Data extraction device and abnormality monitoring device

Abstract

Provided is a technique capable of improving the accuracy of abnormality determination compared to the prior art. The data extraction device uses, as abnormality determination data, an envelope spectrum, which is a spectrum of an envelope-processed wave obtained by enveloping a vibration wave, and therefore, the data extraction device is less susceptible to noise and improves the accuracy of abnormality determination compared with the prior art in which the spectrum of the vibration wave itself is used as the abnormality determination data. The envelope spectrum of a defined frequency region including the frequency of some specific waves among the plurality of types of waves included in the envelope-processed waves is extracted as abnormality determination data instead of using the entire envelope spectrum of the vibration waves as abnormality determination data, so that data not used in abnormality determination is excluded, and abnormality determination becomes easy. Even if the frequency of some of the specific waves varies with a change in the rotational speed of the bearing, the frequency of some of the specific waves is calculated from the rotational speed of the bearing to determine the defined frequency region, so that abnormality determination can be easily performed without being affected by the change in the rotational speed of the bearing.

Description

Technical Field

[0001] The present invention relates to a data extraction device for extracting data for abnormality determination of a rotating part of a monitored machine and an abnormality monitoring device having such a data extraction device. Background Art

[0002] As such an abnormality monitoring device, there is known an abnormality monitoring device that performs abnormality determination based on a frequency spectrum of a vibration wave during rotation of a rotating component of a monitoring target machine (for example, refer to Patent Document 1).

[0003] Prior art literature

[0004] Patent Literature

[0005] Patent Document 1: Japanese Patent Application Publication No. 2001-99757 (Claim 1, paragraph 0005) Summary of the invention

[0006] Problems to be solved by the invention

[0007] However, in the conventional abnormality monitoring device, it is difficult to make an accurate abnormality determination due to the large influence of noise included in the vibration wave. Therefore, in the present application, a technology capable of improving the accuracy of abnormality determination compared with the conventional technology is disclosed.

[0008] Solutions to Solve Problems

[0009] One aspect of the present invention is a data extraction device that extracts data for abnormality determination of a rotating part of a monitored machine, and includes: an envelope processing unit that performs envelope processing on a vibration wave during rotation of the rotating part to obtain an envelope processed wave; a spectrum calculation unit that calculates a spectrum of the envelope processed wave, i.e., an envelope spectrum; a basic data storage unit that stores ratio data for determining a ratio of a frequency of a specific part of a plurality of waves included in the envelope processed wave when the rotating part rotates at a predetermined reference speed to the reference speed; a frequency calculation unit that calculates the frequency of the specific part of the waves when the rotating part rotates at the actual speed as a real special frequency based on the actual speed of the rotating part and the ratio data; and a data extraction unit that extracts the envelope spectrum of a limited frequency region including the real special frequency in the envelope spectrum when the rotating part rotates at the real speed as the data for abnormality determination. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 This is a schematic diagram of the machine to be monitored.

[0011] Figure 2 It is a circuit diagram showing the structure of the abnormality monitoring device.

[0012] Figure 3 It is a block diagram showing the structure of the abnormality monitoring device.

[0013] Figure 4 This is a graph of the envelope spectrum of the vibration wave of the monitored machine.

[0014] Figure 5 This is a graph showing an example of a degradation shock wave. DETAILED DESCRIPTION

[0015] Reference Figures 1 to 5 An abnormality monitoring device 10 according to an embodiment of the present invention will be described. Figure 1 90 to be monitored by the abnormality monitoring device 10 is shown in FIG. The monitored machine 90 is, for example, a grinding device, and includes: a rotating shaft 91 extending in the vertical direction; a disc-shaped grinding stone 92 fixed to the upper end of the rotating shaft 91; a supporting base 93 that supports two locations in the vertical direction of the rotating shaft 91 via a pair of bearings 94 so as to be rotatable; a motor 89 that is a driving source of the monitored machine 90; and a controller 88 (see FIG. 1 ) that controls the motor 89. Figure 2 ).

