one-day increments, representing days 1 to 200. In this case, the learning unit 303 sets the training data for fault condition 1 and the physical quantities and operating states at each time point t70 to t93 in the first data set. Furthermore, the learning unit 303 sets the training data for fault condition 2 and the physical quantities and operating states at each time point t170 to t193 in the second data set. Furthermore, the learning unit 303 sets the training data for fault condition 2 and the physical quantities and operating states at each time point t110 to t133 in the third data set. Similarly, for fault conditions occurring at other times in household appliances 10a and 10b, and for fault conditions occurring in household appliances 10c whose occurrence can be predicted using learning model #1, the learning unit 303 sets the training data for each fault condition and the physical quantities and operating states for 24 days, from 30 days to 7 days past the time of each fault condition's occurrence.

[0055] Next, the learning unit 303 divides the teacher data into training data and verification data. The learning unit 303 inputs the training data into the learning model. The learning unit 303 adjusts the coefficients of the learning model so that the output of the learning model is consistent with the actual bad condition corresponding to the input training data. In this way, the learning unit 303 determines the coefficients in the learning model. Next, the learning unit 303 inputs the verification data into the learning model whose coefficients have been determined. The learning unit 303 determines whether the output of the learning model is the content of the actual bad condition that has been associated with the input verification data. When the learning unit 303 determines that the output of the learning model is consistent with the content of the actual bad condition for the verification data, the learning model with the coefficient is set as the learned model. In addition, when the learning unit 303 determines that the output of the learning model is different from the content of the actual bad condition for the verification data, the teacher data is newly prepared and divided into training data and verification data, and the above-mentioned process is repeated until the learned model can be obtained. The learning unit 303 generates the learned model in this way.

[0056] The learning unit 303 also determines coefficients for each learning model other than learning model #1 using the same method as for learning model #1. Each time the learning unit 303 determines coefficients, it transmits the learned model with the determined coefficients and the identifier of the home appliance 10 associated with the learned model to the interface device 20.

[0057] If the output of the learned model generated in this manner indicates the occurrence of a fault condition, the time at which the fault condition occurred is the time that has advanced from the present time by the time difference between the present time and the time closest to the present time within the predetermined period. For example, if the predetermined period is 24 days from the past 30 days to the past 7 days, and the output of the learned model indicates the occurrence of a fault condition, the time at which the fault condition occurred is 7 days from the present time, which is the time difference between the present time and the past 7 days.

[0058] Therefore, it is necessary to set the time closest to the current time within the specified period as the time at which the undesirable condition can be addressed before it occurs. However, the closer the time is to the current time, the closer it is to the time when the undesirable condition will occur. Therefore, it is believed that as the time closest to the current time within the specified period approaches the current time, the characteristics of the physical quantity within the specified period become more apparent. As a result, it is believed that setting the time closest to the current time within the specified period closer to the current time allows for more accurate judgment of the occurrence of an undesirable condition using the learned model. Therefore, the time closest to the current time within the specified period is preferably a time at which the undesirable condition can be addressed before it occurs, and preferably a time as close to the current time as possible.

[0059] In one embodiment, the learned models for the same type of household appliance include a first learned model M1 and a second learned model M2. The first learned model M1 is generated for each large group classified based on a first criterion (for each large category), and the second learned model M2 is generated for each sub-group (for each small category) that is subdivided into the large group based on a second criterion different from the first criterion. The second criterion is based on the model of the household appliance 10. On the other hand, the first criterion is based on the basic form and components of the household appliance 10. For example, if the household appliance 10 is a refrigerator, the first criterion may include whether the cooler (evaporator) is of one or two types, the layout of the storage compartments (such as the placement of the vegetable compartment and the cooling compartment), or whether the installed compressors are the same. For example, if the household appliance 10 is a washing machine, the first criterion may include whether it is a drum-type washing machine or a vertical-type washing machine, and whether it has a drying function. For example, if the household appliance 10 is an air conditioner, the first criterion may include whether it has an automatic internal cleaning function (such as a mold removal function).

