Device abnormality diagnosis system, device abnormality diagnosis method, and device abnormality diagnosis program

Abstract

To diagnose abnormality of a device by also using a program other than a learning model.SOLUTION: A device abnormality diagnosis system includes: an input unit to which driving information of a home electric appliance 11; a factor output unit which determines an abnormality factor of the home electric appliance 11 on the basis of a learning model which outputs the abnormality factor of the home electric appliance 11 according to input of the driving information of the home electric appliance 11, and outputs the abnormality factor; and a countermeasure output unit which, in response to having received input of the abnormality factor, outputs component information for dealing with the abnormality factor or operation information for evading the abnormality factor.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to a device abnormality diagnosis system, a device abnormality diagnosis method, and a device abnormality diagnosis program. [Background technology]

[0002] In devices such as home appliances, various malfunctions can occur due to a variety of causes. In some cases, the causal relationship between the malfunction and the detection value of the device's sensor is unclear. For this reason, there have been cases where it has been difficult to perform appropriate fault diagnosis using diagnosis based on calculations of sensor information.

[0003] For this reason, there is a movement to improve diagnostic accuracy by utilizing trained models (learning models) that have been trained using training data.

[0004] The invention of Patent Document 1 includes a diagnostic unit that performs fault diagnosis of a home appliance based on a trained model that has been trained to output a fault diagnosis result for the home appliance when information based on the detection results of one or more sensors that the home appliance has is input. In Patent Document 1, the trained model is configured to output, for example, that a malfunction has occurred in a component (paragraphs 0042 and 0056 of Patent Document 1). It is not necessarily clear what specific situation is output as a malfunction. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Publication No. 2021-177319 Summary of the Invention [Problem to be solved by the invention]

[0006] The invention described in Patent Document 1 uses the detection results of various sensors as input and outputs whether or not a malfunction has occurred in various parts. However, the estimation results (output) of the learning model are not always correct, and the service agent performing the repair is expected to proceed with the process, including checking the actual home appliance being diagnosed and examining whether the estimation of the learning model is correct. In this case, it is desirable for the output of the diagnosis unit to not only output the defective part (e.g., a valve is faulty), but also include the abnormality factors (e.g., gas leak) that explain why the part malfunctioned, from the perspective of improving the efficiency and accuracy of the service agent's inspection work.

[0007] If all such outputs are to be output by the learning model, there is a risk that the required amount of learning and the accuracy of the learning data will become too high. It is therefore necessary to provide an efficient diagnostic configuration that allows a diagnostic unit including a learning model to output both the parts that need to be replaced or repaired and the causes of the anomaly. [Means for solving the problem]

[0008] In order to solve the above-mentioned problems, the equipment abnormality diagnosis system of the present invention is characterized by comprising an input unit to which operation information of the equipment is input, a factor output unit that executes processing based on a learning model to output abnormality factors of the equipment in response to the input operation information of the equipment, and a countermeasure output unit that receives the input of the abnormality factor and outputs part information corresponding to the abnormality factor or operation information for avoiding the abnormality factor.

[0009] The equipment abnormality diagnosis method of the present invention is characterized by comprising the steps of: an input unit receiving input of equipment operation information; a factor output unit determining and outputting an abnormality factor of the equipment using a learning model that outputs an abnormality factor of the equipment in response to the input of the equipment operation information; and a countermeasure output unit outputting, in response to the input of the abnormality factor, part information corresponding to the abnormality factor or operation information for avoiding the abnormality factor.

[0010] The equipment abnormality diagnosis program of the present invention causes a computer to execute the following steps: accepting input of equipment operation information; determining and outputting abnormality factors of the equipment using a learning model that outputs abnormality factors of the equipment in response to the input of the equipment operation information; and outputting part information corresponding to the abnormality factors or operation information for avoiding the abnormality factors by a method other than using the learning model. Other means will be described in the detailed description of the invention. [Brief explanation of the drawings]

[0011] [Figure 1] FIG. 1 is a configuration diagram of a device abnormality diagnosis system. [Figure 2] FIG. 1 is a configuration diagram of a device abnormality diagnosis system. [Figure 3] FIG. 10 is a configuration diagram of another device abnormality diagnosis system. [Figure 4] 10 is a flowchart of a device abnormality diagnosis process. [Figure 5] FIG. 10 is a diagram showing an example of a master flow definition table used when narrowing down abnormality factors of a refrigeration cycle failure. [Figure 6A] FIG. 10 is a diagram illustrating an example of a display template for a diagnosis result. [Figure 6B] FIG. 10 is a diagram illustrating an example of a display template for a diagnosis result. [Figure 7] FIG. 10 is a diagram illustrating an example of a process table for narrowing down and determining factors. [Figure 8] 10 is a flowchart of a replacement part selection process. [Figure 9] FIG. 10 is a diagram showing a replacement part correspondence master table. [Figure 10] FIG. 10 is a diagram showing a replacement part details dialogue master table. [Figure 11] FIG. 10 is a diagram showing a fault diagnosis screen displayed on the mobile terminal. [Figure 12] 10 is a flowchart showing a consistent production process of a learning device abnormality diagnosis system. [Figure 13]1 is a flowchart showing a learning-type abnormality diagnosis method and development of a diagnostic program. DETAILED DESCRIPTION OF THE INVENTION

[0012] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. FIG. 1 is a configuration diagram of a device abnormality diagnosis system 1. The device abnormality diagnosis system 1 includes a server 2, an interface 12, and a mobile terminal 13. The device abnormality diagnosis system 1 is a system for diagnosing abnormalities in a home electric appliance 11. Note that the object to be diagnosed by the device abnormality diagnosis system 1 is not limited to the home electric appliance 11, and may be any type of device.

