CN109633270A - The identification of fault energy zone boundary and feature extracting method based on instantaneous spectrum entropy and noise energy difference - Google Patents
The identification of fault energy zone boundary and feature extracting method based on instantaneous spectrum entropy and noise energy difference Download PDFInfo
- Publication number
- CN109633270A CN109633270A CN201910001146.0A CN201910001146A CN109633270A CN 109633270 A CN109633270 A CN 109633270A CN 201910001146 A CN201910001146 A CN 201910001146A CN 109633270 A CN109633270 A CN 109633270A
- Authority
- CN
- China
- Prior art keywords
- time
- frequency
- fault
- boundary
- energy
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R23/00—Arrangements for measuring frequencies; Arrangements for analysing frequency spectra
- G01R23/16—Spectrum analysis; Fourier analysis
Abstract
The present invention discloses a kind of fault energy zone boundary identification based on instantaneous spectrum entropy and noise energy difference and feature extracting method, the present invention can be under water in the time-frequency power density spectrum of Robotic Dynamic signal, effectively determine the time domain boundary and frequency domain boundary in area of energy concentration domain, and using the gross energy in the boundary as fault signature, extracted time-frequency energy failure feature and fault degree mapping relations are unique；Moreover, constructing fault sample using the fault signature, when for the classification of propeller fault degree, the fault degree nicety of grading of test sample is 100%.
Description
Technical field
The invention belongs to underwater robot propeller fault diagnosis technology fields, and in particular to one kind is based on instantaneous spectrum entropy
The identification of fault energy zone boundary and feature extracting method with noise energy difference.
Background technique
Propeller is underwater robot critical component and power unit, and load is most heavy, is the major failure of underwater robot
One of source, so propeller monitoring running state is very important.Fault diagnosis is to monitor commonly using for propeller operating status
Technological means.The energy of underwater robot Dynamic Signal singular behavior is a kind of important fault signature.Known signal amplitude is flat
Side and method are from the energy feature of the angle extraction signal singularity behavior of time domain, and known wavelet energy method is from frequency domain angle extraction
The energy feature of signal singularity behavior, however, rarely having energy feature of the method from time-frequency domain angle extraction signal singularity behavior.
Summary of the invention
Goal of the invention: it is an object of the invention to solve the deficiencies in the prior art, provides a kind of based on instantaneous frequency
Compose the identification of fault energy zone boundary and the feature extracting method of entropy and noise energy difference.
Technical solution: a kind of fault energy region time domain Boundary Recognition method based on instantaneous spectrum entropy of the invention, packet
Include following steps:
The first step, acquisition length are L1Underwater robot speed signal data；
Second step, calculating speed signal time-frequency power density spectrum；
Smoothing Pseudo Eugene Wigner-Willie spectrum of speed signal is sought using conventional smooth puppet Wigner-Ville Distribution Algorithm
SPWVD (n, m), n are the serial number on time shaft, n=1,2,3 ..., L1, m is the serial number on frequency axis, m=1,2,3 ..., N3,
N3For frequency axis demarcation interval number, such as setting N1=512；Pass through the conventional signed magnitude arithmetic(al) to SPWVD (n, m) and then obtains
Speed signal time-frequency power spectral density composes SPWVDA (n, m)；
Third step calculates the distribution of instantaneous spectrum entropy；
In time-frequency power density spectrum SPWVDA (n, m), at time shaft serial number n=1, when constructing this by formula (1)
Between probability density function p (1, m) at point, and then calculate the Shannon entropy at the time point, calculated result H by formula (2)
(1), the instantaneous spectrum entropy as at time shaft serial number n=1；
This step content is repeated, calculating time shaft serial number n is respectively 2,3 ..., L1When instantaneous spectrum entropy H (n), obtain
The distribution of instantaneous spectrum entropy；
In formula, SPWVDA (1, m) is time shaft serial number n=1, m-th of frequency in time-frequency power density spectrum SPWVDA (n, m)
The energy taken, p (1, m) are probability density function, N3For frequency axis demarcation interval number, H (1) is at time shaft serial number n=1
Instantaneous spectrum entropy；
4th step determines fault energy region time domain boundary, i.e., determines distortion section in the distribution of instantaneous spectrum entropy；
Determine minimum value position in the distribution of instantaneous spectrum entropy, then thus position extends to adjacent poles to both sides simultaneously
Big value point continues to extend to both sides if this maximum is less than the minimum in other regions, until the maximum encountered is big
Until the minimum in other regions, then the region between two final maximum points is section of distorting；Then with left side pole
Big value point position is the time domain lower boundary N in fault energy regionD, using right side maximum point position as fault energy area