[0060] Figure 5 This figure schematically illustrates a first learned model M1 and a second learned model M2. Input information for these learned models M1 and M2 includes, for example, the most recent detection value of the first sensor S1, the detection value from one hour ago, the detection value from two hours ago, ..., the most recent detection value of the second sensor S2, the detection value from one hour ago, the detection value from two hours ago, ..., the most recent detection value of the third sensor S3, the detection value from one hour ago, the detection value from two hours ago, and the operating status of the household appliance 10 at the time these detection values ​​were detected. On the other hand, output information from learned models M1 and M2 includes the probability of fault condition A (e.g., a fault condition related to the first component C1) occurring after the specified period (e.g., one week later), the probability of fault condition B (e.g., a fault condition related to the second component C2) occurring after the specified period, the probability of fault condition C occurring after the specified period, and so on.

[0061] [Composition of the fault diagnosis device]

[0062] The fault diagnosis device 40 predicts the occurrence of a fault condition of each household appliance 10. Figure 1 As shown, the fault diagnosis device 40 includes a receiving unit 401, a diagnosis unit 402, an accumulation unit 403, and an update unit 404. The accumulation unit 403 is an example of a "first information accumulation unit". Figure 1 The “notification unit 405” in FIG. 4 is described in the second embodiment.

[0063] The receiving unit 401 receives data transmitted from the interface device 20. If the data received by the receiving unit 401 from the interface device 20 is a learned model, the diagnostic unit 402 uses the learned model as a learning model for predicting the occurrence of a fault condition. Furthermore, if the data received by the receiving unit 401 from the interface device 20 is a physical quantity sensed by the sensor 102 and the operating state of the household appliance 10, the diagnostic unit 402 inputs the physical quantity and operating state into the learned model corresponding to the identifier. Based on the output of the learned model, the diagnostic unit 402 predicts the occurrence of a fault condition in the household appliance 10 corresponding to the identifier. The predicted occurrence of a fault condition includes the time when the fault condition will occur and the nature of the fault condition that will occur.

[0064] Figure 6 FIG. 1 is a diagram showing an example of the second data table TBL2 of the first embodiment. Figure 6 As shown, when the receiving unit 401 receives the physical quantity sensed by the sensor 102, the operating state of the household appliance 10, and the identifier of the household appliance 10 from the interface device 20, the accumulation unit 403 establishes an association with the learned model corresponding to the identifier of the household appliance 10 and records the physical quantity sensed by the sensor 102 and the operating state of the household appliance 10 received by the receiving unit 401, as well as the time of occurrence and content of the adverse condition predicted by the diagnosis unit 402 as a second data table TBL2.

[0065] If a fault condition occurs a predetermined number of times during fault diagnosis using the same learned model, where the diagnosis unit 402 fails to predict the occurrence of the fault condition, or where the diagnosis unit 402 predicts the occurrence of the fault condition but fails to eliminate the fault condition, the update unit 404 transmits data for a predetermined period of time from each set of data consisting of the fault condition and the physical quantities and operating states associated with the fault condition at each time point recorded by the accumulation unit 403 as teacher data to the interface device 20. The predetermined number of times refers to the number of times that a sufficient amount of teacher data has been used as training data and verification data to improve the learning model for predicting the occurrence of the fault condition. That is, when a sufficient amount of training data and verification data has been accumulated, the update unit 404 transmits data for a predetermined period of time as teacher data to the interface device 20.

[0066] [Processing for predicting the occurrence of malfunctions in home appliances]

[0067] Next, the process (fault diagnosis) performed by the home appliance system 1 to predict the occurrence of a malfunction in the home appliance 10 will be described. In this embodiment, the learned model is learned using information accumulated over a certain period (e.g., three weeks) as described above. This period is the period from the time the malfunction actually occurred in the home appliance 10 to a predetermined period (e.g., one week) in the past. The diagnostic unit 402 then performs fault diagnosis based on the information accumulated over the most recent certain period (e.g., the last three weeks).

[0068] Figure 7 This is a diagram showing a process flow for predicting the occurrence of a malfunction by the home appliance system 1 according to the first embodiment. Figure 7 The first processing flow of the first embodiment shown in FIG.

