MODEL GENERATION DEVICE, MODEL GENERATION METHOD, ABNORMALITY SIGN DETECTION DEVICE, ABNORMALITY SIGN DETECTION METHOD, AND ABNORMALITY SIGN DETECTION SYSTEM

Abstract

To provide a technology capable of detecting an abnormality and an abnormality sign of a production facility.SOLUTION: A model generation device according to an aspect of the present disclosure comprises: a preprocessing unit for acquiring feature amount data of production performance data for training; an abnormality degree calculation unit that calculates an abnormality degree of the acquired feature amount data using an abnormality degree calculation model for representing a feature amount group under a normal state; and a training unit for determining a state of the feature amount data based on the calculated abnormality degree, and for training an abnormality sign detection model for detecting the state of the feature amount data from the feature amount data using training data configured of the feature amount data and the state of the feature amount data.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to a model generation device, a model generation method, an abnormality sign detection device, an abnormality sign detection method, and an abnormality sign detection system. [Background technology]

[0002] In order to prevent deterioration or failure of a certain facility and maintain normal operation, it is common to provide an anomaly sign detection device that uses data acquired from the facility. Patent Document 1 discloses a condition monitoring device that performs machine learning using current values ​​continuously acquired from a temperature control device to detect anomalies or signs of anomalies. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2018-204940 A Summary of the Invention [Problem to be solved by the invention]

[0004] In the technology disclosed in Patent Document 1, machine learning is performed to determine anomalies or signs of anomalies. However, since there are generally few cases in which anomalies occur in equipment, there is also little abnormal data that can be used for machine learning to generate a model. Even if abnormal data is obtained, data in which features related to anomalies or signs of anomalies in equipment are not present may be mixed in. There is no problem if learning is performed after accumulating a large amount of data as learned by general machine learning. However, data accumulation takes a long time. In addition, production may end by the time sufficient data is accumulated. Therefore, learning after accumulating a large amount of data is not practical. Therefore, it is desirable to learn data in which features related to anomalies or signs of anomalies are appropriately present at a stage when there is little abnormal data, and generate a highly accurate trained anomaly sign detection model.

[0005] An object of the present disclosure is to provide a technology capable of detecting abnormalities and signs of abnormalities in production equipment. [Means for solving the problem]

[0006] A model generation device according to an aspect of the present disclosure includes a preprocessing unit that acquires feature data of production performance data for training, an anomaly degree calculation unit that calculates an anomaly degree of the acquired feature data by using an anomaly degree calculation model that represents a group of features in a normal state, and a training unit that determines a state of the feature data based on the calculated anomaly degree, and trains an anomaly sign detection model that detects the state of the feature data from the feature data using training data composed of the feature data and the state of the feature data. Each feature data of the training data is composed of a group of overlapping features on a time axis. do.

[0007] A model generation method according to an aspect of the present disclosure includes a step of acquiring feature data of production performance data for training, a step of calculating an anomaly degree of the acquired feature data by using an anomaly degree calculation model that represents a group of feature values ​​in a normal state, and a step of determining a state of the feature data based on the calculated anomaly degree, and training an anomaly sign detection model that detects the state of the feature data from the feature data by using training data composed of the feature data and the state of the feature data, the step being executed by a computer. Each feature data of the training data is composed of a group of overlapping features on a time axis. do.

[0008] An anomaly sign detection device according to one aspect of the present disclosure includes a pre-processing unit that acquires feature data of production performance data to be detected, and an anomaly sign detection unit that detects a state of the feature data from the feature data by using a trained anomaly sign detection model, the anomaly sign detection model being trained by training data composed of feature data of training production performance data and a state of the feature data, and the state of the feature data being determined based on an anomaly level of the feature data calculated by using an anomaly level calculation model that represents a group of features in a normal state. Each feature data of the training data is composed of a group of overlapping features on the time axis. do.

[0009] According to one aspect of the present disclosure, there is provided an anomaly sign detection method, the anomaly sign detection model is trained by training data including feature data of production performance data for training and a state of the feature data, the state of the feature data being determined based on an anomaly level of the feature data calculated by using an anomaly level calculation model that represents a group of features in a normal state. Each feature data of the training data is composed of a group of overlapping features on the time axis. do. Effect of the Invention

[0010] According to the present disclosure, it is possible to provide a technology capable of detecting abnormalities and signs of abnormalities in production equipment. [Brief description of the drawings]

[0011] [Figure 1] Schematic diagram showing an anomaly detection system [Diagram 2] FIG. 2 is a block diagram showing the hardware configuration of a model generating device and an anomaly sign detecting device. [Diagram 3] Schematic diagram showing a configuration example of a winding section [Figure 4] Schematic diagram of a wound body [Diagram 5] FIG. 13 is a diagram showing an example of correspondence between measurement results and judgment results of a first sheet material; [Figure 6] FIG. 1 is a block diagram illustrating a functional configuration of a model generating device according to a first embodiment; [Figure 7] An example of a production performance database [Figure 8] An example of a feature database [Figure 9] An example of an anomaly database [Figure 10] A diagram showing an example of a state label database. [Figure 11] FIG. 1 is a block diagram illustrating a functional configuration of an abnormality sign detection device according to a first embodiment; [Figure 12] Flowchart illustrating an example of pre-processing [Figure 13] 1 is a flowchart illustrating an example of a process for generating an anomaly degree calculation model. [Figure 14] 1 is a flowchart illustrating an example of an abnormality degree calculation process. [Figure 15] A flowchart for explaining an example of an adoption determination process for an anomaly degree calculation model. [Figure 16] Flowchart illustrating an example of a state label setting process [Figure 17] 1 is a flowchart illustrating an example of a training process for an anomaly sign detection model. [Figure 18] A diagram to explain the learning data list [Figure 19] A flowchart for explaining an example of an adoption determination process for an anomaly sign detection model. [Figure 20] 1 is a flowchart for explaining an abnormality sign detection process according to a first embodiment. [Figure 21] FIG. 11 is a block diagram illustrating a functional configuration of a model generating device according to a second embodiment. [Figure 22] A flowchart for explaining a state label setting process according to a second embodiment. [Figure 23] FIG. 13 is a block diagram illustrating a functional configuration of a model generating device according to a third embodiment. [Figure 24] A flowchart for explaining a state label setting process according to a third embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] Hereinafter, each embodiment of the present disclosure will be described in detail with reference to the drawings. However, more detailed explanation than necessary, for example, detailed explanation of already well-known matters and duplicate explanation of substantially the same configuration, may be omitted.

[0013] It should be noted that the following description and the referenced drawings are provided to enable those skilled in the art to understand the present disclosure, and are not intended to limit the scope of the claims of the present disclosure.

[0014] (First embodiment) FIG. 1 is a schematic diagram showing an abnormality sign detection system 10 according to the first embodiment.

[0015] As shown in FIG. 1, the abnormality sign detection system 10 includes a model generation device 60 that generates an abnormality sign detection model, and an abnormality sign detection device 100 that performs abnormality sign detection using the generated abnormality sign detection model.

[0016] The abnormality sign detection system 10 uses the model generation device 60 to generate in advance an abnormality sign detection model for detecting whether there is an abnormality or a sign of an abnormality in the production facility 200. After generating the abnormality sign detection model, the abnormality sign detection system 10 uses the abnormality sign detection device 100 that can access the trained abnormality sign detection model to detect abnormalities and signs of an abnormality in the production facility 200 based on actual production data of the detection target collected while the production facility 200 is in operation.

[0017] The anomaly sign detection model according to the following embodiment distinguishes the state of the production facility 200 into a normal period, a warning period, or an abnormal period based on production performance data. However, the anomaly sign detection system 10 according to the present disclosure is not limited to this, and may distinguish any other appropriate state of the production facility 200.

[0018] The model generation device 60 extracts training production performance data from the training database 20, and uses the extracted training production performance data to train an anomaly sign detection model through supervised machine learning.

[0019] For example, the production performance data for training may be time-series production performance data related to products, etc., collected during operation of the production facility 200. In reality, it is desirable to be able to generate an anomaly sign detection model before the production facility 200 starts operating, but it is generally difficult to obtain production performance data indicating anomalies or signs of anomalies due to aging deterioration, etc., before the production facility 200 starts operating.

[0020] For this reason, immediately after the operation of the production facility 200 (for example, one week after the start of operation), it is assumed that the production facility 200 is in a normal state, and an anomaly calculation model for calculating an anomaly degree as a deviation from a normal sample is derived based on the production performance data in that period, as described in detail below. Such an anomaly calculation model can be derived by applying an anomaly sign detection technique that detects anomalies and signs of anomaly based on normal samples, such as the well-known MT method (Mahalanobis Taguchi Method). The model generation device 60 utilizes this anomaly calculation model to label the production performance data with a "normal period," a "sign period," and an "abnormal period," even during a period in which the number of collected anomaly samples and sign samples is considered to be relatively small, such as immediately after the operation of the production facility, and can efficiently generate an anomaly sign detection model using the labeled production performance data.

[0021] In one embodiment, the training database 20 stores unlabeled production performance data, and as described below, the model generation device 60 calculates the degree of anomaly of the unlabeled production performance data using the anomaly degree calculation model, and labels the state of the production performance data as training data for the anomaly sign detection model. Typically, according to a known anomaly sign detection technique such as the MT method, although the training data for training the anomaly degree calculation model needs to be state-labeled, the anomaly degree calculation model can be generated with significantly less training data than the anomaly sign detection model. Therefore, even if only a relatively small amount of state-labeled production performance data can be prepared, it is possible to generate the anomaly sign detection model.

[0022] Such an anomaly degree calculation model may be generated by a method described in detail below and may be provided in advance in the model generation device 60, or may be accessible via a network, etc. However, the present disclosure is not limited to this, and the model generation device 60 may generate not only the anomaly sign detection model but also the anomaly degree calculation model.

[0023] When training of the anomaly sign detection model is completed, the model generation device 60 provides the trained anomaly sign detection model to the anomaly sign detection device 100.

[0024] The abnormality sign detection device 100 detects an abnormality or an abnormality sign of the production facility 200 based on the production performance data of the production facility 200, using the trained abnormality sign detection model generated by the model generation device 60. Specifically, the abnormality sign detection device 100 inputs the production performance data of the product acquired during the operation of the production facility 200 into the trained abnormality sign detection model, acquires the detection result (e.g., "normal period", "sign period", or "abnormal period") from the abnormality sign detection model, and notifies a user such as a manager of the production facility 200 of various information including the detection result for the maintenance of the production facility 200. In this embodiment, one abnormality sign detection device 100 is applied to one production facility 200, but one abnormality sign detection device 100 may be applied to multiple production facilities 200.

[0025] In addition, in this embodiment, the model generation device 60 and the anomaly sign detection device 100 are described as computing devices such as a server, but each component may be realized in the form of a server system connected via a network. Through the description of the configurations of the model generation device 60 and the anomaly sign detection device 100, the method for generating an anomaly sign detection model and the method for generating training data according to the present disclosure will be explained and understood.

[0026] <Hardware configuration of the model generating device 60 and the abnormality sign detection device 100> FIG. 2 is a block diagram showing the hardware configuration of the model generation device 60 and the anomaly sign detection device 100 according to an embodiment of the present disclosure.

[0027] 2, the model generating device 60 and the abnormality sign detection device 100 are each realized by any computing device or information processing device such as a server, and have a storage unit 11, a processing unit 12, a communication unit 13, and an input / output unit 14 that are interconnected via a bus B. Note that the model generating device 60 and the abnormality sign detection device 100 may each be composed of a plurality of computers that perform distributed processing, or may be realized by a virtual computer virtually constructed by software.

[0028] The storage unit 11 includes a main storage device such as a Read Only Memory (ROM) and a Random Access Memory (RAM), and an auxiliary storage device such as a Hard Disk Drive (HDD), a Solid State Drive (SSD), a flash memory, etc. The storage unit 11 stores programs, instructions, data, etc. for executing functions and processes of the model generating device 60 and the anomaly sign detection device 100 described later.

