JP6822242B2 - Diagnostic equipment, diagnostic systems, diagnostic methods and programs - Google Patents

Description

The present invention relates to diagnostic devices, diagnostic systems, diagnostic methods and programs.

In machines such as image forming devices and machine tools, a technique for determining a machine abnormality or the like by using detection information for detecting a physical quantity such as sound or vibration and detection information prepared in advance is already known. ing.

As a technique for detecting such a physical quantity and determining a machine abnormality, etc., a machine abnormality is detected by using the operation sound data collected by the sound collecting means and the normal sound data prepared in advance. A technique for detecting is disclosed (Patent Document 1).

However, in the technique described in Patent Document 1, it is necessary to manually label the data to be prepared in advance, for example, normal or abnormal, and for a large amount of data. There was a problem that it was not realistic in terms of time to label manually.

The present invention has been made in view of the above problems, and is a diagnostic device, a diagnostic system, a diagnostic method, and a program capable of easily adding a corresponding label to the training data to be prepared in advance. The purpose is to provide.

In order to solve the above-mentioned problems and achieve the object, the present invention relates to a first acquisition unit that acquires predetermined information corresponding to the current operation of the target device from the target device, and a member of the target device. An extraction that extracts a second acquisition unit that acquires the first detection information output from the detection unit that detects a physical quantity as learning data, and a first feature information that indicates the characteristics of the learning data acquired by the second acquisition unit. The unit and the duration from the start of use of the member in a new state to the inability to use the member by executing the operation are divided into a plurality of predetermined sections, and the learning data corresponding to each section is divided. Then, based on the first addition part to which the label information indicating the state of the member is added, the label information added to the learning data corresponding to each section, and the first feature information. The model for determining which of the label information the second detection information for which the label information acquired by the second acquisition unit is unknown corresponds to, and the predetermined information corresponding to the training data are specified as a model. It includes a generation unit that is associated and generated as information.

According to the present invention, a corresponding label can be easily added to the learning data to be prepared in advance.

FIG. 1 is a diagram showing an example of the overall configuration of the diagnostic system according to the first embodiment. FIG. 2 is a diagram showing an example of the hardware configuration of the processing machine of the first embodiment. FIG. 3 is a diagram showing an example of the hardware configuration of the diagnostic apparatus according to the first embodiment. FIG. 4 is a diagram showing an example of the configuration of the functional block of the diagnostic system according to the first embodiment. FIG. 5 is a diagram for explaining that label information is added to the learning data at intervals of equal intervals in time. FIG. 6 is a flowchart showing an example of the model generation process according to the first embodiment. FIG. 7 is a flowchart showing an example of the state determination process according to the first embodiment. FIG. 8 is a diagram illustrating a specific example of the model generation process and the state determination process in the first embodiment. FIG. 9 is a diagram illustrating an example in which a common model is used in another processing process. FIG. 10 is a diagram showing an example of the configuration of the functional block of the diagnostic system according to the second embodiment. FIG. 11 is a diagram illustrating that label information is added based on the relationship between the tool usage time and the degree of deterioration. FIG. 12 is a flowchart showing an example of the model generation process in the second embodiment. FIG. 13 is a diagram showing an example of the configuration of the functional block of the diagnostic system according to the third embodiment. FIG. 14 is a graph showing an example of the relationship between the usage time of a plurality of tools and the degree of deterioration. FIG. 15 is a diagram illustrating that label information is added based on the relationship between the tool usage time and the degree of deterioration. FIG. 16 is a flowchart showing an example of the model generation process according to the third embodiment.

Hereinafter, embodiments of a diagnostic apparatus, a diagnostic system, a diagnostic method, and a program according to the present invention will be described in detail with reference to FIGS. 1 to 16. Further, the present invention is not limited by the following embodiments, and the components in the following embodiments include those easily conceived by those skilled in the art, substantially the same, and so-called equivalent ranges. Is included. Furthermore, various omissions, substitutions, changes and combinations of components can be made without departing from the gist of the following embodiments.

[First Embodiment]
(Overall configuration of diagnostic system)
FIG. 1 is a diagram showing an example of the overall configuration of the diagnostic system according to the first embodiment. The overall configuration of the diagnostic system 1 according to the present embodiment will be described with reference to FIG.

As shown in FIG. 1, the diagnostic system 1 according to the present embodiment includes a sensor 57 installed in the processing machine 200 and a diagnostic device 100. The processing machine 200 is connected so as to be able to communicate with the diagnostic device 100. Although FIG. 1 shows an example in which one processing machine 200 is connected to the diagnostic device 100, the present invention is not limited to this, and a plurality of processing machines 200 are connected to the diagnostic device 100. They may be connected so that they can communicate with each other.

The processing machine 200 is a machine tool that uses a tool to perform processing such as cutting, grinding, or polishing on a processing object. The processing machine 200 is an example of a target device to be diagnosed by the diagnostic device 100. The target device is not limited to the processing machine 200, and may be any machine that can be the target of diagnosis. For example, a machine such as an assembly machine, a measuring machine, an inspection machine, or a washing machine is the target device. May be good. Hereinafter, the processing machine 200 will be described as an example of the target device.

The diagnostic device 100 is a device that is connected to the processing machine 200 so as to be communicable and diagnoses the operation of the processing machine 200.

The sensor 57 is a vibration or sound generated when a tool such as a drill, an end mill, a bite tip or a grindstone installed in the processing machine 200 and a processing target come into contact with each other during the processing operation, or the tool or the processing machine 200 itself. It is a sensor that detects a physical quantity such as vibration or sound generated and outputs the information of the detected physical quantity as detection information (sensor data) to the diagnostic apparatus 100. The sensor 57 is composed of, for example, a microphone, a vibration sensor, an acceleration sensor, an AE sensor, or the like, and is installed in the vicinity of a tool capable of detecting, for example, vibration or sound.

The processing machine 200 and the diagnostic apparatus 100 may be connected in any connection form. For example, the processing machine 200 and the diagnostic apparatus 100 may be connected by a dedicated connection line, a wired network such as a wired LAN (Local Area Network), a wireless network, or the like.

Further, the number of sensors 57 may be arbitrary. Further, a plurality of sensors 57 for detecting the same physical quantity may be provided, or a plurality of sensors 57 for detecting different physical quantities may be provided.

Further, the sensor 57 may be provided in advance in the processing machine 200, or may be attached to the processing machine 200 which is a completed machine later.

(Hardware configuration of processing machine)
FIG. 2 is a diagram showing an example of the hardware configuration of the processing machine of the first embodiment. The hardware configuration of the processing machine 200 of the present embodiment will be described with reference to FIG.

As shown in FIG. 2, the processing machine 200 is driven by a CPU (Central Processing Unit) 51, a ROM (Read Only Memory) 52, a RAM (Random Access Memory) 53, a communication I / F (interface) 54, and the like. The control circuit 55 and the control circuit 55 are connected to each other so as to be communicable by the bus 58. The sensor 57 is communicably connected to the diagnostic device 100.

The CPU 51 is an arithmetic unit that controls the entire processing machine 200. For example, the CPU 51 controls the operation of the entire processing machine 200 and realizes a processing function by executing a program stored in the ROM 52 or the like with the RAM 53 as a work area (work area).

The communication I / F 54 is an interface for communicating with an external device such as the diagnostic device 100. The drive control circuit 55 is a circuit that controls the drive of the motor 56. The motor 56 is a motor that drives a tool used for machining a drill, an end mill, a tool tip, a grindstone, or the like, and a table or the like on which a machining target is placed and moved according to the machining. The sensor 57 is as described above.

(Hardware configuration of diagnostic device)
FIG. 3 is a diagram showing an example of the hardware configuration of the diagnostic apparatus according to the first embodiment. The hardware configuration of the diagnostic apparatus 100 according to the present embodiment will be described with reference to FIG.

As shown in FIG. 3, the diagnostic device 100 includes a CPU 61, a ROM 62, a RAM 63, a communication I / F 64, a sensor I / F 65, an auxiliary storage device 66, an input device 67, and a display 68. It is configured so that it can be communicated with 69.