The time domain coboundary N in domainU。
The fault energy region frequency domain Boundary Recognition method based on noise energy difference that invention additionally discloses a kind of, including it is following
Step:
The first step, acquisition length are L1Underwater robot speed signal data；
Second step is carried out the wavelet function feedback of multiple scales to speed signal using conventional wavelet changing method, obtained
To multiple small echo approximation component ujA(n), j is wavelet decomposition scales, j=0,1,2 ..., 8, wherein j=0 indicates speed signal not
Wavelet decomposition is carried out, n is the serial number on time shaft, n=1,2,3 ..., L1；
Third step calculates time-frequency power density spectrum；It is calculated using conventional smooth puppet Wigner-Ville Distribution Algorithm and absolute value
Method calculates j-th of multi-scale wavelet approximation component ujA(n) time-frequency power density spectrum SPWVDAj(n,m)；
4th step calculates the distribution of noise energy difference；
The time domain lower boundary N in fault energy region is determined based on instantaneous spectrum entropyDWith time domain coboundary NU；
In time-frequency power density spectrum SPWVDAjIn (n, m), by time domain boundary NDWith NUBetween time-frequency power density spectrum
SPWVDAj(n, m) sums, acquired results ESjAs fault energy, by time domain boundary NDWith NUExcept time-frequency power density
Compose SPWVDAj(n, m) sums, acquired results ENjAs noise energy, by E△SNj=ESj-ENjAs noise energy difference；Meter
J=0,1,2 is calculated ..., 8 noise energy difference E△SNj, obtain the distribution of noise energy difference；
5th step determines fault signature region frequency domain boundary；In noise energy difference E△SNjIn distribution, noise energy difference is determined
Maximum value determines the corresponding wavelet decomposition scales of the maximum value, determines the corresponding small echo approximation component of the decomposition scale, then this is small
Frequency band [M corresponding to wave approximation componentD MU] it is fault energy region frequency domain boundary, wherein MDFor frequency domain lower boundary, MUFor
Frequency domain coboundary.
The present invention discloses a kind of identification of fault energy zone boundary and feature based on instantaneous spectrum entropy and noise energy difference
Extracting method, comprising the following steps:
The first step, acquisition length is L respectively1Underwater robot speed signal and propeller control voltage change ratio signal
Data；
Second step determines fault energy region time domain lower boundary N based on instantaneous spectrum entropyDAnd time domain coboundary NU；
Third step determines fault energy region frequency domain lower boundary M based on noise energy differenceDAnd frequency domain coboundary MU；
4th step extracts propeller time-frequency energy failure feature；
Speed signal, control voltage are calculated separately using conventional smooth puppet Wigner-Ville Distribution Algorithm, absolute value algorithm
The time-frequency power density spectrum SPWVDA of change rate signalU(n, m) and SPWVDAC(n, m), in time domain lower boundary ND, time domain coboundary
NU, frequency domain lower boundary MDWith frequency domain coboundary MUIn the rectangular area surrounded, sum to time-frequency power density spectrum, gained
As a result FU、FCRespectively as speed signal time-frequency energy failure feature and control signal time-frequency energy failure feature.
The utility model has the advantages that the invention patent can be under water in the time-frequency power density spectrum of Robotic Dynamic signal, effectively really
The time domain boundary and frequency domain boundary of concentrated area are surely measured, and using the gross energy in the boundary as fault signature.It is extracted
Time-frequency energy failure feature and fault degree mapping relations are unique.Moreover, the fault sample constructed using the fault signature, it can be with
Realize the Accurate classification of propeller fault degree.
Detailed description of the invention
Fig. 1 is that the present invention is based on the fault energy region time domain Boundary Recognition method flow diagrams of instantaneous spectrum entropy；
Fig. 2 is that the present invention is based on the fault energy region frequency domain Boundary Recognition method flow diagrams of noise energy difference；
Fig. 3 is whole flow process figure of the invention；
Fig. 4 is underwater robot speed signal and propeller control signal data time domain waveform；
Fig. 5 is underwater robot speed signal instantaneous spectrum entropy distribution map；
Fig. 6 is speed signal fault energy region time domain boundary demarcation schematic diagram；
Fig. 7 is speed signal noise energy difference Butut；
Fig. 8 is speed signal fault energy region time-frequency boundary demarcation schematic diagram；
Fig. 9 is the speed signal time-frequency energy feature distribution map for not setting boundary；
Figure 10 is the speed signal time-frequency energy feature distribution map that time-frequency boundary is arranged；
Figure 11 propeller fault sample distribution map；
Figure 12 propeller different faults Tachistoscope sample and each hyper-sphere model relative distance distribution map.
Specific embodiment
Technical solution of the present invention is described in detail below, but protection scope of the present invention is not limited to the implementation
Example.