[0069] In addition, the intermediary unit 201 determines that the stored identifiers do not contain an identifier corresponding to the identifier of the home appliance 10 received from the home appliance 10. In addition, the intermediary unit 201 sends the physical quantity sensed by the sensor 102, the operating state of the home appliance 10, and the identifier of the home appliance 10 to the learning device 30. In addition, the learning unit 303 has generated a learned model through the above-mentioned processing. In addition, the learning unit 303 sends the generated learned model together with the identifier of the corresponding home appliance 10 to the interface device 20. In addition, the intermediary unit 201 stores the learning model and the identifier received from the learning device 30 together. In addition, the intermediary unit 201 sends the received learning model and the identifier to the fault diagnosis device 40. In addition, the receiving unit 401 receives the learning model and the identifier from the interface device 20 together. In addition, the diagnosis unit 402 already has the learned model.

[0070] In each home appliance 10, the sensor 102 senses a physical quantity (step S1). The transmitter 101 transmits the physical quantity sensed by the sensor 102 of the home appliance 10, the operating state of the home appliance 10, and the identifier of the home appliance 10 to the interface device 20 at predetermined intervals (step S2).

[0071] The intermediary unit 201 receives the physical quantity sensed by the sensor 102, the operating state of the home appliance 10, and the identifier of the home appliance 10 from the home appliance 10. The intermediary unit 201 determines whether there is an identifier among its stored identifiers that is identical to the identifier of the home appliance 10 received from the home appliance 10. If the intermediary unit 201 determines that there is no identifier among its stored identifiers that is identical to the identifier of the home appliance 10 received from the home appliance 10, the intermediary unit 201 transmits the physical quantity sensed by the sensor 102, the operating state of the home appliance 10, and the identifier of the home appliance 10 to the learning device 30. Furthermore, if the intermediary unit 201 determines that there is an identifier among its stored identifiers that is identical to the identifier of the home appliance 10 received from the home appliance 10, the intermediary unit 201 transmits the physical quantity sensed by the sensor 102, the operating state of the home appliance 10, and the identifier of the home appliance 10 to the fault diagnosis device 40.

[0072] In the example shown here, the diagnostic unit 402 already has a learned model. Therefore, the intermediary unit 201 determines that one of the stored identifiers matches the identifier of the home appliance 10 received from the home appliance 10 (step S3). The intermediary unit 201 then transmits the physical quantity sensed by the sensor 102, the operating status of the home appliance 10, and the identifier of the home appliance 10 to the fault diagnosis device 40 (step S4).

[0073] The receiving unit 401 receives the physical quantity sensed by the sensor 102, the operating state of the household appliance 10, and the identifier of the household appliance 10 from the interface device 20. The accumulating unit 403 records the physical quantity sensed by the sensor 102, the operating state of the household appliance 10, and the identifier of the household appliance 10 received by the receiving unit 401, along with the learned model corresponding to the identifier and the time, in the second data table TBL2 (step S5) before recording data for a certain period (e.g., the most recent three weeks). Furthermore, the accumulating unit 403 does not have enough data input into the learned model before recording data for the certain period. Therefore, the physical quantity sensed by the sensor 102 and the operating state of the household appliance 10 are not input to the diagnosing unit 402. Consequently, the diagnosing unit 402 does not predict the occurrence time and content of the malfunction. Furthermore, the fact that the diagnosing unit 402 does not predict the occurrence time and content of the malfunction is indicated by a "-" in the second data table TBL2.

[0074] When the accumulation unit 403 records the data for the predetermined period, the diagnosis unit 402 inputs the data for the predetermined period into the learned model corresponding to the identifier of the home appliance 10. Furthermore, the diagnosis unit 402 predicts the occurrence time of a malfunction and the content of the malfunction based on the output of the learned model (step S6).

[0075] For example, it is assumed that the above-mentioned fixed period is 24 days from the past 30 days to the past 7 days relative to the time when each failure occurred. Figure 3 The times t1001 to t1200 in the second data table TBL2 shown are in one-day increments, representing the period from 207 days ago to 7 days ago based on the present time. Furthermore, the diagnostic unit 402 inputs the physical quantities and operating states at each of the times t1177 to t1200 into the learned model. Furthermore, the output of the learned model indicates the occurrence of Fault 2. In this case, the diagnostic unit 402 predicts the occurrence of Fault 2 7 days from now. The diagnostic unit 402 repeats the fault diagnosis based on the above process at a predetermined period (e.g., every day).