[0013] The home appliance 11 is, for example, a refrigerator installed in the home of a customer 14. When the customer discovers that the home appliance 11 is malfunctioning, the following series of operations are initiated.

[0014] A customer 14 requests repairs from an operator 15. The operator 15 contacts a service center 16 located near the address of the customer 14 to request repairs. The service center 16 dispatches a service agent 17 to the home of the customer 14. The service agent 17 visits the home of the customer 14 with an interface 12 and a mobile terminal 13, and connects the interface 12 to the home appliance 11 by wire. The interface 12 is, for example, a Bluetooth (registered trademark) interface module, which enables wireless connection between the home appliance 11 and the mobile terminal 13.

[0015] The mobile terminal 13 is, for example, a tablet terminal equipped with a touch panel display. The mobile terminal 13 further functions as an input unit into which operation information of the home electric appliance 11 is input.

[0016] The mobile terminal 13 acquires driving information from the home electric appliance 11 via the interface 12 and transmits the driving information to the server 2 via the Internet.

[0017] The server 2 is a server computer installed in, for example, a data center. The server 2 includes a model-specific board definition information acquisition / update unit 21, a fault diagnosis unit 22, a cause narrowing-down determination program 222, a cause narrowing-down determination unit 220, a part information output unit 23, and a countermeasure output unit 20. A repair part selection program 24 is installed in the server 2, and when a CPU of the server 2 executes the repair part selection program 24, the model-specific board definition information acquisition / update unit 21, the fault diagnosis unit 22, the part information output unit 23, and the countermeasure output unit 20 are realized. The server 2 outputs the cause of an abnormality in the home appliance 11, and part information corresponding to the cause of the abnormality or operation information for avoiding the cause of the abnormality.

[0018] The model-specific board definition information acquisition / update unit 21 acquires information on the mounted board for each model of the home appliance 11 and updates the information on the mounted board. This enables the part information output unit 23 to output replacement parts for each model that correspond to the failure.

[0019] The fault diagnosis unit 22 is a factor output unit that determines and outputs an abnormality factor of the home appliance 11 based on a single learning model that is based on an average value calculated from a time series of operation information of the home appliance 11. In other words, the fault diagnosis unit 22 executes processing based on the learning model that outputs an abnormality factor of the home appliance 11 in response to input of operation information of the home appliance 11.

[0020] Note that if the cause of the abnormality has been narrowed down to one, the abnormality cause is output, but if there are two or more abnormality causes (i.e., if the cause of the abnormality has been narrowed down to a certain "fault phenomenon" but not to any of the "causes of abnormality" (fault factors) belonging to this fault phenomenon; see FIG. 7, etc.), the cause narrowing down determination unit 220 narrows down the abnormality causes of the home appliance 11 based on a predetermined definition table (see FIG. 5, etc.) and outputs the narrowed down cause. The cause narrowing down determination unit 220 is realized by a computer executing the cause narrowing down determination program 222. The fault diagnosis unit 22 also determines whether the cause of the abnormality of the home appliance 11 is caused by an abnormality in the home appliance 11 or is caused by, for example, a user operation rather than an abnormality in the home appliance. The cause narrowing down determining unit 220 obtains information relating to the results of manual checking of the home electric appliance 11, and narrows down the causes of the abnormality using the input of this information.

[0021] Countermeasure output unit 20 outputs part information corresponding to the cause of the abnormality or operation information for avoiding the cause of the abnormality. When the cause of the abnormality is an abnormality in the home appliance, countermeasure output unit 20 acquires part information corresponding to the cause of the abnormality from part information output unit 23. Part information output unit 23 outputs part information corresponding to the cause of the abnormality in home appliance 11. Based on model number information of home appliance 11, part information output unit 23 outputs information on replacement parts corresponding to that model number.

[0022] If the abnormality factor is caused by the home appliance 11, the countermeasure output unit 20 outputs part information corresponding to the normal factor based on a predetermined rule, not a learning model. Furthermore, if the abnormality factor is not caused by the home appliance 11, the countermeasure output unit 20 outputs operation information for avoiding this abnormality factor. The execution timing of the factor narrowing down determination unit 220 can be selected from before or after the output of information by the countermeasure output unit 20. When the factor narrowing down determination unit 220 executes processing after the countermeasure output unit 20 outputs information, the factor narrowing down determination unit 220 outputs information including a message that there is an unknown abnormality factor.

[0023] 1, the fault diagnosis unit 22, the cause narrowing down determination program 222, the cause narrowing down determination unit 220, the countermeasure output unit 20, the part information output unit 23, and the repair part selection program 24 may be installed in the mobile terminal 13 instead of the server 2. By installing these components, the mobile terminal 13 can have the same functions as the server 2. Furthermore, when various programs installed in the mobile terminal 13 are updated in the server 2, the programs can be newly acquired from the server 2 and updated. The function of updating the programs of the server 2 on the mobile terminal 13 side may be executed when the mobile terminal 13 and the server 2 are capable of communicating with each other. In this case, an increase in the amount of communication between the mobile terminal 13 and the server 2 can be suppressed.

[0024] FIG. 2 is a configuration diagram of the device abnormality diagnosis system 1A. The device abnormality diagnosis system 1A includes a server 2A that may be the same as or different from the device abnormality diagnosis system 1 in Fig. 1, and a service parts system 25. This device abnormality diagnosis system 1A is a system for diagnosing abnormalities in a home electric appliance 11. The service parts system 25 is a computer equipped with a CPU.