[0016] The above-mentioned motor 89, rotating shaft 91, a pair of bearings 94 and other rotating parts included in the monitored object machine 90 rotate at an arbitrary rotation speed (rotation speed) corresponding to the operation relative to the controller 88. In addition, due to the offset between the center of gravity and the center of rotation of each rotating part, a vibration wave of a natural frequency is generated in each rotating part (hereinafter, it is appropriately referred to as "eccentric wave"). In addition, the vibration wave in the operation of the monitored object machine 90 also includes various types of waves caused by the non-roundness or surface roughness at the contact parts between the rotating parts (hereinafter, it is appropriately referred to as "processing distortion wave"). In addition, the vibration wave in the operation of the monitored object machine 90 also includes waves caused by the deterioration of the rotating contact parts.

[0017] Specifically, the pair of bearings 94 are common ball bearings of the same type, such as Figure 1As shown, for example, this pair of bearings 94 is composed of an inner ring 95, an outer ring 96, a plurality of balls 97 arranged between their circumferential surfaces, a retaining frame 98 that retains the plurality of balls 97 at equal intervals, and a pair of sealing rings 99 for retaining grease between the inner ring 95 and the outer ring 96, and these components are in rotational contact with each other. Furthermore, in the case where a portion of the inner ring 95 or the outer ring 96 has a deteriorated portion such as damage, indentation, or peeling, a shock wave (hereinafter, appropriately referred to as a "deterioration shock wave") is generated each time the ball 97, retaining frame 98, or sealing ring 99 collides with the deteriorated portion. As shown in FIG. Figure 5 As conceptually shown in FIG. 1 , such a deterioration shock wave is generated at intervals of a generation period T10 corresponding to the relative rotation speed of the inner ring 95 and the outer ring 96 (hereinafter simply referred to as "the rotation speed of the bearing 94"), and becomes a frequency inherent to a certain part of the inner ring 95, the retainer 98, and the like. It should be noted that although the bumps and grooves caused by the regular processing errors may also generate shock waves of a frequency inherent to each component, they are smaller than the deterioration shock wave.

[0018] However, if the position of the deteriorated portion when the bearing 94 starts to crack is substantially the same and the bearing 94 is of the same type, the above-mentioned generation period T10 is the same, and the natural frequency of the component parts of the bearing 94 that resonate due to the collision is also the same. Therefore, if the type of the bearing 94 is known, the progress of the degradation can be investigated by investigating the change in the size of the degradation shock wave in advance.

[0019] However, as described above, the vibration waves in the operation of the monitored machine 90 include various waves such as eccentric waves and processing distortion waves in addition to the deterioration shock waves. In addition, in the power spectrum obtained by performing FFT processing on the vibration waves in the operation of the monitored machine 90, Figure 5 The information of the degradation shock wave as illustrated does not appear significantly, so it is difficult to extract the information. In addition, the rotation speed of the bearing 94 is also changed arbitrarily, and the generation period T10 of the degradation shock wave also changes accordingly. Therefore, even if the information of the degradation shock wave appears significantly in the power spectrum, it is difficult to extract the information of the degradation shock wave. That is, in the previous device, it is difficult to monitor the abnormality of the bearing 94 of the monitored object machine 90. Therefore, the abnormality monitoring device 10 of the present embodiment is used to monitor the abnormality of the bearing 94 of the monitored object machine 90.

[0020] In order to monitor the abnormality by the abnormality monitoring device 10, an acceleration sensor 70 for detecting the acceleration in the direction orthogonal to the axial direction of the rotating shaft 91 is installed on the support base 93 of the monitored machine 90. In addition, a clamp ammeter 71 is installed on the power line 89A of the motor 89. Then, the detection results of the acceleration sensor 70 and the clamp ammeter 71 are input to the abnormality monitoring device 10.

