CN113465913A - Fault feature extraction and optimization method for nuclear power valve - Google Patents
Fault feature extraction and optimization method for nuclear power valve Download PDFInfo
- Publication number
- CN113465913A CN113465913A CN202110651899.3A CN202110651899A CN113465913A CN 113465913 A CN113465913 A CN 113465913A CN 202110651899 A CN202110651899 A CN 202110651899A CN 113465913 A CN113465913 A CN 113465913A
- Authority
- CN
- China
- Prior art keywords
- signal
- feature
- characteristic
- frequency domain
- index
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M7/00—Vibration-testing of structures; Shock-testing of structures
- G01M7/02—Vibration-testing by means of a shake table
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention relates to the field of feature extraction of nuclear power valves, in particular to a fault feature extraction and optimization method of a nuclear power valve. The method effectively solves the problems that the traditional fault characteristic types are few and the sensitive characteristic evaluation system is incomplete. The method is mainly characterized in that original time domain characteristics and frequency domain characteristic parameters are extracted from a vibration signal, an envelope signal is extracted from the original vibration signal by Hilbert transformation, characteristic parameters of the envelope signal are extracted to obtain all characteristic sets to be optimized, a characteristic evaluation index is established by constructing a target optimization function, evaluation and optimal screening of the characteristic parameters are completed, and objectivity and reliability of characteristic selection are guaranteed.
Description
Technical Field
The invention relates to the field of feature extraction of nuclear power valves, effectively solves the problem of few features, and particularly relates to a method for optimizing and evaluating sensitive features of multi-feature parameters based on original signals and Hilbert extracted envelope signals.
Background
The construction of the nuclear power station is an important construction project for driving technological innovation, improving the livelihood and promoting welfare, and the valve is key equipment and a device which mainly need to be maintained in the normal operation process of the nuclear power station.
In the process of actual nuclear power plant operation, the working environment composition is very complex, the medium, the safety level and the environmental condition are higher in requirements, and the structure and the internal structure are different from those of a common valve, so that the probability of the failure of a nuclear power valve is very high in the normal operation process of a nuclear power project, and the failure prediction of the valve is very necessary.
Before fault prediction is carried out, an important step is to extract fault characteristics as sensitive characteristics, and for fault extraction of traditional valve equipment, time domain and frequency domain characteristics are usually extracted from a monitored original signal, for example, a vibration original signal, and only experience and artificial evaluation are used as indexes for screening the sensitive characteristics, so that the demonstration and verification are lacked, and meanwhile, the workload is also slightly inconvenient. And only the time domain and frequency domain characteristics of the original signal are extracted, often the extracted parameters cannot completely reflect the fault information, and the fault prediction precision is influenced to a great extent.
Disclosure of Invention
The invention relates to the field of extraction of valve fault characteristics, provides a fault sensitive characteristic extraction and optimal evaluation method based on an original signal and an envelope signal after Hilbert transformation, and effectively solves the problems of few types of traditional fault characteristics and incomplete sensitive characteristic evaluation system.
The invention adopts the technical method that a method for extracting and optimizing the fault characteristics of a nuclear power valve comprises the following steps:
s1: at equal intervals TnRecording and acquiring a vibration signal of the valve, storing sampling points, extracting characteristics in a time domain and a frequency domain aiming at an original vibration signal, and obtaining an original time domain characteristic set D _ t and a frequency domain characteristic set D _ f;
s2: extracting an envelope curve of the vibration signal by using Hilbert transform to obtain an envelope signal, and extracting characteristics in a time domain and a frequency domain aiming at the envelope signal to obtain characteristic sets D _ T and D _ F;
s3: establishing a multi-index sensitive feature evaluation and feature optimization screen system, constructing a multi-objective optimization function W according to multi-feature evaluation indexes, determining the weight of each index of the optimization function by combining an entropy weight method, finding out an optimal parameter, and preferably selecting high-score features by scoring and sequencing each feature of the feature set in the step S2;
further, the specific method involved in step S3 includes:
s31: establishing multiple evaluation indexes of a sensitive characteristic optimization system;
Wherein X is (X)1,x2,x3,…,xN) A feature sequence representing a degradation feature, N is the number of collected sample points, and T ═ T1,t2,t3,…tN) A sequence of time instants representing monitoring of the corresponding signature sequence;
calculating monotonicity indexWherein the content of the first and second substances,representing a unit step function;
calculating a dispersion indicatorWherein x ismaxIs the maximum value of the signature sequence, xminRepresents the minimum value of the signature sequence, σ (x) represents the standard deviation of the signature sequence,is the sample standard deviation size of the monitored signature sequence;
s32: constructing a multi-optimization parameter objective function, comprehensively considering the combination of three indexes of correlation, monotonicity and discreteness, and designing a multi-objective combination function, wherein W is omega1Corr(X)+ω2Mon(X)+ω3Dis(X);ωiWeight, ω, representing an index parameteriNot less than 0 and
wherein x isijIs a standardized parameter that is, for example,the number of the indexes is shown, and j is 1,2,3 … n certain characteristic sequence number;
s34: the characteristic W high is selected as the preferred characteristic.
