Abnormality diagnosis device and method

Abstract

This abnormality diagnosis device that estimates the degree of abnormality and abnormality factors of a target from data makes the presenting of an explanation easily understandable by a worker, by presenting an explanation on the reason for a diagnosis in association with the diagnostic knowledge of the worker. To this end, the abnormality diagnosis device comprises: an abnormality diagnosis unit that diagnoses the abnormality of a diagnosis target by inputting data related to the diagnosis target; an explanation presenting unit that presents an explanation corresponding to the abnormality of the diagnosis target, on the basis of the result of the diagnosis by the abnormality diagnosis unit; and a display unit that displays, on a screen, an explanation corresponding to the abnormality of a diagnosis target presented by the explanation presenting unit.

Description

[Technical field]

[0001] The present invention relates to an abnormality diagnosis device and method. [Background technology]

[0002] In an abnormality diagnosis device that estimates the degree of abnormality and abnormality factors of an object from data, there is a technology for presenting an explanation of the diagnosis reason (hereinafter, explanation presentation technology) described in Patent Document 1. This publication states that "the device includes an extraction means for extracting a plurality of feature amounts from an image including the object to be inspected, a determination means for determining the degree of abnormality of the object to be inspected based on the extracted feature amounts, a map generation means for generating a plurality of score maps representing the scores of each of the feature amounts from the image, and an image generation means for generating a defect display image representing a defect that is the basis for determining that the object to be inspected is abnormal by integrating the plurality of score maps based on the contribution of each of the feature amounts to the degree of abnormality determined by the determination means." [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2016-114592 A Summary of the Invention [Problem to be solved by the invention]

[0004] The explanation presentation technology that explains the results of an abnormality diagnosis presents the reason for the abnormality diagnosis, which can be useful for users of the abnormality diagnosis device (hereinafter referred to as operators) to judge the validity of the diagnosis results and create diagnosis reports.

[0005] However, conventional explanation presentation techniques have the problem that when the explanation presented by the device does not correspond to the diagnostic knowledge of the worker regarding the abnormality, the explanation can be difficult for the worker to understand.

[0006] For example, in Patent Document 1, an explanation is presented using feature amounts that contributed to diagnosis. The feature amounts are data related to the target of abnormality diagnosis, such as images or sensor data acquired from the target of abnormality diagnosis, or modified versions of these, and there are various design methods. The feature amounts are not necessarily designed to be easy for the operator to understand, but are designed by the designer of the abnormality diagnosis device to increase the accuracy of abnormality diagnosis. Therefore, in the method of Patent Document 1, there is a possibility that the operator can only understand the explanation when the operator has knowledge about the feature amounts.

[0007] The present invention provides an abnormality diagnosis device and method that solves the problems of the conventional technology described above and presents an explanation of the diagnostic reason in correspondence with the operator's diagnostic knowledge, thereby enabling the operator to easily understand the explanation. [Means for solving the problem]

[0008] In order to solve the problems of the conventional techniques described above, the present invention provides an abnormality diagnosis device including: a diagnostic model storage unit that stores, as a diagnostic model, a machine learning model that has learned the relationship between the feature amount of data, which is diagnostic information related to a diagnosis target, and a diagnostic result; an abnormality diagnosis unit that estimates a diagnostic result of the diagnosis target using the feature amount and the diagnostic model extracted from the diagnostic information data at the time of abnormality diagnosis; a definition unit that holds a plurality of diagnostic knowledge items; an explanation model storage unit that stores, as an explanation model, a machine learning model that has learned the relationship between the diagnostic knowledge items and processing information that reflects a diagnostic process in which the diagnostic model was executed for past abnormality cases; an explanation presentation unit that estimates diagnostic knowledge items related to the diagnostic result output by the abnormality diagnosis unit using the explanation presentation model and the processing information when the abnormality diagnosis unit executed the diagnostic model for the diagnostic information; and a display unit that displays on a screen the diagnostic result of the diagnosis target estimated by the abnormality diagnosis unit and the diagnostic knowledge items estimated by the explanation presentation unit. The diagnostic model stored in the diagnostic model storage unit is a neural network, and the processing information is vector information of an intermediate layer obtained when the neural network is executed on the diagnostic information. do.

