Abnormality diagnosis device and abnormality diagnosis method

Abstract

The abnormality diagnosis device of the present invention includes an operation signal collection unit that collects a plurality of sensor signals, an operation interval operation signal extraction unit that extracts a sensor signal for each predetermined operation interval of the equipment from the plurality of sensor signals collected by the operation signal collection unit, a signal level acquisition unit that acquires a signal level for each predetermined frequency band for each operation interval for the sensor signal for each operation interval extracted by the operation interval operation signal extraction unit, a deviation calculation unit that calculates a deviation from the signal level during normal operation of the same equipment, operation interval, and frequency band for the signal level for each operation interval and frequency band acquired by the signal level acquisition unit, and an abnormality part identification unit that identifies an abnormal part of the equipment based on the deviation calculated by the deviation calculation unit.

Description

[Technical field]

[0001] The present invention relates to an abnormality diagnosis device and an abnormality diagnosis method for identifying an abnormal portion of equipment using a plurality of different types of sensor signals during operation of the equipment. [Background technology]

[0002] Patent Document 1 describes a diagnostic device that diagnoses an object to be diagnosed by comparing an observed waveform observed from the object to be diagnosed with a reference waveform. Patent Document 2 describes a state estimation device that estimates a change in the state of equipment based on waveforms and frequency information detected by multiple sensors installed in the equipment. Patent Document 3 describes a diagnostic device that detects chattering of rolling rolls using the frequency of an acoustic signal detected by an acoustic sensor installed in a rolling mill. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2008-164490 A [Patent Document 2] International Publication No. 2019 / 017345 [Patent Document 3] International Publication No. 2020 / 157818 Summary of the Invention [Problem to be solved by the invention]

[0004] The techniques described in Patent Documents 1 to 3 are techniques in which abnormality diagnosis by sound, abnormality detection by electrical signals, and abnormality detection by vibration sensors are each performed independently. However, when detecting abnormalities in equipment, for example, abnormality detection using an AE sensor is effective for initial abnormalities such as cracks and minute scratches in bearings, and abnormality detection using a vibration sensor is effective for abnormalities such as wear and rattle, and the sensors capable of abnormality detection may be limited depending on the type and degree of abnormality.

[0005] The present invention has been made to solve the above-mentioned problems, and an object of the present invention is to provide an abnormality diagnosis device and an abnormality diagnosis method that can accurately identify an abnormal portion of equipment. [Means for solving the problem]

[0006] The abnormality diagnosis device of the present invention is an abnormality diagnosis device that identifies abnormal parts of equipment using multiple sensor signals detected using multiple different types of sensors during operation of the equipment, and includes an operation signal collection unit that collects the multiple sensor signals, an operation interval operation signal extraction unit that extracts sensor signals for each predetermined operation interval of the equipment from the multiple sensor signals collected by the operation signal collection unit, a signal level acquisition unit that acquires signal levels for each frequency band predetermined for each operation interval for the sensor signals for each operation interval extracted by the operation interval operation signal extraction unit, a deviation calculation unit that calculates a deviation from the signal level during normal operation of the same equipment, operation interval, and frequency band for the signal levels for each operation interval and for each frequency band acquired by the signal level acquisition unit, and an abnormality part identification unit that identifies abnormal parts of the equipment based on the deviation calculated by the deviation calculation unit.

[0007] The signal level acquisition unit classifies each sensor signal based on predetermined operating conditions of the equipment to acquire a signal level, and the deviation calculation unit calculates a deviation from a signal level during normal operation of the same equipment, operating section, frequency band, and operating conditions, for the signal level acquired by the signal level acquisition unit for each operating section, frequency band, and operating condition.

[0008] The abnormal part identifying unit may identify an abnormal part of the equipment based on a pattern of deviation degrees that is determined for each of the operation sections and that has as its elements the deviation degrees in each of the frequency bands.

[0009] The equipment may further include a database unit that stores conditions for extracting the multiple sensor signals of the equipment and data necessary for identifying abnormal parts of the equipment, and the abnormal part identification unit may identify the abnormal part of the equipment by referring to a table that includes information indicating the relationship between the operating interval and the deviation pattern stored in the database unit and the abnormal part of the equipment.

[0010] The table includes information indicating the operating section, the pattern of the deviation degree, and the relationship between the operating conditions of the equipment and abnormal parts of the equipment, and the abnormal part identification unit refers to the table and identifies the abnormal part of the equipment by taking into consideration the operating section, the pattern of the deviation degree, and the operating conditions of the equipment.

[0011] It is preferable to include a normal state signal level acquisition unit that acquires a signal level for each frequency band predetermined for each operation interval for the sensor signal for each operation interval extracted by the operation interval operation signal extraction unit, and a normal distribution determination and recording unit that classifies the signal levels acquired by the normal state signal level acquisition unit for each equipment, each operation interval, and each frequency band, and records them as signal levels during normal operation of the equipment.