[0029] The processing unit 12 is equipped with a hardware processor such as a CPU (Central Processing Unit), and executes programs, instructions, etc. stored in the memory unit 11 to perform the functions and processing of the model generation device 60 and the anomaly sign detection device 100 described below.

[0030] The communication unit 13 executes various communication processes for wired or wireless communication with external devices (e.g., the training database 20, the production equipment 200, etc.) via a network such as the Internet, a LAN (Local Area Network), or a WLAN (Wireless LAN).

[0031] The input / output unit 14 includes various interfaces such as a Universal Serial Bus (USB) and a High-Definition Multimedia Interface (HDMI (registered trademark)) for connecting to various peripheral devices such as a keyboard, a mouse, a microphone, a display, a printer, a speaker, a sensor, and an alarm (e.g., a buzzer, a lamp, etc.). For example, the input / output unit 14 may notify a user of the abnormality sign detection device 100 of the detection of an abnormality and an abnormality sign. In this embodiment, the user of the abnormality sign detection device 100 may include an administrator of the abnormality sign detection device 100, or an operator who produces a wound body 204 (see FIG. 3 described later) using the production equipment 200, or an operator who maintains the production equipment 200, etc.

[0032] The storage unit 11 and the processing unit 12 do not have to be configured as an integrated server or computer. In other words, the storage unit 11 and the processing unit 12 may be configured separately from each other and located at separate locations as long as they are configured to be able to communicate with each other. In addition, the model generating device 60 and the anomaly sign detection device 100 may have an operation unit not shown in FIG. 2 and may receive operation input from the outside. The details of the storage unit 11 and the processing unit 12 will be described later.

[0033] <Production facilities 200> In this embodiment, the production facility 200 is a production facility for lithium ion secondary batteries. As shown in Fig. 3, the production facility 200 has a winding unit 201 and an inspection machine 207. However, the production facility 200 according to the present disclosure is not limited to this and may be a production facility for any other product, etc.

[0034] 3 is a schematic diagram showing a configuration example of the winding unit 201 and the inspection machine 207. The winding unit 201 has a first supply reel 50, a second supply reel 51, a cutting unit 208, a first laminating roll 205A, a second laminating roll 205B, a winding core 206, and a winding core rotation drive unit 206M. The inspection machine 207 has a sensor 40 for measuring the length of the first sheet material after cutting and a sensor 41 for measuring the length of the second sheet material after cutting.

[0035] The winding section 201 laminates a first sheet material 202 supplied from a first supply reel 50 and a second sheet material 203 supplied from a second supply reel 51 using a first laminating roll 205A and a second laminating roll 205B, and winds them up around a winding core 206 to produce a wound body 204.

[0036] 4 is a schematic diagram of the wound body 204 produced in the winding section 201. For convenience, FIG. 4 shows a state in which the ends of the first sheet material 202 and the second sheet material 203 constituting the wound body 204 are not wound. As shown in FIG. 4, the second sheet material 203 is formed to have a larger width (length along the axial direction of the wound body 204) than the first sheet material 202. The first sheet material 202 is, for example, a sheet-like member (positive electrode sheet) coated with a positive electrode material, and the second sheet material 203 is, for example, a sheet-like member (negative electrode sheet) coated with a negative electrode material.

[0037] Returning to Fig. 3, the explanation will be continued. The core rotation drive unit 206M drives the core 206 at a desired rotation speed. The cutting unit 208 cuts the first sheet material 202 and the second sheet material 203 when winding on the core 206 is completed. Note that, although Fig. 3 shows the cutting unit 208 being disposed at a position upstream of the first laminating roll 205A and the second laminating roll 205B, the cutting unit 208 may be disposed at a position downstream of these laminating rolls.

[0038] The inspection machine 207 is a device that inspects the wound body 204 produced by the winding section 201 .

[0039] During the production of the wound body 204 in the winding section 201, defective products may occur. Defective products may occur, for example, when the first sheet material 202 or the second sheet material 203 wound around the winding core 206 is excessively long or insufficient in length after cutting. If the first sheet material 202 or the second sheet material 203 is excessively long or insufficient in length, the quality of the lithium ion secondary battery deteriorates. Therefore, the lengths of the first sheet material 202 and the second sheet material 203 are measured using a first sheet material post-cut length measuring sensor 40 and a second sheet material post-cut length measuring sensor 41.

[0040] FIG. 3 shows the first sheet material post-cut length measuring sensor 40 and the second sheet material post-cut length measuring sensor 41 positioned downstream of the first laminating roll 205A and the second laminating roll 205B, but each sensor may also be positioned upstream of these laminating rolls.

[0041] The measurement results of the first sheet material post-cut length measuring sensor 40 and the second sheet material post-cut length measuring sensor 41 are judged to be either "good" or "bad". If the measurement result is within a predetermined target range, it is judged to be "good", and if it is not within the target range, it is judged to be "bad". FIG. 5 shows an example of the correspondence between the measurement result of the first sheet material 202 and the judgment result. That is, the measurement data I1 to I3 in which the length after cutting of the first sheet material 202 is equal to or longer than the target length is judged to be "good", and the measurement data I4 and I5 in which the length after cutting of the first sheet material 202 is shorter than the target length are judged to be "bad". In this embodiment, the measurement result is judged to be either "good" or "bad", but it may be judged to be any of three or more types.

[0042] In addition, in FIG. 5, the length measurement data I1 to I5 are arranged in a time series. That is, FIG. 5 shows the state in which the length of the first sheet material 202 after cutting gradually shortens with the deterioration of the production equipment 200 over time. However, in reality, the way in which the length of the first sheet material 202 changes becomes complicated due to various factors other than the deterioration of the production equipment 200 over time, and it is not desirable to judge the abnormality of the production equipment 200 simply by the length. For this reason, it is desirable for the model generation device 60 to train a machine learning model, specifically, to read time series data in which the change in the length of the first sheet material 202 is characterized using a neural network model or the like, thereby training the machine learning model in advance as to which of the time series data corresponds to the "normal period", the "premonition period", and the "abnormal period". In this case, the abnormality sign detection device 100 uses a machine learning model trained by the model generation device 60 to detect abnormalities and signs of abnormalities in the production equipment 200 based on time-series data in which changes in the length of the first sheet material 202 and the second sheet material 203 have been converted into feature quantities. Note that, for the sake of convenience, the following description will be given of a case in which changes in the length of the first sheet material 202 are used, but it goes without saying that changes in the length of the second sheet material 203 or changes in the lengths of both sheet materials may also be used.

[0043] <Details of the model generating device 60> Next, the functional configuration and operation of the model generating device 60 according to an embodiment of the present disclosure will be described in detail.

[0044] The model generation device 60 generates an anomaly sign detection model that detects signs of anomaly and anomalies in the production facility 200 based on actual production data related to the production of products by the production facility 200. After generating the anomaly sign detection model, the model generation device 60 provides the generated anomaly sign detection model to the anomaly sign detection device 100 that detects signs of anomaly and anomalies in the operating production facility 200. Note that, in this embodiment, anomaly sign detection refers to detecting signs of anomaly before an anomaly occurs in the production facility 200 in order to prevent the occurrence of anomalies such as deterioration and failure of the production facility 200 and to maintain normal operation.

[0045] FIG. 6 is a block diagram illustrating a functional configuration of a model generating device 60 according to the first embodiment of the present disclosure.

[0046] As shown in the figure, the model generating device 60 has a preprocessing unit 61, an anomaly degree calculation unit 62, and a training unit 63. For example, one or more functional units of the preprocessing unit 61, the anomaly degree calculation unit 62, and the training unit 63 may be realized by a processing unit 12 having one or more processors executing a program or instructions stored in a storage unit 11 having one or more memories.

[0047] The pre-processing unit 61 acquires feature amount data of the training production performance data. Specifically, the pre-processing unit 61 may receive the training production performance data from the training database 20, and extract the feature amount data by performing a predetermined pre-processing on the received production performance data. As an example, the pre-processing unit 61 may extract production performance unit data of a predetermined number of units or unit time from the time-series production performance data, and perform statistical processing on each of the extracted production performance unit data, and acquire one or more acquired statistics as the feature amount data.

[0048] The pre-processing unit 61 can perform the same pre-processing on both types of production result data, that is, labeled production result data and unlabeled production result data, to derive feature amount data.

[0049] [Production performance database 71] The pre-processing unit 61 may, for example, store the production performance data acquired from the production facility 200 in the production performance database 71 of the storage unit 11. That is, the production performance database 71 may be a database in which the production performance data PD relating to the production performance of the production facility 200 is cumulatively registered. The production performance data PD of a plurality of production facilities 200 may be registered in the production performance database 71.

[0050] Fig. 7 is a diagram showing an example of the production history database 71. In the example shown in Fig. 7, "production date and time", "equipment", "first sheet material measurement value", "second sheet material measurement value", "first sheet material measurement value judgment result", and "second sheet material measurement value judgment result" are registered in association with each other in the production history database 71. These data are associated one by one to configure one piece of production history data PD.

[0051] "Production date and time" is data relating to the production date and time when the wound body 204 was produced.

[0052] "Facility" is data for identifying the facility that produced the wound body 204 when there are multiple production facilities 200. In Fig. 7, as an example, "A", "B", and "C" are shown as identifiers of different production facilities 200.

[0053] The "measurement value of the first sheet material" and the "measurement value of the second sheet material" are the measurement value data of the first sheet material 202 and the measurement value data of the second sheet material 203 used in producing the wound body 204. These measurement value data are acquired by the first sheet material post-cut length measuring sensor 40 and the second sheet material post-cut length measuring sensor 41.

[0054] The "first sheet material measurement value judgment result" and the "second sheet material measurement value judgment result" are data showing the judgment results (see FIG. 5) of the "first sheet material measurement value" and the "second sheet material measurement value". For example, when the target length of the first sheet material 202 is 761 mm, the first sheet material measurement values ​​of 762.2 mm and 761.4 mm are judged as "good", and the first sheet material measurement value of 760.1 mm is judged as "bad". Also, when the target length of the second sheet material 203 is 692 mm, the second sheet material measurement values ​​of 693.4 mm and 693.5 mm are judged as "good", and the first sheet material measurement value of 691.1 mm is judged as "bad". These judgments are automatically performed by the processing unit 120 or the inspection machine 207, for example.

[0055] Such production performance data PD is input automatically or manually by an operator, for example, every time a wound body 204 is produced in the production facility 200, and is registered in the training database 20. Typically, since it is desired to generate an anomaly sign detection model before the production facility 200 starts full-scale operation, the production performance data for training is collected in an experimental stage, a test operation stage, or the like before the production facility 200 starts full-scale operation, and stored in the training database 20.

[0056] In the above-described embodiment, the training production record data is stored in the training database 20, and the pre-processing unit 61 extracts the production record data from the training database 20 and extracts the feature amount data from the extracted production record data, but the present disclosure is not limited to this. For example, the training database 20 may store the training feature amount data extracted from the training production record data. In this case, the pre-processing unit 51 does not need to hold the production record database 61.

[0057] [Feature database 72] The pre-processing unit 61 may store, for example, feature amounts extracted from the production performance data in the feature amount database 72 of the storage unit 11. That is, the feature amount database 72 is a database in which feature amount data FD is registered. The feature amount data FD may be data obtained by collecting a plurality of consecutive production performance data PD for a predetermined number of counting units W and performing data processing such as statistical processing on the collected production performance data PD to convert the collected production performance data PD into feature amounts. The feature amount database 72 may register feature amount data FD obtained from the production performance data PD of a plurality of production facilities 200.

[0058] Fig. 8 is a diagram showing an example of the feature database 72. In the example shown in Fig. 8, the feature database 72 registers "production date and time", "equipment", "counting unit", "maximum value of the first sheet material measurement value", "minimum value of the first sheet material measurement value", "average value of the first sheet material measurement value", "standard deviation of the first sheet material measurement value", and "number of defectives of the first sheet material measurement value" in association with each other. These data are associated one by one to form one piece of feature data FD.