[0009] In addition, in order to solve the above-mentioned problems, the present invention provides an abnormality diagnosis method for diagnosing an abnormality of a diagnosis target using an abnormality diagnosis device, the abnormality diagnosis device comprising a diagnostic model storage unit, a definition unit, an explanation presentation model storage unit, an abnormality diagnosis unit, an explanation presentation unit, and a display unit, the diagnostic model storage unit stores, as a diagnostic model, a machine learning model that has learned a relationship between a feature amount of data that is diagnostic information related to the diagnosis target and a diagnostic result, the definition unit holds a plurality of diagnostic knowledge items, the explanation presentation model storage unit stores, as an explanation presentation model, a machine learning model that has learned a relationship between processing information that reflects a diagnostic process in which a diagnostic model in a past abnormality case was executed and the diagnostic knowledge items, the abnormality diagnosis unit estimates a diagnostic result of the diagnosis target using the feature amount and the diagnostic model extracted from the data that is diagnostic information at the time of abnormality diagnosis, the explanation presentation unit estimates diagnostic knowledge items related to the diagnostic result output by the abnormality diagnosis unit using the processing information and the explanation presentation model when the abnormality diagnosis unit executed the diagnostic model on the diagnostic information, and the display unit displays, on a screen, the diagnostic result of the diagnosis target estimated by the abnormality diagnosis unit and the diagnostic knowledge items estimated by the explanation presentation unit together. The diagnostic model stored in the diagnostic model storage unit is a neural network, and the processing information is vector information of an intermediate layer obtained when the neural network is executed on the diagnostic information. do. Effect of the Invention

[0010] According to the present invention, by presenting an explanation of the diagnosis reason in association with the existing knowledge of the worker, it is possible to present an explanation that is easy for the worker to understand.

[0011] Problems, configurations and effects other than those described above will become apparent from the following description of the embodiments. [Brief description of the drawings]

[0012] [Figure 1] 1 is a block diagram showing a schematic configuration of an abnormality diagnosis device according to a first embodiment of the present invention; [Diagram 2] 4 is a flowchart showing a flow of processing executed by an explanation presentation model executing unit according to the first embodiment of the present invention. [Figure 3A]FIG. 2 is a diagram showing an example of diagnostic knowledge according to the first embodiment of the present invention, and is a diagram showing the relationship between check items and diagnostic results when the check items are abnormal vibrations. [Figure 3B] FIG. 13 is a diagram showing an example of diagnostic knowledge according to the first embodiment of the present invention, and is a diagram showing the relationship between the check items and the diagnosis results when the check items are high-frequency abnormality in the motor current. [Figure 3C] FIG. 13 is a diagram showing an example of diagnostic knowledge according to the first embodiment of the present invention, and is a diagram showing the relationship between check items and diagnostic results when the check items are the magnitude relationship between a position error and a threshold value. [Figure 4] 3 is a front view of a screen showing an example of an explanation presentation screen according to the first embodiment of the present invention; FIG. [Diagram 5] 4 is a flowchart illustrating a processing flow of an abnormality diagnosis process and an explanation presentation process according to the first embodiment of the present invention. [Figure 6] FIG. 11 is a front view of a screen showing an example of an explanation presentation screen according to a second embodiment of the present invention. [Figure 7] FIG. 11 is a block diagram showing a schematic configuration of an abnormality diagnosis device according to a third embodiment of the present invention. [Figure 8] 13 is a flowchart illustrating a flow of a learning process in a learning unit according to a third embodiment of the present invention. [Figure 9] FIG. 2 is a schematic diagram illustrating a hardware configuration of an information processing device. [Figure 10] 2 is a diagram showing an example of programs and data stored in a storage device in the abnormality diagnosis device according to the first or second embodiment of the present invention; FIG. [Figure 11] FIG. 11 is a diagram showing an example of programs and data stored in a storage device in an abnormality diagnosis device according to a third embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] In an anomaly diagnosis device that identifies the presence or absence of anomalies, their severity, and their causes through AI (Artificial Intelligence) diagnosis based on acquired data and past cases, it is required to present an explanation of the diagnosis reason to serve as a reference for judging the validity of the diagnosis result and creating a diagnosis report. In this invention, an explanation presentation model that predicts the check items and check results that the operator referred to during manual diagnosis from the processing information of the diagnosis model is used, and an explanation that is easy to understand is presented to the operator by showing the reason for the diagnosis including countermeasures in association with the check items during manual diagnosis.

[0014] That is, in the present invention, the abnormality diagnosis device is provided with a definition unit that defines diagnostic knowledge related to abnormality diagnosis, and is configured to obtain processing information of a diagnostic model that estimates the degree of abnormality and abnormality factors of a target from features, and to present an explanation of the diagnosis reason of the diagnostic model by linking it to the diagnostic knowledge using an explanation presentation model that predicts diagnostic knowledge items from the processing information, and to present an explanation of the diagnosis reason including countermeasures in association with the operator's existing knowledge, thereby making it possible to present an explanation that is easy for the operator to understand.

