System and method for detecting abnormal feature of rotating machine

Abstract

To provide a system and a method for detecting an abnormal feature of a rotating machine, with which an abnormal portion in an attachment can be effectively found out.SOLUTION: A system 1 and a method for detecting an abnormal feature of a rotating machine include a sensing module 11, a data processing module 12, a conversion module 13, and an abnormal feature detection module 14. The sensing module obtains vibration data of an attachment of the rotating machine. The data processing module processes the vibration data to generate a plurality of pieces of vibration timing data. The conversion module converts the plurality of pieces of vibration timing data into a plurality of pieces of conversion data. The abnormal feature detection module compares a feature value distribution in the plurality of pieces of conversion data to define the attachment corresponding to the conversion data in which a change occurs in the feature value distribution as abnormal.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an abnormality feature detection system and method, and more particularly to an abnormality feature detection system and method for a rotating machine. [Background technology]

[0002] Attachments (e.g., gears, chains / belts, saw belts, drill bits, tires, etc.) used in rotating machinery (e.g., power transmissions, belt conveyors, power saws, power drills, vehicles, etc.) can break down after a certain period of use, affecting the operation of the entire rotating machinery. Therefore, a technology for detecting attachment damage was developed. Conventional technology directly converts detected attachment vibration data into frequency for analysis. However, with frequency-based analysis, the frequency varies depending on the operating speed, so there is no consistent comparison standard and it cannot effectively respond to different operating speeds of rotating machinery. Furthermore, to prevent resonance during use, the spacing between the multiple saw teeth on the saw belt must be uneven, making it difficult to determine the location of a damaged saw tooth. Summary of the Invention [Problem to be solved by the invention]

[0003] An object of the present invention is to provide a system and method for detecting abnormal characteristics of a rotating machine, which uses the interval of converted data as the X-axis instead of frequency as the X-axis, and thereby can effectively locate abnormal parts of an attachment even under circumstances where (1) the rotating machine is operating at different speeds, or (2) the intervals between multiple saw teeth on the saw belt are uneven to prevent resonance from occurring during use. [Means for solving the problem]

[0004] In order to solve the above-mentioned problems, according to a first form of the present invention, there is provided an abnormality feature detection system for a rotating machine, comprising a sensing module, a data processing module, a conversion module, and an abnormality feature detection module, wherein the sensing module obtains vibration data of an attachment of the rotating machine, the data processing module processes the vibration data to generate a plurality of vibration timing data, the conversion module converts the plurality of vibration timing data into a plurality of converted data, and the abnormality feature detection module compares feature value distributions in the plurality of converted data, and defines an attachment corresponding to the converted data in which a change has occurred in the feature value distribution as abnormal.

[0005] Preferably, the abnormal feature detection module compares whether a change has occurred in the feature value distribution of the same interval section in any two adjacent transformed data, or first defines a plurality of transformed data groups in a manner that N pieces of transformed data form one set, and then compares whether a change has occurred in the feature value distribution of the same interval section in any two adjacent transformed data groups, where N is a natural number.

[0006] The change occurring in the distribution of the characteristic values is preferably such that a plurality of third quartiles corresponding to the same interval section in the plurality of transformed data or the plurality of transformed data groups are different from each other.

[0007] It is preferable that the plurality of pieces of conversion data among the plurality of conversion data groups overlap with each other partially or not at all.

[0008] Preferably, the anomalous feature detection module defines an attachment as anomalous when a plurality of the third quartiles differ from each other by a maximum.

[0009] Preferably, the feature value distribution is a signal-to-noise ratio distribution, and the third quartile is calculated using a plurality of signal-to-noise ratios in an interval section.

[0010] It is preferable that the data processing module first calculates the standard deviation of the vibration data, defines a portion of the vibration data that is greater than the standard deviation as an operating section, and defines a portion of the vibration data that is smaller than the standard deviation as a waiting section, and that the vibration data corresponding to the operating section between two adjacent waiting sections is vibration timing data.

