WO2022260221A1 - 진동 측정 오류 판단 방법 및 이를 이용하는 진동 오류 판별 시스템 - Google Patents
진동 측정 오류 판단 방법 및 이를 이용하는 진동 오류 판별 시스템 Download PDFInfo
- Publication number
- WO2022260221A1 WO2022260221A1 PCT/KR2021/016634 KR2021016634W WO2022260221A1 WO 2022260221 A1 WO2022260221 A1 WO 2022260221A1 KR 2021016634 W KR2021016634 W KR 2021016634W WO 2022260221 A1 WO2022260221 A1 WO 2022260221A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- data
- vibration
- measurement error
- vibration data
- extracted
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
- G05B23/0221—Preprocessing measurements, e.g. data collection rate adjustment; Standardization of measurements; Time series or signal analysis, e.g. frequency analysis or wavelets; Trustworthiness of measurements; Indexes therefor; Measurements using easily measured parameters to estimate parameters difficult to measure; Virtual sensor creation; De-noising; Sensor fusion; Unconventional preprocessing inherently present in specific fault detection methods like PCA-based methods
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
- G01H1/00—Measuring characteristics of vibrations in solids by using direct conduction to the detector
- G01H1/003—Measuring characteristics of vibrations in solids by using direct conduction to the detector of rotating machines
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
- G05B23/0224—Process history based detection method, e.g. whereby history implies the availability of large amounts of data
- G05B23/024—Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
- G01H1/00—Measuring characteristics of vibrations in solids by using direct conduction to the detector
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
- G01H1/00—Measuring characteristics of vibrations in solids by using direct conduction to the detector
- G01H1/04—Measuring characteristics of vibrations in solids by using direct conduction to the detector of vibrations which are transverse to direction of propagation
- G01H1/08—Amplitude
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
- G05B23/0224—Process history based detection method, e.g. whereby history implies the availability of large amounts of data
- G05B23/0227—Qualitative history assessment, whereby the type of data acted upon, e.g. waveforms, images or patterns, is not relevant, e.g. rule based assessment; if-then decisions
- G05B23/0235—Qualitative history assessment, whereby the type of data acted upon, e.g. waveforms, images or patterns, is not relevant, e.g. rule based assessment; if-then decisions based on a comparison with predetermined threshold or range, e.g. "classical methods", carried out during normal operation; threshold adaptation or choice; when or how to compare with the threshold
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/30—Nc systems
- G05B2219/37—Measurements
- G05B2219/37351—Detect vibration, ultrasound
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/30—Nc systems
- G05B2219/37—Measurements
- G05B2219/37434—Measuring vibration of machine or workpiece or tool
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/30—Nc systems
- G05B2219/37—Measurements
- G05B2219/37435—Vibration of machine
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E30/00—Energy generation of nuclear origin
- Y02E30/30—Nuclear fission reactors
Definitions
- the present invention relates to a vibration measurement error determination method and a vibration error determination system using the same, and more particularly, to a vibration measurement error determination method for detecting whether measurement error data is included in vibration data and a vibration error determination using the same It's about the system.
- the current state of the structure and the state of the structure in the near future can be predicted by measuring the vibration generated in the structure.
- vibration of facilities is measured using a mobile vibration measuring device for facility management.
- incorrect vibration measurement data may be collected due to a measurer's inexperience in using the equipment, negligence in management of the measurement equipment, or external signal interference.
- (a) and (b) are measurement data including noise due to residual stress after an impact occurs in a sensor that detects vibration, and (c) is noise of an impact waveform due to problems such as sensor attachment. , and (d) is measurement data including noise due to sensor and/or cable failure.
- vibration data including measurement error data generated in the vibration measurement process deteriorates reliability of diagnosis even if specialized machine learning is used to analyze the state of the vibrating structure.
- An object to be solved by the present invention is to provide a vibration measurement error determination method for detecting whether measurement error data is included in vibration data and a vibration error determination system using the same.
- a vibration measurement error determination method for solving the above problems includes a vibration data acquisition step of acquiring vibration data by measuring vibration generated in a structure, and the vibration data based on a preset error data selection rule.
- a first determination step of determining whether the vibration data is due to a measurement error a second determination step of determining whether the vibration data is due to a measurement error using a machine learning algorithm, and in the first determination step and the second determination step and a final judgment step of determining whether the vibration data is caused by a measurement error based on the determined result.
- the error data selection rule may determine whether the vibration data is due to a measurement error based on a difference between an amplitude value of a low frequency region extracted from the vibration data and a threshold value.
- the low-frequency region may include a region of 3 Hz or less.
- the threshold may be 0.6 mm/s.
- the machine learning algorithm extracts a plurality of sample data from the vibration data, extracts feature information for determining a measurement error from the extracted plurality of sample data, and determines the vibration data as a measurement error based on the extracted feature information. It can be determined whether or not
- the feature information may be extracted through principal component analysis from candidate feature information extracted from a plurality of sample data.