[0059] The "production date and time" is data relating to the production date and time of the production history data PD with the latest production date and time among the collected W pieces of production history data PD, and is a production date and time that represents the production dates and times of the W pieces of production history data PD. In other words, the "production date and time" is the production date and time of the production history data PD corresponding to the last produced wound body 204 among the W wound bodies 204 corresponding to the data-processed W pieces of production history data PD.

[0060] "Facility" is data for identifying the production facility 200 that produced the wound body 204 corresponding to the collected W pieces of production performance data PD.

[0061] The "counting unit" is data related to the number of collected production result data PD. In Fig. 8, "100" is shown as an example.

[0062] The "maximum value of the first sheet material measurement value", "minimum value of the first sheet material measurement value", "average value of the first sheet material measurement value" and "standard deviation of the first sheet material measurement value" are data indicating each statistical value of the first sheet material measurement value included in the collected W pieces of production performance data PD. In addition to the statistical values ​​shown in Fig. 8, the feature database 112 may also register statistical values ​​such as the median, the first quartile, and the third quartile. These statistical values ​​are the feature values.

[0063] The "number of defective first sheet material measurement values" is data indicating the number of wound bodies 204 whose "first sheet material measurement value judgment result" was "defective" included in the W wound bodies 204 corresponding to the collected W pieces of production performance data PD.

[0064] Such feature data FD may be extracted by the pre-processing unit 61 performing a predetermined pre-processing each time production result data PD of the aggregated unit number is registered in the production result database 71, and the extracted feature data may be registered in the feature database 72.

[0065] The abnormality degree calculation unit 62 uses the abnormality degree calculation model to calculate the abnormality degree of the acquired feature amount data.

[0066] Here, the anomaly degree calculation model may be generated from a group of features in a normal state based on any anomaly sign detection technology. When feature data is received as an input, the anomaly degree calculation model outputs the anomaly degree of the feature data. The anomaly degree calculation model according to the present disclosure may be generated according to any known anomaly sign detection technology (e.g., MT method, etc.) using feature data in a normal state derived from production performance data collected during a period immediately after operation under the assumption that the production equipment 200 is in a normal state immediately after operation. For example, the anomaly degree of the feature calculated by the anomaly degree calculation model may be a deviation (e.g., Mahalanobis distance, etc.) from a group of features in a normal state.

[0067] As an example, the anomaly calculation model M1 may be a model that expresses the deviation of the production facility 200 from the normal state with a numerical value called anomaly A. The anomaly A indicates the degree of anomaly of the production facility 200. In other words, the anomaly A is an index indicating how much the feature amount data FD of the anomaly calculation target deviates from the feature amount data FD in a normal period. If the anomaly calculation model M1 is appropriately generated, the greater the anomaly amount A, the greater the degree of anomaly such as deterioration and failure in the production facility 200, and the higher the probability that a defective wound body 204 will be newly produced. The anomaly calculation model M1 calculates the deviation from the normal state as the anomaly amount A by defining the normal state using the feature amount data FD in a normal period in which no defects occur. This makes it possible to calculate the anomaly even when the production performance data for training the anomaly calculation model M1 is collected immediately after the operation of the production facility 200, and the frequency of occurrence of anomalies due to aging or the like is low, and a sufficient number of sample data of the abnormal state cannot be obtained. The normal period is a period during which no deterioration or breakdown occurs in the production equipment 200 and no defects occur in the wound body 204 produced by the production equipment 200.

[0068] The anomaly degree calculation model M1 may be generated using a combination of multiple types of feature amounts included in the feature amount data FD. The selection of the feature amounts to be combined may be performed by a user of the anomaly sign detection device 100, or may be automatically performed by the anomaly degree calculation unit 62 based on a predetermined criterion.

[0069] Furthermore, by changing the combination of feature quantities, it is possible to generate a plurality of mutually different anomaly degree calculation models M1. Note that only one anomaly degree calculation model M1 may be generated.

[0070] The abnormality degree calculation model M1 is generated based on production result data PD produced during a period when the production facility 200 is assumed to be in a normal state, such as immediately after the production facility 200 starts operating.

[0071] Here, the anomaly degree calculation model M1 may be appropriately selected from a number of different formats, such as a statistical model, a machine learning model generated by machine learning (for example, a neural network model), and the like.

[0072] As an example, a known MT method may be used to generate the anomaly calculation model M1. This makes it possible to generate an anomaly calculation model M1 that can calculate an anomaly A that more accurately indicates the degree of deviation from the normal state. When the degree of deviation from the normal state is used as the degree of anomaly, the degree of anomaly calculated for the feature data is compared with an anomaly threshold that indicates an abnormal state, and the state of the feature data is determined according to the comparison result.

[0073] As another example, when generating the anomaly calculation model M1, a known distance-based anomaly sign detection method such as OC-SVM (One Class Support Vector Machine), an Auto Encoder using a neural network, or a K-nearest neighbor method may be adopted.

[0074] The anomaly degree calculation model M1 thus generated may be verified as to whether the calculated anomaly degree is appropriate. For example, the generated anomaly degree calculation model M1 may be verified based on the anomaly degree calculated for the feature amount data and the defect rate in the production performance data corresponding to the feature amount data. That is, when the anomaly degree derived from the anomaly degree calculation model M1 for the input feature amount data exceeds the anomaly threshold, the feature amount data is assumed to correspond to the production performance data of the production facility 200 during the abnormal period, and should have a positive correlation with the defect rate for the production performance data. Here, the anomaly threshold may be set based on the anomaly degree at which the production facility 200 is determined to be in an abnormal state.

[0075] Therefore, when the correlation coefficient between the degree of abnormality equal to or greater than the anomaly threshold derived from the anomaly degree calculation model M1 and the corresponding defect rate is equal to or greater than a predetermined threshold, it is determined that the anomaly degree calculation model M1 is capable of appropriately discriminating between a normal state and an abnormal state of the feature amount data. In this case, the anomaly degree calculation model M1 is determined to have been appropriately trained, and may be used in the subsequent labeling process of the feature amount data.

[0076] On the other hand, if the correlation coefficient between the degree of abnormality equal to or greater than the abnormality threshold derived from the anomaly degree calculation model M1 and the corresponding defective rate is less than a predetermined threshold, it is determined that the anomaly degree calculation model M1 is not capable of appropriately discriminating between a normal state and an abnormal state of the feature amount data. In this case, the anomaly degree calculation model M1 is determined to have not been appropriately trained, and may be regenerated, for example, by using other training data.

[0077] The above-mentioned verification of the anomaly degree calculation model M1 may be performed in advance (manually or the like) before use by the anomaly degree calculation model 62, or may be performed by the anomaly degree calculation unit 62.

[0078] In the above-described embodiment, the anomaly degree calculation model M1 generated in advance is provided in the anomaly degree calculation unit 62, but the present disclosure is not limited to this. For example, the anomaly degree calculation unit 62 may itself generate an anomaly degree calculation model using training data. Specifically, the anomaly degree calculation unit 62 may similarly generate the anomaly degree calculation model M1 in accordance with the MT method or the like as described above, using labeled production performance data or feature amount data stored in the training database 20.

[0079] The selection of the format of the abnormality degree calculation model M1 may be performed by a user of the abnormality sign detection device 100, or may be performed automatically by the abnormality degree calculation unit 62.

[0080] [Abnormality Database 73] The anomaly degree calculation unit 62 may store the calculated degree of anomaly in an anomaly degree database 73 in the storage unit 11. That is, the anomaly degree database 73 is a database in which anomaly degree data AD related to the degree of anomaly A calculated by the anomaly degree calculation model M1 is registered. The anomaly degree database 73 may also register anomaly degree data AD obtained based on production performance data PD of a plurality of production facilities 200.

[0081] Fig. 9 is a diagram showing an example of the abnormality degree database 73. In the example shown in Fig. 9, "production date and time", "equipment", "abnormality degree calculation model ID", "abnormality degree", "counting unit", and "number of defective measurement values ​​of first sheet material" are registered in association with each other in the abnormality degree database 73. These data are associated one by one to configure one piece of abnormality degree data AD.

[0082] The "production date and time" is the production date and time included in the feature amount data FD of the target for anomaly calculation.

[0083] The "abnormality degree calculation model ID" is data for identifying the abnormality degree calculation model M1 used in the abnormality degree calculation. As an example, in Fig. 9, "001", "002", and "003", which are IDs (identification numbers) of the different abnormality degree calculation models M1, are shown.

[0084] The "degree of abnormality" is data indicating the degree of abnormality A calculated from the feature amount data FD of the object of the degree of abnormality calculation using the degree of abnormality calculation model M1.

[0085] The "counting unit" is the counting unit included in the feature amount data FD for which anomaly degree is to be calculated.

[0086] The "number of defective first sheet material measurement values" is data indicating the number of defective first sheet material measurement values ​​included in the feature amount data FD of the target of anomaly degree calculation. This number of defectives is used to evaluate the reliability of the anomaly degree calculation model, as will be described in detail below.

[0087] Such abnormality degree data AD may be generated by the abnormality degree calculation unit 62 and registered in the abnormality degree database 73 every time feature amount data FD is registered in the feature amount database 72.

[0088] The training unit 63 determines the state of the feature data based on the calculated degree of abnormality, and trains an anomaly sign detection model that detects the state of the feature data from the feature data using training data composed of the feature data and the state of the feature data.

[0089] Here, the anomaly sign detection model receives feature data as input and outputs the state of the input feature data. The training data used to generate the anomaly sign detection model may be composed of feature data derived from production performance data and the state of the feature data determined using the anomaly degree calculation model. In other words, the training data used to generate the anomaly sign detection model may be production performance data whose state is not labeled in the training database 20, and the training unit 63 may determine the state of the feature data based on the degree of anomaly calculated by the anomaly degree calculation unit 62, and label the production performance data or the feature data according to the determined state.

[0090] In one embodiment, the training unit 63 may set the warning period based on the average value of the abnormality degree in the normal period and the average value of the abnormality degree in the abnormal period. Specifically, the training unit 63 first identifies a period in which the defect rate is equal to or greater than the abnormality threshold as the abnormal period, and calculates an average value A of the abnormality degree in the normal period and an average value B of the abnormality degree in the abnormal period used to train the abnormality degree calculation model M1. The training unit 63 may then calculate an average value C of the two calculated average values ​​A and B, and set all periods in which the abnormality degree exceeds the average value C between the normal period and the abnormal period as warning periods. By setting the normal period, warning period, and abnormal period in this manner, the training unit 63 can label each piece of feature data by a normal period, warning period, or abnormal period.

[0091] In one embodiment, the training unit 63 may set the warning period based on the start point and the length of the abnormal period. Specifically, the training unit 63 first identifies a period in which the defect rate is equal to or greater than the abnormal threshold as the abnormal period, and identifies a period having the same length as the abnormal period by going back from the start point of the abnormal period. The training unit 63 may then calculate an average value A of the abnormality degree in the normal period and an average value B of the abnormality degree in the abnormal period, and set the identified period as the warning period if an average value D of the abnormality degree in the identified period is within a range between the average value A and the average value B. By setting the normal period, warning period, and abnormal period in this manner, the training unit 63 can label each piece of feature data by the normal period, warning period, or abnormal period.

[0092] In one embodiment, the training unit 63 may set the predictive period based on the start point of the abnormal period, the average value of the abnormality degree in the normal period, and the average value of the abnormality degree in the abnormal period. Specifically, the training unit 63 first identifies a period in which the defect rate is equal to or greater than the abnormality threshold as the abnormal period, and calculates an average value A of the abnormality degree in the normal period and an average value B of the abnormality degree in the abnormal period. Then, the training unit 63 may set, as the predictive period, a period that is a period going back from the start point of the abnormal period and in which the average value E of the abnormality degree in the period is equal to or greater than the average value C of the two calculated average values ​​A and B. By setting the normal period, predictive period, and abnormal period in this manner, the training unit 63 can label each piece of feature data as a normal period, predictive period, or abnormal period.