The CPU 61 is an arithmetic unit that controls the entire diagnostic device 100. The CPU 61 controls the operation of the entire diagnostic device 100 and realizes a diagnostic function by executing a program stored in the ROM 62 or the like with the RAM 63 as a work area (work area), for example.

The communication I / F64 is an interface for communicating with an external device such as a processing machine 200. The communication I / F64 is, for example, a NIC (Network Interface Card) corresponding to TCP (Transmission Protocol) / IP (Internet Protocol).

The sensor I / F65 is an interface for receiving detection information from the sensor 57 installed in the processing machine 200.

The auxiliary storage device 66 includes setting information of the diagnostic device 100, detection information and context information received from the processing machine 200, a model generated by the diagnostic device 100 described later, an OS (Operating System), various data such as an application program, and the like. It is a non-volatile storage device such as an HDD (Hard Disk Drive), an SSD (Solid State Drive), or an EEPROM (Electrically Erasable Program Read-Only Memory) that stores the data. The auxiliary storage device 66 is provided in the diagnostic device 100, but is not limited to this, and may be, for example, a storage device installed outside the diagnostic device 100, or the diagnostic device. It may be a storage device provided with a server device capable of data communication with 100.

The input device 67 is an input device such as a mouse or keyboard for inputting characters and numbers, selecting various instructions, and performing operations such as moving a cursor.

The display 68 is a display device such as a CRT (Cathode Ray Tube) display, an LCD (Liquid Crystal Display), or an organic EL (Electro-Luminesse) display that displays characters, numbers, various screens, operation icons, and the like.

The hardware configuration shown in FIG. 3 is an example, and it is not necessary to include all the constituent devices, or it may be provided with other constituent devices. For example, when the diagnostic device 100 specializes in the diagnostic operation of the processing machine 200 and transmits the diagnostic result to an external server device or the like, the input device 67 and the display 68 may not be provided.

(Configuration and operation of functional blocks of diagnostic system)
FIG. 4 is a diagram showing an example of the configuration of the functional block of the diagnostic system according to the first embodiment. FIG. 5 is a diagram for explaining that label information is added to the learning data at intervals of equal intervals in time. The configuration and operation of the functional blocks of the diagnostic system 1 and the processing machine 200 according to the present embodiment will be described with reference to FIGS. 4 and 5.

As shown in FIG. 4, the processing machine 200 includes a numerical control unit 201, a communication control unit 202, a drive control unit 203, a drive unit 204, and a detection unit 211.

The numerical control unit 201 is a functional unit that executes machining by the drive unit 204 by numerical control (NC: Numerical Control). For example, the numerical control unit 201 generates and outputs numerical control data for controlling the operation of the drive unit 204. Further, the numerical control unit 201 outputs the context information to the communication control unit 202. Here, the context information is a plurality of information defined for each type of operation of the processing machine 200. The context information includes, for example, identification information of a machine tool (machining machine 200), identification information of a drive unit 204 (for example, tool identification information, etc.), a diameter of a tool driven by the drive unit 204, a tool material, and the like. Configuration information, operating status of the tool driven by the drive unit 204, cumulative usage time from the start of use of the drive unit 204, load related to the drive unit 204, rotation speed of the drive unit 204, machining speed of the drive unit 204. This is information indicating information on processing conditions such as.

For example, the numerical control unit 201 sequentially transmits the context information corresponding to the operation of the current processing machine 200 to the diagnostic apparatus 100 via the communication control unit 202. When machining an object to be machined, the numerical control unit 201 changes the type of the drive unit 204 to be driven or the drive state (rotation speed, rotation speed, etc.) of the drive unit 204 according to the machining process. Each time the numerical control unit 201 changes the operation type, the numerical control unit 201 sequentially transmits the context information corresponding to the changed operation type to the diagnostic apparatus 100 via the communication control unit 202. The numerical control unit 201 is realized by, for example, a program running on the CPU 51 shown in FIG.

The communication control unit 202 is a functional unit that controls communication with an external device such as the diagnostic device 100. For example, the communication control unit 202 transmits the context information corresponding to the current operation to the diagnostic device 100. The communication control unit 202 is realized by, for example, a communication I / F 54 shown in FIG. 2 and a program running on the CPU 51.

The drive control unit 203 is a functional unit that drives and controls the drive unit 204 based on the numerical control data obtained by the numerical control unit 201. The drive control unit 203 is realized by, for example, the drive control circuit 55 shown in FIG.

The drive unit 204 is a functional unit that is the target of drive control by the drive control unit 203. The drive unit 204 drives the tool under the control of the drive control unit 203. The drive unit 204 is an actuator that is driven and controlled by the drive control unit 203, and is realized by, for example, the motor 56 shown in FIG. The drive unit 204 may be any actuator as long as it is used for machining and is subject to numerical control. Further, two or more drive units 204 may be provided.

The detection unit 211 is a vibration or sound generated when a tool such as a drill, an end mill, a tool tip or a grindstone installed in the processing machine 200 and a processing target come into contact with each other during the processing operation, or the tool or the processing machine 200 itself. This is a functional unit that detects a physical quantity such as vibration or sound generated by the machine and outputs the detected physical quantity information to the diagnostic apparatus 100 as detection information (sensor data). The detection unit 211 is realized by, for example, the sensor 57 shown in FIG. The number of detection units 211 is arbitrary. For example, a plurality of detection units 211 that detect the same physical quantity may be provided, or a plurality of detection units 211 that detect different physical quantities may be provided. For example, when the blade, which is a tool used for machining, breaks or chipping of the blade occurs, the sound during machining changes. Therefore, by detecting the acoustic data with the detection unit 211 (microphone) and making a judgment using a model or the like for judging the normal sound, it is possible to detect an abnormality in the operation of the processing machine 200.

The numerical control unit 201 and the communication control unit 202 shown in FIG. 4 may be realized by causing the CPU 51 shown in FIG. 2 to execute a program, that is, by software, or by hardware such as an IC (Integrated Circuit). It may be realized by using software and hardware together.

As shown in FIG. 4, the diagnostic device 100 of the diagnostic system 1 includes a communication control unit 101, a detection information receiving unit 102 (second acquisition unit), a processing information acquisition unit 103 (first acquisition unit), and a reception unit. 104, feature extraction unit 105, label addition unit 106 (first addition unit), generation unit 107, determination unit 108, storage unit 109, input unit 110, display control unit 111, and display unit 112. And have.

The communication control unit 101 is a functional unit that controls communication with the processing machine 200. For example, the communication control unit 101 receives context information from the numerical control unit 201 of the processing machine 200 via the communication control unit 202. The communication control unit 101 is realized by, for example, the communication I / F64 shown in FIG. 3 and a program operated by the CPU 61.

The detection information receiving unit 102 is a functional unit that receives detection information from the detection unit 211 installed in the processing machine 200. The detection information receiving unit 102 is realized by, for example, a program that operates on the sensor I / F65 and the CPU 61 shown in FIG.

Here, FIG. 5 shows an example of the waveform of the detection information received by the detection information receiving unit 102 and used as the learning data for generating the model described later. FIG. 5 shows detection information from the timing when a newly replaced new tool is started to be used to the timing when the tool is broken due to deterioration such as wear of the tool due to repeated machining processes. The detection information shown in FIG. 5 includes, for example, waveform data in a machining section in which machining processing of a machining target is executed by a tool, and waveform data in a non-machining section in which the tool does not hit the machining target.

The machining information acquisition unit 103 is a functional unit that acquires context information (machining information) received by the communication control unit 101 from the machining machine 200. The processing information acquisition unit 103 is realized by, for example, a program running on the CPU 61 shown in FIG.

The reception unit 104 is a functional unit that receives input of context information different from the context information received from the processing machine 200 by the communication control unit 101. For example, the cumulative usage time can be configured to be obtained from the processing machine 200. In this case, the processing machine 200 may have, for example, a function of resetting (initializing) the cumulative usage time when the tool is replaced.

It is also possible to configure the reception unit 104 to accept the cumulative usage time without acquiring it from the processing machine 200. The reception unit 104 receives the context information input from the input unit 110 realized by, for example, a keyboard and a touch panel. The context information received by the reception unit 104 is not limited to the cumulative usage time, for example, information on the specifications of the tool to be used (tool diameter, number of blades, material, whether or not the tool is coated, etc.), machining target. Information (material, etc.) may be used. Further, the reception unit 104 may be configured to receive context information from an external device. The reception unit 104 is realized by, for example, a program that operates on the CPU 61 shown in FIG. If it is not necessary to receive context information from other than the processing machine 200, the reception unit 104 may not be provided.