[0015] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In all the drawings for explaining the embodiment, the same members are generally given the same reference numerals, and the repeated description thereof will be omitted. In addition, in the following embodiment, the components (including element steps, etc.) are not necessarily essential, unless otherwise specified or considered to be obviously essential in principle. In addition, when it is said that "consists of A", "consists of A", "has A", or "includes A", it is needless to say that other elements are not excluded, except when it is specifically specified that only the element is included. Similarly, in the following embodiment, when referring to the shape, positional relationship, etc. of the components, etc., it is intended to include those that are substantially similar or similar to the shape, etc., except when it is specifically specified or considered to be obviously not essential in principle. EXAMPLES

[0016] Here, we will describe an anomaly diagnosis device that presents an explanation linked to diagnostic knowledge about the reason for a diagnosis made by a diagnostic model, using anomaly diagnosis for a robot as an example.

[0017] FIG. 1 is a diagram showing an example of the configuration of an abnormality diagnosis device according to a first embodiment of the present invention.

[0018] The robot 20 is a robot that performs gripping and assembly operations and is the subject of abnormality diagnosis, and is configured with an arm 21, a control unit 23, a current sensor 26, a vibration sensor 24, and an image sensor 25.

[0019] The arm 21 is a movable part of the robot, and the motor 22 is a motor for driving the arm. The control unit 23 is a control unit that causes the arm 21 to perform a desired motion. In order to cause the arm 21 to perform a desired motion, the control unit 23 first determines a target movement amount for each part, and then drives the motor 22 to control the arm 21. For example, the control unit 23 can be configured by a control board or a microcomputer equipped with a CPU (Central Processing Unit), RAM (Random Access Memory), and ROM (Read Only Memory).

[0020] The vibration sensor 24 is a sensor that measures the amount of vibration of the arm 21. The image sensor 25 is a sensor that measures the actual value of the amount of movement of the arm 21. The current sensor 26 is a sensor that measures the current value of the motor 22.

[0021] The abnormality diagnosis device 10 is a device that performs an abnormality diagnosis on the robot 20 and presents an explanation of the reason for the diagnosis, and includes an explanation presentation unit 11, an abnormality diagnosis unit 12, an explanation presentation model storage unit 14, a definition unit 15, and a diagnosis model storage unit 16. The explanation presentation unit 11, the abnormality diagnosis unit 12, and the explanation presentation model storage unit 14 may be collectively referred to as an explanation presentation processing unit.

[0022] The abnormality diagnosis device 10 is realized by an information processing device 90 including a processor (CPU) 91, a memory (RAM) 92, a storage device 93, an input device 94, an output device 95, a communication device 96, and a bus 97 as main components as shown in FIG. 9. The processor 91 functions as a functional unit (functional block) that provides a predetermined function by executing processing according to a program loaded in the memory 92. The storage device 93 stores data used in the functional unit in addition to programs that cause the functional unit to function. For the storage device 93, a non-volatile storage medium such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive) is used. The input device 94 is a keyboard, a pointing device, etc., and the output device 95 is a display, etc. The input device 94 and the output device 95 may be integrated using a touch panel. The communication device 96 enables communication with other information processing devices via a network. These are connected to each other by a bus 97 so that they can communicate with each other.

[0023] The abnormality diagnosis device 10 does not have to be realized by one information processing device, but may be realized by multiple information processing devices. Also, some or all of the functions of the abnormality diagnosis device 10 may be realized as an application on the cloud.

[0024] 10 shows the programs and data stored in the storage device 93. The explanation presentation program 11p and the abnormality diagnosis program 12p are each loaded into the memory 92 and executed by the processor 91, causing the processor 91 to function as the explanation presentation unit 11 or the abnormality diagnosis unit 12. The data stored in the explanation presentation model storage unit 14, the definition unit 15, and the diagnostic model storage unit 16 are stored in the storage device 93 as an explanation presentation model database 14d, a definition database 15d, and a diagnostic model database 16d, respectively.

[0025] The abnormality diagnosis unit 12 is a functional unit that performs abnormality diagnosis processing on the robot 20, and includes a diagnosis information input unit 121, a feature extraction unit 122, a diagnosis model execution unit 123, and an output unit 124 as sub-functional units.