- the machine learning algorithm may extract the feature information for each sample data, cluster the sample data based on the feature information, and determine whether the vibration data is due to a measurement error based on a clustering result. .
- a vibration error determination system for solving the above problems is a data acquisition unit that acquires vibration data by measuring vibration generated in a structure, and the vibration data is measured based on a preset error data selection rule.
- a first determination unit that determines whether the vibration data is due to an error
- a second determination unit that determines whether the vibration data is due to a measurement error by using a machine learning algorithm, and the vibration data determined in the first and second determination steps and a final determination unit that determines whether the vibration data is due to a measurement error based on the result.
- the error data selection rule may determine whether the vibration data is due to a measurement error based on a difference between an amplitude value of a low frequency region extracted from the vibration data and a threshold value.
- the low-frequency region may include a region of 3 Hz or less.
- the threshold may be 0.6 mm/s.
- the machine learning algorithm extracts a plurality of sample data from the vibration data, extracts feature information for determining a measurement error from the extracted plurality of sample data, and determines the vibration data as a measurement error based on the extracted feature information. It can be determined whether or not
- the feature information may be extracted through principal component analysis from candidate feature information extracted from a plurality of sample data.
- the machine learning algorithm may extract the feature information for each sample data, cluster the sample data based on the feature information, and determine whether the vibration data is due to a measurement error based on a clustering result. .
- vibration data which is basic data of predictive diagnosis
- FIG. 1 is a flowchart illustrating a method for determining a vibration measurement error according to an embodiment of the present invention.
- FIG. 2 is a flowchart for explaining the first judgment step of FIG. 1 .
- 3 is a graph showing an example of normal vibration data.
- FIG. 4 is a graph illustrating an example of vibration data including measurement errors.
- 5 is a graph illustrating expert judgment results for sample vibration data.
- FIG. 6 is a flowchart for explaining the second judgment step of FIG. 1 .
- FIG. 7 and 8 are graphs illustrating clustering results obtained by determining sample vibration data according to the second determination step.
- FIG. 9 is a block diagram illustrating a vibration measurement error determination system according to an embodiment of the present invention.
- 10 is a graph illustrating examples of vibration data including measurement errors.
- FIG. 1 is a flowchart illustrating a method for determining a vibration measurement error according to an embodiment of the present invention.
- a vibration measurement error determination method includes a vibration data acquisition step (S10), a first determination step (S20), a second determination step (S30), and a final determination step (S40). ).
- vibration data is obtained by measuring vibrations generated in the structure.
- Vibration data may be obtained using a mobile vibration measuring device or obtained from a database (DB) in which vibration data is stored.
- the structure may be a structure including a rotary device that operates while rotating. As the rotating machine operates, vibration occurs in the structure. By measuring the vibration, the current state of the structure and rotating machine can be diagnosed. can do.
- the vibration measurement error determination method according to an embodiment of the present invention can obtain vibration data for various facilities used in nuclear power plants and can be used for facility management of nuclear power plants.
- the first determination step ( S20 ) it may be determined whether data due to measurement errors is included in the vibration data acquired in the vibration data acquisition step ( S10 ) based on a preset error data selection rule. A more detailed description of this will be described later.
- the vibration data obtained in the vibration data acquisition step (S10) includes data due to a measurement error, based on a criterion different from the judgment criterion of the first judgment step (S20).
- the vibration data obtained in the vibration data acquisition step ( S10 ) includes data due to a measurement error by using a machine learning algorithm. A more detailed description of this will be described later.
- the vibration data acquired in the vibration data acquisition step (S10) based on the result determined in the first judgment step (S20) and the result determined in the second judgment step (S30) are determined by a measurement error. Finally, it is determined whether the data is included or not.
- only vibration data determined to have no measurement error in both the first determination step (S20) and the second determination step (S30) may be determined to have no measurement error.
- the vibration data determined to have no measurement error in any one of the first determination step (S20) and the second determination step (S30) may be determined to have no measurement error.
- FIG. 2 is a flowchart for explaining the first judgment step of FIG. 1 .
- the first determination step (S20) includes extracting time waveform data (S21), fast Fourier transform (S22), extracting vibration values in a low frequency region (S23), and comparing vibration values with a threshold value. It may include step S24 and measurement error determination step S25.
- vibration data of a time wave form is extracted from the vibration data obtained in the vibration data acquisition step (S10).
- the vibration data acquired in the vibration data acquisition step (S10) is time wave form data
- the time wave data extraction step (S21) may be omitted.
- vibration data of a time waveform is converted into a frequency domain.
- the vibration data of the time waveform may be sampled and then converted into a frequency domain.
- FFT Fast Fourier transform
- FFT Fast Fourier transform
- step S23 of extracting the vibration value of the low frequency domain a vibration value (amplitude value) of the low frequency domain is calculated and extracted from the vibration data converted to the frequency domain.