The feature extraction unit 105 is a functional unit that extracts feature information used in model generation by the generation unit 107 and determination processing by the determination unit 108 from the detection information. The feature information may be any information as long as it indicates the features of the detection information. For example, the feature extraction unit 105 may extract energy, frequency spectrum, MFCC (mel frequency cepstrum coefficient), and the like as feature information from the detection information. The feature extraction unit 105 is realized by, for example, a program running on the CPU 61 shown in FIG.

The label adding unit 106 is a functional unit that adds (associates) predetermined label information with the detection information (first detection information) as learning data received by the detection information receiving unit 102. Specifically, in the learning data shown in FIG. 5 described above, for example, the label adding portion 106 shows that the tool is worn due to repeated machining processes from the timing when the newly replaced new tool is started to be used. The time until the timing of breakage due to deterioration (hereinafter, may be referred to as "duration") is equally divided into five sections, and the learning data corresponding to the section with the oldest time is in order, "deterioration degree 1", Label information such as "deterioration degree 2", "deterioration degree 3", "deterioration degree 4", and "deterioration degree 5" is added. That is, the label information shown in FIG. 5 is information indicating which of the five stages of deterioration the degree of deterioration of the tool being used is in each stage of the duration, and is "deterioration degree". The degree of deterioration of the tool increases from "1" to "degree of deterioration 5". In this way, the label adding unit 106 adds the label information by using the duration of the learning data. The label adding unit 106 is realized by, for example, a program running on the CPU 61 shown in FIG.

The duration is limited to the timing when the tool is broken, but it is not limited to the state where the tool is broken, and the tool cannot be used due to deformation, chipping, breakage, or chipping from the jig. It may be until the timing of becoming.

Further, the label addition unit 106 shows an example in which the duration is equally divided into five sections, but the present invention is not limited to this, and it goes without saying that the number of sections to be divided may be arbitrary.

Further, the label information added to the learning data by the label addition unit 106 is not limited to the degree of deterioration. For example, the state of a member such as a tool or device whose state is to be determined changes with the passage of time, such as a label for wear of a tool length (for example, "tool length", "wear degree"). In this case, a label indicating the state may be added as label information to the detection information that captures the change in the state of the member.

The generation unit 107 is a model for determining the degree of deterioration of the tool by using the label information added to the training data by the label addition unit 106 and the feature information extracted from the training data by the feature extraction unit 105. It is a functional part that generates. In this case, since label information is added to the training data for each section in the above-mentioned duration, the feature information (first feature information) extracted from the learning data is also included in the feature information corresponding to each section. Is associated with label information. Further, when the detection information receiving unit 102 receives the learning data, the generation unit 107 generates a model in association with the context information acquired by the processing information acquisition unit 103.

The generation of the model by the generation unit 107 is executed based on the so-called supervised learning of machine learning. As an algorithm for generating a model by supervised learning, for example, SVM (Support Vector Machine) or GMM (Gaussian Mixture Model) may be used. For example, in SVM, the feature information extracted from the learning data is used as training data (teacher data), the feature information corresponding to each label information is divided into classes, and the distance of the feature information included in each class is the maximum value (maximum margin). ), And find the parameters that specify the discriminant function that indicates the hyperplane that divides the classes. Obtaining such parameters corresponds to the generation of a model. That is, by using the feature information for which it is unknown which label information corresponds to and the model obtained by SVM, which class the feature information corresponds to, that is, which label information (in the above example, "deterioration") It becomes clear whether it corresponds to "degree 1" to "deterioration degree 5"). Specifically, the feature information is obtained by obtaining an output value indicating which label information is supported from the feature information whose label information is unknown and the model generated by the generation unit 107. It becomes clear which label information corresponds to.

The generation unit 107 is realized by, for example, a program running on the CPU 61 shown in FIG.

The determination unit 108 includes the feature information (second feature information) extracted from the detection information (second detection information) by the feature extraction unit 105 and the model corresponding to the context information acquired by the processing information acquisition unit 103. It is a functional unit that is used to determine which degree of deterioration (label information) the feature information corresponds to. Specifically, as described above, the determination unit 108 obtains, for example, an output value indicating which label information the feature information corresponds to from the feature information and the model, and which deterioration of the output value is obtained. Judge whether it corresponds to the degree (label information). Thereby, it can be determined that the tool used in the processing machine 200 is in the state of the degree of deterioration determined by the determination unit 108. The determination unit 108 is realized by, for example, a program running on the CPU 61 shown in FIG.

The storage unit 109 is a functional unit that stores the model generated by the generation unit 107 in association with the context information. Further, the storage unit 109 may store the detection information received by the detection information receiving unit 102 in association with the context information received by the processing information acquisition unit 103. The storage unit 109 is realized by, for example, the RAM 63 and the auxiliary storage device 66 of FIG.

The input unit 110 is a functional unit for inputting characters and numbers, selecting various instructions, moving the cursor, and the like. The input unit 110 is realized by the input device 67 shown in FIG.

The display control unit 111 is a functional unit that controls the display operation of the display unit 112. Specifically, the display control unit 111 causes the display unit 112 to display, for example, a determination result of the degree of deterioration of the tool by the determination unit 108. The display control unit 111 is realized by, for example, a program running on the CPU 61 shown in FIG.

The display unit 112 is a functional unit that displays various information according to the control by the display control unit 111. The display unit 112 is realized by, for example, the display 68 shown in FIG.

Each functional unit of the diagnostic apparatus 100 shown in FIG. 4 (communication control unit 101, detection information receiving unit 102, processing information acquisition unit 103, reception unit 104, feature extraction unit 105, label addition unit 106, generation unit 107, The determination unit 108 and the display control unit 111) have the CPU 61 of FIG. 3 execute the program, that is, they may be realized by software, hardware such as IC, or software and hardware. May be realized in combination with.

Further, each functional unit of the diagnostic apparatus 100 and the processing machine 200 shown in FIG. 4 conceptually shows the function, and is not limited to such a configuration. For example, a plurality of functional units shown as independent functional units in FIG. 4 may be configured as one functional unit. On the other hand, the function of one functional unit in FIG. 4 may be divided into a plurality of functions and configured as a plurality of functional units.

(Model generation processing by diagnostic equipment)
FIG. 6 is a flowchart showing an example of the model generation process according to the first embodiment. The flow of the model generation process of the diagnostic apparatus 100 of the diagnostic system 1 according to the present embodiment will be described with reference to FIG. It should be noted that the model generation process is, for example, executed in advance before the state determination process. Before executing the following model generation process, it is assumed that a new tool is attached to the processing machine 200.

<Step S101>
The communication control unit 101 of the diagnostic apparatus 100 receives the context information transmitted from the processing machine 200. Then, the processing information acquisition unit 103 of the diagnostic apparatus 100 acquires the context information received by the communication control unit 101. Then, the process proceeds to step S102.

<Step S102>
The detection information receiving unit 102 of the diagnostic apparatus 100 receives the detection information (sensor data) as learning data transmitted from the processing machine 200. The operation of receiving the learning data by the detection information receiving unit 102 continues from the timing when the newly replaced new tool is started to be used to the timing when the tool is broken due to deterioration such as wear of the tool due to repeated machining processes.

The context information and detection information (learning data) received in this way are used to generate the model. Since the model is generated for each context information, the detection information needs to be associated with the corresponding context information. Therefore, for example, the detection information receiving unit 102 may store the received detection information in the storage unit 109 or the like in association with the context information received by the communication control unit 101 at the same timing. In this case, the context information is used as model identification information for identifying the generated model.

When the learning data for the duration is received by the detection information receiving unit 102, the process proceeds to step S103.

<Step S103>
The feature extraction unit 105 of the diagnostic device 100 extracts feature information from the learning data for the duration received by the detection information receiving unit 102. Then, the process proceeds to step S104.