[0026] The diagnostic information input unit 121 is an input unit that receives the vibration amount measured by the vibration sensor 24 of the robot 20, the target movement amount determined by the control unit 23, the actual movement amount measured by the image sensor 25, and the current value measured by the current sensor 26 as diagnostic information, and passes them on to the feature extraction unit.

[0027] The feature extraction unit 122 is a functional unit that performs extraction processing of features in a data format suitable for input to the diagnostic model execution unit 123 from the diagnostic information received from the diagnostic information input unit 121. An example of the feature extraction processing is conversion to a data format that is highly correlated with anomalies so that the diagnostic model execution unit 123 can realize highly accurate anomaly diagnosis, and processing such as applying frequency conversion to a current sensor to extract harmonic components can be considered. In addition, the received data does not need to be modified.

[0028] The diagnostic model execution unit 123 is a functional unit that performs processing to read out a diagnostic model from the diagnostic model storage unit 16, input the features extracted by the feature extraction unit 122 to the diagnostic model read out from the diagnostic model storage unit 16, execute the diagnostic model, and obtain a diagnostic result as an output of the diagnostic model. The diagnostic result is output to the output unit 124 and an explanation presentation model execution unit 111 of the explanation presentation unit 11, which will be described later.

[0029] The diagnostic model stored in the diagnostic model storage unit 16 is, for example, a machine learning model such as a trained neural network, which has learned the relationship between features and diagnostic results based on past abnormality cases. The diagnostic result may be the presence or absence or the degree of an abnormality in the diagnostic target, or an estimated cause. The diagnostic result may also be a measure to be taken against the abnormality.

[0030] The output unit 124 outputs the diagnosis result received from the diagnostic model execution unit to the display unit 30. The display unit 30 is an example of the output device 95 in Fig. 9. In addition to the display unit 30, for example, the diagnosis result may be stored in a storage device or transmitted to an external processing terminal via a communication line, or, when an abnormality is detected in the robot 20, an alert sound may be emitted from an alarm device (not shown) or a notification light (not shown) may be flashed.

[0031] The explanation presentation unit 11 is a functional unit that performs processing to present an explanation of the diagnosis reason of the abnormality diagnosing unit 12, and includes an explanation presentation model executing unit 111 and an output unit 112 as sub-functional units.

[0032] The explanation presentation model executing unit 111 is a process for executing steps S201 to S204 shown in FIG.

[0033] S201: Read out diagnostic knowledge from the definition unit 15. Here, the definition unit 15 is a part that holds diagnostic knowledge consisting of a plurality of diagnostic knowledge items, and a diagnostic knowledge item is, for example, relationship information between a check item and a diagnostic result when an operator manually performs a diagnosis.

[0034] 3A to 3C are examples of such relationship information, and the diagnosis results when the check results of check item 301 in Fig. 3A, check item 302 in Fig. 3B, and check item 303 in Fig. 3C are YES are respectively diagnosis result 305 in Fig. 3A, diagnosis result 306 in Fig. 3B, and diagnosis result 307 in Fig. 3C, which are connected by arrows. For example, relationship information consisting of the pair of check item 301 and diagnosis result 305 in Fig. 3A means a diagnostic knowledge item that reads "If abnormal vibration is detected by vibration sensor 24 of robot 20, contact the robot manufacturer."

[0035] S202: An explanation presentation model is read from the explanation presentation model storage unit 14. Here, the explanation presentation model storage unit 14 holds, as an explanation presentation model, a machine learning model that has learned the correspondence between processing information reflecting the diagnostic process by the diagnostic model execution unit 123 from past abnormality cases and diagnostic knowledge items included in the diagnostic knowledge. As an example of the processing information, when the diagnostic model executed by the diagnostic model execution unit 123 is a neural network, vector information of an intermediate layer when the neural network is executed can be considered as the processing information. As an example of a diagnostic knowledge item that can be associated with the processing information, for example, the pair of the check item 301 and the diagnostic result 305 in FIG. 3A can be given.

[0036] S203: Processing information on the diagnosis to be presented as an explanation is acquired from the diagnostic model execution unit 123, and input to the explanation presentation model execution unit 111, which executes the processing information and obtains diagnostic knowledge items corresponding to the processing information as an output. The obtained diagnostic knowledge items reflect the processing information of the diagnostic model, and can be regarded as the diagnosis reason of the diagnostic model.

[0037] S204: The diagnostic knowledge items obtained in S203 and the diagnostic knowledge read from the definition unit 15 are combined to form an explanation, which is then transferred to the output unit 112.