[0021] like Figure 2 As shown in FIG. 1 , the abnormality monitoring device 10 includes a CPU 11, a memory 12, an A / D converter 13, and an amplifier (AMP) 14. The abnormality monitoring device 10 as a whole is configured by executing a program (not shown) stored in the memory 12 by the CPU 11. Figure 3 The block diagram shows a device having multiple functional units including a rotation speed detection unit 20, an input control unit 21, a vibration information acquisition unit 22, a frequency calculation unit 23, a basic data storage unit 24, a temporary data storage unit 25, a filter generation unit 26, a spectrum calculation unit 27, a data extraction unit 28, and an envelope processing unit 29.

[0022] The rotation speed detection unit 20 is composed of an A / D converter 13 and a CPU 11, and the detection result of the clamp ammeter 71 is taken into the CPU 11 through the A / D converter 13. And, based on the change of the driving current to the motor 89 detected by the clamp ammeter 71, the actual (realistic) rotation speed of the motor 89 being driven, that is, the actual rotation speed is calculated. More specifically, the motor 89 of the monitoring object machine 90 of the present embodiment is an AC motor, which rotates at a rotation speed corresponding to the frequency of the driving current to the motor 89, that is, the AC current. And, the rotation speed detection unit 20 calculates the frequency of the driving current based on the change of the driving current detected by the clamp ammeter 71, and calculates the actual rotation speed of the motor 89 based on the frequency. It should be noted that in the present embodiment, the rotation speed of the bearing 94 is the same as the rotation speed of the motor 89.

[0023] It should be noted that the clamp ammeter 71 may be installed not on the power line 89A of the motor 89 but on a control line for switching the drive circuit of the motor 89, that is, the inverter circuit. In addition, for example, when the controller 88 outputs information on the rotation speed of the motor 89, the information on the rotation speed of the motor 89 output by the controller 88 may be input to the abnormality monitoring device 10 instead of using the clamp ammeter 71. In addition, instead of the rotation speed of the motor 89, a rotation sensor such as a rotary transformer may be provided on the rotating shaft 91, and the detection result of the rotation speed of the rotating shaft 91 may be taken into the abnormality monitoring device 10.

[0024] The input control unit 21 determines whether the monitored machine 90 is in operation based on whether the actual rotation speed calculated by the rotation speed detection unit 20 is less than a predetermined reference value. When the monitored machine 90 is in operation, a data acquisition instruction is given to the vibration information acquisition unit 22 and the frequency calculation unit 23, and then, while the monitored machine 90 continues to operate, a data acquisition instruction is given to the vibration information acquisition unit 22 and the frequency calculation unit 23 every time a predetermined data acquisition cycle (for example, 1 hour) passes.

[0025] The vibration information acquisition unit 22 is composed of the aforementioned amplifier 14, the A / D converter 13, and the CPU 11. Furthermore, when the vibration information acquisition unit 22 receives the data acquisition instruction, after amplifying the detection signal of the acceleration sensor 70 through the amplifier 14, the A / D converter 13 performs sampling at a predetermined sampling frequency (for example, 15 to 30 [kHz]), thereby acquiring a plurality of acceleration data of a predetermined amount of time (for example, 10 to 40 [s]). In addition, the vibration information acquisition unit 22 makes the acquired plurality of acceleration data into an acceleration data file with acquisition date and time information and stores it in the temporary data storage unit 25.

[0026] The envelope processing unit 29 performs envelope processing on the vibration wave determined by the data of the acceleration data file, thereby obtaining an envelope processing wave. Specifically, a plurality of acceleration data are used to determine the change in acceleration over time, and the plurality of acceleration data constitute vibration data. Furthermore, the envelope processing unit 29 performs envelope processing on the vibration wave determined by the vibration data, thereby obtaining an envelope processing wave.

[0027] The spectrum calculation unit 27 performs spectrum analysis processing (FFT) on the envelope processed wave, and calculates an envelope spectrum which is a spectrum of the envelope processed wave.

[0028] The data extracting unit 28 extracts, as abnormality determination data, an envelope spectrum in a limited frequency region from a lower limit frequency Fb to an upper limit frequency Ft determined based on data of the basic data storage unit 24 as follows, from the envelope spectrum calculated by the spectrum calculating unit 27 .