[0093] The above-mentioned labeling of the normal period, the precursor period, and the abnormal period is merely an example, and the labeling of the training data for the abnormality sign detection model according to the present disclosure is not limited thereto. For example, the training unit 63 may perform labeling based on a comparison between the degree of abnormality and any appropriately set precursor threshold and abnormality threshold. Specifically, when the degree of abnormality calculated for the feature data to be discriminated is less than the precursor threshold, the training unit 63 may discriminate the state of the feature data as a normal period and label the feature data as a "normal period."

[0094] In addition, if the degree of abnormality calculated for the feature data to be discriminated is equal to or greater than the prognostic threshold and less than the abnormality threshold, the training unit 63 may discriminate the state of the feature data as a prognostic period and label the feature data as a "prognostic period."

[0095] In addition, if the degree of abnormality calculated for the feature data to be discriminated is equal to or greater than the abnormality threshold, the training unit 63 may discriminate the state of the feature data as being in an abnormal period and label the feature data as an "abnormal period."

[0096] The training unit 63 stores each piece of feature amount data labeled in this manner as state label data, and uses the state label data as training data for the anomaly sign detection model M2.

[0097] [Condition Label Database 74] The training unit 63 may store the generated state label data in the state label database 74 of the storage unit 11. That is, the state label database 74 is a database in which state label data LD is registered to time-series data in which the anomaly degree data AD registered in the anomaly degree database 73 is arranged in order of production date and time. The time-series data is divided for each production facility 200.

[0098] Fig. 10 is a diagram showing an example of the condition label database 74. In the example shown in Fig. 10, a "condition label," "equipment," "start date and time," "end date and time," and an "anomaly degree calculation model ID" are registered in association with each other in the condition label database 74. These data are associated one by one to configure one condition label data.

[0099] The "status label" is data indicating one of three status stages: a "normal period," a "premonitory period," and an "abnormal period." Note that the premonitory period is a period in which a premonitory period of an abnormality appears in the production equipment 200, and the abnormal period is a period in which an abnormality occurs in the production equipment 200.

[0100] "Facility" is an identifier of the production facility 200 corresponding to the time-series data to which the state label L is assigned.

[0101] The "start date and time" and "end date and time" are the date and time of the first (oldest) data and the date and time of the last (newest) data in the anomaly level data AD constituting the time-series data to which the status label L is assigned. In other words, they are the date and time when the period of the status indicated by the status label L starts and ends.

[0102] The state label data LD is generated by the training unit 63 and registered in the state label database 74 .

[0103] The abnormality sign detection model M2 according to an embodiment of the present disclosure is a machine learning model that predicts whether the state of the production facility 200 is a "normal period", a "premonition period", or an "abnormal period", and receives feature data as an input and outputs the state of the feature data. For example, the abnormality sign detection model M2 is a machine learning model for predicting what kind of time-series changes the production performance data PD has undergone before the abnormality occurs in the production facility 200 where many defective products have occurred due to abnormalities such as deterioration and failure. That is, the abnormality sign detection model M2 can be used to detect whether an abnormality such as deterioration and failure has occurred in the production facility 200, or whether a premonition of an abnormality has appeared. When the production performance data PD of the production facility 200 is input to the production performance database 71, the abnormality sign detection model M2 outputs a result of determining whether the state of the production facility 200 at the time when the production performance data PD was obtained is a "normal period", a "premonition period", or an "abnormal period" based on the feature data derived from the input production performance data.

[0104] For example, the training unit 63 uses training data consisting of pairs of feature data and the state of the feature data acquired from the production performance data extracted from the training database 20 to train the abnormality sign detection model M2 to be trained by supervised learning. For example, when the abnormality sign detection model M2 is realized as a neural network, the training unit 63 inputs the feature data of each training data to the neural network and compares the output result from the neural network with the state of the feature data of the training data. Then, the training unit 63 adjusts the parameters of the neural network according to backpropagation or the like according to the error between the output result and the state of the training feature data. The training unit 63 repeats the parameter adjustment process until a predetermined termination condition is satisfied. The predetermined termination condition may be, for example, that the parameter adjustment process has been performed on all prepared training data, that the error has become less than a predetermined threshold, that the error has converged within a predetermined threshold range, or the like.

[0105] When the training of the abnormal sign detection model M2 is completed, the training unit 63 provides the trained abnormal sign detection model M2 to the abnormal sign detection device 100.

[0106] In addition, one abnormality sign detection model M2 may be generated using the production performance data PD of multiple production facilities 200, or one abnormality sign detection model M2 may be generated using the production performance data PD of one production facility 200.

[0107] The format of the anomaly sign detection model M2 is not particularly limited, but in order to further improve the determination accuracy, it is desirable to employ a machine learning model such as a neural network model. The selection of the model to be employed in the anomaly sign detection model M2 may be performed by a user of the anomaly sign detection device 100, or may be automatically performed by the training unit 63.

[0108] <Details of the abnormality sign detection device 100> Next, an abnormality sign detection device 100 according to an embodiment of the present disclosure will be described with reference to Fig. 11. The abnormality sign detection device 100 uses the abnormality sign detection model generated by the model generation device 60 to detect an abnormality or a sign of an abnormality that has occurred in the production facility 200 based on actual production data of the production facility 200 to be detected.

[0109] FIG. 11 is a block diagram showing a functional configuration of an abnormality sign detection device 100 according to one embodiment of the present disclosure.

[0110] As shown in FIG. 11, the abnormality sign detection device 100 has a preprocessing unit 101 and an abnormality sign detection unit 102.

[0111] The pre-processing unit 101 acquires feature quantities of production performance data of the detection target. Specifically, when the pre-processing unit 101 acquires production performance data related to products, etc., produced by the production facility 200, the pre-processing unit 101 executes a predetermined pre-processing on the acquired production performance data to acquire feature quantity data of the production performance data.

[0112] Here, the production performance data acquired by the pre-processing unit 101 is typically of the same type as the data used to generate the abnormality sign detection model by the model generating device 60. For example, when the model generating device 60 uses the production performance data of six data types, namely, "production date and time", "equipment", "first sheet material measurement value", "second sheet material measurement value", "first sheet material measurement value judgment result", and "second sheet material measurement value judgment result", as training data, the pre-processing unit 101 also acquires the production performance data of the detection target of these six data types from the production equipment 200.

[0113] Then, the preprocessing unit 101 performs a predetermined preprocessing on the acquired production performance data to acquire feature amount data of the production performance data. Specifically, the preprocessing unit 101 performs the same processing as the preprocessing in the model generating device 60 on the production performance data to be detected, and derives feature amount data of the production performance data. For example, when the preprocessing in the model generating device 60 is a statistical processing and the feature amount data is composed of eight data types, namely, "production date and time", "equipment", "counting unit", "maximum value of the first sheet material measurement value", "minimum value of the first sheet material measurement value", "average value of the first sheet material measurement value", "standard deviation of the first sheet material measurement value", and "number of defects of the first sheet material measurement value", the preprocessing unit 101 also performs the same statistical processing on the production performance data to be detected, and derives feature amount data of the same type.

[0114] When the feature data is acquired in this manner, the pre-processing unit 101 passes the acquired feature data to the anomaly sign detection unit 102. Note that when the pre-processing unit 101 acquires the feature data of the production result data instead of the production result data of the detection target, the pre-processing unit 101 does not need to execute the process for deriving the feature data described above, and it is sufficient to simply pass the acquired feature data to the anomaly sign detection unit 102.

[0115] The anomaly sign detection unit 102 detects the state of the feature data from the feature data by using a trained anomaly sign detection model. Here, the anomaly sign detection model is trained by training data composed of feature data of training production performance data and the state of the feature data. In addition, the state of the feature data is determined based on the degree of anomaly of the feature data calculated by using an anomaly degree calculation model that represents a group of features in a normal period.

[0116] Specifically, the anomaly sign detection unit 102 inputs the feature amount data received from the pre-processing unit 101 to a trained anomaly sign detection model, and acquires the state of the input feature amount data as an output from the trained anomaly sign detection model. As described above in relation to the model generation device 60, the anomaly sign detection model may be any type of machine learning model, such as a neural network, trained by supervised learning using feature amount data labeled using a trained anomaly degree calculation model as training data.

[0117] In one embodiment, the anomaly calculation unit 62 may evaluate the reliability of the anomaly calculation model by using the anomaly data AD registered in the anomaly database 63. For example, the anomaly calculation unit 62 calculates a correlation coefficient C between the defect rate DR, which is obtained by dividing the number of defects in the anomaly data AD by the number of counting units W, and the anomaly A. When the correlation coefficient C is a predetermined threshold value, a correlation threshold Th c If the correlation coefficient C is greater than the correlation threshold Th, the abnormality degree calculation unit 62 can determine that the abnormality degree A calculated by the abnormality degree calculation model M1 appropriately represents the degree of abnormality of the production equipment 200. c may be adopted as the abnormality calculation model M1 for the production facility 200.

[0118] In addition, the correlation threshold Th c may be the minimum value of the correlation coefficient C at which a correlation is recognized between the degree of abnormality A calculated by the degree of abnormality calculation model M1 and the number of defects of the wound body 204 produced by the production facility 200. In other words, it is the minimum value of the correlation coefficient C assumed when a combination of the features of the feature data FD used to generate the degree of abnormality calculation model M1 represents the characteristics of an abnormality in the production facility 200. In other words, it is the minimum value of the correlation coefficient C at which it can be considered that the degree of abnormality calculation model M1 is appropriately calculating the degree of abnormality in the production facility 200. Note that, when the predetermined correlation threshold Th c may be empirically determined based on, for example, past production performance data PD.

[0119] Then, the training unit 63 performs the following process each time feature data FD is registered in the feature database 72 (i.e., each time anomaly degree data AD is registered in the anomaly degree database 73). That is, the training unit 63 obtains a defect rate DR by dividing the number of defects in the registered feature data FD (or anomaly degree data AD) by the number of counting units W. The training unit 63 calculates the defect rate DR by dividing the number of defects in the registered feature data FD (or anomaly degree data AD) by the number of counting units W. dr The defect rate DR is compared with the defect rate threshold value Th dr If it is greater than , the training unit 63 generates condition label data LD by assigning a condition label L of "normal period", "premonitory period", or "abnormal period" to time-series data in which the abnormality degree data AD registered in the abnormality degree database 73 is arranged in order of production date and time. Then, the training unit 63 registers the condition label data LD in the condition label database 74.

[0120] The training unit 63 then learns time-series data in which the feature data FD are arranged in chronological order of production date and time, and generates an anomaly sign detection model M2. The time-series data used for learning is set as a starting point for feature data FD whose production date and time is included between the start date and time and the end date and time of each of the state label data LD for the "normal period", "sign period", and "abnormal period", and arranges a predetermined number of past feature data FD from the starting point. The anomaly sign detection model M2 generated by the training unit 63 is provided to the anomaly sign detection device 100, as described above.

[0121] In one embodiment, the training unit 63 may compare the determination accuracy of the anomaly degree calculation model M1 adopted by the anomaly degree calculation unit 62 with the anomaly sign detection model M2 generated by the training unit 63, and adopt the anomaly sign detection model M2 if the determination accuracy of the anomaly sign detection model M2 is higher. In other words, the generation of the anomaly sign detection model by the training unit 63 (learning of the learning data list) is continued until the determination accuracy of the anomaly sign detection model M2 exceeds the determination accuracy of the anomaly degree calculation model M1. The adopted anomaly sign detection model M2 is used in the anomaly sign detection device 100 to detect an anomaly or a sign of an anomaly in the production equipment 200.