[0012] The normal distribution determination and recording unit may record, for each piece of equipment, a table including information indicating a relationship between a pattern of deviations of signal levels from a signal level during normal operation in each frequency band obtained for each operating section and an abnormal portion of the equipment.

[0013] The normal distribution determining and recording unit may record a table including information indicating the operation interval, the pattern of the deviation degree, and the relationship between the operating conditions of the equipment and the abnormal portion of the equipment.

[0014] The operation section may be determined in association with an operation time of the facility or an operation position of a component device of the facility.

[0015] The operating section may be determined based on a sensor signal installed in the equipment that can identify a change point in the operation of the equipment, by specifying an assumed number of categories based on knowledge about the equipment operation, and determining the position of the categories by extracting change points by a machine learning technique using data from the sensor signal or by extracting change points by classifying data using clustering technology.

[0016] The abnormality diagnosis method of the present invention is an abnormality diagnosis method for identifying an abnormal part of equipment using a plurality of sensor signals detected using a plurality of different types of sensors during operation of the equipment, and includes an operation signal collection step of collecting the plurality of sensor signals; an operation section operation signal extraction step of extracting sensor signals for each predetermined operation section of the equipment from the plurality of sensor signals collected in the operation signal collection step; a signal level acquisition step of acquiring signal levels for each predetermined frequency band for each operation section for the sensor signals for each operation section extracted in the operation section operation signal extraction step; a deviation calculation step of calculating a deviation from the signal level during normal operation of the same equipment, operation section, and frequency band for the signal levels for each operation section and for each frequency band acquired in the signal level acquisition step; and an abnormal part identification step of identifying an abnormal part of the equipment based on the deviation calculated in the deviation calculation step.

[0017] It is preferable that the method further includes a normal state signal level acquiring step of acquiring a signal level for each predetermined frequency band for each operating section for the sensor signal for each operating section extracted in the operating section operation signal extracting step, and a normal distribution determining / recording step of classifying the signal levels acquired in the normal state signal level acquiring step for each equipment, each operating section, and each frequency band, and recording them as signal levels during normal operation of the equipment.

[0018] The multiple different types of sensors may include acoustic sensors, vibration sensors, and AE sensors. Effect of the Invention

[0019] According to the abnormality diagnosis device and acoustic diagnosis method of the present invention, the variability in multiple different types of sensor signals is evaluated and abnormal parts of the equipment are identified based on the abnormality levels for each operating section and each frequency band of the equipment, making it possible to accurately identify abnormal parts of the equipment. [Brief description of the drawings]

[0020] [Figure 1] FIG. 1 is a block diagram showing the configuration of an abnormality diagnosis device according to one embodiment of the present invention. [Diagram 2] FIG. 2 is a flowchart showing the flow of data recording processing according to an embodiment of the present invention. [Diagram 3] FIG. 3 is a diagram for explaining the data recording process according to an embodiment of the present invention. [Figure 4] FIG. 4 is a diagram for explaining a method of extracting a sound pressure signal when the extraction condition of the sound pressure signal is specified using the crank angle. [Diagram 5] FIG. 5 is a diagram showing an example of change in sound pressure level of operation sound in one frequency band with respect to crank angle. [Figure 6] FIG. 6 is a diagram showing an example of classification of the operation sounds shown in FIG. [Figure 7] FIG. 7 is a flowchart showing the flow of the abnormal portion identification process according to one embodiment of the present invention. [Figure 8] FIG. 8 is a diagram for explaining a specific example of the abnormal portion specifying process according to an embodiment of the present invention. [Figure 9] FIG. 9 is a diagram showing frequency distribution of sound pressure levels when the operating conditions of the press equipment are limited and when they are not limited. [Figure 10] FIG. 10 is a diagram showing the frequency distribution of the AE wave. [Figure 11] FIG. 11 is a diagram for explaining the abnormal portion specifying process of the embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0021] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An abnormality diagnosis device and an abnormality diagnosis method according to an embodiment of the present invention will now be described with reference to the drawings.

[0022] 〔composition〕 Fig. 1 is a block diagram showing the configuration of an abnormality diagnosis device according to one embodiment of the present invention. As shown in Fig. 1, the abnormality diagnosis device 1 according to one embodiment of the present invention is a device that identifies an abnormal part of one or more predetermined diagnosis target facilities (hereinafter, abbreviated as facilities) A ​​by using a sensor signal during operation of the facilities A, and is configured by an information processing device such as a computer. Here, the facilities A are facilities that are configured by a plurality of parts, and examples of such facilities include steel-making facilities such as press facilities and rolling facilities.

