Manufacturing equipment abnormality monitoring system and method

Abstract

A method for detecting abnormalities in manufacturing equipment according to an embodiment of the present invention for achieving the above objective collects data detected by sensors installed in the manufacturing equipment, stores the collected data, and analyzes the stored data to predict abnormalities in the manufacturing equipment, wherein some data are selected based on the stored data and abnormalities in the manufacturing equipment are predicted based on the selected data. Accordingly, among the various sensor data regarding the manufacturing equipment, the data with the greatest significance according to the failure process is selectively utilized to preemptively predict signs of abnormalities in the manufacturing equipment.

Description

Technology Field

[0001] The present invention relates to a technology for detecting abnormalities in manufacturing equipment, and more specifically, to a system and method for monitoring whether there are abnormalities in manufacturing equipment by collecting and analyzing sensor data of manufacturing equipment. Background Technology

[0003] As manufacturing trends shift from a focus on productivity to a current emphasis on safety, the maintenance of manufacturing facilities is emerging as a critical issue. While reactive maintenance—repairing after a breakdown—and preventive maintenance—periodical inspections—were primarily utilized in the past, the industry is now evolving into predictive maintenance, which involves performing maintenance based on forecasts of equipment's future condition.

[0004] In the case of manufacturing equipment maintenance, particularly predictive maintenance, it is necessary to detect signs of abnormalities in the equipment before a failure occurs; therefore, effective monitoring of manufacturing equipment is impossible with analysis based on a single factor alone.

[0005] Accordingly, a monitoring system and method are needed to predict signs of abnormalities in manufacturing facilities more proactively. The problem to be solved

[0007] The present invention has been devised to solve the aforementioned problems, and the objective of the present invention is to provide a method and system for preemptively predicting signs of abnormality in manufacturing equipment by selectively utilizing the data with the greatest significance according to the failure process among various sensor data regarding the manufacturing equipment. means of solving the problem

[0009] A method for detecting abnormalities in manufacturing equipment according to an embodiment of the present invention for achieving the above objective comprises: a step of collecting data detected by sensors installed in the manufacturing equipment; a step of storing the collected data; and a step of analyzing the stored data to predict abnormalities in the manufacturing equipment. The prediction step involves selecting some data based on the stored data and predicting abnormalities in the manufacturing equipment based on the selected some data.

[0010] In the prediction step, if the first data among the data is greater than or equal to a first threshold, the abnormality of the manufacturing equipment can be predicted using a first artificial intelligence model trained to predict the abnormality of the manufacturing equipment from the first data.

[0011] In the prediction step, if the first data is below a first threshold and the second data among the data is above a second threshold, the abnormality of the manufacturing equipment can be predicted using a second artificial intelligence model trained to predict the abnormality of the manufacturing equipment from the second data.

[0012] The first data may be data that has a significant value later than the second data during the failure process of the manufacturing equipment.

[0013] In the prediction step, if the reliability of the prediction result by the first artificial intelligence model is less than the third threshold, one of the prediction result by the first artificial intelligence model and the prediction result by the 'third artificial intelligence model trained to predict abnormalities in manufacturing equipment from the first data and the second data' can be selected to predict abnormalities in manufacturing equipment.

[0014] In the prediction step, the prediction result of the AI ​​model with higher reliability among the first AI model and the third AI model can be selected.

[0015] The first data is one of odor data, noise data, temperature data, and vibration data, and the second data may be another of noise data, temperature data, and vibration data.

[0016] If the first data is odor data, the second data is noise data; if the first data is noise data, the second data is temperature data; and if the first data is temperature data, the second data may be vibration data.

[0017] The second data further includes manufacturing equipment information, and if the first data is vibration data, the second data may be manufacturing equipment information.

[0018] Meanwhile, a manufacturing facility abnormality detection server according to another embodiment of the present invention comprises: a collection unit that collects data detected by sensors installed in the manufacturing facility; a storage unit that stores the collected data; and a prediction unit that analyzes the stored data to predict abnormalities in the manufacturing facility. The prediction unit selects some data based on the stored data and predicts abnormalities in the manufacturing facility based on the selected some data. Effects of the invention

[0020] As explained above, according to the embodiments of the present invention, among various sensor data regarding the manufacturing equipment, the data with the greatest significance according to the failure process can be selectively utilized to preemptively predict signs of abnormality in the manufacturing equipment.