[0025] The server 2A is a server computer installed in, for example, a data center. The server 2A includes a model-specific board definition information acquisition / update unit 21, a fault diagnosis unit 22, a cause narrowing-down determination program 222, a cause narrowing-down determination unit 220, and a countermeasure output unit 20. The server 2A transmits information about the home appliance 11 to the service parts system 25, thereby acquiring part information for dealing with an abnormality in the home appliance 11.

[0026] The model-specific board definition information acquisition / update unit 21, fault diagnosis unit 22, cause narrowing-down determination program 222, cause narrowing-down determination unit 220, and countermeasure output unit 20 in FIG. 2 have the same functions as the respective units in FIG. Countermeasure output unit 20 outputs part information corresponding to the cause of the abnormality of home appliance 11 or operation information for avoiding the cause of the abnormality. When the cause of the abnormality of home appliance 11 is an abnormality in the home appliance, countermeasure output unit 20 acquires part information corresponding to the abnormality from part information output unit 23 of service parts system 25.

[0027] Service parts system 25 is a computer, and has repair part selection program 24 installed. A CPU of service parts system 25 executes repair part selection program 24, thereby realizing part information output unit 23. Part information output unit 23 outputs part information corresponding to the cause of an abnormality in home appliance 11. In FIG. 2 , fault diagnosis unit 22, cause narrowing down determination program 222, cause narrowing down determination unit 220, and countermeasure output unit 20 do not necessarily have to be provided in server 2A, and may be provided in mobile terminal 13.

[0028] FIG. 3 is a configuration diagram of the device abnormality diagnosis system 1B. The device abnormality diagnosis system 1B includes a server 2B, a diagnostic engine 26, a service center monitor 19, and a service parts system 25. The device abnormality diagnosis system 1B is a system for diagnosing abnormalities in a home electric appliance 11 remotely.

[0029] The home electric appliance 11 transmits the operation information to the server 2B via a short-range wireless communication path, the router 18, and the Internet. The server 2B collects and stores the operation information. The server 2B in the second embodiment functions as an input unit to which the operation information of the appliance is input.

[0030] The diagnostic engine 26 is a computer equipped with a CPU and may be included in the server 2B. The diagnostic engine 26 includes a fault diagnosis unit 22, a cause narrowing down determination program 222, and a cause narrowing down determination unit 220, and diagnoses whether or not there is a fault in the home appliance 11. The diagnostic engine 26 evaluates operation information of the home appliance 11, for example, periodically using the fault diagnosis unit 22. The diagnostic engine 26 executes the cause narrowing down determination program 222 based on the evaluation result of the fault diagnosis unit 22 to embody the cause narrowing down determination unit 220. The diagnostic engine 26 transmits result information of the diagnosis made by the cause narrowing down determination unit 220 based on the evaluation result of the fault diagnosis unit 22 to the service center monitor 19. Furthermore, if a customer 14 detects a malfunction in the home appliance 11, the customer 14 requests repairs from an operator 15, for example, by telephone or via the web. The operator 15 notifies a service center monitor 19 located near the customer's 14 home of the request for repairs, and the service center monitor 19 (service center) preferably causes a diagnostic engine 26 to execute a malfunction diagnosis of the target home appliance 11, and transmits the result information and the request for repair to the service agent 17. This allows the latest malfunction diagnosis results to be provided when an inquiry from the customer 14 is triggered.

[0031] Service agent 17 accesses service parts system 25 to obtain parts information for repairing the failure. A repair parts selection program 24 is installed in service parts system 25, and a parts information output unit 23 and a countermeasure output unit 20 are realized by a CPU of service parts system 25 executing repair parts selection program 24. Parts information output unit 23 outputs parts information for repairing the failure to countermeasure output unit 20 based on the failure information.

[0032] The countermeasure output unit 20 outputs operation information for taking countermeasures against the fault or parts information for repairing the fault based on the fault information. The service agent 17 informs the customer 14 of the operation information for taking countermeasures against the fault or parts information for repairing the fault.

[0033] 4 to 11 below, the operations will be explained assuming the first embodiment shown in FIG. 1. In the first modified example and the second embodiment, the parts with corresponding names will be interpreted as executing the operations. For example, the mobile terminal 13 in the first embodiment corresponds to the server 2 or the diagnostic engine 26 in another example.

[0034] FIG. 4 is a flowchart of the device abnormality diagnosis process. First, mobile terminal 13 accepts input of operation information from home appliance 11, which is the device to be diagnosed (step S10). Fault diagnosis unit 22 provided in server 2 and / or mobile terminal 13 may evaluate in advance whether an abnormality is likely to occur based on 24 hours' worth of operation information (step S11). Based on the input of the operation information in step S10 or the evaluation result, fault diagnosis unit 22 diagnoses whether home appliance 11 is malfunctioning or not based on a single learning model based on an average value calculated from a time series of operation information (step S12), diagnoses whether the malfunction is caused by the product (step S13), and whether the cause of the malfunction could be determined (step S16), and transmits the diagnosis result to mobile terminal 13 for display.

[0035] The fault diagnosis unit 22 may perform a general diagnosis to diagnose the cause of the abnormality and the general location of the abnormality before performing the diagnosis in step S13 or step S16. In this case, the fault diagnosis unit 22 performs the general diagnosis and then the fault diagnosis. Such a two-stage diagnosis can provide a highly accurate output.