[0029] The basic data storage unit 24 stores ratio data for determining the ratio of the frequency of a specific part of the multiple waves included in the envelope processing wave to the reference speed when the bearing 94 rotates at a predetermined reference speed. Figure 4, the envelope spectrum when the bearing 94 rotates at the reference speed is shown. The envelope spectrum of the envelope processing wave contains information of multiple waves with frequencies f1 to f9. And, among these waves, for example, the aforementioned degradation shock wave shown by the frequency f6 is included. In addition, in the non-degraded monitoring target machine 90, the frequency of the degradation shock wave when the bearing 94 rotates at the reference speed can be known from the envelope spectrum of the vibration wave of the monitoring target machine 90 using the same type of bearing 94. In the following description, the "degradation shock wave" is assumed before the degradation occurs and the data of the wave with the same frequency as the "degradation shock wave" is extracted. Therefore, in this embodiment, the wave with the same frequency as the "degradation shock wave" is set as "a part of the specific wave". And, when the ratio data is set to K1, the reference speed is set to N1, and the frequency of a part of the specific wave in the reference speed is set to f6, the ratio data K1 obtained by the following formula is stored in the basic data storage unit 24.

[0030] K1=f6 / N1····Formula 1

[0031] It should be noted that as long as the ratio of the base speed and the frequency of the degraded shock wave in the base speed can be determined, the above-mentioned ratio data can be any ratio data. For example, the base speed value and the frequency value can be used as ratio data, or the value obtained by dividing the speed of motor 89 by the frequency of a specific part of the wave can be used as ratio data.

[0032] When receiving the data acquisition instruction, the frequency calculation unit 23 calculates the frequency of a part of the specific wave when the motor 89 rotates at the actual rotation speed as the actual special frequency. Specifically, when the actual rotation speed calculated by the rotation speed detection unit 20 is set to N2, the actual special frequency is set to Fr, and the ratio data as described above is set to K1, the frequency calculation unit 23 calculates the actual special frequency Fr using the following formula 2.

[0033] Fr=K1·N2[Hz]····Formula 2

[0034] The filter generation unit 26 generates a filter by adding a preset allowable error to the actual special frequency Fr. Specifically, the allowable error is stored in the basic data storage unit 24. Here, when the allowable error is set to Fw, the upper limit frequency is set to Ft, and the lower limit frequency is set to Fb, the filter generation unit 26 calculates the upper limit frequency Ft and the lower limit frequency Fb through the following equations 3 and 4.

[0035] Ft=Fr+Fw[Hz]····Formula 3

[0036] Fb=Fr-Fw[Hz]····Formula 4

[0037] Here, the upper limit frequency Ft and the lower limit frequency Fb are set so that the upper limit frequency Ft and the lower limit frequency Fb can be selected from various waves such as Figure 4 The size of the wave of only a certain frequency (frequency f6 in this embodiment) is extracted from a plurality of waves with large spectrum values ​​of the envelope spectrum, such as the waves of frequencies f1 to f9 shown. That is, the "limited frequency region" means a region for extracting a certain frequency wave from a plurality of waves.

[0038] In the present embodiment, the data extraction device 30 is constituted by the above-mentioned rotation speed detection unit 20, input control unit 21, vibration information acquisition unit 22, frequency calculation unit 23, basic data storage unit 24, temporary data storage unit 25, filter generation unit 26, spectrum calculation unit 27, data extraction unit 28, and envelope processing unit 29 in the abnormality monitoring device 10. Moreover, by the data extraction device 30, even if the operation speed of the monitoring object machine 90 is arbitrarily changed, it is possible to select a part of the envelope spectrum of a specific frequency required for the determination of the abnormality of the bearing 94 from a variety of vibrations and extract it as abnormality determination data. Moreover, the abnormality determination data can be used to determine the abnormality by the first feature quantity acquisition unit 31, the second feature quantity acquisition unit 32, the multidimensional data generation unit 33, the accumulated data storage unit 34, the abnormality determination unit 35, the function update unit 36, etc. of the abnormality monitoring device 10.