[0122] Then, the abnormality sign detection unit 102 judges the state of the production facility 200 using the feature amount data FD generated from the production performance data PD of the wound body 204 newly produced in the production facility 200 and the abnormality sign detection model M2 adopted by the model generating device 60. Specifically, the abnormality sign detection unit 102 judges the production performance data of the detection target into three stages, namely, "normal period", "sign period" and "abnormal period". If the judgment result is "abnormal period" or "sign period", the abnormality sign detection unit 102 may notify the user or the like who manages the production facility 200 of the detection of the abnormality or the sign. This notification may include an alarm to attract the user's attention, a display notifying which production facility 200 an abnormality or a sign of an abnormality occurred and when, and the like.

[0123] <Details of each process for model generation> Below, a detailed description will be given of each process for generating an anomaly sign detection model by one implementation of the model generation device 60 shown in Fig. 6. Note that the process described below is merely an example, and the generation of an anomaly sign detection model according to the present disclosure may be realized by a generation process according to the production facility 200, the product, etc.

[0124] (Pretreatment) The following describes the pre-processing executed by the pre-processing unit 61. Fig. 12 is a flowchart for explaining the pre-processing according to an embodiment.

[0125] In step S11, the pre-processing unit 61 determines whether or not a predetermined number, S, of new production performance data PD have been registered in the production performance database 71. If S pieces of new production performance data PD have not been registered (step S11: NO), the pre-processing unit 61 repeats step S11. If S pieces of new production performance data PD have been registered (step S11: YES), the pre-processing unit 61 advances the process to step S12.

[0126] In step S12, the pre-processing unit 61 extracts production performance data PD of the counting unit number W from the newly registered S pieces of production performance data PD, and generates a production performance data list PL in which the extracted production performance data PD is listed. In other words, the pre-processing unit 61 extracts the production performance data list PL. The production performance data list PL includes the latest production performance data PD last registered in the production performance database 71. last First, production performance data PD last The production result data list PL is a list of production result data PD extracted from the production date and time of the last production to the production result data PD of the Wth aggregation unit number. last The number of counting units W is equal to or less than a predetermined number S.

[0127] In step S13, the pre-processing unit 61 calculates statistics such as the maximum value, minimum value, average value, and standard deviation of the first sheet material measurement value from the extracted production performance data list PL, and sets these statistics as feature quantities. Note that in addition to the statistics shown here, statistics such as the median, the first quartile, and the third quartile may be calculated and set as feature quantities. Furthermore, when generating feature quantities, pre-processing such as normalization, binning, logarithmic transformation, and difference transformation may be performed in advance, or lag features may be generated by shifting the data a predetermined number of times in the past.

[0128] In step S14, the pre-processing unit 61 counts the number of production performance data PD in which the first sheet material measurement value judgment result included in the extracted production performance data list PL is "failure", and sets this as the number of defects in the first sheet material measurement value.

[0129] In step S15, the preprocessing unit 61 registers the feature data FD including the calculated feature and the number of defects in the feature database 72. The production date and time of the feature data FD is stored in the latest production performance data PD last The production date and time.

[0130] (Generation process of anomaly calculation model) Next, a description will be given of the generation process of an anomaly degree calculation model executed by the anomaly degree calculation unit 62. Note that, when the anomaly degree calculation unit 62 uses a trained anomaly degree calculation model, the generation process of the anomaly degree calculation model does not need to be executed by the anomaly degree calculation unit 62. Fig. 13 is a flowchart for explaining the generation process of an anomaly degree calculation model according to one embodiment.

[0131] In step S21, the abnormality degree calculation unit 62 judges whether or not a predetermined number or more of feature data FD including a production date and time within a normal period (e.g., one week or a period longer than one week) are registered in the feature database 72. The normal period is a period during which no deterioration or failure occurs in the production equipment 200 and no defects occur in the wound body 204 produced by the production equipment 200.

[0132] If the number of feature data FD including the production date and time within the normal period is less than the predetermined number (step S21: NO), the abnormality calculation unit 62 repeats step S21. If the number of feature data FD including the production date and time within the normal period is equal to or greater than the predetermined number (step S21: YES), the abnormality calculation unit 62 advances the process to step S22. In other words, the abnormality calculation unit 62 waits for the predetermined number of feature data FD in which the "number of defective measurement values ​​of the first sheet material" is 0 to be registered in the feature database 72, and then advances the process to step S22.

[0133] In step S22, the abnormality degree calculation unit 62 extracts a predetermined number of feature amount data FD from the feature amount database 72, and generates a feature amount data list FL in which the extracted feature amount data FD is listed. That is, the abnormality degree calculation unit 62 extracts the feature amount data list FL. The feature amount data list FL is a list obtained by extracting a predetermined number of feature amount data FD whose production date and time falls within the above-mentioned normal period from among the feature amount data FD registered in the feature amount database 72.

[0134] In step S23, the anomaly degree calculation unit 62 generates an anomaly degree calculation model M1 by using (for example, by learning) a combination of multiple features possessed by the feature data FD included in the feature data list FL. For example, the anomaly degree calculation unit 62 may generate the anomaly degree calculation model M1 according to any known anomaly sign detection technology (for example, MT method, etc.). For example, the anomaly degree calculation model M1 may calculate the degree of deviation (for example, Mahalanobis distance, etc.) from a group of features in a normal state as the degree of anomaly of the feature data FD. In this way, the anomaly degree calculation unit 62 may generate multiple anomaly degree calculation models M1 by changing the way in which multiple features are combined.

[0135] In step S24, the abnormality degree calculation unit 62 registers the multiple generated abnormality degree calculation models M1.

[0136] (Abnormality calculation process) Next, a description will be given of the abnormality degree calculation process executed by the abnormality degree calculation unit 62. Fig. 14 is a flowchart for explaining the abnormality degree calculation process according to one embodiment.

[0137] In step S31, the abnormality degree calculation unit 62 adds new feature amount data FD new Determine whether the new feature data FD has been registered. new If the feature amount data FD is not registered (step S31: NO), the abnormality degree calculation unit 62 repeats step S31. new If registered (step S31: YES), the abnormality degree calculation unit 62 advances the process to step S32.

[0138] In step S32, the abnormality degree calculation unit 62 selects the newly registered feature amount data FD from the feature amount database 72. new Extract.

[0139] In step S33, the abnormality degree calculation unit 62 reads out a plurality of registered abnormality degree calculation models M1, and outputs each of the read out abnormality degree calculation models M1 and the extracted feature amount data FD newIn other words, one feature data FD new A plurality of anomaly degrees A are calculated from the above. The type of feature used to calculate the anomaly degree A is the same as the type of feature used (for example, learned) when generating the anomaly degree calculation model M1 used to calculate the anomaly degree A.

[0140] In step S34, the abnormality degree calculation unit 62 calculates the abnormality degree A and the feature amount data FD new The number of defects included in the abnormality data AD new and registers the abnormality degree data AD in the abnormality degree database 73. new The production date and time of the feature data FD used to calculate the degree of anomaly A new The production date and time.

[0141] Anomaly Data AD new includes the ID of the anomaly degree calculation model M1 used to calculate the anomaly degree A. Therefore, it is possible to identify which anomaly degree calculation model M1 was used to calculate the anomaly degree A.

[0142] (Adoption decision process for anomaly calculation model) Next, a description will be given of the adoption determination process of an anomaly degree calculation model executed by the anomaly degree calculation unit 62. Note that, when the anomaly degree calculation unit 62 is equipped with only one anomaly degree calculation model, the adoption determination process does not have to be executed by the anomaly degree calculation unit 62. Fig. 15 is a flowchart for describing the adoption determination process of an anomaly degree calculation model according to one embodiment.

[0143] In step S41, the abnormality calculation unit 62 stores new abnormality data AD new It is determined whether or not new abnormality data AD in which the number of defects in the first sheet material measurement value is 1 or more has been registered. new If the number of defects in the first sheet material measurement value is not registered (step S41: NO), the abnormality degree calculation unit 62 repeats step S41. newIf registered (step S41: YES), the abnormality degree calculation unit 62 advances the process to step S42.

[0144] In step S42, the abnormality degree calculation unit 62 extracts a predetermined number of abnormality degree data AD from the abnormality degree database 73, and generates an abnormality degree data list AL that lists the extracted abnormality degree data AD. In other words, the abnormality degree calculation unit 62 extracts the abnormality degree data list AL. The abnormality degree data list AL includes the abnormality degree data AD newly registered in the abnormality degree database 73. new First, the abnormality data AD new The abnormality degree data AD extracted from the production date and time up to a predetermined number of abnormality degree data AD are listed. new The abnormality degree data AD has the same abnormality degree calculation model ID as the abnormality degree calculation model ID of the abnormality degree A. In other words, the abnormality degree calculation unit 62 extracts an abnormality degree data list AL for each abnormality degree calculation model ID that has calculated the abnormality degree A. The abnormality degree data list AL includes at least the newly registered abnormality degree data AD having one or more defects in the first sheet material measurement value. new Contains:

[0145] In step S43, the abnormality calculation unit 62 calculates a defect rate DR for each abnormality data AD included in the abnormality data list AL. The defect rate DR is a value obtained by dividing the number of defects in the first sheet material measurement value included in the abnormality data AD by the number of aggregation units W. Next, the abnormality calculation unit 62 calculates a correlation coefficient C between the defect rate DR and the abnormality A. The calculation of the correlation coefficient C is performed for each abnormality calculation model M1, that is, for each abnormality calculation model ID. The abnormality calculation unit 62 selects the maximum correlation coefficient C from the multiple correlation coefficients C calculated for each abnormality calculation model ID.

[0146] In step S44, the abnormality degree calculation unit 62 calculates the correlation coefficient C based on the correlation threshold Th c It is judged whether the maximum correlation coefficient C is greater than the correlation threshold Th cIn the following case (step S44: NO), the abnormality degree calculation unit 62 returns the process to step S41. c If it is greater than the correlation threshold value Th (step S44: YES), the abnormality degree calculation unit 62 advances the process to step S45. c As described above, this is the minimum value of the correlation coefficient C at which a correlation is recognized between the degree of abnormality A calculated by the degree of abnormality calculation model M1 and the number of defective wound bodies 204 produced by the production facility 200.

[0147] In step S45, the abnormality degree calculation unit 62 adopts the abnormality degree calculation model M1 that has calculated the maximum correlation coefficient C.

[0148] Note that even after adopting the anomaly degree calculation model M1 in step S45, the anomaly degree calculation unit 62 may repeat the processes from step S41 onwards. As a result, there is a possibility that an anomaly degree calculation model M1' different from the adopted anomaly degree calculation model M1 will calculate the anomaly degree A used to calculate the maximum correlation coefficient C. In this case, the anomaly degree calculation unit 62 may adopt the anomaly degree calculation model M1' instead of the anomaly degree calculation model M1. In this case, each process executed by the training unit 63 after the process executed by the anomaly degree calculation unit 62 is executed based on the replaced anomaly degree calculation model M1'.

[0149] (State label setting process) Next, a description will be given of the state label setting process executed by the training unit 63. Fig. 16 is a flowchart for explaining the state label setting process.

[0150] In step S51, the training unit 63 stores in the feature database 72 the defect rate DR and the defect rate threshold value Th dr It is judged whether the feature quantity data FD is registered that is greater than the defect rate threshold value Th dr If the feature amount data FD larger than the defect rate threshold Th is not registered (step S51: NO), the training unit 63 repeats step S51. drIf feature amount data FD larger than is registered (step S51: YES), the training unit 63 advances the process to step S52.

[0151] In step S52, the training unit 63 sets the period of production date and time of the feature amount data FD used when the anomaly degree calculation model M1 adopted in the anomaly degree calculation model adoption determination process was generated by the anomaly degree calculation unit 62 as the normal period. The training unit 63 associates the identifier of the target production equipment 200, the start date and time and the end date and time of the set normal period, and the anomaly degree calculation model ID of the adopted anomaly degree calculation model M1, and sets the normal period state label data LD normal In addition, a period other than the abnormal period or the warning period described later can be set as the normal period. For example, a period that was not used in generating the abnormality degree calculation model M1 but that precedes the period set as the warning period may be set as the normal period.