- FIG. 3 is a graph showing an example of normal vibration data
- FIG. 4 is a graph showing an example of vibration data including a measurement error.
- the vibration data including the measurement error has a large amplitude in the low frequency region of 3 Hz or less compared to the normal vibration data, and it can be confirmed that the Ski-Slope phenomenon is observed in the low frequency band.
- the low-frequency region may be set to a frequency region of 3 Hz or less.
- the vibration value of the low frequency region extracted in the step of extracting the vibration value of the low frequency region (S23) is compared with the threshold value.
- the present applicant measured vibration data at 1,811 points for 248 facilities of Hanbit 3 Power Plant Units 5 and 6, and for a total of 10,634 sample vibration data, Discrimination was performed on vibration data containing measurement errors.
- 5 is a graph illustrating expert judgment results for sample vibration data.
- FIG. 5 there are many normal vibration data in a low-frequency region (region of 0.006 in/s or less), and normal vibration data and measurement error in a region where the amplitude of the low-frequency region is 0.007 to 0.026 in/s. It can be seen that the vibration data including , and the vibration data including the measurement error appear in a region where the amplitude of the low frequency region is 0.013 in/s or more.
- the threshold When the threshold was set to 0.8 mm/s, normal vibration data was not judged as data containing measurement errors, but data containing measurement errors was often judged as normal vibration data. On the contrary, the threshold was set to 0.4 mm When set to /s, the ratio of determining data containing measurement errors as normal vibration data decreases, but the ratio of determining normal vibration data as data containing measurement errors increases.
- a ratio of determining data including measurement errors as normal vibration data and a ratio of determining normal vibration data as data including measurement errors are in a trade-off relationship according to the threshold value.
- the threshold value (T, see FIG. 5) is set to 0.6 mm/s in order to minimize the determination of normal vibration data as data containing measurement errors and also to minimize overall misjudgment.
- the threshold value of the low frequency amplitude for selecting vibration data containing measurement errors may be selected differently within the range of 0.4 to 0.8 mm/s according to the demand for accuracy in determining measurement errors, etc., and the threshold value is 0.4 It may be set to a value of mm/s or less, or may be set to a value of 0.8 mm/s or more.
- the vibration value of the low frequency region extracted in the step of extracting the vibration value of the low frequency region (S23) is compared with the set threshold value to determine whether the vibration value is less than (or less than) the threshold value. It can be judged whether it is excessive (or abnormal).
- the measurement error determination step (S25) the measurement error in the vibration data obtained in the vibration data acquisition step (S10) based on the difference between the vibration value determined in the step (S24) and the threshold value in comparing the vibration value with the threshold value It is determined whether the data due to is included.
- the vibration value when the vibration value is less than (or less than) the threshold value, it is determined as normal vibration data, and when the vibration value exceeds (or exceeds) the threshold value, it can be determined as vibration data including data due to measurement errors. .
- the first determination step (S20) based on the error data selection rule, that is, the difference between the vibration value and the threshold value in the low frequency region, it is determined whether the vibration data includes data due to measurement errors.
- the error data selection rule that is, the difference between the vibration value and the threshold value in the low frequency region.
- the vibration measurement error determination method goes through a second determination step (S30).
- FIG. 6 is a flowchart for explaining the second judgment step of FIG. 1 .
- the second determination step ( S30 ) it may be determined whether data due to a measurement error is included in the vibration data using a machine learning algorithm.
- the second determination step (S30) includes a time waveform data extraction step (S31), a sample data extraction step (S32), a candidate feature information extraction step (S33), a principal component analysis step (S34), A clustering analysis step (S35) and a measurement error determination step (S36) may be included.
- time wave data extraction step (S31) vibration data of a time wave form is extracted from the vibration data obtained in the vibration data acquisition step (S10).
- the vibration data acquired in the vibration data acquisition step (S10) is time wave form data
- the time wave data extraction step (S31) may be omitted.
- At least one sample data may be extracted from the vibration data of the time waveform.
- the number of sample data to be extracted may be variously selected depending on the embodiment, it is preferable to extract a plurality of sample data from the vibration data in order to secure the robustness of the machine learning algorithm for determining measurement errors in the vibration data.
- the number of sample data increases, considerable resources are required for calculations for extracting feature information of the machine learning algorithm, so selection of an appropriate number of sample data is required. Considering this point, the number of sample data is appropriate to be dozens, more preferably about 20.
- a criterion for extracting sample data may be set in various ways according to embodiments.
- candidate feature information is extracted from the sample data extracted in the sample data extraction step (S32).
- the candidate feature information may include first statistical information calculated from the sample data extracted in the sample data extraction step (S32) and/or time waveform extracted in the time waveform data extraction step (S31) or obtained in the vibration data acquisition step (S10). It can be extracted from the second statistical information calculated from the vibration data of .