[0011] Preferably, the conversion module first converts the plurality of vibration timing data into a plurality of intermediate data, and then converts the plurality of intermediate data into a plurality of converted data.

[0012] Preferably, the conversion module converts the plurality of vibration timing data into a plurality of intermediate data using a first algorithm, the X axis of the intermediate data being frequency, the Y axis being signal-to-noise ratio, and the signal-to-noise ratio being the vibration amount of the vibration timing data divided by the standard deviation.

[0013] Preferably, the first algorithm is a Fourier transform, a fast Fourier transform, a wavelet transform or an empirical mode decomposition.

[0014] Preferably, the conversion module converts the plurality of intermediate data into a plurality of converted data respectively using a second algorithm, and the X axis of the converted data is interval and the Y axis is signal-to-noise ratio.

[0015] The second algorithm is expressed by the following formula (1):

number

[0016] The vibration data is preferably mechanical motion vibration data or audio vibration data.

[0017] In order to solve the above problem, according to a second aspect of the present invention, there is provided an abnormality feature detection method for a rotating machine, characterized by including the steps of: a sensing module obtaining vibration data of an attachment of the rotating machine; a data processing module processing the vibration data to generate a plurality of vibration timing data; a conversion module converting the plurality of vibration timing data into a plurality of converted data; and an abnormality feature detection module comparing feature value distributions in the plurality of converted data and defining as abnormal an attachment corresponding to the converted data in which a change has occurred in the feature value distribution.

[0018] Preferably, the abnormal feature detection module compares whether a change has occurred in the feature value distribution of the same interval section in any two adjacent transformed data, or first defines a plurality of transformed data groups in a manner that N pieces of transformed data form one set, and then compares whether a change has occurred in the feature value distribution of the same interval section in any two adjacent transformed data groups, where N is a natural number.

[0019] The change occurring in the distribution of the characteristic values is preferably such that a plurality of third quartiles corresponding to the same interval section in the plurality of transformed data or the plurality of transformed data groups are different from each other.

[0020] It is preferable that the plurality of pieces of conversion data among the plurality of conversion data groups overlap with each other partially or not at all.

[0021] Preferably, the anomalous feature detection module defines an attachment as anomalous when a plurality of the third quartiles differ from each other by a maximum.

[0022] Preferably, the feature value distribution is a signal-to-noise ratio distribution, and the third quartile is calculated using a plurality of signal-to-noise ratios in an interval section.

[0023] It is preferable that the data processing module first calculates the standard deviation of the vibration data, defines a portion of the vibration data that is greater than the standard deviation as an operating section, and defines a portion of the vibration data that is smaller than the standard deviation as a waiting section, and that the vibration data corresponding to the operating section between two adjacent waiting sections is vibration timing data.

[0024] Preferably, the conversion module first converts the plurality of vibration timing data into a plurality of intermediate data, and then converts the plurality of intermediate data into a plurality of converted data.

[0025] Preferably, the conversion module converts the plurality of vibration timing data into a plurality of intermediate data using a first algorithm, the X axis of the intermediate data being frequency, the Y axis being signal-to-noise ratio, and the signal-to-noise ratio being the vibration amount of the vibration timing data divided by the standard deviation.

[0026] Preferably, the first algorithm is a Fourier transform, a fast Fourier transform, a wavelet transform or an empirical mode decomposition.

[0027] Preferably, the conversion module converts the plurality of intermediate data into a plurality of converted data respectively using a second algorithm, and the X axis of the converted data is interval and the Y axis is signal-to-noise ratio.