Further, the specific method involved in step S2 includes:
s21: the vibration signal of the valve is a continuous signal, a time signal x (t) is set, and after Hilbert transform is carried out on the time signal x (t)To obtain
S22: constructing an analytic signal, taking x (t) as a real part, and obtaining the analytic signal by Hilbert transformFor the imaginary part, the constructed complex signal z (t) is defined as,z (t) is an analytic signal, and a (t) is a modulo operation on the analytic signal, where a (t) is | z (t) |, and a (t) is an envelope signal representing x (t); wherein the content of the first and second substances,
s23: extracting time domain and frequency domain characteristics according to the obtained original signal x (t) and the envelope signal a (t), carrying out frequency domain analysis on the signals by adopting fast Fourier transform to obtain frequency domain signals, and regarding the nth point in any sampling period, the frequencyWherein, FsIs the sampling frequency of the signal and N is the number of sample points collected.
The method has the beneficial effects that the method for extracting the fault characteristics and preferably evaluating the fault characteristics of the nuclear power valve is provided. Aiming at the problem of few traditional fault feature types, the invention performs Hilbert transformation on a vibration signal to obtain an envelope curve of the signal, and performs extraction of time domain and frequency domain features on the envelope signal, thereby expanding feature types on the basis of an original signal; meanwhile, for the problem of incomplete sensitive characteristic evaluation system, the invention constructs a target function with multiple optimized parameters by combining an entropy weight method and a multiple-evaluation index method, further establishes a sensitive characteristic optimal evaluation system with multiple evaluation indexes, accurately and effectively extracts the sensitive characteristic of the valve fault, can ensure the accuracy of the sensitive characteristic, and avoids invalid information quantity and redundancy.
Drawings
FIG. 1 is a schematic diagram of the sensitive feature screening method of the present invention.
Fig. 2 is a graph of a valve portion vibration signal.
Fig. 3 is a time and frequency domain signature graph of an original signal.
Fig. 4 is a graph of the envelope signal on the valve portion signal after the Hilbert transform.
Fig. 5 is a time and frequency domain feature distribution plot of an envelope signal.
FIG. 6 is a graphical representation of the evaluation scores after the fine screening feature.
Detailed Description
The method proposed by the present invention is further described below with reference to the accompanying drawings and practical examples.
FIG. 1 is a schematic diagram of the method of the present invention.
S1: and collecting a valve vibration signal, and extracting time domain and frequency domain characteristics of the original vibration signal.
In the combined practical case, the sensor is used for collecting vibration signals, wherein the vibration signals comprise vibration signals in three directions of an X axis, a Y axis and a Z axis, the sampling frequency is 1000HZ, and the sampling time is the complete process from the start of the valve closing action to the end of the valve closing of one test; the part of the collected vibration signal is shown in fig. 2; the four graphs in the graph are graphs of vibration original signals of the 1 st test, the 12 th test, the 14 th test and the last test respectively, and are intuitive, and the vibration signals obviously change along with the increase of the test times.
Extracting two different characteristic types of a time domain and a frequency domain according to the collected vibration original signal, and performing fast Fourier transform on the time domain signal to obtain frequency domain characteristics; obtaining D _ t and D _ f by extracting features of an original signal; FIG. 3 is a time domain feature distribution and a frequency domain feature distribution of an original signal; the four graphs in the graph are respectively the distribution of two time domain features in the X-axis direction and two frequency domain features in the Y-axis direction, and show that the time domain and frequency domain features change more and more obviously with the increase of the test frequency.