[0076] In this embodiment, the abnormality determination unit 103 of the household appliance 10 uses the detection results of the first sensor S and the first threshold value to detect an abnormality related to the first component C1. In other words, the abnormality determination unit 103 of the household appliance 10 does not use the detection results of the second sensor S2 and the third sensor S3, etc., when detecting an abnormality related to the first component C1. Meanwhile, the diagnosis unit 402 uses the information based on the detection results of the first sensors S1, S2, S3, etc., accumulated over the predetermined period, to diagnose a fault in the first component C1. Furthermore, the diagnosis unit 402 uses the information based on the detection results of the first sensors S1, S2, S3, etc., accumulated over the predetermined period, to diagnose a fault in the second component C2.

[0077] In the present embodiment, the diagnosis unit 402 performs a first fault diagnosis using the above-mentioned first learned model M1 (a learned model generated for each large group) and a second fault diagnosis using the second learned model M2 (a learned model generated for each small group). Here, it is originally preferred to use the learned model by obtaining teacher data for each model. However, in reality, it is expected that there will be a situation where the collection of fault condition information for a specific model has not made progress. Therefore, in the present embodiment, in addition to the above-mentioned second learned model M2, the first learned model M1 is also used for fault diagnosis. Thus, even if sufficient teacher data is not available for each model and the prediction accuracy of the learned model M2 is not high, high-precision fault diagnosis can be performed.

[0078] [Processing of updating the learned model]

[0079] Next, a description will be given of a process of updating the learned model performed by the home appliance system 1 . Figure 8 This is a diagram showing a process flow of updating a learned model performed by the home appliance system 1 according to the first embodiment. Figure 8The second processing flow of the first embodiment shown will be described.

[0080] As mentioned above, the moment closest to the present moment in the above-mentioned certain period is preferably a moment at which the adverse condition can be dealt with before the adverse condition occurs, and is preferably a moment as close to the present moment as possible. However, even in the case where the diagnosis unit 402 predicts the occurrence of an adverse condition, it is not necessarily actually processed to avoid the adverse condition. In addition, the prediction of the occurrence of an adverse condition by the diagnosis unit 402 is not always correct. Therefore, even in a period in which the diagnosis unit 402 determines that no adverse condition has occurred, there is a possibility that an adverse condition has actually occurred. In this case, in order to know the content of the adverse condition that actually occurred, the physical quantity and action state of the specified period recorded by the accumulation unit 403 in this case can be set as teacher data for improving the learning completion model.

[0081] The process of updating the learned model performed by the home appliance system 1 described here is a process of improving the learned model using the content of the actual failure condition and the physical quantities and operating states recorded by the accumulation unit 403 during the predetermined period as training data. Furthermore, this process is performed when, even if the processes of steps S1 to S6 above are performed and a failure condition is predicted, no preventive measures are taken and the failure condition actually occurs.

[0082] After performing steps S1 through S6, until a fault condition actually occurs, the accumulation unit 403 records the physical quantity sensed by the sensor 102, the operating state of the home appliance 10, the identifier of the home appliance 10, the learned model corresponding to the identifier, and the time in the second data table TBL2 for each home appliance 10 (step S11). Furthermore, when a fault condition occurs in the home appliance 10, the accumulation unit 403 records the physical quantity sensed by the sensor 102, the operating state of the home appliance 10, the identifier of the home appliance 10, the learned model corresponding to the identifier, and the time, along with the details of the fault condition, in the second data table TBL2 (step S12).

[0083] The updating unit 404 determines whether or not the failure has occurred for each learned model in the second data table TBL2 more than a predetermined number of times (step S13 ).

[0084] When the updating unit 404 determines that the number of failures that have occurred is less than the predetermined number for each learned model in the second data table TBL2 (NO in step S13 ), the process returns to step S11 .

[0085] In addition, when the update unit 404 determines that the adverse condition has occurred more than the specified number of times for each completed learning model in the second data table TBL2 (yes in step S13), the data of the specified period in the data consisting of the content of the adverse condition, the physical quantities at each moment associated with the content of the adverse condition, and the action state according to the completed learning model determined in this way is sent as teacher data to the interface device 20 (step S14).

[0086] The intermediary unit 201 receives the teacher data from the fault diagnosis device 40. The intermediary unit 201 transmits the received teacher data to the learning device 30 (step S15).

[0087] The receiving unit 301 receives the teacher data (ie, data of a predetermined period of time for each learned model) from the interface device 20. The accumulating unit 302 records the teacher data (step S16).