[0028] The second algorithm is expressed by the following formula (1):

number

[0029] The vibration data is preferably mechanical motion vibration data or audio vibration data. [Effects of the Invention]

[0030] The system and method for detecting abnormal characteristics of a rotating machine according to the present invention uses the interval of the converted data as the X-axis instead of the frequency as the X-axis, and therefore can effectively locate abnormal locations in the attachment even under the following conditions: (1) when the rotating machine is operating at different speeds, and (2) when the intervals between the multiple saw teeth on the saw belt are uneven to prevent resonance during use. [Brief explanation of the drawings]

[0031] [Figure 1] 1 is a configuration diagram showing an abnormality feature detection system for a rotating machine according to an embodiment of the present invention; [Figure 2] 1 is a flow chart illustrating a method of an anomaly feature detection system for a rotating machine according to an embodiment of the present invention. [Figure 3] 1 is a graph showing vibration data according to an embodiment of the present invention. [Figure 4] 10 is a graph showing vibration timing data according to one embodiment of the present invention. [Figure 5] 10 is a graph showing intermediate data according to an embodiment of the present invention. [Figure 6] 1 is a graph illustrating conversion data according to an embodiment of the present invention. [Figure 7] 1 is a graph showing a series of multiple transform data according to an embodiment of the present invention. [Figure 8] 10 is a graph showing a distribution change of a feature value according to an embodiment of the present invention; DETAILED DESCRIPTION OF THE INVENTION

[0032] In order to enable those skilled in the art to fully understand the objectives, features, and effects of the present invention, the system and method for detecting abnormal characteristics of a rotating machine according to the present invention will be described in detail below with reference to suitable embodiments and accompanying drawings.

[0033] Please refer to Fig. 1. Fig. 1 is a configuration diagram showing an abnormality feature detection system for a rotating machine according to one embodiment of the present invention.

[0034] As shown in FIG. 1, an abnormality characteristic detection system 1 for a rotating machine according to one embodiment of the present invention includes a sensing module 11, a data processing module 12, a conversion module 13, and an abnormality characteristic detection module .

[0035] The rotating machinery anomaly characteristic detection system 1 according to one embodiment of the present invention can be implemented in a computer device having a processing unit and a storage unit. The processing unit may be a central processing unit (CPU), a microprocessor, a graphics processing unit (GPU), or an application-specific integrated circuit (ASIC). The storage unit may be any type of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory, hard disk, soft disk, database, or a combination of the above-mentioned components. The computer device may be a mobile phone, tablet computer, laptop computer, desktop computer, server, or cloud server.

[0036] The data processing module 12, the conversion module 13 and the abnormal feature detection module 14 are program code fragments, software or firmware respectively stored in a storage unit and executed by a processing unit, but the present invention is not limited to these. For example, the data processing module 12, the conversion module 13 and the abnormal feature detection module 14 may be realized by other hardware or a combination of software and hardware.

[0037] The sensing module 11 acquires vibration data of an attachment of a rotating machine. In this embodiment, the sensing module 11 is specifically a sensor made of a piezoelectric material (crystal or ceramic) or a general vibration sensor, but the present invention is not limited to these. If the sensing module 11 is a sensor made of a piezoelectric material, the vibration data is audio vibration data. If the sensing module 11 is a general vibration sensor, the vibration data is mechanical motion vibration data.

[0038] In this embodiment, the rotary machine is, for example, a power saw having a saw head, and the corresponding attachment is a saw belt. In other embodiments, the rotary machine is, for example, a power transmission, a belt conveyor, a power drill, a vehicle, and the corresponding attachment is, for example, a gear, a chain / belt, a drill bit, a tire, and the like, but the invention is not limited thereto.

[0039] Please refer to Fig. 2. As shown in Fig. 2, a method for detecting abnormal characteristics of a rotating machine according to one embodiment of the present invention uses the abnormal characteristic detection system 1 for a rotating machine described above, and includes the steps of: obtaining vibration data of an attachment of the rotating machine (S1); processing the vibration data to generate a plurality of vibration timing data (S2); converting the plurality of vibration timing data into a plurality of intermediate data (S3); converting the plurality of intermediate data into a plurality of converted data (S4); and comparing feature value distributions in the plurality of converted data and defining that an abnormality has occurred in an attachment corresponding to converted data in which a change has occurred in the feature value distribution (S5).