S2: extracting an envelope curve of an original signal by using Hilbert transform, and extracting a time domain feature set D _ T and a frequency domain feature set D _ F of the envelope signal; signal envelope signal s (t) extracted by Hilbert transform as shown in fig. 4, upper envelope diagram of extracted partial vibration signal; the four graphs in the graph are respectively the graphs of envelope signals of the 1 st time, the 13 th time, the 15 th time and the last time, and the envelope signals show obvious variation trends in the frequency domain along with the increase of the test times.
Similarly, the time domain and frequency domain features of the envelope signal s (t) are extracted, and the time domain feature distribution and the frequency domain feature distribution of the envelope signal are partially shown in fig. 5; the four graphs in the graph are respectively the distribution of four frequency domain characteristics extracted from the envelope signal, and the frequency domain characteristics obviously change along with the increase of the test frequency.
S3: design W ═ omega1Corr(X)+ω2Mon(X)+ω3Dis (X) Multi-objective function, and determining the weight ω of the Multi-objective optimization functioniAccording to the size of W, four different feature sets of D _ T, D _ F, D _ T and D _ F obtained in S1 and S2 are sorted and evaluated.
In this example, the weight determined by the entropy weight method, ω ═ 0.183, 0.355, 0.462 };
through calculation, the sorting results of the four types of characteristic parameter sets of D _ T, D _ F, D _ T and D _ F are shown in tables 1,2,3 and 4.
TABLE 1 Dt characteristic parameter ordering
TABLE 2D _ f characteristic parameter ordering
TABLE 3D _ T characteristic parameter ordering
TABLE 4D _ F characteristic parameter ordering
According to the scores and the sorting results of the four tables, for the object of the example, an envelope curve of the vibration signal is extracted by Hilbert transform, the original value of the vibration signal and the envelope signal of the vibration signal are used as targets for feature extraction, indexes are screened according to feature evaluation indexes, a weight is determined by using an entropy weight method during feature evaluation, finally, feature parameters of the first three are selected as corresponding sensitive features, the size of the sensitive features and the size of a W value are obtained, and Top1, Top2 and Top3 in the graph represent the features of the first three in sorting and corresponding score sorting of four different feature sets of D _ T, D _ F, D _ T and D _ F.
Claims (2)
1. A fault feature extraction and optimization method for a nuclear power valve comprises the following steps:
s1: at equal intervals TnRecording and acquiring a vibration signal of the valve, storing sampling points, extracting characteristics in a time domain and a frequency domain aiming at an original vibration signal, and obtaining an original time domain characteristic set D _ t and a frequency domain characteristic set D _ f;
s2: extracting an envelope curve of the vibration signal by using Hilbert transform to obtain an envelope signal, and extracting characteristics in a time domain and a frequency domain aiming at the envelope signal to obtain characteristic sets D _ T and D _ F;
s3: establishing a multi-index sensitive feature evaluation and feature optimization screen system, constructing a multi-objective optimization function W according to multi-feature evaluation indexes, determining the weight of each index of the optimization function by combining an entropy weight method, finding out an optimal parameter, and preferably selecting high-score features by scoring and sequencing each feature of the feature set in the step S2;
s31: establishing multiple evaluation indexes of a sensitive characteristic optimization system;
Wherein X is (X)1,x2,x3,...,xN) A feature sequence representing a degradation feature, N is the number of collected sample points, and T ═ T1,t2,t3,...tN) A sequence of time instants representing monitoring of the corresponding signature sequence;
calculating monotonicity indexWherein the content of the first and second substances,representing a unit step function;
calculating a dispersion indicatorWherein x ismaxIs the maximum value of the signature sequence, xminRepresents the minimum value of the signature sequence, σ (x) represents the standard deviation of the signature sequence,is the sample standard deviation size of the monitored signature sequence;
s32: constructing a multi-optimization parameter objective function, comprehensively considering the combination of three indexes of correlation, monotonicity and discreteness, and designing a multi-objective combination function, wherein W is omega1Corr(X)+ω2Mon(X)+ω3Dis(X);ωiWeight, ω, representing an index parameteriNot less than 0 and
wherein x isijIs a standardized parameter that is, for example,n represents the number of index numbers, j is 1,2,3.. n certain feature sequence number;
s34: the characteristic W high is selected as the preferred characteristic.