- the first statistical information may include root mean square (RMS), mean, variance, and standard deviation of sample data. If there are a plurality of sample data extracted in the sample data extraction step ( S12 ), the first statistical information may include RMS, average, variance, and standard deviation for each sample data.
- RMS root mean square
- mean mean
- variance variance
- standard deviation standard deviation
- the second statistical information may include RMS, average, variance, and standard deviation of the vibration data of the time waveform.
- Normal vibration data shows a pattern of periodically changing + and -. However, vibration data including measurement errors show low periodicity unlike normal vibration data.
- the average value of the statistical information of the data may be an index capable of distinguishing normal vibration data from vibration data including measurement errors.
- an RMS value may be an index representing information related to amplitude of data.
- the variance and standard deviation of the data may be indicators that can distinguish normal vibration data having periodicity from vibration data including measurement errors having non-periodicity.
- sample data including a measurement error has a higher variance and/or standard deviation than other sample data. Therefore, the variance and standard deviation between sample data can be indicators by which data related to measurement errors included in vibration data can be distinguished.
- a difference in variance and/or standard deviation between the vibration data and the sample data may also be an index by which data related to a measurement error included in the vibration data can be distinguished.
- Candidate feature information may include anomalous features of a signal form that appear relatively high in vibration data including measurement errors, features with high responsiveness to dispersion, features with low responsiveness to signals generated by defects or periodic components, and the like. have.
- the RMS of sample data which is the first statistical information, is referred to as s_rms
- the mean of sample data is referred to as s_mean
- the variance of sample data is referred to as s_var
- the standard deviation of sample data is referred to as s_std .
- the RMS of vibration data which is the second statistical information, is referred to as d_rms
- the average of vibration data is referred to as d_mean
- the variance of vibration data is referred to as d_var
- the standard deviation of vibration data is referred to as d_std.
- the first candidate feature information may be a ratio (s_mean/s_rms) of the mean (s_mean) of the sample data to the RMS (s_rms) of the sample data.
- the second candidate feature information may be a ratio (s_std/d_rms) of the standard deviation (s_std) of the sample data to the RMS (d_rms) of the vibration data.
- the third candidate feature information may be a ratio (s_std/d_std) of the standard deviation (s_std) of the sample data to the standard deviation (d_std) of the vibration data.
- the fourth candidate feature information may be a ratio (s_mean/d_var) of an average (s_mean) of sample data to a variance (d_var) of vibration data.
- the fifth candidate feature information may be a ratio (s_mean/d_std) of an average (s_mean) of sample data to a standard deviation (d_std) of vibration data.
- the sixth candidate feature information may be a ratio (s_mean/d_rms) of an average (s_mean) of sample data to an RMS (d_rms) of vibration data.
- the seventh candidate feature information may be a ratio (d_mean/s_rms) of the mean (d_mean) of the vibration data to the RMS (s_rms) of the sample data.
- the eighth candidate feature information may be a ratio (s_std/s_rms) of the standard deviation (s_std) of the sample data to the RMS (s_rms) of the sample data.
- the ninth candidate feature information may be a ratio (s_var/d_var) of the variance (s_var) of sample data to the variance (d_var) of vibration data.
- the tenth candidate feature information may be a ratio (s_std/s_var) of the standard deviation (s_std) of the sample data to the variance (s_var) of the sample data.
- the eleventh candidate feature information may be a ratio (s_var/s_std) of the variance (s_var) of the sample data to the standard deviation (s_std) of the sample data.
- the twelfth candidate feature information may be a ratio (s_var/s_rms) of the variance (s_var) of the sample data to the RMS (s_rms) of the sample data.
- feature information may be extracted by reducing dimensions through principal component analysis (PCA) for 12 candidate features. This is for efficient analysis using machine learning.
- PCA principal component analysis
- the principal component analysis step (S34) may be omitted according to embodiments.
- sampling data is clustered based on feature information.
- Clustering is a type of unsupervised learning, which evaluates clustering based on the similarity between unlabeled data.
- the method according to the present embodiment aims to distinguish between normal vibration data and vibration data including measurement errors, the structural condition to be determined through clustering is simple, and a small number of measurement error data must be found within a large amount of data.
- a clustering technique is suitable as a machine learning method for distinguishing between normal vibration data and vibration data including measurement errors.
- Hierarchical clustering can be divided into a bottom-up type of aggregation (merge hierarchical clustering) and a top-down type of separation type (split hierarchical clustering).
- the aggregation type is a method of sequentially purifying similar data starting from each data.
- single linkage complete linkage, centroid linkage, average linkage, Ward's linkage, etc. exist as algorithms of agglomeration clustering.
- Ward's linkage performs cluster merging in a way that minimizes the deviation within the cluster.
- Ward's linkage is less sensitive to noise and outliers than Single linkage, and since other algorithms tend to cluster clusters of similar size, this is a model for distinguishing normal vibration data from vibration data that includes measurement errors. It is judged that it is more suitable for the Example.