[0088] The learning unit 303 improves the learned model based on the teacher data recorded by the accumulation unit 302 in the process of step S16 (step S17). Specifically, the learning unit 303 divides the teacher data into training data and verification data. The learning unit 303 inputs the training data into the learning model. The learning unit 303 adjusts the coefficients of the learning model so that the actual defect corresponding to the input training data is consistent with the output of the learning model. In this way, the learning unit 303 determines the coefficients in the learning model. Next, the learning unit 303 inputs the learning model with the determined coefficients into the verification data. The learning unit 303 determines whether the output of the learning model is the content of the actual defect associated with the input verification data. If the learning unit 303 determines that the output of the learning model is consistent with the content of the actual defect, it sets the learning model with the coefficients as the learned model. If the learning unit 303 determines that the output of the learning model is different from the content of the actual defect, it repeats the processes of steps S11 to S15 until new teacher data is obtained. In this way, the learned model is improved.

[0089] The learning unit 303 sends the improved learned model to the interface device 20 (step S18). The intermediary unit 201 receives the improved learned model from the learning device 30. The intermediary unit 201 sends the received learned model to the fault diagnosis device 40 (step S19).

[0090] The receiving unit 401 receives the improved learned model from the interface device 20. The diagnosing unit 402 stores the learned model received by the receiving unit 401 as a new learned model (step S20).

[0091] The learned model provided in the diagnosis unit 402 is improved in this manner. The home appliance system 1 uses the learned model improved in this manner to refer to the Figure 4 The described processing can more accurately predict the occurrence of a malfunction in the household appliance 10 .

[0092] In the first embodiment, the transmitting units 101 a , 101 b , and 101 c , the intermediary unit 201 , the receiving units 301 and 401 , the accumulating units 302 and 403 , the learning unit 303 , the diagnosing unit 402 , and the updating unit 404 are each software functional units.

[0093] The above describes the home appliance system 1 according to the first embodiment. The home appliance system 1 includes a diagnostic unit 402. The diagnostic unit 402 predicts the occurrence of a failure based on a learned model. This learned model uses the physical quantities and operating states of the home appliance 10 and the failure conditions over a predetermined period of time in the past, based on the occurrence of the failure in the home appliance 10, as training data to determine coefficients.

[0094] This household appliance system 1 can predict the occurrence time and content of a fault condition whose causal relationship with a physical quantity sensed by a sensor is unclear. Furthermore, by adjusting the time closest to the current time within a predetermined period, both the accuracy of the fault condition determination performed by the learned model and the prediction time can be appropriately set.

[0095] The home appliance system 1 includes a learning unit 303. The learning unit 303 determines the coefficient using the physical quantity and the failure status of the home appliance 10 during a predetermined period in the past based on the occurrence of the failure as the training data.

[0096] According to the household appliance system 1 , a learned model for accurately predicting a malfunction can be generated.

[0097] The home appliance system 1 includes an updater 404. When the number of failures in the learned model exceeds a predetermined number, the updater 404 sends teacher data to the learning unit 303. The learning unit 303 then improves the learned model based on the teacher data sent by the updater 404.

[0098] This home appliance system 1 can generate a learned model that more accurately predicts adverse conditions, i.e., it can improve the learned model. Furthermore, by improving the learned model with teacher data containing new adverse condition information, the home appliance system 1 can predict adverse conditions that were not predicted by the learned model.

[0099] In the home appliance system 1 , the details of the failure are determined based on the details of repairs actually performed on the home appliance.

[0100] According to the home appliance system 1, not only the failure conditions generally determined by abnormal values ​​of physical quantities are included, but also failure conditions corresponding to requests from users of the home appliance 10 can be included. As a result, the home appliance system 1 can also predict the occurrence of failure conditions other than those determined by abnormal values ​​of physical quantities.

[0101] (Second embodiment)

[0102] [Composition of home appliance systems]

[0103] The configuration of the household appliance system 1 according to the second embodiment is the same as that of the household appliance system 1 according to the first embodiment. However, the second embodiment differs in that the fault diagnosis device 40 includes a notification unit 405 .