[0036] In step S13, the process branches depending on whether the failure is caused by the product. If the failure is not caused by the product (No), the process proceeds to step S14, where the countermeasure output unit 20 performs a usage instruction process, and when the mobile terminal 13 displays and provides instructions on how to use the product and important points as operation information (step S15), the process of FIG. 4 ends. In this usage instruction process, the countermeasure output unit 20 outputs operation information for avoiding the failure. An example of the operation information is "Please reduce the frequency with which the refrigerator door is opened and closed." If the failure is caused by the product in step S13 (Yes), the process proceeds to step S16.

[0037] In step S16, the process branches depending on whether the cause of the abnormality could be determined. If the cause of the abnormality could be determined (Yes), the process proceeds to step S18. If the cause of the abnormality could not be determined (No), a cause narrowing down process is executed (step S17). Here, "the cause of the abnormality could be determined" means that the cause of the abnormality has been narrowed down to one. "The cause of the abnormality could not be determined" means that the cause of the abnormality has been narrowed down to only the defect phenomenon.

[0038] In the cause narrowing down and determination process, causes are narrowed down, for example, in a chat format, through a terminal used by the user (service agent) of the mobile terminal 13 or the customer 14, based on a predetermined master flow definition table 31, an example of which is shown in Figure 5. That is, the process of step S17 is started when, in terms of the table in Figure 7, the learning model outputs that the abnormality cause is unknown or could not be narrowed down and only the defect phenomenon was identified.

[0039] Addressing an abnormality caused by usage does not require replacement parts, but there are various causes. For example, the refrigerator door may be half-open (half-open to the extent that the sensor that detects door open cannot detect it), or the door may be opened too frequently. In such a case, if only the defective part is set as the output of the diagnosis result, the output will simply say "no replacement parts required," but the cause will not be known, making it difficult for the service agent 17 to explain the situation to the customer 14. In this embodiment, for example, the output will say "no replacement parts required" and the abnormality cause "the door was open for a long time," making it easier for the service agent 17 to explain the situation to the customer 14.

[0040] The determination in step S13 may be performed after step S17. That is, assuming that the learning model cannot determine whether a defective phenomenon is a malfunction caused by the product, a chat scenario may be prepared so that the cause narrowing-down process can narrow down malfunctions caused by operations rather than the product.

[0041] In step S18, the CPU of mobile terminal 13 displays the estimated cause of the abnormality. Then, the CPU of mobile terminal 13 calls part information output unit 23 of server 2 and executes replacement part selection processing (step S19). Then, countermeasure output unit 20 causes mobile terminal 13 to display the countermeasure. The CPU of mobile terminal 13 or server 2 displays replacement parts and precautions and provides guidance on countermeasures for the abnormality (step S20), and ends the processing of FIG. 4.

[0042] 5 is a diagram showing an example of a master flow definition table 31 used when narrowing down the abnormality factors of a refrigeration cycle malfunction. Here, we will explain the case where, in step S16, No, the learning model outputs the malfunction phenomenon "refrigeration cycle abnormality," but the cause of the abnormality is unknown. The number of malfunction phenomena output may be two or more types, not just one.

[0043] The master flow definition table 31 includes an ID column, a button display text column, an explanation / response method column, a branch question column, a display button number column, and a next button column. The ID field stores the identifier for each button. The Button Display Text field stores the string to be displayed on the button. The Explanation / Response Method field stores the string of the explanation or response method to be displayed in the dialog along with the button. The Branching Question field stores the string of the question that accompanies branching with the Next button. The Number of Display Buttons field stores the number of display buttons related to the branching question. The Next Button field stores the identifier of the button related to the next option.

[0044] In response to the start of step S17, the CPU of the mobile terminal 13 first displays the button display text "Isolate the cause of refrigeration cycle abnormality" button in the ID field S1 corresponding to the fault phenomenon "refrigeration cycle abnormality." When the button in the ID field S1 is tapped, the CPU displays the text "Is there frost on the S pipe?" in the corresponding branch question field, and also displays and outputs the two buttons set as the corresponding next button fields, the "Yes" button in the button display text A11 and the "No" button in the button display text A12, on the display screen of the mobile terminal 13, thereby continuing to narrow down the causes. A question to be displayed in the branch question field may, for example, be one that requests the service agent or the customer to check the condition of the home appliance to be diagnosed and to input the results of that check.

[0045] When the "Included" button as the button display text A11 is tapped, the cause narrowing down and determination program 222 narrows down the causes of the abnormality in FIG. 7 to "excessive gas filling." The CPU of the mobile terminal 13 outputs "Excessive refrigerant filling is suspected. Please reseal the refrigerant." The cause narrowing down and determination program 222 determines to include the output of "excessive gas filling" as a presumed abnormality cause (step S18). If there are any other malfunction phenomena remaining to be processed in step S17, the program proceeds to narrow down the malfunction causes corresponding to those malfunction phenomena.

[0046] If the "Not attached" button in button display text A12 is tapped instead of the "Attached" button in button display text A11, the CPU of mobile terminal 13 displays "Check for gas leaks" and displays a branching question, "Is there oil on the evaporating dish?", as well as the "Attached" button in button display text A21 and the "Not attached" button in button display text A22. By continuing to ask branching questions in this manner, the cause of the abnormality is narrowed down, preferably to one.

[0047] Note that execution of the cause narrowing down determination program 222 in step S17 is not necessarily required, and the service agent may be able to select whether or not to execute it. If execution or no execution is selected, step S17 is not executed, and the process proceeds to steps S18, S19, and S20. FIG. 11 is an example of a screen on which these displays are made. The "Cause / Status" column 52 and the like indicates that an unknown abnormality cause exists. In this case, when the "Narrow Down Causes" button 532 is tapped, the cause narrowing down determination program 222 in step S17 is executed again. Furthermore, when the "Remedial Part Information" button 535 is tapped, the part information output by the countermeasure output unit 20 is displayed.