- 7 and 8 are graphs illustrating clustering results obtained by determining existing vibration data according to the second determination step.
- vibration data C1 and vibration data C2 including measurement errors are clustered into different clusters.
- 1,188 pieces of data among 10,634 samples of vibration data were determined as vibration data C2 including measurement errors.
- the measurement error determination step S36 based on the analysis result derived in the clustering analysis step S35, it may be determined whether data due to measurement errors is included in the vibration data obtained in the vibration data acquisition step S10. .
- the measurement error judgment of the vibration data according to the second judgment step (S30) is an area that is difficult to accurately judge even by experts, but normal vibration data and In a region where vibration data including measurement errors coexist, it is possible to supplement the judgment error in the first judgment step S20 by presenting an accurate judgment basis.
- the vibration measurement error determination method determines whether data due to measurement errors is included in vibration data using two different determination methods, and determines according to each determination method. Since the results are collected and finally it is determined whether the vibration data includes data due to measurement errors, the accuracy of the determination can be improved.
- vibration data which is basic data of predictive diagnosis
- FIG. 9 is a block diagram illustrating a vibration measurement error determination system according to an embodiment of the present invention.
- the vibration measurement error determination system includes a data acquisition unit 10, a first determination unit 20, a second determination unit 30, and a final determination unit 40. and a machine learning analysis database 50 .
- the data acquisition unit 10 performs the aforementioned vibration data acquisition step (S10). That is, the data acquisition unit 10 may acquire vibration data from a mobile vibration measuring device or obtain vibration data from a database DB in which vibration data is stored.
- the first determination unit 20 performs the above-described first determination step (S20). That is, the first determination unit 20 may determine whether data due to a measurement error is included in the vibration data acquired in the vibration data acquisition step ( S10 ) based on a preset error data selection rule.
- the second determination unit 30 performs the above-described second determination step (S30). That is, the second determination unit 30 determines whether the vibration data acquired in the vibration data acquisition step (S10) includes data due to a measurement error as a criterion different from the criterion of the first determination step (S20). can judge For example, the second determination unit 30 may determine whether data due to a measurement error is included in the vibration data obtained in the vibration data acquisition step (S10) by using a machine learning algorithm.
- the final determination unit 40 performs the above-described final determination step (S40). That is, the final determination unit 40 collects the determination result of the first determination unit 20 and the determination result of the second determination unit 30 and determines the vibration data obtained in the vibration data acquisition step (S10) due to a measurement error. Finally, it is determined whether the data is included or not.
- the final determination unit 40 may determine that there is no measurement error only for the vibration data determined by both the first determination unit 20 and the second determination unit 30 to have no measurement error. Alternatively, the final determination unit 40 may determine that there is no measurement error only for the vibration data for which either the first determination unit 20 or the second determination unit 30 determines that there is no measurement error.
- the final determination unit 40 may provide a user with a notification that a measurement error exists in the vibration data determined to have a measurement error.
- the first determination unit 20, the second determination unit 30, and the final determination unit 40 have been described as separate configurations, but this is the first determination unit 20, the second determination unit 30 ) and the final determination unit 40 are functionally classified and described, and the first determination unit 20, the second determination unit 30, and the final determination unit 40 are physically composed of one or a plurality of calculation units. can be configured.
- the machine learning analysis database 50 may include machine learning algorithms that perform analysis on vibration data.
- the final determination unit 40 may provide the vibration data determined to have no measurement error to the machine learning analysis database 50, and the machine learning analysis database 50 may provide the vibration data received from the final determination unit 40. Data can be updated.
- the vibration measurement error determination system determines whether data due to a measurement error is included in vibration data using two different determination methods, and determines according to each determination method. Since the results are collected and finally it is determined whether the vibration data includes data due to measurement errors, the accuracy of the determination can be improved.
- vibration data which is basic data of predictive diagnosis
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Mathematical Physics (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
Description
| 기준 진폭 | 측정 오류 → 정상 판정 | 정상 → 측정 오류 판정 | 합계 | ||
| 건수 | 비율 | 건수 | 비율 | ||
| 0.8 mm/s | 103 | 44.40% | 0 | 0.00% | 103 |
| 0.7 mm/s | 82 | 35.34% | 1 | 0.01% | 83 |
| 0.6 mm/s | 48 | 20.69% | 10 | 0.10% | 58 |
| 0.5 mm/s | 24 | 10.34% | 29 | 0.29% | 53 |
| 0.4 mm/s | 7 | 3.02% | 113 | 1.11% | 120 |
| No. | Feature | No. | Feature |
| 1 | s_mean/s_rms | 7 | d_mean/s_rms |
| 2 | s_std/d_rms | 8 | s_std/s_rms |
| 3 | s_std/d_std | 9 | s_var/d_var |
| 4 | s_mean/d_var | 10 | s_std/s_var |
| 5 | s_mean/d_std | 11 | s_var/s_std |
| 6 | s_mean/d_rms | 12 | s_var/s_rms |
Claims (14)
- 구조물에서 발생하는 진동을 측정하여 진동 데이터를 획득하는 진동 데이터 획득 단계;미리 설정된 오류 데이터 선정 규칙에 기초하여 상기 진동 데이터가 측정 오류에 의한 것인지 여부를 판단하는 제1 판단 단계;기계 학습 알고리즘을 이용해 상기 진동 데이터가 측정 오류에 의한 것인지 여부를 판단하는 제2 판단 단계; 및상기 제1 판단 단계와 상기 제2 판단 단계에서 판단된 결과를 기반으로 상기 진동 데이터가 측정 오류에 의한 것인지 여부를 판단하는 최종 판단 단계;를 포함하는, 진동 측정 오류 판단 방법.