[0023] The abnormality diagnosis device 1 includes an operation signal collection unit 11, an operation section operation signal extraction unit 12, a normal state signal level acquisition unit 13, a normal distribution determination and recording unit 14, a signal level acquisition unit 15, a deviation degree calculation unit 16, an abnormal part identification unit 17, and a database unit 18. The operation signal collection unit 11, the operation section operation signal extraction unit 12, the normal state signal level acquisition unit 13, the normal distribution determination and recording unit 14, the signal level acquisition unit 15, the deviation degree calculation unit 16, and the abnormal part identification unit 17 are functional blocks realized by an arithmetic processing unit such as a CPU inside an information processing device executing a computer program. The functions of each unit will be described later.

[0024] The database unit 18 is configured by a storage device such as an HDD that constitutes an information processing device, and includes an extraction condition database 18a, a frequency band database 18b, a normal distribution database 18c, an operating condition database 18d, and a deviation pattern database 18e. Details of each database will be described later. The database unit 18 may be arranged outside the abnormality diagnosis device 1, and the abnormality diagnosis device 1 may acquire data from the database unit 18 via an electric communication line.

[0025] The abnormality diagnosis device 1 having such a configuration executes the data recording process and the abnormal part identification process described below to identify the abnormal part of the equipment A. The operation of the abnormality diagnosis device 1 when executing the data recording process and the abnormal part identification process will be described below.

[0026] [Data recording processing] Fig. 2 is a flowchart showing the flow of data recording processing according to one embodiment of the present invention. The flowchart shown in Fig. 2 starts when an execution command for data recording processing is input to the abnormality diagnosis device 1, and the data recording processing proceeds to step S1.

[0027] In the process of step S1, the operation signal collector 11 uses a plurality of different types of sensors 20 installed near the facility A to collect a plurality of sensor signals (operation signals) of the facility A during a predetermined period of normal operation including operation intervals of k types (k≧1) of the facility A classified in advance, at a predetermined time interval. Examples of the plurality of different types of sensors 20 include an acoustic sensor, a vibration sensor, and an AE (Acoustic Emission) sensor. With this, the process of step S1 is completed, and the data recording process proceeds to the process of step S2.

[0028] In the process of step S2, the movement section operation signal extraction unit 12 extracts sensor signals for each of k types of movement sections from the multiple sensor signals collected in the process of step S1. Specifically, the movement section operation signal extraction unit 12 reads out extraction conditions (extraction start time and extraction end time, etc.) for the sensor signals for each of k types of movement sections from the extraction condition database 18a. For example, in the example shown in FIG. 3(a), the movement section operation signal extraction unit 12 reads out information on the start time and end time of movement section 1, where the load rises and returns to a predetermined value, and the start time and end time of movement section 2, where the load maintains a predetermined value, as extraction conditions. Then, the movement section operation signal extraction unit 12 extracts sensor signals that satisfy the extraction conditions from the sensor signals collected in the process of step S1, thereby extracting sensor signals for each of k types of movement sections. For example, in the example shown in FIG. 3(b), the movement section movement signal extraction unit 12 extracts sound pressure signals that satisfy extraction conditions from the sound pressure signals collected using the acoustic sensor in the processing of step S1, thereby extracting sound pressure signals for each of k types of movement sections.

[0029] The extraction condition of the sensor signal is not limited to the method of designation using the start time and end time of the operation section, and may be designated based on the operation position of the target equipment. For example, in the case of a press equipment that processes a workpiece by pressing a die against the workpiece using a crank mechanism, the operation section of the press equipment can be designated by the magnitude of the crank angle at which the workpiece comes into contact. Specifically, if the crank angle when the workpiece first comes into contact with the die is α and the crank angle when the workpiece leaves the die is β, the press operation section can be designated by the crank angles α and β. In addition, if there is a process after the press operation (for example, the operation of a conveyor, etc.), the operation section of the process after the press can be designated by using a crank angle larger than the crank angle β. Therefore, in this case, the extraction condition of the sensor signal can be designated using the crank angle. When the extraction condition of the sensor signal is designated based on the operation position of the target equipment, information that associates the extraction condition with the operation position of the equipment may be recorded and stored in the extraction condition database 18a.

[0030] FIG. 4 is a diagram for explaining a method of extracting a sound pressure signal when the extraction condition of the sound pressure signal of the acoustic sensor as a sensor signal is specified using a crank angle. The crank angle α when the workpiece first contacts the die can be derived from the geometric relationship between the workpiece and the crank mechanism. In the example shown in FIG. 4, the press operation section in which the press load is applied to the workpiece is divided into a plurality of sections (subdivision sections 1 to 3), and a sound pressure signal is extracted in each subdivision section. When extracting the sound pressure signal, since the crank angle uniquely corresponds to the start time and end time of each subdivision section, it is sufficient to extract the sound pressure signal within the range of the crank angle.