<Step S104>
The label adding unit 106 of the diagnostic apparatus 100 adds (associates) predetermined label information with the learning data received by the detection information receiving unit 102. Specifically, as in the learning data shown in FIG. 5 described above, for example, the label addition unit 106 divides the duration into five sections equally, and in order from the learning data corresponding to the section with the oldest time, " Label information such as "deterioration degree 1", "deterioration degree 2", "deterioration degree 3", "deterioration degree 4", and "deterioration degree 5" is added. In this case, since the label information is added to the training data for each section in the duration, the label information is associated with the feature information corresponding to each section of the feature information extracted from the learning data. It will be. Then, the process proceeds to step S105.

<Step S105>
The generation unit 107 of the diagnostic apparatus 100 determines the degree of deterioration of the tool by using the label information added to the learning data by the label addition unit 106 and the feature information extracted from the learning data by the feature extraction unit 105. Generate a model to do. In this case, the generation unit 107 generates a model in association with the context information acquired by the processing information acquisition unit 103 when the detection information reception unit 102 receives the learning data. As an algorithm for generating this model, for example, SVM or GMM of supervised learning is used. By using the model generated in this way, which label information corresponds to which class of characteristic information, that is, which label information (in the above example, "deterioration degree 1" to "deterioration") It becomes clear whether it corresponds to degree 5 "). Then, the process proceeds to step S106.

<Step S106>
The generation unit 107 associates the generated model with the context information (model specific information) acquired by the processing information acquisition unit 103, and stores the generated model in the storage unit 109.

The model generation process is executed by the above steps S101 to S106.

(Status judgment processing by diagnostic device)
FIG. 7 is a flowchart showing an example of the state determination process according to the first embodiment. The flow of the state determination process of the diagnostic apparatus 100 of the diagnostic system 1 according to the present embodiment will be described with reference to FIG. 7.

<Step S201>
The numerical control unit 201 of the processing machine 200 sequentially transmits the context information transmitted from the processing machine 200 to the diagnostic apparatus 100. The communication control unit 101 of the diagnostic apparatus 100 receives the context information transmitted from the processing machine 200 in this way. Then, the processing information acquisition unit 103 of the diagnostic apparatus 100 acquires the context information received by the communication control unit 101. Then, the process proceeds to step S202.

<Step S202>
Further, the detection unit 211 of the processing machine 200 sequentially outputs the detection information. The detection information receiving unit 102 of the diagnostic apparatus 100 receives the detection information (sensor data) transmitted from the processing machine 200 in this way. Then, the process proceeds to step S203.

<Step S203>
The feature extraction unit 105 of the diagnostic device 100 extracts the feature information from the detection information received by the detection information receiving unit 102. Then, the process proceeds to step S204.

<Step S204>
The determination unit 108 of the diagnostic apparatus 100 acquires a model corresponding to the context information (model identification information) acquired by the processing information acquisition unit 103 from the storage unit 109. Then, the process proceeds to step S205.

<Step S205>
The determination unit 108 uses the feature information extracted from the detection information by the feature extraction unit 105 and the model corresponding to the context information (model identification information) acquired by the processing information acquisition unit 103 to obtain the feature information. Determine which degree of deterioration (label information) corresponds to. Specifically, the determination unit 108 obtains, for example, an output value indicating which label information the feature information corresponds to from the feature information and the model, and which degree of deterioration (label information) the output value corresponds to. Determine if it corresponds to. Thereby, it can be determined that the tool used in the processing machine 200 is in the state of the degree of deterioration determined by the determination unit 108. Then, the process proceeds to step S206.

<Step S206>
The determination unit 108 outputs the determination result. Any method may be used to output the determination result. The determination unit 108 may cause the display control unit 111 of the diagnostic apparatus 100 to display the determination result on the display unit 112, for example. Alternatively, the determination unit 108 may output the determination result to the server device and an external device such as a PC (Personal Computer).

The state determination process is executed by the above steps S201 to S206.

(Specific examples of model generation processing and diagnostic processing)
Next, specific examples of model generation processing and state processing according to the present embodiment will be described. FIG. 8 is a diagram illustrating a specific example of the model generation process and the state determination process in the first embodiment.

FIG. 8 shows, for example, a model generation process for determining the degree of deterioration of a tool based on detection information for a part of a process of machining a certain machining target, and a state determination process. In the model generation process, learning data (detection information 711 in FIG. 8) for the duration of the tool corresponding to the context information 701 received together with the context information 701 is used.

The context information 701 indicates an operation of driving the motor B. The detection information 711 does not have to be received in a state where context information indicating an operation for driving the motor B is constantly output from the processing machine 200, and some processing steps (for example, four motors (for example, four motors)) need to be received. In the process of driving the motor A, the motor B, the motor C, and the motor D), when the context information indicates the driving of the motor B, the received detection information may be stitched together. That is, the context information 701 shown in FIG. 8 does not indicate that it is always output as information indicating the operation of driving the motor B.

The feature extraction unit 105 extracts the feature information from the detection information which is the learning data for the duration received by the detection information receiving unit 102. The generation unit 107 corresponds to the context information (information indicating the drive of the “motor B” in FIG. 8), and the label information added to the learning data by the label addition unit 106 and the features extracted from the detection information. Generate a model using information. The generated model is stored in the storage unit 109 or the like in association with the corresponding context information. FIG. 8 shows an example in which a model (“motor B”) generated for context information when the motor B is driven is stored in the storage unit 109. The stored model is referred to in the subsequent state determination process.

In the state determination process, the detection information 721 is received together with the context information 702. As shown in FIG. 8, the content of the context information 702 is changed every time the machining process is switched, and the context information 702 is output from the machining machine 200. When the context information 702 indicates that "the motor B is being driven", the determination unit 108 detects, for example, that the context information 702 indicating that "the motor B is being driven" is received during the period of reception. Using the information and the model "motor B" stored in the storage unit 109, the degree of deterioration of the tool of the processing machine 200 (the tool driven by the motor B) is determined.

Similarly, when context information indicating another operation is received, the state determination process by the determination unit 108 is executed by using the corresponding detection information and the corresponding model.

Further, even if the processing processes are different from each other, for example, when the same motor or the like is used, the state determination process may be executed by commonly using the corresponding models. FIG. 9 is a diagram illustrating an example in which a common model is used in another processing process.

The context information 901 in FIG. 9 indicates that the machining process includes an operation of driving four motors (motor X, motor Y, motor Z, motor B). The point that the motor B is used is common to, for example, the processing process shown in FIG. Therefore, even in the processing step of FIG. 9, the determination unit 108 can execute the state determination process using the same model “motor B” stored in the storage unit 109 and the detection information 921.

As described above, in the diagnostic system 1 according to the present embodiment, the time (duration) from the new state to the unusable state of the tool is divided into several sections, and the learning data corresponding to each section is divided. Label information indicating a specific state is added. That is, the label information is added by using the duration. Then, using the label information added in this way and the feature information extracted from the learning data, a model for determining the state of the tool is generated. As a result, it is not necessary to manually determine the state of the tool and add the label information, and the label information indicating the corresponding state can be easily added. Then, the accuracy of the model generated in this way is improved, and for the detection information whose state is unknown, the state determination process for determining what kind of state (label information) is by using the model. The accuracy can also be improved.

As described above, the model is generated from the added label information and the feature information of the learning data, and it is determined whether the unknown feature information corresponds to any of a plurality of types of label information. Although described as one model for this, the present invention is not limited to this. For example, the above-mentioned five label information of "deterioration degree 1" to "deterioration degree 5" is added to the divided sections, and the label information added to the section is obtained from the feature information of the learning data corresponding to each section. The models corresponding to may be generated respectively. That is, in the above example, five models of a model corresponding to "deterioration degree 1", a model corresponding to "deterioration degree 2", ..., And a model corresponding to "deterioration degree 5" may be generated. Good. In this case, for example, it is assumed that the likelihood for the "deterioration degree 1" is calculated as the output value by using the unknown feature information and the model corresponding to the "deterioration degree 1", and the unknown feature information and each of them. Of the respective likelihoods calculated from the model corresponding to the label information, the label information corresponding to the highest likelihood may be determined to be the label information (deterioration degree) corresponding to the feature information. ..