[0021] In addition, according to embodiments of the present invention, if the reliability of the prediction result is low, the reliability can be improved by supplementing the prediction of abnormalities in the manufacturing facility using additional sensor data. Brief explanation of the drawing

[0023] FIG. 1 is a diagram illustrating the configuration of a manufacturing facility monitoring system according to one embodiment of the present invention. FIG. 2 is a drawing illustrating the detailed configuration of the manufacturing facility abnormality detection server shown in FIG. 1. FIG. 3 is a diagram illustrating a failure process model of another manufacturing facility in an embodiment of the present invention. FIG. 4 is a diagram showing artificial intelligence models used by the anomaly prediction unit, FIG. 5 is a flowchart provided for explaining a method for monitoring manufacturing facilities according to another embodiment of the present invention, FIG. 6 is a diagram showing the detailed steps of step S240. FIG. 7 is a diagram showing the detailed steps of step S260. FIG. 8 is a diagram showing the detailed steps of step S280, FIG. 9 is a diagram showing the detailed steps of step S280, FIG. 10 is a diagram illustrating the configuration of a manufacturing facility monitoring system according to another embodiment of the present invention. FIG. 11 is a diagram showing artificial intelligence models used by the anomaly prediction unit, FIG. 12 is a flowchart provided for explaining a method for monitoring manufacturing facilities according to another embodiment of the present invention. Specific details for implementing the invention

[0024] The present invention will be described in more detail below with reference to the drawings.

[0025] FIG. 1 is a diagram illustrating the configuration of a manufacturing facility monitoring system according to an embodiment of the present invention. The manufacturing facility monitoring system according to an embodiment of the present invention is a system for predicting abnormal signs of manufacturing facilities by collecting and analyzing various sensor data from manufacturing facilities provided in factories, plants, etc.

[0026] A manufacturing facility monitoring system according to an embodiment of the present invention that performs such functions is constructed such that sensors (11-14) and a manufacturing facility abnormality detection server (100) are connected to enable mutual communication, as shown in FIG. 1.

[0027] Sensors (11 to 14) are installed in the manufacturing facility to detect / measure conditions and generate sensor data. 1) A vibration sensor (11) detects vibrations generated in the manufacturing facility, 2) A temperature sensor (12) measures heat generated in the manufacturing facility, 3) A sound sensor (12) measures noise generated in the manufacturing facility, and 4) A smell sensor (14) detects smells caused by smoke, etc. generated in the manufacturing facility.

[0028] Sensor data generated by the sensors (11-14) is transmitted to the manufacturing equipment abnormality detection server (100) via wireless communication through a mobile communication network. Mobile communication methods such as 3G / 4G (LTE) / 5G can be applied without restriction.

[0029] The manufacturing equipment anomaly detection server (100) stores sensor data collected from sensors (11-14) and analyzes it in real-time or periodically to predict signs of anomalies in the manufacturing equipment.

[0030] FIG. 2 is a diagram illustrating the detailed configuration of the manufacturing equipment anomaly detection server (100) illustrated in FIG. 1. As illustrated, the manufacturing equipment anomaly detection server (100) is configured to include a data collection unit (110), a data storage unit (120), and an anomaly prediction unit (130).

[0031] The data collection unit (110) collects sensor data detected by sensors (11~14) installed in the manufacturing facility, and the data storage unit (120) builds and manages the sensor data collected by the data collection unit (110) into a DB.

[0032] The anomaly prediction unit (130) analyzes the data stored in the DB of the data storage unit (120) to predict signs of an anomaly in the manufacturing equipment. To this end, the anomaly prediction unit (130) selects some data based on rules from the data stored in the data storage unit (120) and uses the selected data to predict an anomaly in the manufacturing equipment based on artificial intelligence.

[0033] In an embodiment of the present invention, the failure process of the manufacturing equipment was modeled as shown in FIG. 3. The presented failure process assumes that when an abnormality occurs in the manufacturing equipment, 1) vibration occurs first, 2) heat is generated, 3) if it worsens further, noise is generated, 4) an odor is generated due to the appearance of smoke, and 5) eventually, damage occurs.