[0031] As described above, the time and the operating position for specifying the operation section can be determined by considering the relationship between the manufacturing process of the product for each equipment and the operation of the equipment, but it is not easy to specify the section where the characteristic sound can be effectively extracted by the time and the operating position. In such a case, the operation section may be specified by using information on the change point of the sensor signal indicating the change of the equipment operation. For example, in the case of the above-mentioned press equipment, the start timing of the press operation can be detected by the change of the load of the press load, so the subdivision section may be specified by using the load cell signal that detects the press load. Specifically, in the example shown in FIG. 4, the load signal indicates the change of the load cell signal, and the subdivision section can be specified by including not only each rising and falling position of the load signal but also the section where the load signal is maximum. In this way, the position of each operation section can be specified by relating the time and the operating position such as the above-mentioned crank angle to the change point of the sensor signal such as the load cell that indicates the change of the equipment operation as described above.

[0032] In addition, the operation section can also be specified using a classification technique based on a combination of knowledge about the equipment operation and data on the location of occurrence of physical changes (e.g., operation sounds obtained by an acoustic sensor) observed in a sensor signal associated with the equipment operation. A single equipment operation may generate multiple operation sounds, and even in such cases, diagnostic accuracy can be improved by classifying the occurrence patterns of operation sounds associated with the equipment operation. For example, in the case of press equipment, the operation from pressing the workpiece with a die to returning is considered to be one equipment operation, and the operation sounds can be classified by subdividing the occurrence location of the operation sounds in that process by the crank angle, etc. Specifically, the operation section can be specified as follows based on prior knowledge of the press operation.

[0033] (1) Section where the workpiece is transported: Operating noise is generated by the transport machine. (2) Section where the workpiece and die come into contact: Operating noise is generated by friction and collision vibrations when the workpiece and die come into contact. The operating noise is the sound of the workpiece and die coming into contact. (3) Section where the workpiece and die are separated: Operating noise is generated by the movement and vibration of the crank mechanism. The operating noise corresponds to the vibration and friction noise caused by the movement of the crank mechanism and die.

[0034] Although the frequency of the operation sound may differ in each of the above sections, the sound pressure level of the operation sound will differ greatly because the section includes a period in which the workpiece is pressed. Therefore, it is advisable to divide the operation section by focusing on the sound pressure level. FIG. 5 is a diagram showing an example of the change in the sound pressure level of the operation sound of one frequency band with respect to the crank angle. FIG. 5 shows the change in the sound pressure level in three press operations. As shown in FIG. 5, peaks in the sound pressure level can be confirmed near crank angles of 50°, 150°, and 250°. Therefore, since the peaks and the valleys of the peaks are the change points of the operation sound, in the example shown in FIG. 5, the operation sound is divided into three operation sounds. The change points of the operation sound can be obtained from the data, and a machine learning method can be used to detect the change points. In addition, for example, the K-means algorithm can be used as a method of classifying the operation sound. The K-means algorithm is a method of classifying data into a plurality of predetermined clusters. In the example shown in FIG. 5, there are three change points of the operation sound, so an example of classifying the operation sound into three is shown in FIG. 6(a) to (c).

[0035] In this way, when classifying the operation sound, the number of divisions of the equipment operation is assumed based on prior knowledge and taking into account the equipment mechanism, the change point of the sound pressure data is calculated based on the assumed number of divisions using a machine learning method or a clustering method, and the division position is determined based on the calculated change point. When determining the division position, data capable of expressing the characteristics or change point information of the operation, such as load data (the above-mentioned load cell data), other than the sound pressure data, may be used. In the above description, the accurate division of the operation section is obtained by focusing on the sound pressure level of a specific frequency band, but the final operation section may be determined by selecting several candidates of the frequency band of interest and performing the same processing for each of them. In the above example, an example was shown in which the sound pressure level signal of an acoustic sensor was used as the sensor signal, but the same processing can also be performed with the signal of an AE sensor or a vibration sensor.

[0036] As described above, it is possible to accurately distinguish between operation sections with different characteristics based on sensor signals that can identify changing points in the operation of the equipment among sensor signals installed in the target equipment or its vicinity (sensor signals such as press load signals and sound pressure levels of acoustic signals in the case of press equipment). It is advisable to register in advance in the extraction condition database 18a conditions for distinguishing the operation sections thus determined based on the time period from the start time of the equipment or the operation position of the equipment (for example, the crank angle of the press equipment) as extraction conditions. This completes the process of step S2, and the data recording process proceeds to the process of step S3.