[Second Embodiment]
The diagnostic system according to the second embodiment will be described focusing on the differences from the diagnostic system 1 according to the first embodiment. In the present embodiment, an operation of manually measuring the actual deterioration degree at regular time intervals within the duration, specifying the relationship between the usage time and the deterioration degree, and then adding the label information will be described. .. The overall configuration of the diagnostic system according to the present embodiment and the hardware configuration of the diagnostic apparatus and the processing machine 200 are the same as the configurations described in the first embodiment, respectively. Further, the state determination process of the diagnostic system according to the present embodiment is the same as the state determination process described in the first embodiment.

(Configuration and operation of functional blocks of diagnostic system)
FIG. 10 is a diagram showing an example of the configuration of the functional block of the diagnostic system according to the second embodiment. FIG. 11 is a diagram illustrating that label information is added based on the relationship between the tool usage time and the degree of deterioration. The configuration and operation of the functional block of the diagnostic system 1a according to the present embodiment will be described with reference to FIGS. 10 and 11. The configuration of the functional block of the processing machine 200 is the same as the configuration described in the first embodiment described above.

As shown in FIG. 10, the diagnostic device 100a of the diagnostic system 1a includes a communication control unit 101, a detection information receiving unit 102 (second acquisition unit), a processing information acquisition unit 103 (first acquisition unit), and a reception unit. 104, feature extraction unit 105, deterioration degree identification unit 113 (specific unit), label addition unit 106a (first addition unit), generation unit 107, determination unit 108, storage unit 109, and input unit 110. And a display control unit 111, and a display unit 112. Of the functional units included in the diagnostic apparatus 100a, the operations of the functional units other than the deterioration degree specifying unit 113 and the label adding unit 106a are the same as the operations described in the first embodiment described above.

The graph shown in FIG. 11 is a graph showing the relationship between the degree of deterioration and the usage time until the tool gradually deteriorates from a new product and finally breaks. For example, the user of the processing machine 200 starts using the tool from a new state, takes a micrograph or the like of the tool at regular intervals, and measures the degree of deterioration at the discretion of the user. For example, as the degree of deterioration, the length at which the cutting edge of the tool is scraped may be treated as the degree of deterioration as it is. The “x” mark shown in the graph in FIG. 11 indicates the degree of deterioration measured by the user at regular intervals. In this way, the usage time and the degree of deterioration may not always be in proportion to each other.

The deterioration degree specifying unit 113 starts using the tool from a new state, and as described above, the deterioration degree of the section where the deterioration degree is not measured is calculated from the deterioration degree actually measured at each timing of the use time. It is a functional unit that specifies the relationship of the degree of deterioration with respect to the usage time by interpolation (for example, linear interpolation or polynomial interpolation). The deterioration degree specifying unit 113 is realized by, for example, a program running on the CPU 61 shown in FIG.

The label addition unit 106a is a functional unit that adds (associates) label information for each learning data corresponding to the interval between the timings at which the degree of deterioration is measured. In the example shown in FIG. 11, similarly to the first embodiment, for each learning data corresponding to the divided sections, “deterioration degree 1”, “deterioration degree 2”, ... In the example of 11, label information is added such as "7 label information of" deterioration degree 1 "to" deterioration degree 7 "). As a result, for example, when the deterioration degrees measured at the timings at both ends of the section corresponding to the "deterioration degree 5" are "0.5" and "0.7", respectively, the label information is the state of the unknown detection information. When the determination process is performed and the corresponding label information is determined to be "deterioration degree 5", it can be recognized that the deterioration degree of the tool is in the range of "0.5" to "0.7". .. That is, the degree of deterioration can be recognized more strictly in the present embodiment as compared with the first embodiment. The label addition unit 106a is realized by, for example, a program that operates on the CPU 61 shown in FIG.

As described above, the label addition unit 106a provides label information such as "deterioration degree 1", "deterioration degree 2", ..., From the one with the smallest deterioration degree, as in the first embodiment. Although it is added, it is not limited to such label information. That is, since the actual deterioration degree of the timing at both ends of the divided section is measured, for example, the label information may be added using the measurement value of the deterioration degree at the rear end timing of the section itself. Good. For example, when the degree of deterioration at the rear end timing of a certain section is "0.9", label information such as "degree of deterioration 0.9" may be added as the label information corresponding to the section. As a result, when the label information is unknown and the state determination process is performed on the detected information and the corresponding label information is determined to be "deterioration degree 0.9", the tool has "deterioration degree 0.9 at maximum". It can be recognized as a "state".

Further, as described above, the label addition unit 106a adds label information for each learning data corresponding to the section between the timings at which the degree of deterioration is measured, but the method of dividing the section is limited to this. It's not a thing. For example, the degree of deterioration of the graph shown in FIG. 11 may be equally divided into a predetermined number, and label information may be added for each section of usage time corresponding to each range of the equally divided degree of deterioration.

Further, each functional unit (communication control unit 101, detection information receiving unit 102, processing information acquisition unit 103, reception unit 104, feature extraction unit 105, deterioration degree specifying unit 113, label adding unit) of the diagnostic device 100a shown in FIG. 10 The 106a, the generation unit 107, the determination unit 108, and the display control unit 111) may be realized by causing the CPU 61 of FIG. 3 to execute the program, that is, by software or by hardware such as an IC. Alternatively, it may be realized by using software and hardware together.

Further, each functional unit of the diagnostic apparatus 100a and the processing machine 200 shown in FIG. 10 conceptually shows the function, and is not limited to such a configuration. For example, a plurality of functional units shown as independent functional units in FIG. 10 may be configured as one functional unit. On the other hand, the function of one functional unit in FIG. 10 may be divided into a plurality of functions and configured as a plurality of functional units.

(Model generation processing by diagnostic equipment)
FIG. 12 is a flowchart showing an example of the model generation process in the second embodiment. The flow of the model generation process of the diagnostic apparatus 100a of the diagnostic system 1a according to the present embodiment will be described with reference to FIG. It should be noted that the model generation process is, for example, executed in advance before the state determination process. Before executing the following model generation process, it is assumed that a new tool is attached to the processing machine 200.

<Step S301>
The communication control unit 101 of the diagnostic apparatus 100a receives the context information transmitted from the processing machine 200. Then, the processing information acquisition unit 103 of the diagnostic apparatus 100a acquires the context information received by the communication control unit 101. Then, the process proceeds to step S302.

<Step S302>
The detection information receiving unit 102 of the diagnostic apparatus 100a starts receiving the detection information (sensor data) as learning data transmitted from the processing machine 200. The operation of receiving the learning data by the detection information receiving unit 102 starts from the timing when the newly replaced new tool is started to be used, and finally the timing when the tool is broken due to deterioration such as wear of the tool due to repeated machining processes. Continue until. When the reception of the training data is started, the process proceeds to step S303.

<Step S303>
If the relationship between the usage time and the degree of deterioration of the tool currently being driven by the machining machine 200 is defined according to the context information acquired by the machining information acquisition unit 103 (step S303: Yes), the step The process proceeds to S306, and if it is not defined (step S303: No), the process proceeds to step S304.

<Step S304>
The user of the processing machine 200 starts using the tool from a new state and measures the degree of deterioration at a predetermined usage time timing (for example, every predetermined time). This user measurement of the degree of deterioration continues until the tool breaks. Then, the process proceeds to step S305.

<Step S305>
The deterioration degree specifying unit 113 of the diagnostic apparatus 100a interpolates the deterioration degree of the section in which the deterioration degree is not measured from the deterioration degree actually measured at each timing of the use time when the tool is started from the new state. By doing so, the relationship between the degree of deterioration and the usage time is specified. Then, the process proceeds to step S306.

<Step S306>
The feature extraction unit 105 of the diagnostic device 100a extracts feature information from the learning data for the duration received by the detection information receiving unit 102. Then, the process proceeds to step S307.

<Step S307>
The label addition unit 106a of the diagnostic apparatus 100a adds (associates) label information for each learning data corresponding to the interval between the timings at which the degree of deterioration is measured. Specifically, in the graph showing the relationship between the usage time and the degree of deterioration shown in FIG. 11 described above, for example, the label addition unit 106a is "for each learning data corresponding to the divided sections, from the one with the smallest degree of deterioration". Label information is added such as "deterioration degree 1", "deterioration degree 2", ..., "deterioration degree 7". In this case, since the label information is added to the training data for each section in the duration, the label information is associated with the feature information corresponding to each section of the feature information extracted from the learning data. It will be. Then, the process proceeds to step S308.