[0038] The output unit 112 outputs the explanation received from the explanation presentation model executing unit 111 as a diagnosis result to the display unit 30. In addition to the display unit 30, the explanation may be stored in a storage device or transmitted to an external processing terminal via a communication line, for example.

[0039] Fig. 4 shows an example of an explanation presentation model output as an explanation presentation screen 400 on the display unit 30. As relationship information of diagnostic knowledge, a pair of check item 301 and diagnostic result 305 shown in Fig. 3A, a pair of check item 302 and diagnostic result 306 shown in Fig. 3B, and a pair of check item 303 and diagnostic result 307 shown in Fig. 3C are displayed, and pairs corresponding to diagnostic knowledge items that are diagnosis reasons of the diagnostic model are indicated by being enclosed in a frame 31.

[0040] FIG. 5 shows an example of a flowchart of the process according to this embodiment.

[0041] First, in the abnormality diagnosis process 500 executed by the abnormality diagnosis unit 12, when the operation starts, data from the vibration sensor 24, image sensor 25, current sensor 26, and control unit 23 of the robot 20 are input to the diagnosis information input unit 121 (S501). The feature extraction unit 122 extracts features from the data input to the diagnosis information input unit 121 (S502).

[0042] Next, in the diagnostic model execution unit 123, the features extracted by the feature extraction unit 122 are input into the diagnostic model read from the diagnostic model storage unit 16 to estimate a diagnostic result (S503), which is then output from the output unit 124 to the display unit 30 and displayed on the screen of the display unit 30 (S504), thereby completing the processing.

[0043] In this case, the screen of the display unit 30 displays, as the estimated diagnosis results, 301, 302, and 303, one of which is surrounded by a frame 31, without displaying 305, 306, and 307 in Figure 4 and the arrows connecting them.

[0044] On the other hand, in the explanation presentation process 510 executed by the explanation presentation unit 11, when operation is started, the explanation presentation model execution unit 111 acquires information on the diagnostic result estimated by the diagnostic model execution unit 123, inputs it into the explanation presentation model read out from the explanation presentation model storage unit 14, and estimates the corresponding diagnostic knowledge item (S511).

[0045] The output unit 112 outputs the diagnostic knowledge items estimated in the explanation presentation model execution unit 111 together with the diagnostic knowledge read out from the definition unit 15 to the display unit 30, and displays them on the screen of the display unit 30 in a format such as that described in Figure 4 (S512), and then ends the processing.

[0046] The diagnosis results are displayed as information associating the diagnosis results with the check items that an operator would check when performing a manual diagnosis, as explained in Figure 4. In other words, by presenting an explanation of the reason for the diagnosis in association with the operator's existing diagnostic knowledge, the cause of the abnormality and the countermeasures to be taken can be associated with the information, making it possible to present an explanation that is easy for the operator to understand.

[0047] As described above, according to this embodiment, an explanation of the reason for diagnosis of an abnormality in an object to be diagnosed by the abnormality diagnosis unit can be presented in association with the operator's existing diagnostic knowledge, making it possible to present an explanation that is easy for the operator to understand. EXAMPLES

[0048] As a second embodiment of the present invention, similarly to the first embodiment, an anomaly diagnosis device will be described that takes an anomaly diagnosis of a robot as an example and can present an explanation linked to diagnostic knowledge including the combination and order of multiple check items by associating the diagnostic reason of the diagnostic model with a diagnostic flow that represents the procedure for a worker to manually diagnose the diagnosis.

[0049] The configuration of the abnormality diagnostic device 10 according to this embodiment is the same as that described in the first embodiment with reference to FIG. 1, and therefore a description thereof will be omitted.

[0050] In the first embodiment, as an example of the relationship information held by the definition unit 15, information as shown in Figures 3A to 3C in which the causes of anomalies are individually associated with their countermeasures is used, and an example is shown in which information as shown in the explanation presentation screen 400 in Figure 4 is displayed on the screen as information associating check items and diagnosis results when an operator manually performs a diagnosis. In this embodiment, an example is described in which an abnormality diagnosis flow 601 as shown in Figure 6 is displayed on the screen, and the causes of anomalies and their countermeasures are displayed on this abnormality diagnosis flow 601 along the flow of the abnormality diagnosis process.