- 제1항에 있어서,상기 오류 데이터 선정 규칙은,상기 진동 데이터에서 추출한 저주파 영역의 진폭값과 임계값의 차이를 기준으로 상기 진동 데이터가 측정 오류에 의한 것인지 여부를 판단하는, 진동 측정 오류 판단 방법.
- 제2항에 있어서,상기 저주파 영역은 3Hz 이하의 영역을 포함하는, 진동 측정 오류 판단 방법.
- 제2항에 있어서,상기 임계값은 0.6 mm/s인, 진동 측정 오류 판단 방법.
- 제1항에 있어서,상기 기계 학습 알고리즘은,상기 진동 데이터로부터 복수의 샘플 데이터를 추출하고, 추출된 복수의 상기 샘플 데이터로부터 측정 오류 판별을 위한 특징 정보를 추출하고, 추출된 특징 정보를 기초로 상기 진동 데이터가 측정 오류에 의한 것인지 여부를 판단하는, 진동 측정 오류 판단 방법.
- 제5항에 있어서,상기 특징 정보는 복수의 상기 샘플 데이터로부터 추출된 후보 특징 정보들로부터 주성분 분석(principal component analysis)을 통해 추출되는, 진동 측정 오류 판단 방법.
- 제5항에 있어서,상기 기계 학습 알고리즘은,상기 특징 정보는 상기 샘플 데이터마다 추출되며, 상기 특징 정보를 기반으로 하여 샘플 데이터들을 클러스터링하고, 클러스터링 결과를 기초로 상기 진동 데이터가 측정 오류에 의한 것인지 여부를 판단하는, 진동 측정 오류 판단 방법.
- 구조물에서 발생하는 진동을 측정하여 진동 데이터를 획득하는 데이터 획득부;미리 설정된 오류 데이터 선정 규칙에 기초하여 상기 진동 데이터가 측정 오류에 의한 것인지 여부를 판단하는 제1 판단부;기계 학습 알고리즘을 이용해 상기 진동 데이터가 측정 오류에 의한 것인지 여부를 판단하는 제2 판단부; 및상기 제1 판단 단계와 상기 제2 판단 단계에서 판단된 결과를 기반으로 상기 진동 데이터가 측정 오류에 의한 것인지 여부를 판단하는 최종 판단부;를 포함하는, 진동 오류 판별 시스템.
- 제8항에 있어서,상기 오류 데이터 선정 규칙은,상기 진동 데이터에서 추출한 저주파 영역의 진폭값과 임계값의 차이를 기준으로 상기 진동 데이터가 측정 오류에 의한 것인지 여부를 판단하는, 진동 측정 오류 판별 시스템.
- 제9항에 있어서,상기 저주파 영역은 3Hz 이하의 영역을 포함하는, 진동 측정 오류 판별 시스템.
- 제9항에 있어서,상기 임계값은 0.6 mm/s인, 진동 측정 오류 판별 시스템.
- 제8항에 있어서,상기 기계 학습 알고리즘은,상기 진동 데이터로부터 복수의 샘플 데이터를 추출하고, 추출된 복수의 상기 샘플 데이터로부터 측정 오류 판별을 위한 특징 정보를 추출하고, 추출된 특징 정보를 기초로 상기 진동 데이터가 측정 오류에 의한 것인지 여부를 판단하는, 진동 측정 오류 판별 시스템.
- 제12항에 있어서,상기 특징 정보는 복수의 상기 샘플 데이터로부터 추출된 후보 특징 정보들로부터 주성분 분석(principal component analysis)을 통해 추출되는, 진동 측정 오류 판별 시스템.
- 제12항에 있어서,상기 기계 학습 알고리즘은,상기 특징 정보는 상기 샘플 데이터마다 추출되며, 상기 특징 정보를 기반으로 하여 샘플 데이터들을 클러스터링하고, 클러스터링 결과를 기초로 상기 진동 데이터가 측정 오류에 의한 것인지 여부를 판단하는, 진동 측정 오류 판별 시스템.