[0037] In the process of step S3, first, the normal state signal level acquisition unit 13 converts the sensor signal for each operation section cut out in the process of step S2 into a signal level for a frequency using a Fourier transform, an octave band filter, or the like. Next, the normal state signal level acquisition unit 13 reads information (minimum and maximum values ​​of frequency, etc.) on l types (l≧1) of frequency bands used for abnormality detection predetermined for each operation section from the frequency band database 18b. Then, the normal state signal level acquisition unit 13 acquires the signal level for each of the l types of frequency bands (signal level of normal state operation signal) from the converted signal level data using a band pass filter, or the like. For example, in the example shown in FIG. 3(c), the normal state signal level acquisition unit 13 acquires the sound pressure level of the normal state operation sound for each of the predetermined frequency bands 1 to 3. This completes the process of step S3, and the data recording process proceeds to the process of step S4.

[0038] In the process of step S4, the normal distribution determination and recording unit 14 classifies the signal levels of the normal operation signals acquired in the process of step S3 by operation section and by frequency band, and records the data of the signal level distribution of the normal operation signals and their feature quantities (average value, standard deviation, etc.) in the normal distribution database 18c. In addition, the normal distribution determination and recording unit 14 may further classify the signal levels of the normal operation signals for each operating condition, taking into account a plurality of operating conditions (e.g., m types of operating conditions). This completes the process of step S4, and the series of data recording processes ends.

[0039] In the data recording process, in addition to the process of recording data such as operating sounds during normal operation, a data table that organizes the relationship between deviation patterns from normal operation and abnormal parts (described later) is stored in advance in the deviation pattern database 18e. The deviation patterns and their registered contents will be explained after the definition of the deviation degree in the "Abnormal Part Identification Process" below.

[0040] [Abnormal part identification process] Fig. 7 is a flowchart showing the flow of the abnormal part identification process according to one embodiment of the present invention. The flowchart shown in Fig. 7 starts when an execution command for the abnormal part identification process is input to the abnormality diagnosis device 1, and the abnormal part identification process proceeds to step S11.

[0041] In the process of step S11, the operation signal collector 11 collects, at predetermined time intervals, a plurality of sensor signals (operation signals) of the equipment A during a predetermined period including operation sections of k types (k≧1) of the equipment A classified in advance, using a plurality of different types of sensors 20 installed near the equipment A. This completes the process of step S11, and the abnormal part identification process proceeds to the process of step S12.

[0042] In the process of step S12, the operation section operation signal extraction unit 12 extracts sensor signals for each of k types of operation sections from the multiple sensor signals collected in the process of step S11. The process flow of this step S12 is the same as the process flow of step S2 described above, so a detailed description thereof will be omitted. This completes the process of step S12, and the abnormal part identification process proceeds to the process of step S13.

[0043] In the process of step S13, the signal level acquisition unit 15 acquires the signal level for each of k types of operation intervals and l types of frequency bands from the sensor signal extracted in the process of step S12. The process flow of this step S13 is the same as the process flow of step S3 described above, so a detailed description thereof will be omitted. The signal level acquisition unit 15 may read information on m types of operation conditions that affect the operation of the equipment A from the operation condition database 18d, and acquire the signal level for each of k types of operation intervals, l types of frequency bands, and m types of operation conditions. This completes the process of step S13, and the abnormal part identification process proceeds to the process of step S14.

[0044] In the process of step S14, the deviation calculation unit 16 reads out information on the signal level distribution during normal operation and its feature quantities from the normal distribution database 18c for each of k types of operation intervals of equipment A and for each of l types of frequency bands. Next, using the read out information, the deviation calculation unit 16 calculates two or more deviations of the signal level acquired in the process of step S13 with respect to the signal level during normal operation for each of k types of operation intervals and l types of frequency bands. Then, using the calculated deviations, the deviation calculation unit 16 judges whether equipment A is normal (0) or abnormal (1) for each deviation, and generates an array pattern vector of the judgment results (0,1) for each of l types of frequency bands in each operation interval as a deviation pattern.

[0045] The deviation can be calculated, for example, by the following formula (1). Furthermore, the measurement value of the sensor signal level in the target operating section and target frequency band may be a statistical value such as the average value of the measurement values ​​in the operating section. After the deviation is calculated, if the deviation exceeds a predetermined threshold, it is determined to be abnormal (1), and otherwise it is determined to be normal (0). The threshold may be set, for example, to 3σ, which is about three times the standard deviation σ, but is not limited to this and may be adjusted and determined while observing the characteristics of the operational data.