[0040] Below, detailed operational content of the rotating machine anomaly feature detection system 1 will be sequentially explained based on the order of the rotating machine anomaly feature detection method. However, parts of the technical content of the rotating machine anomaly feature detection method described above that are the same as the technical content of the rotating machine anomaly feature detection system 1 will not be repeated.

[0041] Please refer to Fig. 3. Fig. 3 is a graph showing vibration data 2 relating to an attachment of a rotating machine obtained by the sensing module 11. As shown in Fig. 3, in this embodiment, the sensing module 11 connects the leading and trailing ends of each vibration data recorded for a fixed time length (e.g., 5 seconds) to generate continuous, complete vibration data 2. The X-axis of Fig. 3 represents time (minutes), and the Y-axis represents the vibration magnitude (G).

[0042] The data processing module 12 processes the vibration data 2 to generate multiple vibration timing data. In this embodiment, the data processing module 12 first calculates the standard deviation of the vibration data 2 and defines the portion of the vibration data 2 greater than the standard deviation as the operating section 21, and the portion of the vibration data 2 smaller than the standard deviation as the waiting section 22. Therefore, the operation start point and the operation end point can be distinguished based on the operating section 21 and the waiting section 22. For example, the leftmost waiting section 22 in FIG. 3 is considered the operation start point, and the second waiting section 22 from the left in FIG. 3 is considered the operation end point. The vibration data of the operating section 21 between these two waiting sections 22 is used as vibration timing data. The vibration data 2 shown in FIG. 3 generates seven vibration timing data (e.g., an electric saw with a saw head performs seven cutting operations using a saw belt). The X-axis unit of each vibration timing data is minutes, and the Y-axis unit is the vibration magnitude (G) converted into a peak-to-peak value. When displayed, this can be represented as vibration timing data 3 shown in FIG. 4.

[0043] In this embodiment, all vibration data 2 within the motion section 21 may be used as vibration timing data 3, and vibration data 2 within a certain period of time before and after the midpoint of the motion section 21 (for example, 5 minutes before and after) may be used as vibration timing data 3, but the present invention is not limited to this.

[0044] The conversion module 13 converts each of the plurality of vibration timing data 3 into each of the plurality of conversion data 5. The conversion module 13 of this embodiment first converts each of the plurality of vibration timing data 3 into each of the plurality of intermediate data 4 (shown in FIG. 5) using a first algorithm, and then converts the plurality of intermediate data 4 into each of the plurality of conversion data 5 (shown in FIG. 6) using a second algorithm.

[0045] In this embodiment, the first algorithm is a Fourier transform, a fast Fourier transform, a wavelet transform, or an empirical mode decomposition. Taking the fast Fourier transform as an example, the first algorithm converts the vibration timing data 3 shown in FIG. 4 from the time domain to the frequency domain. That is, it converts from time (minutes) to frequency (Hz). Next, the vibration magnitude (G) on the Y axis in the vibration timing data 3 is divided by the standard deviation of the vibration data 2. That is, it converts the vibration magnitude (G) on the Y axis in the vibration timing data 3 into a signal-to-noise ratio (dB). After that, the intermediate data 4 is represented by the frequency (Hz) on the X axis and the signal-to-noise ratio (dB) on the Y axis (shown in FIG. 5).

[0046] In this embodiment, the second algorithm can be expressed by the following equation (1).