Priority Applications (4)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2023576190A JP7583191B2 (ja) | 2021-06-11 | 2021-11-15 | 振動測定エラー判断方法及びこれを利用する振動エラー判別システム |
| CN202180099247.1A CN117480461A (zh) | 2021-06-11 | 2021-11-15 | 振动测量误差确定方法以及使用该方法的振动误差辨别系统 |
| US18/568,732 US20240271990A1 (en) | 2021-06-11 | 2021-11-15 | Vibration measurement error determination method and vibration error discernment system using the same |
| EP21945288.5A EP4354241A4 (en) | 2021-06-11 | 2021-11-15 | Vibration measurement error determination method and vibration error discernment system using same |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR10-2021-0075933 | 2021-06-11 | ||
| KR1020210075933A KR102504421B1 (ko) | 2021-06-11 | 2021-06-11 | 진동 측정 오류 판단 방법 및 이를 이용하는 진동 오류 판별 시스템 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2022260221A1 true WO2022260221A1 (ko) | 2022-12-15 |
Family
ID=84425270
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/KR2021/016634 Ceased WO2022260221A1 (ko) | 2021-06-11 | 2021-11-15 | 진동 측정 오류 판단 방법 및 이를 이용하는 진동 오류 판별 시스템 |
Country Status (6)
| Country | Link |
|---|---|
| US (1) | US20240271990A1 (ko) |
| EP (1) | EP4354241A4 (ko) |
| JP (1) | JP7583191B2 (ko) |
| KR (1) | KR102504421B1 (ko) |
| CN (1) | CN117480461A (ko) |
| WO (1) | WO2022260221A1 (ko) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN120010007B (zh) * | 2025-04-18 | 2025-08-15 | 华北电力大学 | 覆层内部结构的勘测装置和方法 |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100199105B1 (ko) * | 1996-11-05 | 1999-06-15 | 이해규 | 학습기능을 구비한 회전체의 진동 진단장치 |
| KR100377882B1 (ko) * | 1995-08-15 | 2003-06-09 | 오므론 가부시키가이샤 | 진동감시장치및진동감시조건결정장치 |
| JP5900296B2 (ja) * | 2012-11-13 | 2016-04-06 | トヨタ自動車株式会社 | 振動解析装置、振動解析方法、及び振動解析プログラム |
| KR102073810B1 (ko) * | 2019-07-23 | 2020-03-11 | 한화시스템(주) | 머신러닝을 이용하여 함정 추진 장비의 고장을 예측하는 방법 및 시스템 |
| KR102144975B1 (ko) * | 2019-11-08 | 2020-08-14 | 주식회사 알체라 | 머신 러닝 시스템 및 머신 러닝 시스템의 동작 방법 |
| JP2020134479A (ja) * | 2019-02-26 | 2020-08-31 | 日本電産株式会社 | 異常検出方法および異常検出装置 |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS59296B2 (ja) | 1981-01-07 | 1984-01-06 | 工業技術院長 | 球形可撓管継手の製造方法 |
| US5521840A (en) * | 1994-04-07 | 1996-05-28 | Westinghouse Electric Corporation | Diagnostic system responsive to learned audio signatures |
| KR101764540B1 (ko) | 2016-06-21 | 2017-08-02 | 두산중공업 주식회사 | 풍력발전기 진동 상태감시 및 진단 시스템 |
| JP6823576B2 (ja) * | 2017-10-26 | 2021-02-03 | 株式会社日立産機システム | 異常検出システムおよび異常検出方法 |
| KR102196611B1 (ko) * | 2019-02-12 | 2020-12-30 | 한국수력원자력 주식회사 | 원전설비의 예측 진단 방법 및 시스템 |
| JP2020192015A (ja) | 2019-05-27 | 2020-12-03 | 株式会社ジェイテクト | 情報処理システム |
| CN110793622B (zh) * | 2019-11-15 | 2021-10-15 | 北京天泽智云科技有限公司 | 一种振动数据电磁干扰的异常检测方法 |
| KR102226687B1 (ko) * | 2019-11-20 | 2021-03-11 | (주)위세아이텍 | 딥러닝을 활용한 시계열 예측 기반의 정비 주기 예측 장치 및 방법 |
-
2021
- 2021-06-11 KR KR1020210075933A patent/KR102504421B1/ko active Active
- 2021-11-15 EP EP21945288.5A patent/EP4354241A4/en active Pending
- 2021-11-15 US US18/568,732 patent/US20240271990A1/en active Pending
- 2021-11-15 CN CN202180099247.1A patent/CN117480461A/zh active Pending
- 2021-11-15 WO PCT/KR2021/016634 patent/WO2022260221A1/ko not_active Ceased
- 2021-11-15 JP JP2023576190A patent/JP7583191B2/ja active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100377882B1 (ko) * | 1995-08-15 | 2003-06-09 | 오므론 가부시키가이샤 | 진동감시장치및진동감시조건결정장치 |