[0034] Accordingly, the anomaly prediction unit (130) selects meaningful data among the vibration data, temperature data, noise data, and odor data of the manufacturing equipment, and selects data that becomes more meaningful as the failure progresses.

[0035] For example, 1) if the odor data has a significant value, an abnormality in the manufacturing equipment is predicted based on the odor data (and noise data), 2) if the odor data is not a significant value but the noise data has a significant value, an abnormality in the manufacturing equipment is predicted based on the noise data (and temperature data), 3) if the odor data and noise data are not significant values ​​but the temperature data has a significant value, an abnormality in the manufacturing equipment is predicted based on the temperature data (and vibration data), and 4) if the odor data, noise data, and temperature data are not all significant values, an abnormality in the manufacturing equipment is predicted based on the vibration data (and manufacturing equipment information).

[0036] Artificial intelligence models are used to predict abnormalities in manufacturing equipment based on selected data. FIG. 4 shows artificial intelligence models (131 to 138) used by the abnormality prediction unit (130), specifically as follows.

[0037] Smell-based anomaly prediction model (131): An artificial intelligence model trained to predict whether there are abnormalities in manufacturing equipment by analyzing odor data.

[0038] Smell / Noise-based Anomaly Prediction Model (132): An artificial intelligence model trained to predict whether there is an anomaly in manufacturing equipment by analyzing odor data and noise data.

[0039] Noise-based anomaly prediction model (133): An artificial intelligence model trained to predict whether there are abnormalities in manufacturing equipment by analyzing noise data.

[0040] Noise / Temperature-Based Anomaly Prediction Model (134): An artificial intelligence model trained to predict whether there is an anomaly in manufacturing equipment by analyzing noise data and temperature data.

[0041] Temperature-based anomaly prediction model (135): An artificial intelligence model trained to predict whether there is an anomaly in manufacturing equipment by analyzing temperature data.

[0042] Temperature / Vibration-based Anomaly Prediction Model (136): An artificial intelligence model trained to predict whether there is an anomaly in manufacturing equipment by analyzing temperature data and vibration data.

[0043] Vibration-based anomaly prediction model (137): An artificial intelligence model trained to predict whether there are abnormalities in manufacturing equipment by analyzing vibration data.

[0044] Vibration / Manufacturing Equipment Information-Based Anomaly Prediction Model (138): An artificial intelligence model trained to predict whether there is an anomaly in manufacturing equipment by analyzing vibration data and manufacturing equipment information.

[0045] Here, manufacturing equipment information includes specification information (type, specifications, etc.) and operational information (total operating time, average operating time, failure frequency, etc.) of the manufacturing equipment.

[0046] The process of operation of the manufacturing facility abnormality detection server (100) will be described in detail below with reference to FIG. 5. FIG. 5 is a flowchart provided for explaining a manufacturing facility monitoring method according to another embodiment of the present invention.

[0047] As described above, first, the data collection unit (110) of the manufacturing equipment abnormality detection server (100) collects data detected by the sensors (11~14) (S210), and the data storage unit (120) stores the data collected in step S210 (S220).

[0048] Then, the abnormality prediction unit (130) analyzes the data collected / stored by steps S210 / S220 to predict abnormalities in the manufacturing equipment (S230 to S290), which will be explained in detail below.

[0049] When the odor data exceeds a threshold, that is, when smoke is generated in the manufacturing facility and a specific odor is produced (S230-Y), the anomaly prediction unit (130) predicts an anomaly in the manufacturing facility using an odor-based anomaly prediction model (131) and an odor / noise-based anomaly prediction model (132) (S240). The detailed steps of step S240 are shown in FIG. 6.

[0050] As described above, the anomaly prediction unit (130) first predicts an anomaly in the manufacturing equipment from the odor data using an odor-based anomaly prediction model (131) (S241). When step S241 is performed, the odor-based anomaly prediction model (131) outputs the prediction result along with the reliability of the prediction result.

[0051] If this confidence level is greater than or equal to a threshold (e.g., 80%) (S242-Y), the anomaly prediction unit (130) outputs the prediction result from step S241 as the final prediction result (S243).