[0104] The notification unit 405 notifies the user of the home appliance 10 of the predicted result of the occurrence of a fault condition. For example, the notification unit 405 sends a predetermined control command to the home appliance 10 or the terminal device UT of the user of the home appliance 10, thereby notifying the user of the predicted result of the occurrence of the fault condition via the notification device (display device, etc.) of the home appliance 10 or the terminal device UT. The terminal device UT can be a portable terminal device such as a smartphone or tablet terminal, a personal computer, or a smart speaker.

[0105] In this embodiment, notification unit 405 outputs a notification via home appliance 10 or terminal device UT, prompting the user to confirm whether the symptom corresponding to the fault diagnosis result of diagnosis unit 402 has actually occurred in home appliance 10. For example, if the home appliance 10 experiencing a malfunction is a refrigerator, notification unit 405 may notify the user based on the predicted malfunction, such as "Can you hear unusual noises from the refrigerator?" or "Has cooling performance deteriorated?" If the user responds in the affirmative, diagnosis unit 402 may request repair of home appliance 10 from a service center or provide the user with contact information for requesting repair from a service center.

[0106] The home appliance system 1 according to the second embodiment has been described above. The home appliance system 1 includes the diagnosis unit 402. The diagnosis unit 402 notifies the user of the home appliance 10 of the predicted result of the occurrence of a malfunction.

[0107] According to the home appliance system 1 , it is possible to quickly respond to repairs of the home appliance 10 in which a malfunction has occurred.

[0108] Furthermore, the processing order of the processes in the embodiments may be changed within the scope of performing appropriate processing.

[0109] In addition, in the embodiment, the learning device 30 is described as generating a learning model such as a neural network according to the type and model of the household appliance determined to be capable of using the same learning model. The determination of whether the same learning model can be used can also be performed in the following manner. For example, a learning model is generated using a cluster analysis algorithm. The learning unit 303 inputs an unlabeled data set into the learning model to generate a learned model. Furthermore, the learning unit 303 can input data of the household appliance 10 that is determined to be an object capable of using the same learning model into the learned model to determine a learning model that predicts the occurrence of an adverse condition represented by the output result of the learned model.

[0110] In the above embodiment, the household appliances 10a, 10b, and 10c, the interface device 20, the learning device 30, and the fault diagnosis device 40 are described as separate devices. However, in other embodiments, the components of the learning device 30 and the fault diagnosis device 40 may be implemented in a single device (e.g., a cloud server). Furthermore, the household appliance 10 may include part or all of the fault diagnosis device 40.

[0111] Each of the accumulators 302, 403 and other storage devices (including registers and latches) in the embodiments can be provided at any location within the range of appropriately transmitting and receiving information. Furthermore, each of the accumulators 302, 403 and other storage devices in the embodiments can be provided in multiple locations within the range of appropriately transmitting and receiving information, thereby distributing and storing data.

[0112] While the embodiments have been described above, the household appliances 10a, 10b, and 10c, the interface device 20, the learning device 30, the fault diagnosis device 40, and other control devices may each include a computer system. Furthermore, the aforementioned processing procedures may be stored in the form of a program on a computer-readable recording medium, and the program may be read and executed by a computer to perform the aforementioned processing. A specific example of a computer system is shown below.

[0113] Next, a computer configuration according to at least one embodiment will be described.

[0114] The computer 5 includes a CPU, a main memory, a storage, and an interface.

[0115] For example, the aforementioned home appliances 10a, 10b, and 10c, interface device 20, learning device 30, fault diagnosis device 40, and other control devices are each installed in a computer. Furthermore, the operations of each of the aforementioned processing units are stored in the form of a program in a memory. The CPU reads the program from the memory, expands it in the main memory, and executes the aforementioned processing according to the program. Furthermore, the CPU allocates storage areas corresponding to each of the aforementioned storage units in the main memory according to the program.

[0116] Examples of memory include HDD (Hard Disk Drive), SSD (Solid State Drive), magnetic disk, optical disk, CD-ROM (Compact Disc Read Only Memory), DVD-ROM (Digital Versatile Disc Read Only Memory), and semiconductor memory. The memory can be an internal medium directly connected to the bus of the computer, or an external medium connected to the computer via an interface or communication line provided by the computer. In addition, when the program is delivered to the computer via a communication line, the computer receiving the delivery can also expand the program in the main memory and execute the above-mentioned processing. In at least one embodiment, the memory is a non-temporary tangible storage medium.