[0047]

number

[0048] Here, d is the spacing (mm), v is the rotating machine operating speed (m / min), and f is the frequency (Hz). The converted data 5 is shown in a format where the X-axis represents spacing (mm) and the Y-axis represents the signal-to-noise ratio (dB) (see Figure 6). The purpose of converting the vibration timing data 3 into converted data 5 with spacing (mm) on the X-axis is to align the comparison criteria for each of the multiple vibration timing data 3. Using only fast Fourier transform, only intermediate data 4, where the X-axis in Figure 5 represents frequency, is obtained. However, if the operating speeds of the rotating machines are different and the X-axis frequencies do not match, comparisons may not be possible. For example, when comparing the same peak in the vibration timing data 3, the peak may be located at a position corresponding to 60 Hz when the operating speed is slow, while the peak may be located at a position corresponding to 120 Hz when the operating speed is fast. This lack of a common reference makes comparison impossible. Therefore, after the second algorithm converts the vibration timing data 3 into converted data 5 with the interval (mm) as the X-axis, even if the rotating machine is operating at different speeds, the obtained multiple converted data 5 are still compared based on the same interval.

[0049] The abnormal feature detection module 14 compares the feature value distributions of the multiple transformed data 51, 52, 53, 54, 55, 56, 57, and 58, and defines an attachment corresponding to transformed data in which a change in the feature value distribution has occurred as abnormal. Specifically, the multiple transformed data 51, 52, 53, 54, 55, 56, 57, and 58 are represented as shown in FIG. 7, and the multiple transformed data 51, 52, 53, 54, 55, 56, 57, and 58 have the same X-axis, i.e., the same intervals. The abnormal feature detection module 14 first compares the feature value distributions of the transformed data 51 and 52. The feature value distributions are specifically signal-to-noise ratio distributions. The abnormal feature detection module 14 calculates the third quartiles corresponding to the signal-to-noise ratio distributions of the transformed data 51 and 52 in the same intervals, and then compares whether a change has occurred between the third quartiles of the transformed data 51 and 52. If the third quartile of transformed data 51 and transformed data 52 should have similar or identical signal-to-noise ratios, it is determined that no change has occurred (e.g., the saw belt in a power saw with a saw head is not yet broken). After comparing transformed data 51 and transformed data 52, if it is found that no change has occurred, the next comparison is made: after comparing transformed data 52 and transformed data 53, the same steps are repeated.

[0050] The following situations may be considered to determine whether a change has occurred. For example, when comparing converted data 54 and converted data 55, as shown in FIG. 8, if the interval is 22.1 to 22.2 mm, there is a signal-to-noise ratio between the third quartile 541 of converted data 54 and the third quartile 551 of converted data 55, and converted data 55 is deemed to have change point 6 (see FIG. 7). A change occurs in the distribution of feature values of converted data 55, and the attachment corresponding to converted data 55 is defined as abnormal (for example, indicating that the saw belt in the electric saw of the saw table has a damaged saw blade). Next, based on the interval corresponding to change point 6 (for example, 22.1 to 22.2 mm), the position of the attachment is inferred to be abnormal. For example, if the attachment is a saw belt, the position of the damaged saw blade on the saw belt is inferred.

[0051] In this embodiment, an attachment is defined as abnormal when the difference between the multiple third quartiles is the greatest (i.e., the difference is the maximum value). However, in other embodiments, a range may be defined and an attachment may be defined as abnormal when the difference between the multiple third quartiles falls within that range, or a threshold may be defined and an attachment may be defined as abnormal when the difference between the multiple third quartiles exceeds the threshold, but the present invention is not limited to these.

[0052] In other embodiments, instead of using the third quartile, the first quartile or median of the transformed data may be calculated to determine whether a change has occurred in the distribution of feature values, but the present invention is not limited to this.

[0053] In the above-described embodiment, a single transformed data is compared with the next single transformed data, but the present invention is not limited to this. For example, multiple transformed data 51, 52, 53, 54, 55, 56, 57, and 58 may be compared using a matched feature value distribution in a set of N pieces (N is a natural number).