[0152] In step S53, the training unit 63 performs a test on the feature amount database 72 to determine whether the defect rate DR is a defect rate threshold value Th dr The period of the production date and time of the feature amount data FD that is greater than the abnormal period is set as the abnormal period. The identifier of the target production equipment 200, the start date and time and the end date and time of the set abnormal period, and the abnormality degree calculation model ID of the adopted abnormality degree calculation model M1 are associated with each other, and the abnormality period state label data LD anormaly Generate.

[0153] In step S54, the training unit 63 extracts the abnormal period state label data LD anormaly The start date and time of the prediction period is set as the end date and time, and a time going back a predetermined period from the end date and time is set as the warning period. The identifier of the target production equipment 200, the start date and time and the end date and time of the set warning period, and the anomaly degree calculation model ID of the adopted anomaly degree calculation model M1 are associated with each other, and the warning period state label data LD sign Generate.

[0154] In step S55, the training unit 63 performs the following process based on the abnormality degree data AD that matches the identifier of the target equipment and the abnormality degree calculation model ID. normal All the anomaly data AD whose production date and time are included between the start date and end date and time of the above are extracted from the anomaly database 73, and the average value of the anomaly A (i.e., the average value of the anomaly A in the normal period) MA normal In addition, the training unit 63 calculates the state label data LD sign All the anomaly data AD whose production date and time are included between the start date and the end date and time of the above are extracted from the anomaly database 73, and the average value of the anomaly A of the data (i.e., the average value of the anomaly A in the warning period) MA sign Furthermore, the training unit 63 calculates the abnormal period state label data LD anormaly All the anomaly data AD whose production date and time are included between the start date and the end date and time of the anomaly database 73 are extracted, and the average value of the anomaly A (i.e., the average value of the anomaly A during the anomaly period) MA anormaly Calculate.

[0155] In step S56, the training unit 63 calculates the average value MA normal and the average value of the abnormality level A during the warning period, MA sign and the average value of abnormality level A during the abnormal period, MA anormaly The relationship between MA and normal <MA sign <MA anormaly It is judged whether or not the MA normal <MA sign <MA anormaly If not (step S56: NO), the training unit 63 ends the process and normal , L.D. sign and L.D. anormaly Discard.

[0156] On the other hand, M.A. normal <MA sign <MA anormalyIf so (step S56: YES), the training unit 63 advances the process to step S57. In step S57, the training unit 63 normal , L.D. sign and L.D. anormaly is registered in the state label database 74.

[0157] (Generation process of anomaly detection model) Next, a description will be given of the generation process of an abnormality sign detection model executed by the training unit 63. Fig. 17 is a flowchart for explaining the generation process of an abnormality sign detection model according to an embodiment.

[0158] In step S 61 , the training unit 63 extracts all the state label data LD from the state label database 74 .

[0159] In step S62, the training unit 63 extracts the normal period state label data LD normal The training unit 63 extracts a predetermined number of feature data FD, each of which includes a production date and time between the start date and time and the end date and time of the production process, from the feature database 72. The training unit 63 uses the extracted predetermined number of feature data FD to create a normal feature data list FL normal If the number of feature data FD does not reach a predetermined number, the training unit 63 generates the extracted normal period state label data LD normal A list of normal feature data FL is created by using all feature data FD that includes the production date and time between the start date and the end date. normal Generate.

[0160] In step S63, the training unit 63 extracts the state label data LD of the prognostic period. sign The training unit 63 extracts a predetermined number of feature data FD, which includes the production date and time between the start date and time and the end date and time of the prediction feature data list FL sign If the number of feature data FD does not reach a predetermined number, the training unit 63 generates the extracted state label data LD signA predictive feature data list FL is created by using all feature data FD that includes the production date and time between the start date and time and the end date and time of the sign Generate.

[0161] In step S64, the training unit 63 extracts the abnormal period state label data LD anormaly The training unit 63 extracts a predetermined number of feature data FD, each of which includes a production date and time between the start date and time and the end date and time of the production process, from the feature database 72. The training unit 63 uses the extracted predetermined number of feature data FD to generate an abnormal feature data list FL anormaly If the number of feature data FD does not reach a predetermined number, the training unit 63 generates the extracted abnormal period state label data LD anormaly Extract all feature data FD that includes the production date and time between the start date and time and the end date and time, and create an abnormal feature data list FL anormaly Generate.

[0162] In step S65, the training unit 63 generates a normal feature data list FL normal , Predictive feature data list FL sign and the abnormal feature data list FL anormaly By assigning three-stage state labels L, namely "normal period", "premonitory period" and "abnormal period", to all feature data FD contained in labeled The training unit 63 generates each feature data FD labeled Based on the above, a learning data list TL is extracted from the feature database 62. The learning data list TL is the feature data FD arranged in time series and assigned with a state label L. labeled Feature data FD with state label L assigned from the feature data FD a certain number of times before labeled In other words, the feature data FD with state label L is labeled is the latest feature data FD among the feature data FD included in the learning data list TL. labeledIn any case, the learning data list TL may include at least the feature amount data FD labeled The number of generated learning data lists TL is the same as the number of normal feature data lists FL normal , Predictive feature data list FL sign and the abnormal feature data list FL anormaly The feature data FD included in labeled ) may be equal to the number of

[0163] Referring to Figure 18, the feature quantity data FD with the state label L of "premonition period" added labeled The learning data list TL will be described below by taking as an example a case where the learning data list TL is generated. Fig. 18 is a diagram for explaining the learning data list.

[0164] In the illustrated example, the predictive feature data list FL sign A plurality of feature data FD are included in the feature data FD. A state label L of the warning period is assigned to each of the plurality of feature data FD. labeled That is, each piece of feature data labeled for training data may be composed of a group of features that overlap with respect to the time axis. In this way, feature data labeled with multiple "premonitory periods" and "abnormal periods" can be generated from one precursor period and abnormal period, and the amount of training data corresponding to states that occur less frequently can be increased. A predetermined number of feature data FD arranged in chronological order are selected from the feature data FD arranged in chronological order, regardless of which feature data list FL they are included in. Among the feature data FD selected at this time, the feature data FD with the latest production date and time is selected as the feature data FD FD labeled The predetermined number of feature data FD selected in this manner constitute a learning data list TL.

[0165] Normal feature data list FL normal and the abnormal feature data list FL anormalyThe feature data FD included in the predictive feature data list FL sign As with the feature data FD included in labeled Then, these feature data FD labeled A learning data list TL including the above is generated and extracted.

[0166] That is, the learning data list TL includes a plurality of pieces of time-series feature quantity data FD generated from each of a plurality of pieces of time-series production result data PD. Also, the learning data list TL includes the feature quantity data FD labeled Feature data FD labeled is representative feature data that represents the multiple feature data FD included in the learning data list TL. labeled is assigned a state label L of “normal period”, “premonitory period”, or “abnormal period”.

[0167] Referring again to FIG. 17, in step S66, the training unit 63 performs machine learning using all of the learning data list TL to generate an anomaly sign detection model M2. The feature amounts and their combinations used at this time are the feature amounts and their combinations used at the time of generating the anomaly degree calculation model M1 adopted by the anomaly degree calculation unit 62. In learning, the representative feature amount data FD labeled The state label L assigned to the learning data list TL is taken as the correct answer, and supervised learning is performed in which all feature data FD included in the learning data list TL is accepted as input.

[0168] As described above, the abnormality sign detection model M2 is a model that learns the characteristics of time-series changes until the production equipment 200 becomes abnormal or a sign of an abnormality. The abnormality sign detection model M2 is used to detect abnormalities and signs of an abnormality in the production equipment 200.

[0169] In addition, a single abnormality sign detection model M2 may be generated by learning the learning data list TL generated based on the production performance data PD of multiple production facilities 200, or a different abnormality sign detection model M2 may be generated for each production facility 200.

[0170] In step S67, the training unit 63 registers one or more generated anomaly sign detection models M2.

[0171] (Processing to determine whether anomaly detection models are adopted) Next, a description will be given of the adoption determination process of the anomaly sign detection model executed by the training unit 63. Note that when the training unit 63 generates only one anomaly sign detection model, the adoption determination process is not necessary. FIG. 19 is a flowchart for describing the process executed by the anomaly sign detection model adoption determination unit 127.

[0172] In step S 71 , the training unit 63 extracts all state label data LD from the state label database 74 .

[0173] In step S72, the training unit 63 extracts the state label data LD normal The training unit 63 extracts a predetermined number of feature data FD, each of which includes a production date and time between the start date and time and the end date and time of the process, from the feature database 72. The training unit 63 uses the extracted predetermined number of feature data FD to generate a normal feature data list EFL for evaluation. normal If the number of feature data FD does not reach a predetermined number, the training unit 63 generates the extracted normal period state label data LD normal Using all feature data FD whose production date and time are included between the start date and the end date, a normal feature data list EFL for evaluation is created. normal Generate.

[0174] In addition, the normal feature data list EFL for evaluation generated in step S72 normal The feature amount data FD included in the normal feature amount data list FL generated in step S62 of FIG. normal It is preferable that the feature amount data FD included in the

[0175] In step S73, the training unit 63 extracts the state label data LD of the extracted prognostic period. signThe training unit 63 extracts a predetermined number of feature data FD, which include a production date and time between the start date and time and the end date and time of the process, from the feature database 72. The training unit 63 uses the extracted predetermined number of feature data FD to generate a predictive feature data list EFL for evaluation. sign If the number of feature data FD does not reach a predetermined number, the training unit 63 generates the extracted state label data LD sign Using all feature data FD that includes the production date and time between the start date and the end date, a predictive feature data list EFL for evaluation is created. sign Generate.

[0176] In addition, the predictive feature data list EFL for evaluation generated in step S73 sign The feature amount data FD included in the predictive feature amount data list FL generated in step S63 of FIG. sign It is preferable that the feature amount data FD included in the

[0177] In step S74, the training unit 63 extracts the state label data LD of the extracted “abnormal period”. anormaly The training unit 63 extracts a predetermined number of feature data FD, which include a production date and time between the start date and time and the end date and time of the production process, from the feature database 72. The training unit 63 uses the extracted predetermined number of feature data FD to generate an abnormal feature data list EFL for evaluation. anormaly If the number of feature data FD does not reach a predetermined number, the training unit 63 generates the extracted abnormal period state label data LD anormaly Using all feature data FD that includes the production date and time between the start date and the end date, an abnormal feature data list EFL for evaluation anormaly Generate.