[0051] That is, in this embodiment, an example of a diagnostic knowledge item associated with processing information in an explanation presentation model stored in the explanation presentation model storage unit 14 is a diagnostic path 602 on an abnormality diagnosis flow 601 when performed manually by an operator as shown in the explanation display screen 600 in Fig. 6, which is a diagnostic knowledge item expressed as a combination taking into consideration the order of check items when an operator performs a manual diagnosis. That is, it is a diagnostic flow that branches depending on the check items and the check results and leads to a diagnostic result.

[0052] In this embodiment, the information output from the output unit 112 and displayed on the display unit 30 is displayed in accordance with the flow of the abnormality diagnosis process, such as a diagnostic path 602, as shown in the explanatory display screen 600 in Figure 6, on an abnormality diagnosis flow 601 that is configured to match the order of check items when an operator manually performs a diagnosis. The abnormality detected by the abnormality diagnosis unit 12 and measures associated with the cause of the abnormality by the explanation presentation unit 11 are displayed as abnormality diagnosis results.

[0053] According to this embodiment, in addition to the effects described in the first embodiment, the diagnosis result can be easily understood because the diagnosis flow used for manual diagnosis is used to explain the reason for diagnosis. Also, the diagnosis result can be easily interpreted because the diagnosis reason can be explained by linking it with the diagnosis knowledge including the combination and order of multiple check items on the abnormality diagnosis processing flow. EXAMPLES

[0054] As a third embodiment of the present invention, an abnormality diagnosis device that can create an explanation presentation model by learning from an abnormality history when no explanation presentation model exists in advance will be described.

[0055] The configuration of an abnormality diagnosis device 100 according to this embodiment is shown in Fig. 7. In Fig. 7, the same components as those of the abnormality diagnosis device 10 in Fig. 1 described in the first embodiment are given the same numbers and their description is omitted. The abnormality diagnosis device 100 shown in Fig. 7 is also the same as the configuration shown in Fig. 1 of the first embodiment in that it receives signals from various sensors attached to the robot 20 that is the target of abnormality diagnosis, so in Fig. 7, the display of the configuration of the robot 20 shown in Fig. 1 is omitted.

[0056] The abnormality diagnosis device 100 is also realized by an information processing device 90 as shown in Fig. 9, and Fig. 11 shows the programs and data stored in the storage device 93. The explanation presentation program 110p, the abnormality diagnosis program 120p, and the learning program 60p are each loaded into the memory 92 and executed by the processor 91, causing the processor 91 to function as the explanation presentation unit 110, the abnormality diagnosis unit 120, or the learning unit 60. The data stored in the explanation presentation model storage unit 14, the definition unit 15, the diagnostic model storage unit 16, and the abnormality history storage unit 63 are stored in the storage device 93 as an explanation presentation model database 14d, a definition database 15d, a diagnostic model database 16d, and an abnormality history database 63d, respectively.

[0057] The abnormality diagnosis device 100 according to the present embodiment shown in FIG. 7 differs from the abnormality diagnosis device 10 in the first embodiment shown in FIG. 1 in that a learning unit 60 and an abnormality history storage unit 63 are provided.

[0058] The abnormality history storage unit 63 is a storage unit for holding past abnormality cases, and stores past diagnostic information and the corresponding diagnostic results.

[0059] The configuration of the abnormality diagnosis unit 120 is the same as that of the abnormality diagnosis unit 12 described in the first embodiment, but some of the processing is different. Only the different processing will be described below.

[0060] The diagnostic information input unit 1201 is an input unit that reads past diagnostic information stored in the abnormality history storage unit 63 and passes it to the feature amount extraction unit 1202 .

[0061] The process of the diagnostic model execution unit 1203 is the same as that in the first embodiment, but the process information is sent to the explanation presentation model learning unit 62 of the learning unit 60.

[0062] The learning unit 60 is a functional unit that learns an explanation presentation model and executes the process of storing this learned explanation presentation model in the explanation presentation model storage unit 14, and has a diagnostic knowledge item acquisition unit 61 and an explanation presentation model learning unit 62 as sub-functional units.

[0063] The diagnostic knowledge item acquisition unit 61 reads out diagnostic knowledge from the definition unit 15, and based on the read out diagnostic knowledge, acquires diagnostic knowledge items corresponding to past diagnostic information and diagnostic results stored in the abnormality history storage unit 63. For example, when the diagnostic knowledge read out from the definition unit 15 is in the form of an abnormality diagnosis flow 601 in Fig. 6 and the past diagnostic result is "subject to regular maintenance", the diagnostic knowledge item is obtained as a diagnostic path 602 connecting "gripping error occurrence", which is the route on the diagnostic flow, with the diagnostic result "subject to regular maintenance".