2. The method for extracting and optimizing the fault characteristics of the nuclear power valve as claimed in claim 1, wherein the specific method of the step S2 is as follows:
s21: the vibration signal of the valve is a continuous signal, a time signal x (t) is set, and the time signal x (t) is obtained after Hilbert transform is carried out on the vibration signal
S22: constructing an analytic signal, taking x (t) as a real part, and obtaining the analytic signal by Hilbert transformFor the imaginary part, the constructed complex signal z (t) is defined as,z (t) is an analytic signal, and a (t) is a modulo operation on the analytic signal, where a (t) is | z (t) |, and a (t) is an envelope signal representing x (t); wherein the content of the first and second substances,
s23: extracting time domain and frequency domain characteristics according to the obtained original signal x (t) and the envelope signal a (t), carrying out frequency domain analysis on the signals by adopting fast Fourier transform to obtain frequency domain signals, and regarding the nth point in any sampling period, the frequencyWherein, FsIs the sampling frequency of the signal and N is the number of sample points collected.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110651899.3A CN113465913A (en) | 2021-06-11 | 2021-06-11 | Fault feature extraction and optimization method for nuclear power valve |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110651899.3A CN113465913A (en) | 2021-06-11 | 2021-06-11 | Fault feature extraction and optimization method for nuclear power valve |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113465913A true CN113465913A (en) | 2021-10-01 |
Family
ID=77869702
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110651899.3A Pending CN113465913A (en) | 2021-06-11 | 2021-06-11 | Fault feature extraction and optimization method for nuclear power valve |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113465913A (en) |
Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140118003A1 (en) * | 2012-06-13 | 2014-05-01 | Jiangsu Electric Power Company | Fault diagnosis and preliminary location system and method for transformer core looseness |
CN104655380A (en) * | 2015-03-16 | 2015-05-27 | 北京六合智汇技术有限责任公司 | Method for extracting fault features of rotating mechanical equipment |
CN105699080A (en) * | 2015-12-18 | 2016-06-22 | 华北电力大学(保定) | Wind turbine generator set bearing fault feature extraction method based on vibration data |
CN109932179A (en) * | 2019-04-09 | 2019-06-25 | 东南大学 | A kind of rolling bearing fault testing method based on the reconstruct of DS Adaptive spectra |
CN110160791A (en) * | 2019-06-27 | 2019-08-23 | 郑州轻工业学院 | Based on small echo-spectrum kurtosis induction machine bearing failure diagnosis system and diagnostic method |
US20190353703A1 (en) * | 2018-05-16 | 2019-11-21 | Wuhan University | Analog circuit fault feature extraction method based on parameter random distribution neighbor embedding winner-take-all method |
CN111598438A (en) * | 2020-05-14 | 2020-08-28 | 哈尔滨工业大学(威海) | Civil aviation engine gas circuit abnormity detection method based on segmented fitting analysis and evaluation |
CN111680661A (en) * | 2020-06-19 | 2020-09-18 | 哈尔滨工业大学 | Mechanical rotating part performance degradation tracking method based on multi-feature fusion |
CN112229633A (en) * | 2020-09-03 | 2021-01-15 | 中国长江三峡集团有限公司福建分公司 | Fan bearing fault diagnosis method based on multivariate feature fusion |
CN112287776A (en) * | 2020-10-15 | 2021-01-29 | 上海明略人工智能(集团)有限公司 | Bearing performance index analysis method and system, readable storage medium and electronic equipment |
CN112347637A (en) * | 2020-11-04 | 2021-02-09 | 中国人民解放军陆军装甲兵学院 | Engine state evaluation method based on improved K _ means clustering algorithm |
CN112507769A (en) * | 2020-08-10 | 2021-03-16 | 北京化工大学 | Bearing fault diagnosis method based on simulated sensor resonance enhancement features |
-
2021
- 2021-06-11 CN CN202110651899.3A patent/CN113465913A/en active Pending
Patent Citations (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140118003A1 (en) * | 2012-06-13 | 2014-05-01 | Jiangsu Electric Power Company | Fault diagnosis and preliminary location system and method for transformer core looseness |
CN104655380A (en) * | 2015-03-16 | 2015-05-27 | 北京六合智汇技术有限责任公司 | Method for extracting fault features of rotating mechanical equipment |