[0117] Furthermore, the program may partially realize the aforementioned functions. Furthermore, the program may be a file that can realize the aforementioned functions by combining it with a program that has already been recorded in the computer system, that is, a so-called differential file (differential program).

[0118] In the above embodiments, the transmitting units 101a, 101b, and 101c, the intermediary unit 201, the receiving units 301 and 401, the accumulating units 302 and 403, the learning unit 303, the diagnostic unit 402, and the updating unit 404 are software functional units. However, these units may be hardware functional units such as LSIs. Furthermore, some of the transmitting units 101a, 101b, and 101c, the intermediary unit 201, the receiving units 301 and 401, the accumulating units 302 and 403, the learning unit 303, the diagnostic unit 402, and the updating unit 404 may be software functional units, while the remainder may be hardware functional units such as LSIs.

[0119] According to at least one of the above-described embodiments, the presence of a diagnosis unit makes it possible to predict the occurrence of a malfunction in a household electrical appliance.

[0120] While several embodiments of the present invention have been described, these embodiments are provided as examples and are not intended to limit the scope of the invention. These embodiments may be implemented in various other forms, and various omissions, substitutions, and modifications may be made without departing from the spirit of the invention. These embodiments and their variations are intended to be encompassed by the invention set forth in the claims and their equivalents, as long as they fall within the scope and spirit of the invention.

[0121] Description of Reference Numerals

[0122] 1…home appliance system, 10a, 10b, 10c…home appliance equipment, 20…interface device, 30…learning device, 40…fault diagnosis device, 101a, 101b, 101c…transmitting unit, 102a, 102b, 102c…sensor, 103…abnormality determination unit, 201…intermediary unit, 301, 401…receiving unit, 302, 403…accumulation unit, 303…learning unit, 402…diagnosis unit, 404…updating unit, 405…notification unit.

Claims

1. A household appliance system, wherein: have: a diagnostic unit that performs fault diagnosis of a home appliance based on a learned model, wherein the learned model is learned to output a fault diagnosis result related to the home appliance when information based on detection results of one or more sensors of the home appliance is input; The learned model includes a first learned model and a second learned model, wherein the first learned model is generated for each large group of household appliances of the same type classified based on a first criterion, and the second learned model is generated for each small group of household appliances of the same type classified based on a second criterion different from the first criterion. The diagnosis unit performs a first fault diagnosis using the first learned model and a second fault diagnosis using the second learned model.

2. The home appliance system according to claim 1, wherein: Also features: an information accumulation unit for accumulating the information, The learned model is learned to output the result of the fault diagnosis when the information accumulated for a certain period of time is input. The diagnosis unit performs the fault diagnosis based on the information accumulated by the information accumulation unit for the certain period.

3. The home appliance system according to claim 2, wherein: The learned model is learned using the information accumulated over the certain period, wherein the certain period is a period from the time when the malfunction of the home appliance actually occurred to the time when the malfunction was traced back to the time when the malfunction was a predetermined period. The diagnosis unit performs the fault diagnosis based on the information accumulated over the most recent fixed period.

4. The household appliance system according to claim 2, wherein: The one or more sensors include a first sensor and a second sensor, The home appliance includes an abnormality determination unit configured to determine an abnormality of a first component included in the home appliance based on a detection result of the first sensor and a threshold value. The diagnosis unit performs a fault diagnosis of the first component using the information based on the detection results of the first sensor and the second sensor accumulated over the certain period.

5. The home appliance system according to claim 2, wherein: The learned model is learned using, as teacher data, a combination of the information accumulated for each of the plurality of home appliances over the predetermined period and the details of the failures actually occurring in each of the plurality of home appliances.

6. The household appliance system according to claim 5, wherein: The details of the failure include details determined based on the details of repairs actually performed.

7. The household appliance system according to claim 1, wherein: Also features: An updating unit updates the learned model.

8. The household appliance system according to claim 1, wherein: The learned models are generated for each group of home appliances of the same type classified based on the model, basic form, or component parts.

9. The household appliance system according to claim 1, wherein: Also features: The notification unit outputs a notification through the home appliance or a terminal device, the notification prompting a user to confirm whether a symptom corresponding to the result of the fault diagnosis by the diagnosis unit has actually occurred in the home appliance.

Images