[0039] The first feature quantity acquisition unit 31 acquires, as feature quantity 1 , a result of normalizing the peak value of the abnormality determination data (that is, the peak value of the envelope spectrum in a limited frequency region).

[0040] The second feature quantity acquisition unit 32 calculates the kurtosis in the statistical distribution of the plurality of acceleration data in the acceleration data file as the feature quantity 2 .

[0041] The multidimensional data generating unit 33 generates two-dimensional multidimensional data having the feature quantity 1 and the feature quantity 2 based on the common acceleration data file as components and stores the generated data in the storage data storage unit 34, and deletes the acceleration data file that is the generation source of the multidimensional data from the temporary data storage unit 25. Thus, in the storage data storage unit 34, a plurality of multidimensional data are stored as the operation time of the monitoring target machine 90 passes.

[0042] The abnormality determination unit 35 determines whether the newly generated multidimensional data is abnormal. To this end, the abnormality determination unit 35 determines whether the new multidimensional data is abnormal using a function for abnormality determination generated in advance by machine learning using OCSVM (One Class Support Vector Machine).

[0043] Here, OCSVM is a well-known machine described in, for example, "Support Vector Machine" (co-authored by Ichiro Takeuchi and Masayuki Toriyama, ISBN 978-4-06-152906-9) as a non-patent document, which acquires a plurality of multidimensional data (also referred to as "vector data" because it is multidimensional) having a plurality of feature quantities as components, and generates a function for abnormality determination as a vector function. At this time, each feature quantity is preferably standardized based on each accumulated feature quantity and used.

[0044] The abnormality determination unit 35 inputs the new multidimensional data as a function for abnormality determination. If the output of the function is "negative", the new multidimensional data is abnormal, that is, the determination result of "monitoring object machine 90 is abnormal" is output. If the output of the function is not "negative", the multidimensional data is normal, that is, the determination result of "monitoring object machine 90 is normal" is output. In addition, the output of the abnormality determination unit 35 is taken into the PLC 15 of the system including the monitoring object machine 90 (refer to Figure 2 ) and it is determined that there is an abnormality, the warning light 16 lights up.

[0045] It should be noted that, when the abnormality monitoring device 10 is shipped together with the monitored machine 90, it is preferred that the monitored machine 90 be operated for a predetermined time or more before shipping so that the abnormality determination unit 35 has a function for abnormality determination by mechanical learning, but the function for abnormality determination that has been mechanically learned by operating other monitored machines 90 of the same structure may be copied to a new abnormality monitoring device 10. In addition, after the abnormality monitoring device 10 and the monitored machine 90 are shipped, the monitored machine 90 may be operated for a predetermined time or more so that the abnormality determination unit 35 may mechanically learn the function for abnormality determination and then use it for abnormality determination of the abnormality monitoring device 10.

[0046] The function updating unit 36 ​​adds the multidimensional data newly stored in the storage data storage unit 34 to the plurality of multidimensional data already stored in the storage data storage unit 34 , and calculates and corrects the abnormality determination function.

[0047] The above is the description of the structure of the abnormality monitoring device 10 of the present embodiment. Next, the effect of the abnormality monitoring device 10 of the present embodiment will be described. As described above, in the data extraction device 30 of the present embodiment, the frequency spectrum of the envelope processed wave obtained by enveloping the vibration wave, that is, the envelope spectrum is used as the abnormality determination data. Therefore, compared with the prior art that uses the frequency spectrum of the vibration wave itself as the abnormality determination data, it is not easily affected by noise, and the accuracy of abnormality determination is improved. In addition, instead of using the entire envelope spectrum of the vibration wave as the abnormality determination data, the envelope spectrum of the limited frequency region including the frequency of a part of a specific wave among the multiple waves included in the envelope processed wave is extracted as the abnormality determination data, so that data not used in the abnormality determination is excluded, and the abnormality determination becomes easy. In addition, even if the frequency of a part of the specific wave changes with the change of the rotation speed of the bearing 94, the frequency of the part of the specific wave is calculated according to the rotation speed of the bearing 94 to determine the limited frequency region. Therefore, if the data extraction device 30 of the present embodiment is used, it is possible to easily perform abnormality determination without being affected by the change of the rotation speed of the bearing 94.