CN105699080A (en) * | 2015-12-18 | 2016-06-22 | 华北电力大学(保定) | Wind turbine generator set bearing fault feature extraction method based on vibration data |
US20190353703A1 (en) * | 2018-05-16 | 2019-11-21 | Wuhan University | Analog circuit fault feature extraction method based on parameter random distribution neighbor embedding winner-take-all method |
CN109932179A (en) * | 2019-04-09 | 2019-06-25 | 东南大学 | A kind of rolling bearing fault testing method based on the reconstruct of DS Adaptive spectra |
CN110160791A (en) * | 2019-06-27 | 2019-08-23 | 郑州轻工业学院 | Based on small echo-spectrum kurtosis induction machine bearing failure diagnosis system and diagnostic method |
CN111598438A (en) * | 2020-05-14 | 2020-08-28 | 哈尔滨工业大学(威海) | Civil aviation engine gas circuit abnormity detection method based on segmented fitting analysis and evaluation |
CN111680661A (en) * | 2020-06-19 | 2020-09-18 | 哈尔滨工业大学 | Mechanical rotating part performance degradation tracking method based on multi-feature fusion |
CN112507769A (en) * | 2020-08-10 | 2021-03-16 | 北京化工大学 | Bearing fault diagnosis method based on simulated sensor resonance enhancement features |
CN112229633A (en) * | 2020-09-03 | 2021-01-15 | 中国长江三峡集团有限公司福建分公司 | Fan bearing fault diagnosis method based on multivariate feature fusion |
CN112287776A (en) * | 2020-10-15 | 2021-01-29 | 上海明略人工智能(集团)有限公司 | Bearing performance index analysis method and system, readable storage medium and electronic equipment |
CN112347637A (en) * | 2020-11-04 | 2021-02-09 | 中国人民解放军陆军装甲兵学院 | Engine state evaluation method based on improved K _ means clustering algorithm |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111853555B (en) | Water supply pipe network blind leakage identification method based on dynamic process | |
CN111664365B (en) | Oil and gas pipeline leakage detection method based on improved VMD and 1DCNN | |
CN110968069B (en) | Fault prediction method of wind generating set, corresponding device and electronic equipment | |
CN103245907B (en) | A kind of analog-circuit fault diagnosis method | |
CN104454785B (en) | A kind of engineering machinery hydraulic cylinder fault diagnosis system and the fault sample signal acquisition method being suitable for therewith | |
CN107153759B (en) | A kind of aviation turbofan engine method for predicting residual useful life of multisource data fusion | |
CN109061426B (en) | Transformer partial discharge fault diagnosis method and online monitoring device | |
CN110096810A (en) | A kind of industrial process flexible measurement method based on layer-by-layer data extending deep learning | |
CN110020637A (en) | A kind of analog circuit intermittent fault diagnostic method based on more granularities cascade forest | |
CN111079861A (en) | Power distribution network voltage abnormity diagnosis method based on image rapid processing technology | |
Zhang et al. | A generic data analytics system for manufacturing production | |
CN112906764A (en) | Communication safety equipment intelligent diagnosis method and system based on improved BP neural network | |
CN112418682A (en) | Security assessment method fusing multi-source information | |
CN117078077A (en) | Ecological vulnerability evaluation method for expressway road domain | |
CN113465734B (en) | Real-time estimation method for structural vibration | |
CN107463872A (en) | A kind of rotating machinery Crack Fault Diagnosis in Shaft method | |
CN110782007A (en) | Flight data missing supplement method based on neural network | |
Li et al. | A coupling model based on grey relational analysis and stepwise discriminant analysis for wood defect area identification by stress wave | |
CN113465913A (en) | Fault feature extraction and optimization method for nuclear power valve | |
CN108959721B (en) | Method for calculating reliability of long-span bridge member under action of windmill load | |
JP2013168020A (en) | State prediction method for process | |
CN110543675A (en) | Power transmission line fault identification method | |
CN112685562B (en) | XGboost model-based multidimensional index integration technical evaluation method | |
CN113281229A (en) | Multi-model self-adaptive atmosphere PM based on small samples2.5Concentration prediction method | |
Dong et al. | Hydraulic system fault diagnosis based on EMD and improved PSO-Elman ANN |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20211001 |