[0178] In addition, the abnormality feature amount data list EFL for evaluation generated in step S74 anormaly The feature amount data FD included in the abnormal feature amount data list FL generated in step S64 of FIG. anormaly It is preferable that the feature amount data FD included in the

[0179] In step S75, the training unit 63 generates a normal feature data list EFL for evaluation. normal ,Evaluation predictive feature data list EFL sign and the abnormal feature data list EFL for evaluation anormaly By assigning three-stage state labels L, namely "normal period", "premonitory period" and "abnormal period", to all feature data FD contained in labeled The training unit 63 generates each feature data FD labeled Based on the above, an evaluation data list EL is extracted from the feature database 72. The evaluation data list EL is the feature data FD arranged in time series and assigned with a state label L. labeled Feature data FD with state label L assigned from the feature data FD a certain number of times before labeled In other words, the evaluation data list EL contains at least the feature amount data FD labeled The number of generated evaluation data lists EL is determined by the number of normal feature data lists EFL normal ,Evaluation predictive feature data list EFL sign and the abnormal feature data list EFL for evaluation anormaly The feature data FD included in labeled ) may be equal to the number of

[0180] That is, the evaluation data list EL is a list of the same format as the learning data list TL described with reference to Fig. 18. Also, the feature amount data FD labeled The evaluation data list EL is also common to the learning data list TL in that it includes the following:

[0181] In step S76, the training unit 63 performs the training on the newly registered abnormality sign detection model M2 new Input all the extracted evaluation data list EL into the anomaly sign detection model M2 newThe training unit 63 outputs a state label L of the number of the input evaluation data list EL. labeled The state label L assigned to and the anomaly detection model M2 new The number of times (number of matches) C2 that match the state label L output from

[0182] In step S77, the training unit 63 trains the abnormality sign detection model M2 new Extract the anomaly calculation model M1 that calculated the anomaly A used for learning. new The abnormality degree A used for learning is the abnormality degree A included in the abnormality degree data AD extracted by the training unit 63 in step S55. The training unit 63 uses the feature amount data FD included in all the evaluation data lists EL to labeled The extracted anomaly calculation model M1 then calculates the anomaly degree based on the input feature amount data FD labeled The degree of anomaly A calculated by the degree of anomaly calculation model M1 is classified as follows: anormaly If the abnormality level A is greater than a predetermined threshold ThA, the abnormality level A is classified as an "abnormal period." sign is greater than the abnormality threshold ThA anormaly If the abnormality level A is equal to or lower than the warning threshold TA sign If the abnormality degree A is equal to or less than the normal period, the abnormality degree A is classified as a "normal period." labeled The number of cases (number of matches) C1 where the state label L assigned to the anomaly level A matches the classification result of the anomaly level A calculated by the anomaly level calculation model M1 (i.e., the correct answer) is tallied.

[0183] In addition, the abnormality threshold ThA anormaly For example, the average value of the abnormality degree A during the abnormal period calculated in step S55, MA anormaly In addition, the prediction threshold ThA signFor example, the average value MA of the abnormality degree A during the warning period calculated in step S55 is sign This allows the training unit 63 to compare the number of matches C1 and the number of matches C2 with the same criterion in step S78 described later.

[0184] In step S78, the training unit 63 compares whether the number of matches C2 is greater than the number of matches C1. If the number of matches C2 is greater than the number of matches C1 (step S78: YES), the training unit 63 proceeds to step S79. If the number of matches C2 is equal to or less than the number of matches C1 (step S78: NO), the training unit 63 new is not adopted, and the training unit 63 ends the process.

[0185] In step S79, the training unit 63 trains the abnormality sign detection model M2 new will be adopted.

[0186] The anomaly detection model M2 was adopted new is provided to the abnormality sign detection device 100 and used to detect abnormalities and signs of abnormalities in the production equipment 200.

[0187] In addition, the abnormality prediction detection model M2 new If the model is not adopted (step S78: NO), a new anomaly sign detection model M2 is trained again by the training unit 63 as the feature amount data FD is accumulated. new is generated, and the new anomaly sign detection model M2 new It is determined whether to adopt the anomaly sign detection model M2. That is, machine learning for generating the anomaly sign detection model M2 is continued until the number of matches C2 becomes larger than the number of matches C1. In this way, the model generation device 60 can collect production performance data even after the production equipment 200 is in operation, and retrain the anomaly sign detection model M2 to improve detection accuracy by using the production performance data collected during operation.

[0188] <Details of each process for anomaly detection> The following describes in detail the abnormality sign detection process according to one embodiment of the abnormality sign detection device 100 shown in Fig. 11. Here, the pre-processing of the abnormality sign detection device 100 is similar to that of the model generation device 60, and therefore will not be described to avoid duplication. Note that the process described below is merely an example, and the abnormality sign detection process according to the present disclosure may be realized by different processes depending on the abnormality sign detection model generated by the model generation device 60, the production equipment 200, the product, etc.

[0189] (Abnormality detection process) Next, a description will be given of the abnormality sign detection process executed by the abnormality sign detection unit 102. Fig. 20 is a flowchart for explaining the abnormality sign detection process according to one embodiment.

[0190] In step S81, the abnormality sign detection unit 102 adds new feature amount data FD new Determine whether the new feature data FD has been registered. new If the feature amount data FD is not registered (step S81: NO), the abnormality sign detection unit 102 repeats step S81. new If registered (step S81: YES), the abnormality sign detection unit 102 advances the process to step S82.

[0191] In step S82, the abnormality sign detection unit 102 detects the newly registered feature amount data FD new Based on the above, a feature data list FL is extracted from the feature database 72. The feature data list FL is created by arranging the feature data FD in chronological order and adding the newly registered feature data FD new From the feature amount data FD that is a certain number of data before, the newly registered feature amount data FD new In other words, the feature data list FL contains at least the newly registered feature data FD new Contains:

[0192] That is, the feature amount data list FL is a data list in the same format as the learning data list TL described with reference to FIG.

[0193] In step S83, the abnormality sign detection unit 102 inputs the feature amount data list FL to the abnormality sign detection model M2. Then, the abnormality sign detection model M2 inputs a plurality of state labels L new Output.

[0194] In step S84, the abnormality sign detection unit 102 detects the state label L new Determine whether the state label L is in the "premonition period". new If it is the "premonition period" (step S84: YES), the abnormality sign detection unit 102 proceeds to step S85. Note that the premonition period is a period during which no abnormality has occurred in the production equipment 200, but a premonition of an abnormality is expected to appear.

[0195] In step S85, the abnormality sign detection unit 102 notifies a manager or the like of the production equipment 200 that a sign of an abnormality has appeared. In step S85, the abnormality sign detection unit 102 may display the production equipment 200 determined to be in the "sign period" on a terminal of the manager or the like. For example, the abnormality sign detection unit 102 may display the production equipment 200 determined to have a sign of an abnormality along with a message such as "There are signs of an abnormality in the following production equipment!"

[0196] On the other hand, state label L new If the abnormality prediction period is not the "premonition period" (step S84: NO), the abnormality prediction detection unit 102 advances the process to step S86.

[0197] In step S86, the abnormality sign detection unit 102 detects the state label L new It is determined whether the abnormal period is an “abnormal period.” Note that the abnormal period is a period in which the production equipment 200 is assumed to have an abnormality beyond the stage of a symptom of an abnormality.

[0198] Condition label L newIf it is the "abnormal period" (step S86: YES), the abnormality sign detection unit 102 advances the process to step S87. In step S87, the abnormality sign detection unit 102 may display the production equipment 200 determined to be in the "abnormal period" on the terminal of the manager or the like. For example, the abnormality sign detection unit 102 may display the production equipment 200 determined to have an abnormality together with a message such as "An abnormality has occurred in the following production equipment!". In addition, the abnormality sign detection unit 102 may issue an alarm to notify the manager or the like that an abnormality has occurred in the target production equipment 200. A situation in which an abnormality has occurred in the production equipment 200 is a situation requiring emergency maintenance work. For this reason, the abnormality sign detection unit 102 may promptly notify the user of the abnormality sign detection device 100 of the occurrence of an abnormality not only by displaying a message but also by outputting a sound or light by an alarm.

[0199] Condition label L new If it is not an "abnormal period" (step S86: NO), the abnormality sign detection unit 102 ends the process.

[0200] <Functions and Effects of the Model Generation Device 60 and the Anomaly Sign Detection Device 100 of the First Embodiment> As described above, the model generation device 60 and the anomaly sign detection device 100 according to the first embodiment perform the following steps: generating feature amount data FD from the production performance data PD of the production facility 200; generating an anomaly degree calculation model M1 that calculates the anomaly degree A of the production facility 200 from the feature amount data FD by learning or the like; determining whether the generated anomaly degree calculation model M1 appropriately calculates the degree of anomaly of the production facility 200; and generating a state label that sets the state of the production facility 200 in three stages, namely, a “normal period”, a “premonition period”, and an “abnormal period”, based on the anomaly degree A calculated by the anomaly degree calculation model M1. a process of generating an anomaly sign detection model M2 that detects an abnormality or a sign of an abnormality in the production equipment 200 based on the set state label L by learning from the feature data FD; a process of comparing the judgment accuracy of the anomaly degree calculation model M1 and the anomaly sign detection model M2 and adopting the anomaly sign detection model M2 if the judgment accuracy of the anomaly sign detection model M2 is higher; and a process of detecting whether or not an abnormality or a sign of an abnormality has occurred in the production equipment 200 using the adopted anomaly sign detection model M2 and notifying the user if an abnormality or a sign of an abnormality has occurred.

[0201] Therefore, even when the production equipment 200 has few abnormalities such as deterioration and failure, feature amount data that appropriately represents features related to abnormalities such as deterioration and failure of the production equipment 200 or signs of abnormalities can be extracted, and a highly accurate trained model (anomaly sign detection model M2) can be generated using the extracted feature amount data. In addition, when an abnormality in the production equipment 200 is detected using the trained model, an alarm can be issued to allow the user to take emergency action. In addition, when a sign of an abnormality is detected using the trained model, the production equipment 200 that has detected the sign of an abnormality can be notified to the user, and maintenance work can be performed to prevent the deterioration and failure of the production equipment 200 from progressing while the rate of occurrence of defective products is low.

[0202] The abnormality sign detection model M2 is a trained model generated by machine learning using the learning data list TL. When a feature data list FL, which is a data list in the same format as the learning data list TL, is input, the abnormality sign detection model M2 outputs a result of determining whether the production equipment 200 is in a "normal period," a "sign period," or an "abnormal period."

[0203] According to the present disclosure, by labeling unlabeled training data using an anomaly degree calculation model that can be generated from a relatively small amount of labeled training data, the anomaly sign detection model M2 can be generated before a large amount of production performance data PD during the warning period and the abnormality period is collected. Therefore, even if a warning sign of an anomaly appears or an anomaly occurs in a relatively short time after the production facility 200 starts producing products, a trained anomaly sign detection model that can reliably detect them can be used. Therefore, it is possible to detect the appearance of a warning sign of an anomaly and the occurrence of an anomaly in a relatively short time after the production facility 200 starts producing products.

[0204] (Second embodiment) A second embodiment of the present disclosure will be described below. FIG. 21 is a block diagram illustrating a functional configuration of a model generation device 60A according to the second embodiment. The model generation device 60A according to the second embodiment differs from the model generation device 60 according to the first embodiment in the following respects. That is, the process performed by the training unit 63A is different from the process performed by the training unit 63.

[0205] The following describes the differences from the first embodiment. The same components as those in the first embodiment are described with the same reference numerals as those in the first embodiment, and the components different from those in the first embodiment are described with the letter "A" added to the reference numerals.

[0206] In the first embodiment, the training unit 63 sets a certain period of warning period immediately before the abnormality period on the premise that a warning of an abnormality appears during a certain period immediately before the occurrence of an abnormality such as deterioration and a breakdown in the production equipment 200. This method is effective when the deterioration pattern of the production equipment 200 is always the same. However, in reality, due to various factors, deterioration may progress more rapidly than expected, and the period during which a warning of an abnormality appears may become shorter. In this case, the feature amount data FD in which no warning of an abnormality appears will be included in the feature amount data FD in the warning period more than expected, and the performance of the generated abnormality warning detection model M2 to detect a warning of an abnormality will be insufficient. The second embodiment is intended to generate an abnormality warning detection model M2 that can handle such a case.

[0207] FIG. 22 is a flowchart illustrating the state label setting process according to the second embodiment.

[0208] In step S91, the training unit 63A stores in the feature database 72 the defect rate DR and the defect rate threshold value Th dr It is judged whether the feature quantity data FD is registered that is greater than the defect rate threshold value Th dr If the feature amount data FD larger than the defect rate threshold value Th is not registered (step S91: NO), the training unit 63A repeats step S91. dr If feature amount data FD larger than is registered (step S91: YES), the training unit 63A advances the process to step S92.