[0048] In addition, simply by installing the clamp ammeter 71 on the power line 89A of the motor 89 serving as the driving source of the monitored machine 90 , information on the actual rotation speed N2 of the rotation action of the motor 89 can be acquired. Therefore, there is no need to significantly change the monitored machine 90 in order to install the abnormality monitoring device 10 .

[0049] In addition, the abnormality determination unit 35 of the abnormality monitoring device 10 has a function for abnormality determination obtained by mechanical learning using the peak of the envelope spectrum containing the above-mentioned limited frequency region as one of the multiple feature quantities for multiple times. And, whether the monitored object machine 90 is abnormal is determined by the output of the function when new multidimensional data is input to the function. As a result, it is possible to easily perform abnormality determination of the monitored object machine 90 whose movement speed changes. In addition, the function for abnormality determination is updated by using the new multidimensional data and OCSVM through the function updating unit 36, so the longer the operating time of the monitored object machine 90, the higher the accuracy of abnormality determination. In addition, by including the kurtosis of the statistical distribution of multiple acceleration data measured as vibration data in the multiple feature quantities, the accuracy of abnormality determination can be improved.

[0050] [Other embodiments]

[0051] The data extraction device 30 of the above-described embodiment extracts the envelope spectrum of one wave included in the multiple vibration waves generated in the monitored machine 90 when the motor 89 is driven as data for abnormality determination, but it can also be configured to extract the envelope spectrum of a part of the multiple waves as data for abnormality determination.

[0052] The abnormality monitoring device 10 of the above embodiment has a function for abnormality determination generated by using OCSVM, but it may also have a function for abnormality determination generated by multivariate analysis other than OCSVM. In addition, it is also possible to set a predetermined reference value to perform abnormality determination without performing multivariate analysis. Specifically, for example, it is also possible to configure the abnormality determination to be performed by determining whether the peak value of the abnormality determination data extracted by the data extraction device 30 is greater than a preset reference value.

[0053] The abnormality monitoring device 10 of the above-described embodiment uses multidimensional data including the characteristic quantities 1 and 2 related to the output of the monitoring target machine 90 (specifically, the vibration output by the monitoring target machine 90) as components to perform abnormality determination using multivariate analysis (specifically, OCSVM), but it may be configured to use multidimensional data including the characteristic quantities related to both the output and input of the monitoring target machine 90 as components to perform abnormality determination using multivariate analysis. Specifically, it is also possible to use multidimensional data including the characteristic quantity related to the input to the monitoring target machine 90, that is, the amount of electric power supplied to the motor 89, and the characteristic quantity related to the output of the monitoring target machine 90, that is, the vibration, as components to perform abnormality determination using multivariate analysis.

[0054] In addition, the multidimensional data in the above-described embodiment is two-dimensional data, but it may be data of three or more dimensions.

[0055] The abnormality monitoring device 10 of the above embodiment is used to determine the abnormality of the bearing 94 of the monitored machine 90, but it can also be used to determine the abnormality of a rotating part other than the bearing. Specifically, the abnormality monitoring device can also be used to determine the abnormality of a rotating part rotatably supported by a fluid bearing or a metal bearing, or a rotating part such as a screw or a propeller rotating in a fluid.

[0056] Furthermore, in the above-described embodiment, a grinding machine is exemplified as the monitoring target machine 90 , but it may be, for example, a press machine.

[0057] <Notes>

[0058] Hereinafter, the feature groups extracted from the above-mentioned embodiments will be described while showing effects etc. as needed. It should be noted that, in order to facilitate understanding, the corresponding structures in the above-mentioned embodiments are appropriately shown with brackets etc., but these feature groups are not limited to the specific structures shown with brackets etc.