[0052] On the other hand, if the reliability is below the threshold value (S242-N), the anomaly prediction unit (130) predicts an anomaly in the manufacturing equipment by using the odor / noise-based anomaly prediction model (132) and additionally using noise data in addition to the odor data (S244). Even when performing step S243, the reliability of the prediction result is output along with the prediction result from the odor / noise-based anomaly prediction model (132).

[0053] The anomaly prediction unit (130) compares the prediction reliability of the odor-based anomaly prediction model (131) with the prediction reliability of the odor / noise-based anomaly prediction model (132) and selects the prediction result of the anomaly prediction model with greater prediction reliability (S245).

[0054] Specifically, if the prediction reliability of the odor-based anomaly prediction model (131) is 65% and the prediction reliability of the odor / noise-based anomaly prediction model (132) is 70%, the prediction result of the odor / noise-based anomaly prediction model (132) is selected in step S245.

[0055] On the other hand, if the prediction reliability of the odor-based anomaly prediction model (131) is 65% and the prediction reliability of the odor / noise-based anomaly prediction model (132) is 60%, the prediction result of the odor-based anomaly prediction model (131) is selected in step S245.

[0056] The next anomaly prediction unit (130) outputs the prediction result selected in step S245 as the final prediction result (S243).

[0057] Referring again to FIG. 5, the following explanation is provided. When the odor data does not exceed a threshold value, i.e., no smoke is generated in the manufacturing facility and no specific odor is produced (S230-N), but when the noise data exceeds a threshold value, i.e., noise is generated in the manufacturing facility (S250-Y), the anomaly prediction unit (130) predicts an anomaly in the manufacturing facility using a noise-based anomaly prediction model (133) and a noise / temperature-based anomaly prediction model (134) (S260). The detailed steps of step S260 are shown in FIG. 7.

[0058] As described above, the anomaly prediction unit (130) first predicts an anomaly in the manufacturing equipment from noise data using a noise-based anomaly prediction model (133) (S261).

[0059] At this time, if the prediction reliability in step S261 is greater than or equal to the threshold value (S262-Y), the anomaly prediction unit (130) outputs the prediction result in step S261 as the final prediction result (S263).

[0060] On the other hand, if the prediction reliability is below the threshold (S262-N), the anomaly prediction unit (130) predicts an anomaly in the manufacturing equipment by using a noise / temperature-based anomaly prediction model (134) and additionally using temperature data in addition to noise data (S264).

[0061] The next anomaly prediction unit (130) compares the prediction reliability of the noise-based anomaly prediction model (133) with the prediction reliability of the noise / temperature-based anomaly prediction model (134), and selects the prediction result of the anomaly prediction model with greater prediction reliability (S265).

[0062] The next anomaly prediction unit (130) outputs the prediction result selected in step S265 as the final prediction result (S263).

[0063] Referring again to FIG. 5, the following explanation is provided. When the noise data does not exceed the threshold value, i.e., no noise is generated in the manufacturing facility (S250-N), but when the temperature data exceeds the threshold value, i.e., heat is generated in the manufacturing facility (S270-Y), the anomaly prediction unit (130) predicts an anomaly in the manufacturing facility using a temperature-based anomaly prediction model (135) and a temperature / vibration-based anomaly prediction model (136) (S280). The detailed steps of step S280 are shown in FIG. 8.

[0064] As described above, the anomaly prediction unit (130) first predicts an anomaly in the manufacturing equipment from temperature data using a temperature-based anomaly prediction model (135) (S281).

[0065] At this time, if the prediction reliability in step S281 is greater than or equal to the threshold value (S282-Y), the anomaly prediction unit (130) outputs the prediction result in step S281 as the final prediction result (S283).

[0066] On the other hand, if the prediction reliability is below the threshold value (S282-N), the anomaly prediction unit (130) predicts an anomaly in the manufacturing equipment by using a temperature / vibration-based anomaly prediction model (136) and additionally using vibration data in addition to temperature data (S284).

[0067] The next anomaly prediction unit (130) compares the prediction reliability of the temperature-based anomaly prediction model (135) with the prediction reliability of the temperature / vibration-based anomaly prediction model (136), and selects the prediction result of the anomaly prediction model with greater prediction reliability (S285).

[0068] The next anomaly prediction unit (130) outputs the prediction result selected in step S285 as the final prediction result (S283).