<Step S308>
The generation unit 107 of the diagnostic apparatus 100a determines the degree of deterioration of the tool by using the label information added to the learning data by the label addition unit 106a and the feature information extracted from the learning data by the feature extraction unit 105. Generate a model to do. In this case, the generation unit 107 generates a model in association with the context information acquired by the processing information acquisition unit 103 when the detection information reception unit 102 receives the learning data. As an algorithm for generating this model, for example, SVM or GMM of supervised learning is used. By using the model generated in this way, which label information corresponds to which class of characteristic information, that is, which label information (in the above example, "deterioration degree 1" to "deterioration") It becomes clear whether it corresponds to degree 7 "). Then, the process proceeds to step S309.

<Step S309>
The generation unit 107 associates the generated model with the context information acquired by the processing information acquisition unit 103, and stores the generated model in the storage unit 109.

The model generation process is executed by the above steps S301 to S309.

As described above, in the diagnostic system 1a according to the present embodiment, when the relationship between the usage time and the degree of deterioration of the tool is not defined, at the timing of the predetermined usage time (for example, at regular time intervals) of the duration. Based on the degree of deterioration measured by the user or the like, interpolation or the like is performed to specify the relationship between the actual degree of deterioration and the usage time. Further, the label addition unit 106a divides the duration into a plurality of sections (for example, sections between the timings at which the degree of deterioration is measured), and adds label information to each learning data corresponding to each section. Then, using the label information added in this way and the feature information extracted from the learning data, a model for determining the state of the tool is generated. As a result, for the first tool before the model is generated, it is necessary for the user to measure the degree of deterioration at a predetermined timing of the duration, but after identifying the relationship between the usage time and the degree of deterioration. Since the label information is added, the label information determined by the model indicates the specific deterioration degree for the detection information whose state (deterioration degree) is unknown, so that the deterioration degree is recognized more strictly. be able to. Then, the accuracy of the model generated in this way is improved, and for the detection information whose state is unknown, the state determination process for determining what kind of state (label information) is by using the model. The accuracy can also be improved.

[Third Embodiment]
The diagnostic system according to the third embodiment will be described focusing on the differences from the diagnostic system 1 according to the first embodiment. In the present embodiment, for the first tool (n = 1), the relationship between the usage time and the degree of deterioration is specified by manually measuring the degree of deterioration at a predetermined timing within the duration. For the second and subsequent tools (n ≧ 2), the operation of adding label information will be described assuming that the shape is similar to the graph showing the relationship between the usage time of the first tool and the degree of deterioration. The overall configuration of the diagnostic system according to the present embodiment and the hardware configuration of the diagnostic apparatus and the processing machine 200 are the same as the configurations described in the first embodiment, respectively. Further, the state determination process of the diagnostic system according to the present embodiment is the same as the state determination process described in the first embodiment.

(Configuration and operation of functional blocks of diagnostic system)
FIG. 13 is a diagram showing an example of the configuration of the functional block of the diagnostic system according to the third embodiment. FIG. 14 is a graph showing an example of the relationship between the usage time of a plurality of tools and the degree of deterioration. FIG. 15 is a diagram illustrating that label information is added based on the relationship between the tool usage time and the degree of deterioration. The configuration and operation of the functional block of the diagnostic system 1b according to the present embodiment will be described with reference to FIGS. 13 to 15. The configuration of the functional block of the processing machine 200 is the same as the configuration described in the first embodiment described above.

The diagnostic device 100b of the diagnostic system 1b shown in FIG. 13 includes a communication control unit 101, a detection information receiving unit 102 (second acquisition unit), a processing information acquisition unit 103 (first acquisition unit), and a reception unit 104. , Feature extraction unit 105, deterioration degree specifying unit 113b (specific unit), first label addition unit 114 (first addition unit), measurement unit 115, and second label addition unit 116 (second addition unit). , A generation unit 107, a determination unit 108, a storage unit 109, an input unit 110, a display control unit 111, and a display unit 112. Of the functional units included in the diagnostic apparatus 100b, the operations of the functional units other than the deterioration degree specifying unit 113b, the first label addition unit 114, the measurement unit 115, and the second label addition unit 116 are described in the first embodiment described above. It is the same as the operation described.

With respect to the relationship between the usage time and the degree of deterioration of a tool with n = 1 (first tool) (an example of the first member), a tool with n ≧ 2 (second and subsequent tools) (an example of the second member) It is assumed that the relationship between the usage time and the degree of deterioration of is relatively unchanged. That is, if the types of tools are the same, it is assumed that the degree of deterioration (relative shape of the graph) is the same, although the life (duration) varies depending on individual differences. In this case, as shown in FIG. 14, the shape of the graph showing the relationship between the usage time of n = 1 and the degree of deterioration and the graph showing the relationship between the usage time of n ≧ 2 and the degree of deterioration (in the example of FIG. 14). , N = 2 and n = 3) and the shapes are similar to each other. Based on this point, label information for the training data is added by the first label addition unit 114 and the second label addition unit 116 by a method as described later.

The graph shown in FIG. 15 is a graph showing the relationship between the usage time and the degree of deterioration until the first tool (n = 1) for generating a model gradually deteriorates from a new tool and finally breaks. As described above in FIG. 11 of the second embodiment, for example, the user of the processing machine 200 starts using the tool from a new state and takes a photomicrograph or the like of the tool at regular intervals. The degree of deterioration is measured at the discretion of the user. For example, as the degree of deterioration, the length at which the cutting edge of the tool is scraped may be treated as the degree of deterioration as it is. The “x” mark shown in the graph in FIG. 15 indicates the degree of deterioration measured by the user at regular intervals.

The deterioration degree specifying unit 113b is a section in which the deterioration degree is not measured from the deterioration degree actually measured at each timing of the use time, as described above, when the tool with n = 1 is started to be used from a new state. It is a functional unit that specifies the relationship of the degree of deterioration with respect to the usage time by interpolating the degree of deterioration (for example, linear interpolation or polynomial interpolation). The deterioration degree specifying unit 113b is realized by, for example, a program running on the CPU 61 shown in FIG.

The first label addition unit 114 is a deterioration of the graph shown in FIG. 15 for the tool with n = 1 (a graph showing the relationship of the degree of deterioration with respect to the usage time for the tool with n = 1 specified by the degree of deterioration specifying unit 113b). It is a functional unit that divides the degree equally into a predetermined number and adds (associates) label information for each training data corresponding to the interval of the usage time corresponding to each range of the equally divided deterioration degree. In the example shown in FIG. 15, "deterioration degree 1", "deterioration degree 2", and so on, from the smallest deterioration degree for each learning data corresponding to the usage time interval corresponding to each range of the equally divided deterioration degree. -, Five label information such as "deterioration degree 5" is added. Further, as shown in FIG. 15, the section corresponding to "deterioration degree 1" is section A, the section corresponding to "deterioration degree 2" is section B, the section corresponding to "deterioration degree 3" is section C, and "deterioration degree 3". The section corresponding to "degree 4" is referred to as section D, and the section corresponding to "deterioration degree 5" is referred to as section E. As a result, for example, when the deterioration degrees measured at the timings at both ends of the section corresponding to the "deterioration degree 3" are "0.3" and "0.5", respectively, the label information is unknown for the detection information. When the determination process is performed and the corresponding label information is determined to be "deterioration degree 3", it can be recognized that the deterioration degree of the tool is in the range of "0.3" to "0.5". .. The first label addition unit 114 is realized by, for example, a program that operates on the CPU 61 shown in FIG.

As described above, the first label addition unit 114 divides the degree of deterioration into a predetermined number and divides the degree of deterioration into equal parts, and provides label information for each learning data corresponding to the interval of the usage time corresponding to each range of the degree of deterioration. It is supposed to be added, but it is not limited to this. For example, label information may be added for each learning data corresponding to the interval between the timings at which the degree of deterioration is measured.