[0059] [Feature 1]

[0060] A data extraction device (30) is a data extraction device (30) for extracting data for abnormality determination of a rotating component (94) of a monitored machine (90), comprising: an envelope processing unit (29) for performing envelope processing on a vibration wave of the rotating component (94) during rotation, thereby obtaining an envelope processed wave; a spectrum calculation unit (27) for calculating the spectrum of the envelope processed wave, i.e., the envelope spectrum; and a basic data storage unit (24) for storing ratio data (K1), wherein the ratio data (K1) is used to determine a specific part of a plurality of waves included in the envelope processed wave when the rotating component (94) rotates at a predetermined reference speed (N1). The invention relates to a method for determining a ratio of a frequency (f6) of a wave of a rotating component (94) to a reference rotational speed (N1); a frequency calculation unit (23) configured to calculate, based on the actual rotational speed (N2) of the rotating component (94) and the ratio data (K1), the frequency (f6) of the specific wave when the rotating component (94) rotates at the actual rotational speed (N2) as a real special frequency (Fr); and a data extraction unit (28) configured to extract, as the data for abnormality determination, the envelope spectrum of a limited frequency region including the real special frequency (Fr) in the envelope spectrum when the rotating component (94) rotates at the real rotational speed (N2).

[0061] [Feature 2]

[0062] Based on the data extraction device (30) described in Feature 1, the monitored machine (90) has a motor (89) as a driving source, and a clamp ammeter is installed on the power line (89A) or control line of the motor (89), and the data extraction device (30) has a speed detection unit (20), which calculates the actual speed (N2) of the rotating part (94) based on the change of the current detected by the clamp ammeter (71).

[0063] [Feature 3]

[0064] An abnormality monitoring device (10) includes an abnormality determination unit (35) which obtains a plurality of multidimensional data including a plurality of feature quantities based on the output or input of a monitored object machine (90) having a rotating part (94) as components, and completes mechanical learning of a function for abnormality determination for determining whether the multidimensional data is abnormal, wherein the abnormality monitoring device (10) determines whether the monitored object machine (90) has an abnormality based on the output of the function when new multidimensional data is input to the function, wherein the abnormality monitoring device (10) has a data extraction device (30) recorded in feature 1 or 2, and the peak value of the envelope spectrum in the limited frequency region is included in the plurality of feature quantities.

[0065] [Feature 4]

[0066] In the abnormality monitoring device (10) according to feature 3, the function is a function obtained by machine learning using OCSVM.

[0067] [Feature 5]

[0068] In the abnormality monitoring device (10) according to feature 4, the abnormality monitoring device (10) includes a function updating unit (36) for updating the function using the OCSVM by utilizing the newly acquired multidimensional data.

[0069] [Feature 6]

[0070] In the abnormality monitoring device (10) described in any one of the features 3 to 5, the plurality of feature quantities include a kurtosis, a standard deviation, and other statistical values ​​for determining a statistical distribution of vibration waves during rotation of the rotating component (94).

[0071] In the data extraction device of feature 1, the spectrum of the envelope processed wave obtained by enveloping the vibration wave, i.e., the envelope spectrum, is used as the data for abnormality determination. Therefore, compared with the prior art that uses the spectrum of the vibration wave itself, it is less susceptible to noise and the accuracy of abnormality determination is improved. In addition, instead of using the entire envelope spectrum of the vibration wave as the data for abnormality determination, the envelope spectrum of a limited frequency region including the frequency of a certain wave among the multiple waves included in the envelope processed wave is extracted as the data for abnormality determination. Therefore, data not used in abnormality determination is excluded, and abnormality determination becomes easy. Furthermore, even if the frequency of a certain wave changes with the change of the rotation speed of the rotating part, the frequency of the certain wave is calculated according to the rotation speed of the rotating part to determine the limited frequency region. Therefore, if the data extraction device of feature 1 is used, abnormality determination can be easily performed without being affected by the change of the rotation speed of the rotating part.