[0048] 6A and 6B are diagrams showing examples of the diagnosis result display template 32. Use of the diagnosis result display template 32 is optional. The diagnosis result display template 32 includes a diagnosis result field, a cause / condition field, a diagnosis result explanation field, and a repair part information field. The diagnosis result field includes a diagnosis result field using operation information for the past seven days, for example, and an overall judgment field.

[0049] The 7-day diagnostic result column stores information on diagnostic results for 7 consecutive days. The overall judgment column stores information on the overall judgment corresponding to the 7-day diagnostic results. The cause / condition column stores the cause or condition corresponding to the 7-day diagnostic results. The diagnostic result explanation column stores a detailed explanation of the diagnostic results.

[0050] If the diagnosis result for the seven days is "normal," the overall judgment is normal, and both the internal temperature and the refrigeration cycle are within the normal range. In this case, the mobile terminal 13 displays the explanation of the diagnosis result, "There is no abnormality in the internal temperature or the refrigeration cycle."

[0051] If the diagnosis result for 7 days is "normal (temporary disturbance in the refrigeration cycle)," the overall judgment is "normal," and both the internal temperature and the refrigeration cycle are within the normal range. In this case, the mobile terminal 13 displays the explanation of the diagnosis result, "A temporary disturbance was observed in the refrigeration cycle, but it is within the normal range and there is no abnormality."

[0052] If the diagnosis result for the seven days is "Normal (Door gap suspected)," the overall judgment is normal (Door gap suspected), and the temperature has temporarily risen due to a suspected door gap. At this time, the mobile terminal 13 displays the following explanation of the diagnosis result: "There may have been a gap in the door of the freezer compartment or other compartment, causing a temporary temperature rise. The temperature inside the freezer has recovered, so please explain that there is no problem if there is no gap in the door."

[0053] If the seven-day diagnosis result is "disturbance in the refrigeration cycle" or "fault in the refrigeration cycle," the overall judgment is that there is an abnormality in the refrigeration cycle system, and the cause / condition is gas leakage / abnormal operation of the refrigerant flow path switching valve / blockage in the refrigeration cycle. In this case, the mobile terminal 13 displays the explanation of the diagnosis result, "There is a possibility of gas leakage / abnormal operation of the refrigerant flow path switching valve / blockage in the refrigeration cycle, and repair is required." The mobile terminal 13 then recommends the cooler and compressor as parts that need to be repaired.

[0054] When the diagnosis result for seven days is "unstable power supply," the overall judgment is that the power supply is unstable, and the cause / condition is that the power supply is frequently interrupted temporarily / the device is set to store display mode. In this case, the mobile terminal 13 displays the following explanation of the diagnosis result: "There may be frequent disconnections of the power cord or frequent interruptions of the power supply, or the device may be set to store display mode."

[0055] If the seven-day diagnosis result is "Air Path Abnormality (Refrigerator / Switchable Compartment)," the overall judgment is that there is an abnormality in the refrigerator / switchable compartment air duct, and the cause (possibility) / condition is an operational abnormality in the refrigerator / switchable compartment damper, or a cold air leak or air duct abnormality in the switchable compartment. In this case, the mobile terminal 13 displays the explanation of the diagnosis result as "There is an abnormality in the wiring of the vegetable / switchable compartment damper, and there is a possibility of a cold air leak or air duct abnormality in the switchable compartment. Repair is required." The mobile terminal 13 then recommends the damper as the part to be repaired.

[0056] If the diagnosis result for the seven days is "Air Path Abnormality (Vegetable Compartment)," the overall judgment is that there is an abnormality in the crisper compartment air duct, and the cause (possibility) / condition is that the crisper compartment damper is malfunctioning. In this case, the mobile terminal 13 displays "There is a possibility that the crisper compartment damper wiring is abnormal" as an explanation of the diagnosis result. The mobile terminal 13 then recommends the damper as the part to be repaired.

[0057] If the diagnosis result over the seven days is "air path abnormality (air path blocked)," the overall judgment is air path abnormality (suspected blockage) (cold air not reaching the refrigerator compartment). For this diagnosis result, the cause (possibility) / condition is one of the following: refrigerator damper wiring abnormality, the refrigerator return air path is blocked, frost buildup in the refrigerator / freezer air path, or the refrigerator / freezer fan not rotating. At this time, the mobile terminal 13 displays the explanation of the diagnosis result: "There is a possibility that the refrigerator damper wiring abnormality, the refrigerator return air path is blocked, frost buildup in the refrigerator / freezer air path, or the refrigerator / freezer fan not rotating is present. Repair is required." The mobile terminal 13 then recommends the damper and fan as parts to be repaired.

[0058] If the diagnosis results for the seven days are other than those mentioned above, the overall judgment is "cannot be judged," and the cause (possibility) / condition is that the power-on time is short. In this case, the mobile terminal 13 displays the explanation of the diagnosis result as follows: "Diagnosis cannot be performed if the operation time is short, such as immediately after power-on, when the power is cut off midway, or immediately after switching the setting of the switching room."

[0059] In the first embodiment, the diagnosis result display template 32 is used for messages and various buttons displayed on the screen of the mobile terminal 13. The service agent 17 can read the messages and tap various buttons to obtain the judgment result, an explanation of the judgment result, information on repair parts, etc.