| KR100199105B1 (ko) * | 1996-11-05 | 1999-06-15 | 이해규 | 학습기능을 구비한 회전체의 진동 진단장치 |
| JP5900296B2 (ja) * | 2012-11-13 | 2016-04-06 | トヨタ自動車株式会社 | 振動解析装置、振動解析方法、及び振動解析プログラム |
| JP2020134479A (ja) * | 2019-02-26 | 2020-08-31 | 日本電産株式会社 | 異常検出方法および異常検出装置 |
| KR102073810B1 (ko) * | 2019-07-23 | 2020-03-11 | 한화시스템(주) | 머신러닝을 이용하여 함정 추진 장비의 고장을 예측하는 방법 및 시스템 |
| KR102144975B1 (ko) * | 2019-11-08 | 2020-08-14 | 주식회사 알체라 | 머신 러닝 시스템 및 머신 러닝 시스템의 동작 방법 |
Non-Patent Citations (1)
| Title |
|---|
| See also references of EP4354241A4 * |
Also Published As
| Publication number | Publication date |
|---|---|
| JP7583191B2 (ja) | 2024-11-13 |
| KR20220166977A (ko) | 2022-12-20 |
| JP2024521454A (ja) | 2024-05-31 |
| US20240271990A1 (en) | 2024-08-15 |
| EP4354241A4 (en) | 2025-06-11 |
| KR102504421B1 (ko) | 2023-02-28 |
| EP4354241A1 (en) | 2024-04-17 |
| CN117480461A (zh) | 2024-01-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2014193040A1 (ko) | 센싱 데이터 분석 시스템 및 방법 | |
| WO2013027970A1 (ko) | 네트워크의 이상증후를 탐지하는 장치 및 방법 | |
| WO2014193041A1 (ko) | 제조 설비의 센서 데이터를 활용한 수율 분석 시스템 및 방법 | |
| WO2023128668A1 (ko) | 부분방전 모니터링 시스템 및 부분방전 모니터링 방법 | |
| WO2017022882A1 (ko) | 의료 영상의 병리 진단 분류 장치 및 이를 이용한 병리 진단 시스템 | |
| WO2022145519A1 (ko) | 딥러닝을 이용한 심전도 시각화 방법 및 장치 | |
| CN114460439A (zh) | 一种数字集成电路测试系统 | |
| WO2023101089A1 (ko) | 머신 러닝을 이용하여 심전도 파형 분류 방법 및 장치 | |
| WO2019177239A1 (ko) | 구동부의 정밀 예지 보전방법 | |
| WO2023128669A1 (ko) | 부분방전 모니터링 시스템 및 부분방전 모니터링 방법 | |
| WO2019177234A1 (ko) | 구동부의 정밀 예지 보전방법 | |
| WO2022114653A1 (ko) | 데이터 경계 도출 시스템 및 방법 | |
| WO2018030733A1 (ko) | 계측-수율 상관성 분석 방법 및 시스템 | |
| WO2022260221A1 (ko) | 진동 측정 오류 판단 방법 및 이를 이용하는 진동 오류 판별 시스템 | |
| CN117932530A (zh) | 一种基于边缘智能安防入侵感知方法及系统 | |
| WO2018043936A1 (ko) | 공정/장비 계측 데이터의 미세 변동 감지 방법 및 시스템 | |
| CN113672658B (zh) | 基于复相关系数的电力设备在线监测错误数据识别方法 | |
| WO2019103188A1 (ko) | 뇌파 분석을 통한 외상성 뇌손상 평가 시스템 및 방법 | |
| WO2012023681A1 (en) | Method of detecting adverse reaction to drug based on retrospective cohort study using electronic medical record, method of detecting adverse reaction to intervention, and method of detecting adverse reaction to drug or intervention based on retrospective cohort study using electronic medical record and electrocardiogram data | |
| WO2014193057A1 (ko) | 불량 샘플의 결함 맵을 이용한 문제 설비 판정 방법 및 그 장치 | |
| CN119199358A (zh) | 基于电鸿物联操作系统的变电站电力设备监测系统及方法 | |
| WO2021100967A1 (ko) | 초음파 대역의 음향 신호를 이용한 설비 고장 예측 시스템 및 그 방법 | |
| WO2023249277A1 (ko) | 전기차 진단 및 예지 시스템 | |
| WO2022169020A2 (ko) | Ppg 기반 스펙트로그램 및 cnn에 기반하는 통증 평가 방법 및 장치 | |
| WO2021261901A1 (ko) | 함수 호출 패턴 분석을 통한 이상 검출 장치 및 방법 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 21945288 Country of ref document: EP Kind code of ref document: A1 |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 18568732 Country of ref document: US |
|
| ENP | Entry into the national phase |
Ref document number: 2023576190 Country of ref document: JP Kind code of ref document: A |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 202180099247.1 Country of ref document: CN |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 2021945288 Country of ref document: EP |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| ENP | Entry into the national phase |
Ref document number: 2021945288 Country of ref document: EP Effective date: 20240111 |