[0054] Specifically, first, multiple transformed data groups 71 and 75 are defined by grouping multiple transformed data 51, 52, 53, 54, 55, 56, 57, and 58 into groups of four. The transformed data groups 71 and 75 are adjacent to each other. The transformed data 51, 52, 53, and 54 included in the transformed data group 71 and the transformed data 55, 56, 57, and 58 included in the transformed data group 75 do not overlap with each other. The distribution of feature values at the same intervals within the transformed data groups 71 and 75 is calculated. For example, the third quartile corresponding to the transformed data 51, 52, 53, and 54 is calculated, and the third quartile corresponding to the transformed data 55, 56, 57, and 58 is calculated. Since the transformed data 55, 56, 57, and 58 have change point 6, a change occurs in the feature value distribution of the transformed data group 75 compared to the feature value distribution of the transformed data group 71, and the difference reaches a maximum value. At this time, the attachment corresponding to the transformed data 55 in the transformed data group 75 is defined as abnormal.

[0055] In this embodiment, multiple converted data groups 71, 72, 73, 74, and 75 are defined, with multiple converted data 51, 52, 53, 54, 55, 56, 57, and 58 grouped into groups of, for example, four. Converted data group 71 includes converted data 51, 52, 53, and 54. Converted data group 72 includes converted data 52, 53, 54, and 55. Converted data group 73 includes converted data 53, 54, 55, and 56. Converted data group 74 includes converted data 54, 55, 56, and 57. Converted data group 75 includes converted data 55, 56, 57, and 58. The multiple converted data 51, 52, 53, 54, 55, 56, 57, and 58 in the multiple converted data groups 71, 72, 73, 74, and 75 partially overlap with each other. Because the transformed data 55 included in the transformed data group 72 has change point 6, a change occurs in the feature value distribution of the transformed data group 72 compared to the transformed data group 71. To prevent erroneous judgment due to noise, the feature value distributions of the transformed data group 73 and the transformed data group 71 may be compared again. Because the transformed data 55 and 56 included in the transformed data group 73 each have change point 6, the amount of change in the feature value distribution of the transformed data group 73 compared to the transformed data group 71 is greater than the amount of change in the feature value distribution of the transformed data group 72 compared to the transformed data group 71. In other words, the third quartile of the transformed data group 73 is greater than the third quartile of the transformed data group 72. As can be inferred from the above, when the feature value distributions of the converted data group 74 and the converted data group 71 are compared sequentially with the feature value distributions of the converted data group 75 and the converted data group 71, it is found that the third quartile of the converted data group 75 is greater than the third quartiles of the converted data groups 71, 72, 73, and 74, respectively. In this case, the attachment corresponding to the converted data 55 in the converted data group 75 is defined as abnormal.

[0056] In the above-described embodiment, the transformed data groups 72, 73, 74, and 75 are each compared with the transformed data group 71, but the present invention is not limited to this. For example, in another embodiment, two adjacent transformed data groups 71, 72, 73, 74, and 75 are compared, and the attachment corresponding to the transformed data 55 in the transformed data group 75 with the largest third quartile is defined as abnormal.

[0057] As can be seen from the above, the system and method for detecting abnormal characteristics of a rotating machine according to the present invention uses the interval of the converted data as the X-axis, rather than the frequency, and therefore can effectively locate abnormal locations in the attachment even under the following conditions (1) and (2). (1) When the rotating machine is operating at different speeds. (2) When the spacing between multiple saw teeth on the saw belt is uneven to prevent resonance during use.