[0069] Referring again to FIG. 5, the following explanation is provided. When the temperature data does not exceed a threshold value, that is, when no heat is generated in the manufacturing facility (S270-N), the anomaly prediction unit (130) predicts an anomaly in the manufacturing facility using a vibration-based anomaly prediction model (137) and a vibration / manufacturing facility information-based anomaly prediction model (138) (S290). The detailed steps of step S290 are shown in FIG. 9.

[0070] As described above, the anomaly prediction unit (130) first predicts an anomaly in the manufacturing equipment from vibration data using a vibration-based anomaly prediction model (137) (S291). At this time, if the prediction reliability in step S291 is greater than or equal to a threshold value (S292-Y), the anomaly prediction unit (130) outputs the prediction result in step S291 as the final prediction result (S293).

[0071] On the other hand, if the prediction reliability is below the threshold value (S292-N), the anomaly prediction unit (130) predicts an anomaly of the manufacturing equipment by using the vibration / manufacturing equipment information-based anomaly prediction model (138) in addition to the vibration data and manufacturing equipment information (S294).

[0072] The next anomaly prediction unit (130) compares the prediction reliability of the vibration-based anomaly prediction model (137) with the prediction reliability of the vibration / manufacturing equipment information-based anomaly prediction model (138), and selects the prediction result of the anomaly prediction model with greater prediction reliability (S295).

[0073] The next anomaly prediction unit (130) outputs the prediction result selected in step S295 as the final prediction result (S293).

[0074] FIG. 10 is a diagram illustrating the configuration of a manufacturing facility monitoring system according to another embodiment of the present invention. The embodiment of the present invention differs from the system presented in FIG. 1 in that the odor sensor (14) is excluded from the sensors (11 to 13).

[0075] Accordingly, the manufacturing equipment abnormality detection server (100) does not use odors caused by smoke, etc. generated from the manufacturing equipment when predicting abnormal signs of the manufacturing equipment, and as shown in FIG. 11, the abnormality prediction unit (130) differs from the previous embodiment presented in FIG. 4 in that the odor-based abnormality prediction model (131) and the odor / noise-based abnormality prediction model (132) are excluded.

[0076] The process of operation of the manufacturing facility abnormality detection server (100) of FIG. 10 will be explained in detail below with reference to FIG. 12. FIG. 12 is a flowchart provided for explaining a manufacturing facility monitoring method according to another embodiment of the present invention.

[0077] As described above, first, the data collection unit (110) of the manufacturing equipment abnormality detection server (100) collects data detected by the sensors (11~13) (S310), and the data storage unit (120) stores the data collected in step S310 (S320).

[0078] Then, the abnormality prediction unit (130) analyzes the data collected / stored by steps S310 / S320 to predict abnormalities in the manufacturing equipment (S330 to S370), which will be explained in detail below.

[0079] When noise data exceeds a threshold value, i.e., when noise occurs in the manufacturing facility (S330-Y), the anomaly prediction unit (130) predicts an anomaly in the manufacturing facility using a noise-based anomaly prediction model (133) and a noise / temperature-based anomaly prediction model (134) (S340). Since the detailed steps of step S340 are the same as those in FIG. 7, a detailed description thereof is omitted.

[0080] When the noise data does not exceed the threshold value, i.e., no noise is generated in the manufacturing facility (S330-N), but when the temperature data exceeds the threshold value, i.e., heat is generated in the manufacturing facility (S350-Y), the anomaly prediction unit (130) predicts an anomaly in the manufacturing facility using a temperature-based anomaly prediction model (135) and a temperature / vibration-based anomaly prediction model (136) (S360). Since the detailed steps of step S360 are the same as those in FIG. 8, a detailed description thereof is omitted.

[0081] When the temperature data does not exceed a threshold value, that is, when no heat is generated in the manufacturing facility (S350-N), the anomaly prediction unit (130) predicts an anomaly in the manufacturing facility using a vibration-based anomaly prediction model (137) and a vibration / manufacturing facility information-based anomaly prediction model (138) (S370). Since the detailed steps of step S370 are the same as those in FIG. 9, a detailed description thereof is omitted.

[0082] So far, preferred embodiments of the manufacturing facility abnormality monitoring system and method have been described in detail.