[0064] The explanation presentation model learning unit 62 executes a process of creating a machine learning model as an explanation presentation model and storing it in the explanation presentation model storage unit 14. The machine learning model is trained so as to receive processing information from the diagnostic model executing unit 1203 for a certain past case as an input and output diagnostic knowledge items acquired by the diagnostic knowledge item acquiring unit 61 for the same past case.

[0065] Fig. 8 shows an example of a flowchart corresponding to the flow of the learning process in the learning unit 60 of this embodiment. This learning process is performed prior to the abnormality diagnosis process 500 and the explanation presentation process 510 described in the first embodiment with reference to Fig. 5. Therefore, the process flow shown in Fig. 8 is executed in a state where the diagnosis information input unit 1201 does not receive input of signals from the various sensors and control unit attached to the robot 20.

[0066] First, the diagnostic information input unit 1201 of the abnormality diagnosis unit 120 receives diagnostic information from the abnormality history storage unit 63 (S801). Next, the diagnostic knowledge item acquisition unit 61 receives a diagnostic result linked to the diagnostic information from the abnormality history storage unit 63, and acquires a corresponding diagnostic knowledge item by referring to the diagnostic knowledge of the definition unit 15 (S802).

[0067] Next, the explanation presentation model learning unit 62 receives processing information processed by the diagnostic model execution unit 1203 based on the features extracted by the feature extraction unit 1202 from the diagnostic information received by the diagnostic information input unit 1201, and diagnostic knowledge items acquired by the diagnostic knowledge item acquisition unit 61 in correspondence with the diagnostic information received by the diagnostic information input unit 1201 (S803).

[0068] It is checked whether all the abnormality histories stored in the abnormality history storage unit 63 have been referred to (S804), and if all the abnormality histories have not been referred to yet (NO in S804), the process returns to S801. On the other hand, if all the abnormality histories have been referred to (YES in S804), the explanation presentation model learning unit 62 learns a machine learning model that estimates diagnostic knowledge items from diagnostic information based on the received pair information of processing information and diagnostic knowledge items, stores it in the explanation presentation model storage unit 14 (S805), and ends the process.

[0069] After the learning process described in FIG. 8 is completed, signals from the various sensors 24 to 26 of the robot 20 and the control unit 23 are input to the diagnosis information input unit 1201, and then the abnormality diagnosis process 500 and explanation presentation process 510 described in the first embodiment using FIG. 5 are executed to perform an abnormality diagnosis of the robot 20.

[0070] According to this embodiment, even if an explanatory presentation model is not stored in advance in the explanatory presentation model storage unit 14, the learning unit 60 can learn an explanatory presentation model by referring to the abnormality history stored in the abnormality history storage unit 63, and store it in the explanatory presentation model storage unit 14.

[0071] According to this embodiment, in addition to the effects described in Examples 1 and 2, an explanation presentation model can be created by machine learning even if an explanation presentation model has not been created in advance, thereby reducing the effort required for creating an explanation presentation model.

[0072] The present invention is not limited to the above-described embodiments, but includes various modified examples. For example, the above-described embodiments have been described in detail to clearly explain the present invention, and the present invention is not necessarily limited to those including all of the configurations described. It is also possible to replace a 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. It is also possible to add, delete, or replace a part of the configuration of each embodiment with another configuration. [Explanation of symbols]

[0073] 10,100 ... abnormality diagnosis device, 11,110 ... explanation presentation unit, 12,120 ... abnormality diagnosis unit, 14 ... explanation presentation model storage unit, 15 ... definition unit, 16 ... diagnostic model storage unit, 30 ... display unit, 60 ... learning unit, 61 ... diagnostic knowledge item acquisition unit, 62 ... explanation presentation model learning unit, 63 ... abnormality history storage unit, 111, 1101 ... explanation presentation model execution unit, 121, 1201 ... diagnostic information input unit, 122, 1202 ... feature extraction unit, 123, 1203 ... diagnostic model execution unit.