[0046]

number

[0047] In the process of step S13, when m types of operating conditions are taken into consideration, the deviation calculation unit 16 calculates the deviation by referring to a data table such as that shown in Table 1 below. The data table shown in Table 1 below shows information on the sound pressure level distribution and its feature amount during normal operation for each of k types of operating sections, each of l types of frequency bands, and each of m types of operating conditions. The data table shown in Table 1 below corresponds to the contents of the extraction condition database 18a, the frequency band database 18b, the normal distribution database 18c, and the operating condition database 18d. The data table shown in Table 1 is for the case where the sensor signal is a sound pressure signal collected using an acoustic sensor, and a similar data table is prepared for each type of sensor. This completes the process of step S14, and the abnormal part identification process proceeds to the process of step S15.

[0048] [Table 1]

[0049] In the process of step S15, the abnormal part identifying unit 17 identifies an abnormal part of the equipment A based on the deviation pattern y' generated in the process of step S14. Specifically, as shown in Figs. 8(a) and (b), the abnormal part identifying unit 17 identifies an abnormal part of the equipment A based on the operation section and deviation pattern y' registered in the deviation pattern database 18e in advance. k From the data table showing the relationship between the abnormal parts of equipment A and deviation pattern y', deviation pattern y similar to deviation pattern y k The data table shown in FIG. 8(b) records deviation patterns obtained when an abnormality occurs in the equipment and organizes them in association with the abnormal parts. This data table is created in advance and stored in the deviation pattern database 18e by the normal distribution determination and recording unit 14. The data table can be configured to identify the abnormal part based on a deviation pattern that takes into account m types of operating conditions in addition to classification for k types of operating sections and l types of frequency bands. Therefore, the abnormal part identifying unit 17 may identify the abnormal part of the equipment A by taking into account m types of operating conditions.

[0050] The abnormal part identifying unit 17 uses the Euclidean distance to identify the deviation pattern y k Calculate the similarity of deviation pattern y' to k The similarity shown in formula (2) is calculated by finding whether deviation pattern y' is larger than deviation pattern y k The more similar the deviation pattern y is, the closer the value is to 1. k 8(a) and (b), since deviation pattern y' generated in the processing of step S14 matches deviation pattern y2, abnormal part identifying unit 17 identifies part B corresponding to deviation pattern y2 as the abnormal part of equipment A. This completes the processing of step S15, and the series of abnormal part identifying processes ends.

[0051]

number

[0052] As is clear from the above description, the abnormality diagnosis device 1 according to one embodiment of the present invention includes an operation signal collection unit 11 that collects a plurality of sensor signals of equipment A, an operation interval operation signal extraction unit 12 that extracts a sensor signal for each predetermined operation interval of equipment A from the plurality of sensor signals collected by the operation signal collection unit 11, a signal level acquisition unit 15 that acquires a signal level for each predetermined frequency band for each operation interval for the sensor signal for each operation interval extracted by the operation interval operation signal extraction unit 12, a deviation calculation unit 16 that calculates a deviation from a signal level during normal operation of the same equipment, operation interval, and frequency band for the signal level for each operation interval and each frequency band acquired by the signal level acquisition unit 15, and an abnormal part identification unit 17 that identifies an abnormal part of the equipment based on the deviation calculated by the deviation calculation unit 16. According to this configuration, the variation of a plurality of different types of sensor signals is evaluated, and an abnormal part of equipment A is identified based on the signal level for each operation interval and each frequency band of equipment A, so that an abnormal part of equipment A can be identified with high accuracy. This also enables the rapid adjustment, repair or replacement of abnormal parts, making it possible to prevent deterioration or defects in the quality of manufactured products and contribute to improving product yields.

[0053] Moreover, the abnormality diagnosis device 1 according to one embodiment of the present invention further includes a normal state signal level acquisition unit 13 that acquires a signal level for each frequency band determined in advance for each operation section for the sensor signal for each operation section during normal operation of the equipment A extracted by the operation section operation signal extraction unit 12, and a normal distribution determination and recording unit 14 that classifies the signal level acquired by the normal state signal level acquisition unit 13 for each equipment, each operation section, and each frequency band, records it as a signal level during normal operation of the equipment A, and records a table in which a deviation pattern obtained when an abnormality occurs in the equipment A for each frequency band is associated with an abnormal part for each operation section. According to this configuration, the variation of a plurality of different types of sensor signals is evaluated, and a table in which a deviation pattern obtained when an abnormality occurs and an abnormal part are associated with a signal level for each equipment, each operation section, and each frequency band for the sensor signal level during normal operation of the equipment A is recorded. As a result, even if there is a change in the signal level of the sensor signal under normal conditions, a new recording process is performed for the normal signal level, and the abnormal part of equipment A is identified based on the signal level for each operating section and each frequency band using the new normal signal level data as a reference, so that the abnormal part of equipment A can be identified with high accuracy.