[0083] In the above embodiment, it was assumed that there is only one manufacturing facility monitored by the manufacturing facility abnormality detection server (100), but this is merely an example for convenience of illustration and explanation. It is, of course, possible to monitor multiple manufacturing facilities with one manufacturing facility abnormality detection server (100).

[0084] Meanwhile, it goes without saying that the technical concept of the present invention may also be applied to a computer-readable recording medium containing a computer program that enables the device and method according to the present embodiment to perform their functions. Furthermore, the technical concept according to various embodiments of the present invention may be implemented in the form of computer-readable code recorded on a computer-readable recording medium. A computer-readable recording medium may be any data storage device that can be read by a computer and store data. For example, a computer-readable recording medium may be a ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical disk, hard disk drive, etc. Additionally, computer-readable code or a program stored on a computer-readable recording medium may be transmitted through a network connected between computers.

[0085] Furthermore, although preferred embodiments of the present invention have been illustrated and described above, the present invention is not limited to the specific embodiments described above. Various modifications are possible by those skilled in the art without departing from the essence of the invention as claimed in the claims, and such modifications should not be understood individually from the technical spirit or perspective of the present invention. Explanation of the symbols

[0087] 11~14: Sensors 100 : Manufacturing equipment anomaly detection server 110: Data Collection Unit 120 : Data storage unit 130 : Anomaly prediction section

Claims

Claim 1 A method for detecting abnormalities in manufacturing equipment, comprising: a step of collecting data detected by sensors installed in the manufacturing equipment; a step of storing the collected data; and a step of analyzing the stored data to predict abnormalities in the manufacturing equipment; wherein the prediction step involves predicting abnormalities in the manufacturing equipment using a first artificial intelligence model trained to predict abnormalities in the manufacturing equipment from the first data if the first data among the stored data is greater than or equal to a first threshold value, comparing the second data among the data with a second threshold value if the first data is less than or equal to a first threshold value, and predicting abnormalities in the manufacturing equipment using a second artificial intelligence model trained to predict abnormalities in the manufacturing equipment from the second data if the second data is greater than or equal to a second threshold value, wherein the first data is data that has a significant value later than the second data during the failure process of the manufacturing equipment. Claim 2 delete Claim 3 delete Claim 4 delete Claim 5 A method for detecting abnormalities in manufacturing equipment according to claim 1, wherein the prediction step is characterized by selecting one of the prediction result by the first artificial intelligence model and the prediction result by the third artificial intelligence model trained to predict abnormalities in manufacturing equipment from the first data and the second data when the reliability of the prediction result by the first artificial intelligence model is less than the third threshold value, and predicting an abnormality in manufacturing equipment. Claim 6 A method for detecting abnormalities in manufacturing equipment according to claim 5, wherein the prediction step is characterized by selecting the prediction result of the artificial intelligence model with higher reliability of the prediction result among the first artificial intelligence model and the third artificial intelligence model. Claim 7 A method for detecting abnormalities in manufacturing equipment according to claim 6, characterized in that the first data is one of odor data, noise data, temperature data, and vibration data, and the second data is the other of noise data, temperature data, and vibration data. Claim 8 A method for detecting abnormalities in manufacturing equipment according to claim 7, characterized in that if the first data is odor data, the second data is noise data; if the first data is noise data, the second data is temperature data; and if the first data is temperature data, the second data is vibration data. Claim 9 A method for detecting abnormalities in manufacturing equipment according to claim 8, wherein the second data further includes manufacturing equipment information, and if the first data is vibration data, the second data is manufacturing equipment information. Claim 10 A manufacturing equipment abnormality detection server comprising: a collection unit for collecting data detected by sensors installed in manufacturing equipment; a storage unit for storing the collected data; and a prediction unit for analyzing the stored data to predict abnormalities in manufacturing equipment; wherein the prediction unit predicts abnormalities in manufacturing equipment using a first artificial intelligence model trained to predict abnormalities in manufacturing equipment from the first data if the first data among the stored data is greater than or equal to a first threshold value, compares the second data among the data with a second threshold value if the first data is less than or equal to the first threshold value, and predicts abnormalities in manufacturing equipment using a second artificial intelligence model trained to predict abnormalities in manufacturing equipment from the second data if the second data is greater than or equal to the second threshold value, and wherein the first data is data that has a significant value later than the second data during the failure process of manufacturing equipment.

Images