[0060] FIG. 7 is a diagram showing an example of the process table 33 for narrowing down and determining the cause of an abnormality. The process table 33 includes a column for a defect phenomenon, a column for a cause of an abnormality (a column for a cause of a failure), a column for an occurrence location, a column for a main cause, and columns for symptoms #1 to #5.

[0061] The defect phenomenon column stores information about the defect phenomenon. The abnormality cause column stores the cause of the abnormality. The occurrence location column stores the location where the abnormality occurred. The main cause column stores the main causes of the defect phenomenon.

[0062] The Symptom #1 column stores the overall symptoms of the refrigerator. The Symptom #2 column stores the symptoms of the compressor. The Symptom #3 column stores specific symptoms of refrigerator malfunctions. The Symptom #4 column stores symptoms of condenser malfunctions. The Symptom #5 column stores symptoms of refrigerant and oil leaks.

[0063] In the first embodiment, a general diagnosis may be performed using each "symptom" column of the process table 33. Apart from processing the learning model, the fault diagnosis unit 22 can diagnose the cause of the abnormality and the general location of the abnormality by applying symptoms #1 to #5 of the home appliance 11.

[0064] FIG. 8 is a flowchart of the replacement part selection process. The CPU of the mobile terminal 13 determines the estimated abnormality factor in step S17 of Fig. 4 (step S30). Then, the mobile terminal 13 selects a replacement part corresponding to the estimated abnormality factor in the replacement part correspondence master table 34 (step S31). The replacement part correspondence master table 34 is created separately from the learning model used in step S12 and the program used in step S17, such as a chat scenario chatbot. This table outputs a replacement part by referring to the estimated abnormality factor output in steps S12 and S17.

[0065] FIG. 9 is a diagram showing an example of the replacement part correspondence master table 34. As shown in FIG. The replacement parts correspondence master table 34 includes a failure phenomenon column corresponding to the estimated abnormality cause, a response column, and a replacement parts column. The failure phenomenon column stores the failure phenomenon. Each failure phenomenon may have the same name as the abnormality cause, or may have a different name as long as they are associated. The response column stores the response to the failure. The replacement parts column stores replacement parts as general classifications when the response is part replacement. This makes it possible to select a method for responding to the failure phenomenon and the replacement parts when part replacement is required. The information in the replacement parts correspondence master table 34 may be included in the process table 33 of FIG. 7.

[0066] Returning to Fig. 8, the explanation will be continued. The CPU of the mobile terminal 13 determines whether or not there is a replacement part corresponding to the cause of the abnormality (step S32). If there is no replacement part corresponding to the cause of the abnormality (No), the processing in Fig. 8 ends. If there is a replacement part corresponding to the cause of the abnormality (Yes), the processing proceeds to step S33.

[0067] In step S33, the CPU of the mobile terminal 13 collates the replacement part and model data with the replacement part detail master table 35, and then displays the replacement part (step S34).

[0068] FIG. 10 is a diagram showing an example of the replacement part details master table 35. As shown in FIG. In the replacement part details master table 35, each row describes a replacement part, and each column stores the model of the home appliance 11. According to the model data of the home appliance being diagnosed and the replacement part, it is possible to identify the part name (part model name) of the replacement part corresponding to the model data.

[0069] The description will continue by returning to Fig. 8. When the service agent 17 replaces the detailed replacement part in step S35, the processing in Fig. 8 is completed.

[0070] FIG. 11 is a diagram showing the fault diagnosis result screen 5 displayed on the mobile terminal 13. As shown in FIG. The fault diagnosis result screen 5 displays an "Overall judgment" column 51, a "Cause / condition" column 52, and a "Diagnosis result details" column 53. The "Diagnosis result details" column 53 displays a "Diagnosis result explanation" button 531, a "Narrow down causes" button 532, a "7-day temperature graph" button 533, an "Operation history information" button 534, and a "Repair part information" button 535.

[0071] FIG. 11 shows the state after the "Explanation of Diagnosis Result" button 531 is tapped. Below the "Diagnosis Result Details" column 53, an explanation 54 of the diagnosis result is displayed. Here, the fault phenomenon is displayed in the "Overall Judgment" column 51. The estimated cause of the abnormality is displayed in the "Cause / Status" column 52. In addition, the overall judgment results for not only the most recent day but also the past few days are displayed. This fault diagnosis result screen 5 allows the user or service agent 17 to know the details of the diagnosis result related to the fault of the home appliance 11.

[0072] FIG. 12 is a flowchart showing the process of creating an abnormality diagnosis system for a learning device according to the present invention. First, the developer creates a diagnostic program that uses machine learning to learn about market defect information and predicts the causes and factors of defects (step S40). Next, the developer creates a program that selects appropriate replacement parts for repairs based on the cause prediction results (step S41), and then installs each program into a diagnostic device such as a mobile terminal or PC (step S42). Then, service agent 17 inputs the operation information of the equipment to be diagnosed into the diagnostic equipment and executes a fault diagnosis (step S43). If the diagnostic result (phenomenon) is not a product malfunction, service agent 17 starts a process to explain how to use the equipment (step S44). The developer links the selected parts to a service parts system and creates a system to display the parts on the diagnostic equipment (step S45). The developer then orders service parts and repairs the equipment (step S46), terminating the process of FIG. 12.

[0073] FIG. 13 is a flowchart showing the development of a learning-type abnormality diagnosis method and a diagnostic program according to the present invention. The developer obtains market malfunction operation information consisting of temperature graphs and tables containing digital data (step S50). Then, the developer analyzes the market malfunction operation information (investigates the cause of the malfunction) (step S51) and attaches the cause of the malfunction to the market malfunction operation information (step S52).