[0072] According to the second feature, information on the actual rotation speed of the rotating component can be acquired simply by installing a clamp ammeter on the power line or control line of the motor serving as the driving source of the monitored machine.

[0073] The abnormality determination unit of the abnormality monitoring device of feature 3 includes a function for abnormality determination obtained by machine learning using the peak value of the envelope spectrum including the above-mentioned limited frequency region as one of the plurality of feature quantities for multiple times. And, it is determined whether the monitored machine is abnormal based on the output of the function when new multidimensional data is input to the function. Thus, it is possible to easily determine the abnormality of the monitored machine whose motion speed has changed.

[0074] Here, the function used for abnormality judgment can be a function obtained by machine learning using OCSVM as in the abnormality monitoring device of feature 4, a function obtained by machine learning through a multivariate analysis method other than OCSVM, or a function obtained by pre-learning through artificial intelligence with a neural network.

[0075] According to the fifth feature, the function updating unit updates the abnormality determination function using OCSVM using new multidimensional data, so the accuracy of abnormality determination increases as the operating time of the monitored machine increases.

[0076] As shown in Feature 6, by including the kurtosis, standard deviation, and other statistical values ​​for specifying the statistical distribution of the vibration wave during the rotation of the rotating component in the plurality of feature amounts, the accuracy of abnormality determination can be improved.

[0077] It should be noted that, while specific examples of the technology included in the technical solution are disclosed in this specification and the accompanying drawings, the technology recorded in the technical solution is not limited to these specific examples, but also includes various deformation and alteration methods of the specific examples, and also includes methods of taking out a part from the specific example separately.

Claims

1. A data extraction device for extracting data for abnormality determination of a rotating part of a monitored machine, wherein: The data extraction device comprises: An envelope processing unit performs envelope processing on the vibration wave of the rotating component during rotation, thereby obtaining an envelope processed wave; A spectrum operation unit, which operates the spectrum of the envelope processed wave, that is, the envelope spectrum; A basic data storage unit storing ratio data for determining a ratio of a frequency of a specific part of a plurality of waves included in the envelope processing wave to the reference rotation speed when the rotating member rotates at a predetermined reference rotation speed; a frequency calculation unit configured to calculate, as a real special frequency, a frequency of the specific part of the wave when the rotating member rotates at the real rotation speed, based on the actual rotation speed of the rotating member and the ratio data; and The data extraction unit is configured to extract, as the abnormality determination data, the envelope spectrum in a limited frequency region including the actual special frequency in the envelope spectrum when the rotating member rotates at the actual rotation speed.

2. The data extraction device according to claim 1, wherein: The monitored machine has a motor as a driving source. A clamp ammeter is installed on the power line or control line of the motor. The data extraction device includes a rotation speed detection unit that calculates the actual rotation speed of the rotating member based on a change in the current detected by the clamp ammeter.

3. An abnormality monitoring device, comprising an abnormality determination unit, the abnormality determination unit acquiring a plurality of multidimensional data including a plurality of feature quantities based on the output or input of a monitoring target machine having a rotating part as components, and completing machine learning of a function for abnormality determination for determining whether the multidimensional data is abnormal, the abnormality monitoring device determining the presence or absence of an abnormality of the monitoring target machine based on the output of the function when new multidimensional data is input to the function, wherein: The abnormality monitoring device comprises the data extraction device according to claim 1 or 2, The plurality of feature quantities include a peak of the envelope spectrum in the limited frequency region.

4. The abnormality monitoring device according to claim 3, wherein: The function is obtained by machine learning using OCSVM.

5. The abnormality monitoring device according to claim 4, wherein: The abnormality monitoring device includes a function updating unit that updates the function using the OCSVM by utilizing the newly acquired multidimensional data.

6. The abnormality monitoring device according to any one of claims 3 to 5, wherein: The plurality of feature quantities include a kurtosis, a standard deviation, and other statistical values ​​for determining a statistical distribution of a vibration wave during rotation of the rotating member.

Images