[0209] In step S92, the training unit 63A sets the period of production dates and times of the feature amount data FD used when the abnormality degree calculation model M1 adopted by the abnormality degree calculation unit 62 was generated by the abnormality degree calculation unit 62 as the normal period. The training unit 63A associates the identifier of the target production equipment 200, the start date and time and the end date and time of the set normal period, and the abnormality degree calculation model ID of the adopted abnormality degree calculation model M1, and sets the normal period state label data LD normalIn addition, a period other than the abnormal period or the warning period described later can be set as the normal period. For example, a period that was not used in generating the abnormality degree calculation model M1 but that precedes the period set as the warning period can be set as the normal period.

[0210] In step S93, the training unit 63A performs a training on the feature amount database 72 by comparing the defect rate DR with the defect rate threshold value Th dr The period of the production date and time of the feature amount data FD that is greater than the abnormal period is set as the abnormal period. The identifier of the target production equipment 200, the start date and time and the end date and time of the set abnormal period, and the abnormality degree calculation model ID of the adopted abnormality degree calculation model M1 are associated with each other, and the abnormality period state label data LD anormaly Generate.

[0211] In step S94, the training unit 63A performs the following process based on the abnormality degree data AD that matches the identifier of the target equipment and the abnormality degree calculation model ID. normal All the anomaly data AD whose production date and time are included between the start date and end date and time of the above are extracted from the anomaly database 73, and the average value of the anomaly A (average value of the anomaly A in the normal period) MA normal In addition, the training unit 63A calculates the abnormal period state label data LD anormaly All the anomaly data AD whose production date and time are included between the start date and the end date and time of the above are extracted from the anomaly database 73, and the average value of the anomaly A (average value of the anomaly A during the anomaly period) MA anormaly Calculate.

[0212] In step S95, the training unit 63A calculates the average value MA normal and the average value of abnormality level A during the abnormal period, MA anormaly The training unit 63A calculates an average value of the calculated average value and the warning threshold value ThA sign Set to.

[0213] In step S96, the training unit 63A sets the start date and time of the abnormal period as the end date and time, sets a period having a date and time going back from the end date and time as the start date and time, as the warning period, and calculates the state label data LD of the warning period. sign The training unit 63A generates an average value MA of the abnormality degree A of all the abnormality degree data AD whose production date and time are included between the start date and time and the end date and time of the warning period. sign is the warning threshold ThA sign In other words, the training unit 63A sets the start date and time of the warning period so as to go back as far into the past as possible, starting from the start date and time of the abnormal period.

[0214] In step S97, the training unit 63A generates each state label data LD normal , L.D. sign and L.D. anormaly is registered in the state label database 74.

[0215] As described above, according to the model generation device 60A of the second embodiment, even when the period during which signs of abnormality appear is not constant and can be short, it is possible to appropriately set the period during which signs of abnormality appear and generate a highly accurate trained model (anomaly sign detection model M2).

[0216] (Third embodiment) The third embodiment of the present disclosure will be described below. FIG. 23 is a block diagram illustrating a functional configuration of a model generation device 60B according to the third embodiment. The model generation device 60B according to the third embodiment differs from the model generation device 60 according to the first embodiment in the following respects. That is, the process performed by the training unit 63B is different from the process performed by the training unit 63.

[0217] The following describes the differences from the first embodiment. The same components as those in the first embodiment are described with the same reference numerals as those in the first embodiment, and the components different from those in the first embodiment are described with the reference numerals followed by the letter "B."

[0218] In the first embodiment, the training unit 63 sets a certain prediction period immediately before the abnormality period on the premise that a sign of an abnormality appears during a certain period immediately before the occurrence of an abnormality such as deterioration and a breakdown in the production equipment 200. This method is effective when the appearance of a sign of an abnormality becomes clear as the deterioration of the production equipment 200 progresses, and the abnormality eventually occurs. However, in reality, due to various factors, a period in which no sign of an abnormality appears may be included in the period immediately before the abnormality period. In this case, the feature amount data FD in which no sign of an abnormality appears will be included in the feature amount data FD in the prediction period, and the performance of the generated abnormality sign detection model M2 to detect a sign of an abnormality will be insufficient. The third embodiment is intended to generate an abnormality sign detection model M2 that can handle such a case.

[0219] FIG. 24 is a flowchart illustrating the state label setting process according to the third embodiment.

[0220] In step S101, the training unit 63B stores in the feature database 62 the defect rate DR and the defect rate threshold value Th dr It is judged whether the feature quantity data FD is registered which is larger than the defect rate threshold value Th dr If no feature amount data greater than 1 is registered (step S101: NO), the training unit 63B repeats step S101. dr If feature amount data FD larger than is registered (step S101: YES), the training unit 63B advances the process to step S102.

[0221] In step S102, the training unit 63B sets the period of production date and time of the feature amount data FD used when the abnormality degree calculation model M1 adopted by the abnormality degree calculation unit 62 was generated by the abnormality degree calculation unit 62 as the normal period. The training unit 63B associates the identifier of the target production equipment 200, the start date and time and the end date and time of the set normal period, and the abnormality degree calculation model ID of the adopted abnormality degree calculation model M1, and generates the state label data LD for the normal period. normalIn addition, a period other than the abnormal period or the warning period described later can be set as the normal period. For example, a period that was not used in generating the abnormality degree calculation model M1 but that precedes the period set as the warning period can be set as the normal period.

[0222] In step S103, the training unit 63B performs a test on the feature amount database 72 to determine whether the defect rate DR is a defect rate threshold value Th dr The period of the production date and time of the feature amount data FD that is greater than the abnormal period is set as the abnormal period. The identifier of the target production equipment 200, the start date and time and the end date and time of the set abnormal period, and the abnormality degree calculation model ID of the adopted abnormality degree calculation model M1 are associated with each other, and the abnormality period state label data LD anormaly Generate.

[0223] In step S104, the training unit 63B performs the following process based on the abnormality degree data AD that matches the identifier of the target equipment and the abnormality degree calculation model ID. normal All the anomaly data AD whose production date and time are included between the start date and end date and time of the above are extracted from the anomaly database 73, and the average value of the anomaly A (average value of the anomaly A in the normal period) MA normal In addition, the training unit 63B calculates the abnormal period state label data LD anormaly All the anomaly data AD whose production date and time are included between the start date and the end date and time of the above are extracted from the anomaly database 73, and the average value of the anomaly A (average value of the anomaly A during the anomaly period) MA anormaly Calculate.

[0224] In step S105, the training unit 63B calculates the average value MA normal and the average value of abnormality level A during the abnormal period, MA anormaly The training unit 63B calculates an average value of the calculated average value and the warning threshold value ThA sign Set to.

[0225] In step S106, the training unit 63B sets the prognostic threshold ThAsign The training unit 63B extracts abnormality degree data AD in which the abnormality degree A is greater than the abnormality degree A and the production date and time is not included in the normal period or the abnormal period. The training unit 63B groups the extracted abnormality degree data AD in units of a predetermined period, for example, one day. The training unit 63B sets a period in which a group contains a predetermined number or more of abnormality degree data AD as a warning period, and generates status label data LD for the warning period. sign Generate.

[0226] In step S107, the training unit 63B generates each state label data LD normal , L.D. sign and L.D. anormaly is registered in the state label database 74.

[0227] As described above, according to the model generation device 60B of the third embodiment, even if the period immediately preceding the abnormal period includes a period in which no signs of abnormality appear, the abnormality precursor period can be appropriately set and a highly accurate trained model (anomaly precursor detection model M2) can be generated.

[0228] In addition, the detection performance of each of the anomaly sign detection models M2 generated in the first, second and third embodiments may be compared in the training sections 63, 63A and 63B, and the anomaly sign detection model M2 with the highest detection performance may be adopted.

[0229] Furthermore, it goes without saying that the production facility 200 to which the model generating devices 60, 60A, 60B and the anomaly sign detection device 100 according to the present disclosure are applied may be a production facility that produces products other than lithium ion secondary batteries.

[0230] The application of the abnormality sign detection model M2 is not limited to the production facility 200 that generated the production performance data PD used in generating the abnormality sign detection model M2, but may be other facilities that produce the same products as the production facility 200. Furthermore, when there is a common tendency for abnormalities to occur, the application of the abnormality sign detection model M2 may be facilities that produce products different from the production facility 200. [Industrial Applicability]

[0231] The present disclosure is useful for an abnormality sign detection device that detects an abnormality or a sign of an abnormality in production equipment. [Explanation of symbols]

[0232] 10. Anomaly detection system 20 Training Database 60, 60A, 60B Model Generator 61 Pretreatment section 62 Abnormality degree calculation section 63,63A,63B Training Department 71 Production performance database 72 Feature Database 73 Anomaly Database 74 Condition Label Database 100 Abnormality Prediction Detection Device 101 Pretreatment section 102 Abnormality Predictor Detection Unit 200 Production facilities 201 Winding section 202 First sheet material 203 Second sheet material 204 Rolled body 205A 1st lamination roll 205B 2nd lamination roll 206 Roll Core 206M Core Rotation Drive Unit 207 Inspection Machine 208 Cutting section 40 First sheet material length measurement sensor after cutting 41 Second sheet material post-cut length measurement sensor 50 1st Supply Reel 51 Second supply reel

Claims

1. A pre-processing unit that acquires feature quantity data of the production performance data for training; an anomaly degree calculation unit that calculates an anomaly degree of the acquired feature amount data by using an anomaly degree calculation model that represents a group of feature amounts in a normal state; a training unit that determines a state of the feature data based on the calculated degree of anomaly, and trains an anomaly sign detection model that detects the state of the feature data from the feature data using training data composed of the feature data and the state of the feature data; having A model generating device in which each feature data of the training data is composed of a group of feature values ​​that overlap with each other on a time axis.

2. The model generating device according to claim 1 , wherein the abnormality sign detection model discriminates a state of the feature amount data into a normal period, a sign period, or an abnormal period.

3. The model generating device according to claim 2 , wherein the training unit determines a state of the feature amount data based on a comparison of the calculated degree of abnormality with a predictive threshold and an abnormality threshold.

4. The model generating device according to claim 1 , wherein the anomaly degree calculation model calculates a deviation from the group of feature amounts in a normal state as the degree of anomaly of the feature amount data.

5. The model generating device according to claim 1 , wherein the preprocessing unit determines the feature amount data based on statistics of the training production performance data.

6. The model generating device according to claim 1 , wherein the anomaly degree calculation unit generates the anomaly degree calculation model using training data configured of the feature amount data and a state of the feature amount data acquired from a training database.

7. A step of acquiring feature quantity data of production performance data for training; calculating an anomaly degree of the acquired feature amount data by using an anomaly degree calculation model that represents a group of feature amounts in a normal state; determining a state of the feature data based on the calculated degree of anomaly, and training an anomaly sign detection model that detects the state of the feature data from the feature data using training data composed of the feature data and the state of the feature data; The computer executes A model generation method in which each feature data of the training data is composed of a group of features that overlap with each other on a time axis.

8. a pre-processing unit for acquiring feature quantity data of the production performance data to be detected; an anomaly sign detection unit that detects a state of the feature data from the feature data by using a trained anomaly sign detection model; having The anomaly sign detection model is trained using training data that is configured from feature amount data of training production performance data and a state of the feature amount data, the state of the feature amount data is determined based on an anomaly degree of the feature amount data calculated using an anomaly degree calculation model that represents a group of feature amounts in a normal state; An anomaly sign detection device, wherein each feature data of the training data is composed of a group of feature amounts that overlap on a time axis.

9. A step of acquiring feature amount data of production performance data of a detection target; Detecting a state of the feature data from the feature data by using a trained anomaly sign detection model; The computer executes The anomaly sign detection model is trained using training data that is configured from feature amount data of training production performance data and a state of the feature amount data, the state of the feature amount data is determined based on an anomaly degree of the feature amount data calculated using an anomaly degree calculation model that represents a group of feature amounts in a normal state; The method for detecting an anomaly, wherein each feature data of the training data is composed of a group of feature amounts that overlap with each other on a time axis.

10. A model generating device according to any one of claims 1 to 6; The abnormality sign detection device according to claim 8 ; An abnormality detection system having the above-mentioned function.

Images