[0058] Although the preferred embodiments of the present invention have been disclosed above so that those skilled in the art can understand, they are not intended to limit the present invention in any way. Various changes and modifications can be made within the scope of the present invention. Therefore, the scope of the claims of the present invention should be broadly interpreted to include such changes and modifications. [Explanation of symbols]

[0059] 1. Rotating Machinery Anomaly Feature Detection System 2. Vibration data 3 Vibration timing data 4 Intermediate data 5. Conversion Data 6 Changes 11 Sensing Module 12 Data Processing Module 13 Conversion Module 14 Anomaly Feature Detection Module 21 Operating Section 22 Waiting area 51 Conversion Data 52 Conversion Data 53 Conversion Data 54 Conversion Data 55 Conversion Data 56 Conversion Data 57 Conversion Data 58 Conversion Data 71 Conversion Data Group 72 Conversion Data Group 73 Conversion Data Group 74 Conversion Data Group 75 Conversion Data Group 541 Third quartile 551 Third quartile

Claims

1. 1. A rotating machine anomaly feature detection system comprising: a sensing module, a data processing module, a conversion module, and an anomaly feature detection module, the sensing module obtains vibration data of an attachment of a rotating machine; the data processing module processes the vibration data to generate a plurality of vibration timing data; The conversion module converts the plurality of vibration timing data into a plurality of conversion data, respectively; The abnormal feature detection system for a rotating machine is characterized in that the abnormal feature detection module compares feature value distributions in multiple pieces of the converted data and defines an attachment corresponding to the converted data in which a change has occurred in the feature value distribution as abnormal.

2. The abnormal feature detection module compares whether a change occurs in the feature value distribution of the same interval section in any two adjacent transformed data, or first defines a plurality of transformed data groups in a manner that N pieces of transformed data are grouped together, and then compares whether a change occurs in the feature value distribution of the same interval section in any two adjacent transformed data groups; 2. The system for detecting abnormal characteristics of a rotating machine according to claim 1, wherein N is a natural number.

3. 3. The system for detecting abnormal characteristics of a rotating machine according to claim 2, wherein the change occurring in the distribution of characteristic values is such that a plurality of third quartiles corresponding to the same interval section in the plurality of transformed data or the plurality of transformed data groups are different from each other.

4. 3. The system for detecting abnormal characteristics of a rotating machine according to claim 2, wherein the plurality of transformed data among the plurality of transformed data groups partially overlap or do not overlap at all with each other.

5. 4. The system for detecting abnormal characteristics of a rotating machine according to claim 3, wherein the abnormal characteristic detection module defines an attachment as abnormal when a difference between a plurality of the third quartiles is maximum.

6. the feature value distribution is a signal-to-noise ratio distribution; 6. The system for detecting abnormal characteristics of a rotating machine according to claim 5, wherein the third quartile is calculated based on a plurality of signal-to-noise ratios in an interval section.

7. 2. The abnormality feature detection system for a rotating machine according to claim 1, wherein the data processing module first calculates a standard deviation of the vibration data, defines a portion of the vibration data that is larger than the standard deviation as an operating section, and defines a portion of the vibration data that is smaller than the standard deviation as a waiting section, and the vibration data corresponding to the operating section between two adjacent waiting sections is vibration timing data.

8. The abnormality characteristic detection system for a rotating machine according to claim 7, characterized in that the conversion module first converts the plurality of vibration timing data into a plurality of intermediate data, respectively, and then converts the plurality of intermediate data into a plurality of converted data.

9. The conversion module converts the plurality of vibration timing data into a plurality of intermediate data using a first algorithm; The X axis of the intermediate data is frequency and the Y axis is signal-to-noise ratio; 9. The system for detecting abnormal characteristics of a rotating machine according to claim 8, wherein the signal-to-noise ratio is a vibration amount of vibration timing data divided by a standard deviation.

10. 10. The system for detecting abnormal characteristics of a rotating machine according to claim 9, wherein the first algorithm is a Fourier transform, a fast Fourier transform, a wavelet transform, or an empirical mode decomposition.

11. the conversion module converts the plurality of intermediate data into a plurality of converted data using a second algorithm; 9. The system for detecting abnormal characteristics of a rotating machine according to claim 8, wherein the X-axis of the transformed data is interval and the Y-axis is signal-to-noise ratio.