The measuring unit 115 has a function of measuring the duration of a model in order to generate a model using another tool (the second and subsequent tools) after the model is generated using the tool of n = 1. It is a department. At this time, as described above, the shape of the graph showing the relationship between the usage time of n = 1 and the degree of deterioration and the shape of the graph showing the relationship between the usage time of n ≧ 2 and the degree of deterioration are similar to each other. Therefore, the deterioration degree specifying unit 113b is for a tool with n ≧ 2 from the ratio of the duration of the tool with n ≧ 2 measured by the measuring unit 115 to the duration of the tool with n = 1. Identify the relationship between the degree of deterioration and the usage time of. The measuring unit 115 is realized by, for example, a program running on the CPU 61 shown in FIG.

The second label addition unit 116 corresponds to each label information added by the first label addition unit 114 with respect to the duration of the tool with n ≧ 2 measured by the measurement unit 115 and the duration of the tool with n = 1. A function of adding the same label information as the label information added by the first label addition unit 114 to the training data corresponding to each section with respect to the ratio of each section to be measured and the duration of the tool of n ≧ 2 obtained from It is a department. For example, as shown in FIG. 15, when the duration of the tool with n = 1 is assumed to be "1", for example, the usage time of the boundary between "deterioration degree 1" and "deterioration degree 2" is "0.3". The usage time of the boundary between the "deterioration degree 2" and the "deterioration degree 3" is "0.6". At this time, when the duration of the tool with n ≧ 2 is “10”, the usage time corresponds to the section from “3” (= 10 × 0.3) to “6” (= 10 × 0.6). Label information of "deterioration degree 2" is added to the training data. The second label addition unit 116 is realized by, for example, a program that operates on the CPU 61 shown in FIG.

In addition, each functional unit (communication control unit 101, detection information receiving unit 102, processing information acquisition unit 103, reception unit 104, feature extraction unit 105, deterioration degree specifying unit 113b, first label) of the diagnostic device 100b shown in FIG. The addition unit 114, the measurement unit 115, the second label addition unit 116, the generation unit 107, the determination unit 108, and the display control unit 111) may be realized by causing the CPU 61 of FIG. 3 to execute the program, that is, by software. , IC or other hardware may be used, or software and hardware may be used in combination.

Further, each functional unit of the diagnostic apparatus 100b and the processing machine 200 shown in FIG. 13 conceptually shows a function, and is not limited to such a configuration. For example, a plurality of functional units shown as independent functional units in FIG. 13 may be configured as one functional unit. On the other hand, the function of one functional unit in FIG. 13 may be divided into a plurality of functions and configured as a plurality of functional units.

(Model generation processing by diagnostic equipment)
FIG. 16 is a flowchart showing an example of the model generation process according to the third embodiment. The flow of the model generation process of the diagnostic apparatus 100b of the diagnostic system 1b according to the present embodiment will be described with reference to FIG. It should be noted that the model generation process is, for example, executed in advance before the state determination process. Before executing the following model generation process, it is assumed that a new tool is attached to the processing machine 200.

<Step S401>
The communication control unit 101 of the diagnostic apparatus 100b receives the context information transmitted from the processing machine 200. Then, the processing information acquisition unit 103 of the diagnostic apparatus 100b acquires the context information received by the communication control unit 101. Then, the process proceeds to step S402.

<Step S402>
The detection information receiving unit 102 of the diagnostic apparatus 100b starts receiving the detection information (sensor data) as learning data transmitted from the processing machine 200. The operation of receiving the learning data by the detection information receiving unit 102 starts from the timing when the newly replaced new tool is started to be used, and finally the timing when the tool is broken due to deterioration such as wear of the tool due to repeated machining processes. Continue until. When the reception of the training data is started, the process proceeds to step S403.

<Step S403>
If the tool currently being driven by the machining machine 200 corresponds to the context information acquired by the machining information acquisition unit 103 and the tool is n = 1 (step S403: Yes (n = 1)), the process proceeds to step S404. If the tool is n ≧ 2 (step S403: No (n ≧ 2)), the process proceeds to step S406.

<Step S404>
The user of the processing machine 200 starts using the tool with n = 1 from a new state, and measures the degree of deterioration at a predetermined usage time timing (for example, every predetermined time). This user measurement of the degree of deterioration continues until the tool breaks. Then, the process proceeds to step S405.

<Step S405>
The deterioration degree specifying unit 113b of the diagnostic apparatus 100b is a section in which the deterioration degree is not measured from the deterioration degree actually measured at each timing of the use time when the tool of n = 1 is started to be used from a new state. By interpolating the degree of deterioration, the relationship of the degree of deterioration with respect to the usage time is specified. Then, the process proceeds to step S408.

<Step S406>
The measuring unit 115 of the diagnostic apparatus 100b measures the duration (time from a new state to breakage) of the tool in order to generate a model using a tool with n ≧ 2 (the second and subsequent tools). .. Then, the process proceeds to step S407.

<Step S407>
The deterioration degree specifying unit 113b has a shape similar to the shape of the graph showing the relationship between the usage time of n = 1 and the degree of deterioration and the shape of the graph showing the relationship between the usage time of n ≧ 2 and the degree of deterioration. As a certain thing, the relationship of the degree of deterioration with respect to the usage time of the tool of n ≧ 2 is specified from the ratio of the duration of the tool of n ≧ 2 measured by the measuring unit 115 to the duration of the tool of n = 1. .. Then, the process proceeds to step S408.

<Step S408>
The feature extraction unit 105 of the diagnostic device 100b extracts the feature information from the learning data for the duration received by the detection information receiving unit 102. Then, the process proceeds to step S409.

<Step S409>
In the case of a tool with n = 1, the first label addition unit 114 of the diagnostic apparatus 100b determines the degree of deterioration of the graph showing the relationship of the degree of deterioration with respect to the usage time of the tool with n = 1 specified by the degree of deterioration specifying unit 113b. It is equally divided into a predetermined number, and label information is added (corresponded) to each learning data corresponding to the interval of usage time corresponding to each range of the equally divided deterioration degree. In the case of a tool with n ≧ 2, the second label adding part 116 is set by the first label adding part 114 with respect to the duration of the tool with n ≧ 2 measured by the measuring unit 115 and the duration of the tool with n = 1. With respect to the ratio of each section corresponding to each added label information and the training data corresponding to each section with respect to the duration of the tool of n ≧ 2 obtained from the label information added by the first label addition unit 114. Add (associate) the same label information. Then, the process proceeds to step S410.

<Step S410>
The generation unit 107 of the diagnostic apparatus 100b combines the label information added to the training data by the first label addition unit 114 or the second label addition unit 116 and the feature information extracted from the training data by the feature extraction unit 105. It is used to generate a model for determining the degree of deterioration of a tool. In this case, the generation unit 107 generates a model in association with the context information acquired by the processing information acquisition unit 103 when the detection information reception unit 102 receives the learning data. As an algorithm for generating this model, for example, SVM or GMM of supervised learning is used. By using the model generated in this way, which label information corresponds to which class of characteristic information, that is, which label information (in the above example, "deterioration degree 1" to "deterioration") It becomes clear whether it corresponds to degree 7 "). Then, the process proceeds to step S411.

<Step S411>
The generation unit 107 associates the generated model with the context information acquired by the processing information acquisition unit 103, and stores the generated model in the storage unit 109.

The model generation process is executed by the above steps S401 to S411.