[0054] In addition, the abnormality diagnosis device 1 according to one embodiment of the present invention identifies an abnormal part of the equipment A by taking into account the operating conditions of the equipment A, and therefore can identify the abnormal part of the equipment A with high accuracy. That is, in general, in steelmaking equipment such as press equipment and rolling equipment, even during normal operation, operating conditions such as the amount of rolling and materials change, and the magnitude and frequency of the sensor signal vary accordingly. For this reason, the normal range of the magnitude and frequency of the sensor signal becomes wider, and the magnitude and frequency of the sensor signal are buried within the normal range, and as a result, the abnormality of the equipment cannot be accurately detected by the magnitude and frequency of the sensor signal alone. In contrast, the abnormality diagnosis device 1 according to one embodiment of the present invention identifies an abnormal part of the equipment A by taking into account the operating conditions of the equipment A, and therefore can identify the abnormal part of the equipment A with high accuracy. EXAMPLES

[0055] In this embodiment, an abnormal part of the press equipment was identified. By installing an acoustic sensor around the press equipment, the spectrum of the impact sound can be obtained at the moment when a press load occurs, and the operation sound of the conveyor can be obtained while the conveyor is operating after the press is completed. In this embodiment, the frequency distribution range of the sound pressure level during normal operation of the press equipment was narrowed by taking into account the operating conditions of the press equipment, and the abnormal part of the press equipment was identified. Figure 9(a) shows the frequency distribution of the sound pressure level when the operating conditions of the press equipment are limited (average value 87.6 dB, standard deviation σ2.19), and Figure 9(b) shows the frequency distribution of the sound pressure level when the operating conditions of the press equipment are not limited (average value 85.7 dB, standard deviation σ1.6).

[0056] Next, assuming a die abnormality, the deviation of the press equipment when the sound pressure level is 91 dB was calculated. The following formula (3) was used to calculate the deviation of the press equipment. As a result, when the operating conditions were not limited, the deviation of the press equipment when the sound pressure level was 91 dB was calculated to be 1.5, and when the operating conditions were limited, the deviation of the press equipment when the sound pressure level was 91 dB was calculated to be 3.3. Therefore, if the deviation threshold for determining that an abnormality has occurred in the press equipment is set to 3, it is possible to accurately determine abnormalities in the press equipment.

[0057]

number

[0058] In addition, an AE sensor was installed on the bearings of the press equipment, and the operating range was limited to measure the frequency distribution of AE waves obtained from the bearings. The measurement results are shown in Fig. 10. Assuming that there are minute defects in the bearings, AE waves capable of detecting minute defects in the bearings were monitored in a high frequency band of 100 kHz or more. When the deviation degree when the assumed abnormality is 60 dB is calculated using the above formula (3), the deviation degree was 8, confirming that there are minute defects in the bearings. In addition, as shown in Fig. 11, by searching for deviation patterns similar to the deviation pattern calculated from deviation patterns registered in advance, it was possible to accurately determine that the abnormal part of the equipment was the bearing.

[0059] Although the embodiments of the present invention have been described above, the present invention is not limited by the descriptions and drawings that form part of the disclosure of the present invention according to the present embodiments. In other words, other embodiments, examples, and operation techniques, etc., made by those skilled in the art based on the present embodiments are all included in the scope of the present invention. [Industrial Applicability]

[0060] According to the present invention, it is possible to provide an abnormality diagnosis device and an abnormality diagnosis method capable of accurately identifying an abnormal portion of equipment. [Explanation of symbols]

[0061] 1. Abnormality diagnosis device 11 Operation signal collection unit 12. Operation section operation signal extraction section 13 Normal signal level acquisition section 14 Normal distribution determination and recording section 15 Signal level acquisition section 16 Degree of deviation calculation section 17 Abnormal part identification part 18 Database Department 18a Extraction condition database 18b Frequency Band Database 18c Normal Distribution Database 18d Operating Conditions Database 18e Deviation Pattern Database 20 Sensors A Equipment to be diagnosed

Claims

1. An abnormality diagnosis device that identifies an abnormal part of equipment using a plurality of sensor signals detected during operation of the equipment using a plurality of different types of sensors, an operation signal collector that collects the plurality of sensor signals; an operation section operation signal extracting unit that extracts a sensor signal for each predetermined operation section of the equipment from the plurality of sensor signals collected by the operation signal collecting unit; a signal level acquisition unit that acquires a signal level for each frequency band that is predetermined for each operation section from the sensor signal for each operation section extracted by the operation section operation signal extraction unit; a deviation calculation unit that calculates a deviation of the signal level for each of the operation sections and each of the frequency bands acquired by the signal level acquisition unit from a signal level during normal operation of the same equipment, operation section, and frequency band; an abnormality portion identifying unit that identifies an abnormal portion of the equipment based on the deviation degree calculated by the deviation degree calculating unit; An abnormality diagnosis device comprising:

2. The signal level acquisition unit classifies each sensor signal based on a predetermined operating condition of the facility and acquires a signal level; The deviation degree calculation unit calculates a deviation degree from a signal level during normal operation of the same equipment, operation section, frequency band, and operation condition for each of the operation sections, the frequency band, and the operation condition acquired by the signal level acquisition unit. The abnormality diagnosis device according to claim 1 .