[0074] The developer converts the on-the-spot malfunction operation information into two-dimensional (digital) data with a time axis (step S53), and selects information that is estimated to affect performance and operation (step S54). Here, the on-the-spot malfunction operation information is one-dimensional time-series information of sensor information, so it is configured as two-dimensional data with a sensor information axis and a time axis.

[0075] The developer then averages the time axis of the market malfunction operation information, which is two-dimensional data, and converts it into one-dimensional data (step S55). Here, the developer separates only the information (explanatory variables) that is estimated to affect performance and operation and converts it into one-dimensional data. Here, one-dimensional data refers to sensor information from which the time axis has been averaged and deleted. Here, the developer is not limited to averaging, and may convert it into one-dimensional data using any statistical value such as the median or mode.

[0076] The developer further clusters the one-dimensional data of multiple pieces of market malfunction driving information using machine learning (unsupervised) (step S56). Here, the developer groups the data regardless of whether the cause is clear or unclear (whether there is a correct answer or not) (step S57). The market malfunction driving information also contains data with known correct labels. Therefore, unlabeled data that belongs to the same group as data with correct labels after grouping can be complemented by assuming that the correct labels are the same. Here, the more one-dimensional data of market malfunction driving information there is, the better, for example, 1,000 to 10,000 items.

[0077] The developer verifies whether individuals in the same group have the same cause (step S58), assigns the same defect cause to individuals whose causes are unclear (step S59), and completes the feature values ​​for each defect cause (step S60). This completes the training data for the learning model.

[0078] The developer creates a single learning model that classifies causes through machine learning using training data (step S61). The developer then places the single learning model on a server (step S62) and installs the learning model from the server into mobile terminals and computers (step S63), completing the process in Figure 13. This allows the developer to use the created learning model for fault diagnosis.

[0079] The learning model used in this embodiment is a learning model that is learned by clustering statistical information on driving information, so it is not affected by small perturbations in the driving information and can perform stable fault diagnosis.In addition, since it is a single learning model that can diagnose faults regardless of the model of multiple devices, it is possible to reduce the development process.

[0080] The configuration and effects of the invention covered by this specification will be described below. Elements in parentheses mainly refer to elements corresponding to the first embodiment.

[0081] [1] an input unit (interface 12, mobile terminal 13) to which operation information of a device (home electric appliance 11) is input; a factor output unit (fault diagnosis unit 22) that executes processing based on a learning model to determine and output an abnormality factor of the appliance in response to input of operation information of the appliance (home electric appliance 11); a countermeasure output unit (20) that receives the input of the abnormality factor and outputs part information corresponding to the abnormality factor or operation information for avoiding the abnormality factor; An equipment abnormality diagnosis system (1) comprising:

[0082] This makes it possible to diagnose equipment abnormalities using programs other than learning models.

[0083] [2] If the abnormality factor is caused by the device (home electric appliance 11), the countermeasure output unit (20) outputs part information corresponding to the abnormality factor based on a predetermined rule, not the learning model. 2. The device abnormality diagnosis system (1) according to claim 1.

[0084] This allows for the confirmation of abnormality factors that can be easily determined using evaluation rules without relying on a learning model.

[0085] [3] If the cause of the abnormality is not caused by the device (home electric appliance 11), the countermeasure output unit (20) outputs operation information for avoiding the cause of the abnormality. 2. The device abnormality diagnosis system according to claim 1.

[0086] This allows the service agent to obtain information on parts corresponding to the cause of the abnormality and quickly deal with the equipment failure.

[0087] [4] a factor narrowing-down determination unit (220) that, when the learning model is unable to narrow down the cause of the abnormality of the device, obtains information relating to the result of a human check of the device and narrows down the cause of the abnormality using the input of the information; 2. The device abnormality diagnosis system according to claim 1, further comprising:

[0088] This allows the cause of the abnormality to be further narrowed down based on information related to the results of human confirmation.

[0089] [5] The execution timing of the factor narrowing down determination unit (220) can be selected from before or after the output of information by the factor output unit, When the factor narrowing-down determination unit (220) executes processing after the factor output unit (22) outputs information, the factor narrowing-down determination unit (220) outputs information including a message indicating that there is an unknown abnormality factor. 5. The device abnormality diagnosis system according to claim 4.

[0090] This makes it possible to obtain information relating to the results of human confirmation at a suitable timing, either before or after the determination by the factor output unit.

[0091] [6] The input unit (interface 12, mobile terminal 13) further receives input of model information of the device, the part information output unit (23) outputs the model information and part information corresponding to the abnormality cause based on a predetermined rule. 4. The device abnormality diagnosis system according to claim 3.

[0092] This allows the service agent to obtain information on parts corresponding to the device model and the cause of the abnormality, enabling the service agent to quickly deal with the device failure.

[0093] [7] The learning model selects explanatory variables that affect the failure from market operation information that has occurred in the market in the past, By clustering using unsupervised learning, the market operation information is grouped into a first pattern in which the failure factor, which is the objective variable, is known and a second pattern in which the failure factor is unknown; generating training data by complementing the failure factors of the first pattern within the same group as failure factors of the second pattern; The training data is learned. 2. The device abnormality diagnosis system according to claim 1.

[0094] This allows a sufficiently large amount of training data to be generated from market operation information, thereby improving the judgment accuracy of the learning model.