As described above, in the diagnostic system 1b according to the present embodiment, when the relationship between the usage time and the degree of deterioration of the first tool (n = 1) for generating the model is not defined, the duration is Of these, based on the degree of deterioration measured by a user or the like at a predetermined timing of use time (for example, every fixed time), interpolation or the like is performed to specify the relationship between the degree of use and the actual degree of deterioration. Further, the first label addition unit 114 divides the graph showing the relationship of the degree of deterioration with respect to the usage time of the tool of n = 1 into a plurality of sections by a predetermined method for the duration, and for each learning data corresponding to each section. Label information is added. Further, in order to generate a model for the second and subsequent tools (n ≧ 2), the measuring unit 115 measures the duration (time from a new state to breakage) of the tool. Then, the second label adding unit 116 provides each label information added by the first label adding unit 114 with respect to the duration of the tool with n ≧ 2 measured by the measuring unit 115 and the duration of the tool with n = 1. The same label information as the label information added by the first label addition unit 114 is added to the training data corresponding to each section with respect to the ratio of each section corresponding to and the duration of the tool of n ≧ 2 obtained from. It is supposed to be done. As a result, for the first tool before the model is generated, it is necessary for the user to measure the degree of deterioration at a predetermined timing of the duration, but the relationship between the usage time and the degree of deterioration can be determined. Since the label information is added after specifying, the label information determined by the model indicates the specific degree of deterioration for the detection information whose state (deterioration degree) is unknown, so that the degree of deterioration is more rigorous. Can be recognized. Further, when a new model is generated using a tool with n ≧ 2, the shape of the graph showing the relationship between the usage time of n = 1 and the degree of deterioration and the relationship between the usage time of n ≧ 2 and the degree of deterioration are shown. The shapes of the graphs shown are treated as having similar shapes to each other, and label information is added to the training data of n ≧ 2 without measuring the degree of deterioration for n ≧ 2. As a result, it is possible to easily add label information indicating the state of the tool to the learning data of the tool with n ≧ 2 without human judgment. Then, the accuracy of the model generated in this way is improved, and for the detection information whose state is unknown, the state determination process for determining what kind of state (label information) is by using the model. The accuracy can also be improved.

The program executed by the diagnostic system of each of the above-described embodiments may be configured to be provided in advance in a ROM or the like.

Further, the programs executed by the diagnostic system of each of the above-described embodiments are files in an installable format or an executable format, such as a CD-ROM (Computer Disc Read Only Memory), a flexible disk (FD), and a CD-R ( It may be configured to be recorded on a computer-readable recording medium such as a Compact Disc-Recordable or a DVD (Digital Versaille Disk) and provided as a computer program product.

Further, the program executed by the diagnostic system of each of the above-described embodiments may be stored on a computer connected to a network such as the Internet and provided by downloading via the network. Further, the program executed by the diagnostic system of each of the above-described embodiments may be configured to be provided or distributed via a network such as the Internet.

Further, the program executed by the diagnostic system of each of the above-described embodiments has a module configuration including each of the above-mentioned functional parts, and as the actual hardware, the CPU (processor) reads the program from the ROM and executes it. As a result, each of the above-mentioned functional units is loaded on the main storage device, and each functional unit is generated on the main storage device.

1,1a, 1b diagnostic system 51 CPU
52 ROM
53 RAM
54 Communication I / F
55 Drive control circuit 56 Motor 57 Sensor 58 Bus 61 CPU
62 ROM
63 RAM
64 Communication I / F
65 Sensor I / F
66 Auxiliary storage device 67 Input device 68 Display 69 Bus 100, 100a, 100b Diagnostic device 101 Communication control unit 102 Detection information reception unit 103 Processing information acquisition unit 104 Reception unit 105 Feature extraction unit 106, 106a Label addition unit 107 Generation unit 108 Judgment Unit 109 Storage unit 110 Input unit 111 Display control unit 112 Display unit 113, 113b Deterioration degree identification unit 114 First label addition unit 115 Measuring unit 116 Second label addition unit 200 Processing machine 201 Numerical control unit 202 Communication control unit 203 Drive control Unit 204 Drive unit 211 Detection unit 701, 702 Context information 711, 721 Detection information 901 Context information 921 Detection information

Japanese Unexamined Patent Publication No. 2010-054558

Claims (11)

A first acquisition unit that acquires predetermined information corresponding to the current operation of the target device from the target device, and
A second acquisition unit that acquires first detection information output as learning data from a detection unit that detects a physical quantity related to a member of the target device, and a second acquisition unit.
An extraction unit that extracts the first feature information indicating the characteristics of the learning data acquired by the second acquisition unit, and an extraction unit.
When the operation is executed, the duration from the start of use of the member in a new state until it becomes unusable is divided into a plurality of predetermined sections, and the learning data corresponding to each section is described. The first additional part to which label information indicating the state of the member is added, and
Based on the label information added to the learning data corresponding to each section and the first feature information, the second detection information in which the label information acquired by the second acquisition unit is unknown is , A generation unit that generates a model for determining which label information corresponds to, in association with the predetermined information corresponding to the training data as model specific information.
Diagnostic device equipped with. The extraction unit extracts the second feature information indicating the feature of the second detection information, and then extracts the second feature information.
The label information indicating which state the second feature information indicates by using the model corresponding to the model specific information acquired by the first acquisition unit and the second feature information corresponding to the model specific information. The diagnostic device according to claim 1, further comprising a determination unit for determining whether or not the information corresponds to. The first additional part divides the duration into a predetermined number of sections and adds label information indicating the state of the member to the learning data corresponding to each of the equally divided sections. Or the diagnostic apparatus according to 2. Further, a specific unit for specifying the relationship between the state of the member and the use time is further provided based on the state of the member measured at a predetermined timing of the use time from the start of use of the member in a new state.
The first addition unit divides the duration into a plurality of sections in which the change in the state is specified by the relationship of the state of the member with respect to the usage time, and with respect to the learning data corresponding to each section. The diagnostic device according to claim 1 or 2, to which the label information is added, respectively. The relationship between the state of the first member and the state of the first member is specified based on the state of the first member measured at a predetermined timing of the use time from the start of use of the first member as the member in a new state. With more specific parts,
The first addition unit divides the duration into a plurality of sections in which changes in the state are specified by the relationship between the state of the first member and the usage time, and with respect to the learning data corresponding to each section. Then, add the label information to each of them.
After the use of the first member, a measuring unit for measuring the duration of the second member different from the first member is further provided.
The specific unit identifies the relationship between the state of the second member and the usage time from the ratio of the duration of the second member to the duration of the first member.
The second member, which is obtained from the duration of the second member and the ratio of each section corresponding to each label information added by the first addition portion to the duration of the first member. Claim 1 or claim 1 is further provided with a second additional part that adds the same label information as the label information added by the first additional part to the learning data corresponding to each section with respect to the duration. 2. The diagnostic device according to 2. Claim 4 or 5 in which the first addition unit divides the duration into sections between timings at which the state is measured, and adds the label information to the learning data corresponding to each section. Diagnostic device according to. The first additional portion divides the state of the member into a predetermined number, and based on the relationship of the state of the member with respect to the use time, sets the interval of the use time corresponding to each range of the equally divided states. The diagnostic device according to claim 4 or 5, wherein the label information is added to the corresponding learning data, respectively. The member of the target device is a tool for machining a machining target.
The diagnostic device according to any one of claims 1 to 7, wherein the state of the member indicated by the label information is the degree of deterioration of the tool. With the detector
The diagnostic device according to any one of claims 1 to 8.
Diagnostic system with. The first acquisition step of acquiring predetermined information corresponding to the current operation of the target device from the target device, and
The second acquisition step of acquiring the first detection information output from the detection unit that detects the physical quantity of the member of the target device as learning data, and
An extraction step for extracting feature information indicating the features of the acquired learning data, and
When the operation is executed, the duration from the start of use of the member in a new state until it becomes unusable is divided into a plurality of predetermined sections, and the learning data corresponding to each section is described. An additional step to add label information indicating the state of the member, and
Based on the label information added to the learning data corresponding to each section and the feature information, which label information the second detection information for which the acquired label information is unknown corresponds to is determined. A generation step of generating a model for determination by associating the predetermined information corresponding to the training data with the model specific information.
Diagnostic method with. On the computer
The first acquisition step of acquiring predetermined information corresponding to the current operation of the target device from the target device, and
The second acquisition step of acquiring the first detection information output from the detection unit that detects the physical quantity of the member of the target device as learning data, and
An extraction step for extracting feature information indicating the features of the acquired learning data, and
When the operation is executed, the duration from the start of use of the member in a new state until it becomes unusable is divided into a plurality of predetermined sections, and the learning data corresponding to each section is described. An additional step to add label information indicating the state of the member, and
Based on the label information added to the learning data corresponding to each section and the feature information, which label information the second detection information for which the acquired label information is unknown corresponds to is determined. A generation step of generating a model for determination by associating the predetermined information corresponding to the training data with the model specific information.
A program to execute.

Images