Claims

1. a diagnostic model storage unit that stores, as a diagnostic model, a machine learning model that has learned the relationship between the feature amount of data, which is diagnostic information regarding a diagnostic target, and a diagnostic result; an abnormality diagnosis unit that estimates a diagnosis result of the diagnosis target by using a feature amount extracted from the data that is the diagnosis information and the diagnosis model during an abnormality diagnosis; a definition section for holding a plurality of diagnostic knowledge items; an explanation presentation model storage unit that stores, as an explanation presentation model, a machine learning model that has learned a relationship between processing information that reflects a diagnostic process in which the diagnostic model is executed for past abnormality cases and the diagnostic knowledge items; an explanation presentation unit that estimates diagnostic knowledge items related to the diagnosis result output by the abnormality diagnosis unit using the explanation presentation model and processing information when the abnormality diagnosis unit executes the diagnostic model on the diagnosis information; a display unit that displays on a screen the diagnosis result of the diagnosis target estimated by the abnormality diagnosis unit and the diagnostic knowledge items estimated by the explanation presentation unit; Equipped with the diagnostic model stored in the diagnostic model storage unit is a neural network; The abnormality diagnosis device according to claim 1, wherein the processing information is vector information of an intermediate layer obtained when the neural network is executed on the diagnosis information.

2. The abnormality diagnosis device according to claim 1, An abnormality diagnosis device, wherein the diagnosis result is the presence or absence, the degree of abnormality, or the cause of an abnormality in the object to be diagnosed, or a measure to be taken.

3. The abnormality diagnosis device according to claim 1, The diagnostic knowledge items stored in the definition unit are relationship information between check items when an operator manually diagnoses the diagnostic object and diagnostic results of the diagnostic object.

4. The abnormality diagnosis device according to claim 1, an abnormality history storage unit that stores past diagnostic information and corresponding diagnostic results regarding the diagnostic object; a diagnostic knowledge item acquisition unit that acquires diagnostic knowledge items for the past diagnostic information stored in the abnormality history storage unit from the definition unit; a learning unit that learns the explanation presentation model by using processing information when the abnormality diagnosis unit executes the diagnostic model on the past diagnostic information and the diagnostic knowledge items acquired by the diagnostic knowledge item acquisition unit; The abnormality diagnosis device further comprising:

5. The abnormality diagnosis device according to claim 1, The abnormality diagnosis device, wherein the display unit displays, as the diagnostic knowledge items, a process flow for when an operator manually diagnoses the diagnostic target on a screen.

6. An abnormality diagnosis method for diagnosing an abnormality in a diagnosis target using an abnormality diagnosis device, comprising: The abnormality diagnosis device includes a diagnosis model storage unit, a definition unit, an explanation presentation model storage unit, an abnormality diagnosis unit, an explanation presentation unit, and a display unit, the diagnostic model storage unit stores, as a diagnostic model, a machine learning model that has learned a relationship between a feature amount of data, which is diagnostic information regarding the diagnostic target, and a diagnostic result; The definition unit holds a plurality of diagnostic knowledge items, the explanation presentation model storage unit stores, as an explanation presentation model, a machine learning model that has learned a relationship between processing information reflecting a diagnostic process in which the diagnostic model was executed in a past abnormality case and the diagnostic knowledge items, the abnormality diagnosis unit estimates a diagnosis result of the diagnosis target by using a feature amount extracted from the data that is the diagnosis information and the diagnosis model during abnormality diagnosis; the explanation presentation unit estimates diagnostic knowledge items related to the diagnosis result output by the abnormality diagnosis unit using processing information when the abnormality diagnosis unit executes the diagnostic model on the diagnostic information and the explanation presentation model; the display unit displays on a screen the diagnosis result of the diagnosis object estimated by the abnormality diagnosis unit and the diagnostic knowledge items estimated by the explanation presentation unit together; the diagnostic model stored in the diagnostic model storage unit is a neural network; The abnormality diagnosis method according to claim 1, wherein the processing information is vector information of an intermediate layer obtained when the neural network is executed on the diagnosis information.

7. The abnormality diagnosis method according to claim 6, The abnormality diagnosis device further includes an abnormality history storage unit, a diagnostic knowledge item acquisition unit, and a learning unit, the abnormality history storage unit stores past diagnostic information related to the diagnostic object and corresponding diagnostic results, the diagnostic knowledge item acquisition unit acquires diagnostic knowledge items for the past diagnostic information stored in the abnormality history storage unit from the definition unit, the learning unit learns the explanation presentation model by using processing information when the abnormality diagnosis unit executes the diagnostic model on the past diagnostic information and the diagnostic knowledge items acquired by the diagnostic knowledge item acquisition unit.

13. An abnormality diagnosis method comprising:

8. The abnormality diagnosis method according to claim 6, The abnormality diagnosis method, wherein the display unit displays, as the diagnostic knowledge items, a process flow for when an operator manually diagnoses the diagnostic target on a screen.

Images