[0095] [8] An input unit (interface 12, mobile terminal 13) receives input of operation information of the equipment; a step in which a cause output unit (fault diagnosis unit 22) determines and outputs an abnormality cause of the equipment based on statistical information of the operation information of the equipment and a single learning model; a step in which a countermeasure output unit (20) outputs part information corresponding to the abnormality factor or operation information for avoiding the abnormality factor; An equipment abnormality diagnosis method comprising:

[0096] This makes it possible to diagnose abnormalities in equipment using a single learning model that has learned from a relatively small amount of training data.

[0097] [9] On the computer, A procedure for accepting input of equipment operation information; a step of determining and outputting an abnormality factor of the equipment based on a single learning model based on statistical information of the operation information of the equipment; a step of outputting part information corresponding to the cause of the abnormality or operation information for avoiding the cause of the abnormality; An equipment abnormality diagnosis program for executing the above.

[0098] This makes it possible to diagnose abnormalities in equipment using a single learning model that has learned from a relatively small amount of training data.

[0099] (Variation) The present invention is not limited to the above-described embodiments and includes various modifications. For example, the above-described embodiments have been described in detail to clearly explain the present invention, and are not necessarily limited to those including all of the described configurations. It is possible to replace part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment. Furthermore, it is also possible to add, delete, or replace part of the configuration of each embodiment with other configurations.

[0100] The above-described configurations, functions, processing units, processing means, etc. may be realized in part or in whole by hardware such as an integrated circuit. The above-described configurations, functions, etc. may be realized by software by a processor interpreting and executing a program that realizes each function. Information such as the programs, tables, and files that realize each function can be stored in a storage device such as a memory, a hard disk, or an SSD (Solid State Drive), or on a storage medium such as a flash memory card or a DVD (Digital Versatile Disk).

[0101] In each embodiment, the control lines and information lines shown are those that are considered necessary for the explanation, and not all control lines and information lines in the product are necessarily shown. In reality, it can be considered that almost all components are interconnected. [Explanation of symbols]

[0102] 1,1A,1B Equipment Abnormality Diagnostic System 12 Interface (part of input section) 13 Mobile terminal (part of input section) 11 Home appliances 14 Customers 15 Operator 16 Service Center 17 Service Agent 18 Router 19 Service Center Monitor 2, 2A, 2B Server (Output section) 21 Model-specific board definition information acquisition and update section 22 Fault diagnosis section (factor output section) 222 Factor Narrowing Decision Program 20 Countermeasure output section 23 Parts information output section 24 Repair Parts Selection Program 25 Service Parts System 26 Diagnostic Engine 31 Master Flow Definition Table 32 Diagnostic result display template 33 Process Table 34 Replacement parts master table 35 Replacement Parts Detail Master Table 5. Fault diagnosis result screen 51 "Overall Judgment" column 52 "Cause / Condition" column 53 "Diagnosis result details" column 531 "Explanation of diagnosis results" button 532 "Refine Factors" Button 533 "7-day temperature graph" button 534 "Driving history information" button 535 "Remedial Parts Information" button 54 Explanation

Claims

1. an input unit to which operation information of the device is input; a factor output unit that executes processing based on a learning model to output an abnormality factor of the equipment in response to input of operation information of the equipment; a countermeasure output unit that receives the input of the abnormality factor and outputs part information corresponding to the abnormality factor or operation information for avoiding the abnormality factor; An equipment abnormality diagnosis system comprising:

2. If the abnormality factor is caused by the device, the countermeasure output unit outputs part information corresponding to the abnormality factor based on a predetermined rule, not based on the learning model.

2. The device abnormality diagnosis system according to claim 1.

3. the countermeasure output unit outputs operation information for avoiding the abnormality factor if the abnormality factor is not caused by the device.

2. The device abnormality diagnosis system according to claim 1.

4. a factor narrowing-down determination unit that, when the learning model is unable to narrow down the cause of the abnormality of the device, obtains information relating to the results of a human check of the device and narrows down the cause of the abnormality using the input of the information; 2. The device abnormality diagnosis system according to claim 1, further comprising:

5. the execution timing of the factor narrowing down determination unit can be selected from before or after the output of information by the factor output unit, When the cause narrowing down determination unit executes processing after the cause output unit outputs information, the cause narrowing down determination unit outputs information including a message that there is an unknown abnormal cause.

5. The device abnormality diagnosis system according to claim 4.

6. the input unit further accepts input of model information of the device; the countermeasure output unit outputs the model information and part information corresponding to the abnormality cause based on a predetermined rule, not based on the learning model.

3. The device abnormality diagnosis system according to claim 2.

7. The learning model selects explanatory variables that affect the failure from market operation information that has occurred in the market in the past, By clustering using unsupervised learning, the market operation information is grouped into a first pattern in which the failure factor, which is a target variable, is known and a second pattern in which the failure factor is unknown; generating training data by complementing the failure factors of the first pattern within the same group as failure factors of the second pattern; The training data is learned.

2. The device abnormality diagnosis system according to claim 1.

8. an input unit receiving input of operation information of the equipment; a step in which a factor output unit determines and outputs an abnormality factor of the equipment using a learning model that outputs an abnormality factor of the equipment in response to input of operation information of the equipment; a step in which a countermeasure output unit outputs, in response to an input of the abnormality factor, component information corresponding to the abnormality factor or operation information for avoiding the abnormality factor; An equipment abnormality diagnosis method comprising:

9. On the computer, A procedure for accepting input of equipment operation information; a step of determining and outputting an abnormality factor of the equipment using a learning model that outputs an abnormality factor of the equipment in response to input of operation information of the equipment; a step of outputting part information corresponding to the abnormality factor or operation information for avoiding the abnormality factor by a method other than using the learning model; An equipment abnormality diagnosis program for executing the above.

Images