3. The abnormality diagnosis device according to claim 1 , wherein the abnormality portion identifying unit identifies an abnormal portion of the equipment based on a pattern of deviation degrees that are elements of the deviation degrees in each frequency band obtained for each of the operation sections.

4. a database unit that stores data necessary for identifying extraction conditions for the plurality of sensor signals of the facility and an abnormal portion of the facility; 4. The abnormality diagnosis device according to claim 3, wherein the abnormality part identifying unit identifies the abnormal part of the equipment by referring to a table including information indicating a relationship between the operation interval and the pattern of the deviation degree stored in the database unit and the abnormal part of the equipment.

5. the table includes information indicating the operation section, a pattern of the deviation degree, and a relationship between an operating condition of the equipment and an abnormal portion of the equipment, the abnormality portion identifying unit identifies an abnormal portion of the equipment by referring to the table and taking into consideration the operation section, the pattern of the deviation degree, and an operating condition of the equipment. The abnormality diagnosis device according to claim 4.

6. a normal state signal level acquisition unit that acquires a signal level for each frequency band that is predetermined for each operation section for the sensor signal for each operation section extracted by the operation section operation signal extraction unit; a normal distribution determination and recording unit that classifies the signal levels acquired by the normal signal level acquisition unit for each piece of equipment, each operating section, and each frequency band, and records the signal levels as signal levels during normal operation of the equipment; The abnormality diagnosis device according to any one of claims 1 to 5, comprising:

7. 7. The abnormality diagnosis device according to claim 6, wherein the normal distribution determination and recording unit records, for each piece of equipment, a table including information indicating a relationship between a pattern of deviations, the pattern being determined by the deviations of signal levels from signal levels during normal operation in each frequency band obtained for each operating section, and abnormal parts of the equipment.

8. 8. The abnormality diagnosis device according to claim 7, wherein the normal distribution determination and recording unit records a table including information indicating the operating section, a pattern of the deviation degree, and a relationship between an operating condition of the equipment and an abnormal portion of the equipment.

9. The abnormality diagnosis device according to any one of claims 1 to 5, wherein the operation section is determined in association with an operation time of the facility or an operation position of a component device of the facility.

10. The abnormality diagnosis device according to any one of claims 1 to 5, wherein the operation section is determined based on specifying an assumed value for the number of sections based on knowledge regarding equipment operation for a sensor signal installed in the equipment that can identify a change point in the operation of the equipment, and determining the section position by extracting change points by a machine learning technique using data of the sensor signal or by extracting change points by data classification using clustering technology.

11. 6. The abnormality diagnosis device according to claim 1, wherein the plurality of different types of sensors include an acoustic sensor, a vibration sensor, and an AE sensor.

12. An abnormality diagnosis method for identifying an abnormal portion of equipment using a plurality of sensor signals detected during operation of the equipment using a plurality of different types of sensors, comprising: an operational signal acquisition step of acquiring the plurality of sensor signals; an operation section operation signal extraction step of extracting a sensor signal for each predetermined operation section of the equipment from the plurality of sensor signals collected in the operation signal collection step; a signal level acquiring step of acquiring a signal level for each frequency band predetermined for each operation section for the sensor signal for each operation section extracted in the operation section operation signal extracting step; a deviation calculation step of calculating a deviation of the signal level for each of the operation sections and each of the frequency bands acquired in the signal level acquisition step from a signal level during normal operation of the same equipment, operation section, and frequency band; an abnormality portion identifying step of identifying an abnormal portion of the equipment based on the deviation degree calculated in the deviation degree calculating step; An abnormality diagnosis method comprising:

13. a normal state signal level acquisition step of acquiring a signal level for each frequency band predetermined for each operation section for the sensor signal for each operation section extracted in the operation section operation signal extraction step; a normal distribution determination and recording step of classifying the signal levels acquired in the normal signal level acquisition step for each piece of equipment, each operating section, and each frequency band, and recording the classified signal levels as signal levels during normal operation of the equipment; The abnormality diagnosis method according to claim 12 , comprising:

14. The abnormality diagnosis method according to claim 12 or 13, wherein the plurality of different types of sensors include an acoustic sensor, a vibration sensor, and an AE sensor.

Images