12. The second algorithm is expressed by the following formula (1): [Equation 1] 12. The system for detecting abnormal characteristics of a rotating machine according to claim 11, wherein d is an interval, v is an operating speed of the rotating machine, and f is a frequency.

13. 2. The system for detecting abnormal characteristics of a rotating machine according to claim 1, wherein the vibration data is mechanical motion vibration data or audio vibration data.

14. a sensing module obtaining vibration data of an attachment of the rotating machine; a data processing module processing the vibration data to generate a plurality of vibration timing data; a conversion module converting the plurality of vibration timing data into a plurality of conversion data, respectively; an abnormality feature detection module comparing the feature value distributions in the plurality of transformed data and defining an attachment corresponding to the transformed data in which a change has occurred in the feature value distribution as abnormal.

15. The abnormal feature detection module compares whether a change occurs in the feature value distribution of the same interval section in any two adjacent transformed data, or first defines a plurality of transformed data groups in a manner that N pieces of transformed data are grouped together, and then compares whether a change occurs in the feature value distribution of the same interval section in any two adjacent transformed data groups; The method for detecting abnormal characteristics of a rotating machine according to claim 14, wherein N is a natural number.

16. The method for detecting abnormal characteristics of a rotating machine according to claim 15, characterized in that the change occurring in the feature value distribution is that a plurality of third quartiles corresponding to the same interval section in the plurality of transformed data or the plurality of transformed data groups are different from each other.

17. 16. The method for detecting abnormal characteristics of a rotating machine according to claim 15, wherein the plurality of transformed data among the plurality of transformed data groups partially overlap or do not overlap at all with each other.

18. 17. The method for detecting an abnormal characteristic of a rotating machine according to claim 16, wherein the abnormal characteristic detection module defines the attachment as abnormal when a difference between the plurality of third quartiles is maximum.

19. the feature value distribution is a signal-to-noise ratio distribution; 19. The method for detecting abnormal characteristics of a rotating machine according to claim 18, wherein the third quartile is calculated according to a plurality of signal-to-noise ratios in an interval section.

20. 15. The method for detecting abnormal characteristics of a rotating machine according to claim 14, wherein the data processing module first calculates a standard deviation of the vibration data, defines a portion of the vibration data that is larger than the standard deviation as an operating section, and defines a portion of the vibration data that is smaller than the standard deviation as a waiting section, and the vibration data corresponding to the operating section between two adjacent waiting sections is vibration timing data.

21. The method for detecting abnormal characteristics of a rotating machine according to claim 20, wherein the conversion module first converts the plurality of vibration timing data into a plurality of intermediate data, respectively, and then converts the plurality of intermediate data into a plurality of converted data.

22. The conversion module converts the plurality of vibration timing data into a plurality of intermediate data using a first algorithm; The X axis of the intermediate data is frequency and the Y axis is signal-to-noise ratio; 22. The method for detecting abnormal characteristics of a rotating machine according to claim 21, wherein the signal-to-noise ratio is a vibration amount of vibration timing data divided by a standard deviation.

23. 23. The method for detecting abnormal characteristics of a rotating machine according to claim 22, wherein the first algorithm is a Fourier transform, a fast Fourier transform, a wavelet transform, or an empirical mode decomposition.

24. the conversion module converts the plurality of intermediate data into a plurality of converted data using a second algorithm; The method for detecting abnormal characteristics of a rotating machine according to claim 21, wherein the X-axis of the transformed data is interval and the Y-axis is signal-to-noise ratio.

25. The second algorithm is expressed by the following formula (1): [Equation 2] The method for detecting abnormal characteristics of a rotating machine according to claim 24, wherein d is an interval, v is an operating speed of the rotating machine, and f is a frequency.

26. The method for detecting abnormal characteristics of a rotating machine according to claim 14, wherein the vibration data is mechanical motion vibration data or audio vibration data.

Images