As shown in Figure 1, a kind of fault energy region time domain Boundary Recognition method based on instantaneous spectrum entropy of the invention, packet
Include following steps:
The first step, acquisition length are L1=400 underwater robot speed signal data；
Second step, calculating speed signal time-frequency power density spectrum；
Smoothing Pseudo Eugene Wigner-Willie spectrum of speed signal is sought using conventional smooth puppet Wigner-Ville Distribution Algorithm
SPWVD (n, m), n are the serial number on time shaft, n=1,2,3 ..., L1, L1=400, m be frequency axis on serial number, m=1,2,
3 ..., N1, N1For frequency axis demarcation interval number, N in the present embodiment1=512；It is transported by the conventional absolute value to SPWVD (n, m)
It adds and obtains speed signal time-frequency power spectral density spectrum SPWVDA (n, m)；
Third step calculates the distribution of instantaneous spectrum entropy；
In time-frequency power density spectrum SPWVDA (n, m), at time shaft serial number n=1, when constructing this by formula (1)
Between probability density function p (1, m) at point, and then calculate the Shannon entropy at the time point, calculated result H by formula (2)
(1), the instantaneous spectrum entropy as at time shaft serial number n=1；
This step content is repeated, calculating time shaft serial number n is respectively 2,3 ..., L1When instantaneous spectrum entropy H (n), obtain
The distribution of instantaneous spectrum entropy；
In formula, SPWVDA (1, m) is time shaft serial number n=1, m-th of frequency in time-frequency power density spectrum SPWVDA (n, m)
The energy taken, p (1, m) are probability density function, N3For frequency axis demarcation interval number, N3=512, H (1) are in time shaft sequence
Instantaneous spectrum entropy at number n=1；
4th step determines fault energy region time domain boundary, i.e., determines distortion section in the distribution of instantaneous spectrum entropy；
Determine minimum value position in the distribution of instantaneous spectrum entropy, then thus position extends to adjacent poles to both sides simultaneously
Big value point continues to extend to both sides if this maximum is less than the minimum in other regions, until the maximum encountered is big
Until the minimum in other regions, then the region between two final maximum points is section of distorting；Then with left side pole
Big value point position is the time domain lower boundary N in fault energy regionD, using right side maximum point position as fault energy area
The time domain coboundary N in domainU。
As shown in Fig. 2, a kind of fault energy region frequency domain Boundary Recognition method based on noise energy difference of the invention, packet
Include following steps:
The first step, acquisition length are L1=400 underwater robot speed signal data；
Second step is carried out the wavelet function feedback of multiple scales to speed signal using conventional wavelet transform method, obtained
To multiple small echo approximation component ujA(n), j is wavelet decomposition scales, j=0,1,2 ..., 8, wherein j=0 indicates speed signal not
Carry out wavelet decomposition；
Third step calculates time-frequency power density spectrum；Using smooth and pseudo Wigner-Ville distribution algorithm and absolute value algorithm meter
Calculate j-th of multi-scale wavelet approximation component ujA(n) time-frequency power density spectrum SPWVDAj(n,m)；
4th step calculates the distribution of noise energy difference；
The time domain lower boundary N in fault energy region is determined based on instantaneous spectrum entropyDWith time domain coboundary NU；
In time-frequency power density spectrum SPWVDAjIn (n, m), by time domain boundary NDWith NUBetween time-frequency power density spectrum
SPWVDAj(n, m) sums, acquired results ESjAs fault energy, by time domain boundary NDWith NUExcept time-frequency power density
Compose SPWVDAj(n, m) sums, acquired results ENjAs noise energy, by E△SNj=ESj-ENjAs noise energy difference；Meter
J=0,1,2 is calculated ..., 8 noise energy difference E△SNj, obtain the distribution of noise energy difference；
5th step determines fault signature region frequency domain boundary；In noise energy difference E△SNjIn distribution, noise energy difference is determined
Maximum value determines the corresponding wavelet decomposition scales of the maximum value, determines the corresponding small echo approximation component of the decomposition scale, then this is small
Frequency band [M corresponding to wave approximation componentD MU] it is fault signature region frequency domain boundary, wherein MDFor frequency domain lower boundary, MUFor
Frequency domain coboundary.
As shown, a kind of fault energy zone boundary based on instantaneous spectrum entropy and noise energy difference of the invention identifies
With feature extracting method, comprising the following steps:
The first step, obtaining length respectively is L1=400 underwater robot speed signal and propeller control voltage change
Rate signal data；
Second step determines fault energy region time domain lower boundary N based on instantaneous spectrum entropyDAnd time domain coboundary NU；
Third step determines fault energy region frequency domain lower boundary M based on noise energy differenceDAnd frequency domain coboundary MU；
4th step extracts propeller time-frequency energy failure feature；
Speed signal, control voltage change are calculated separately using smooth and pseudo Wigner-Ville distribution algorithm, absolute value algorithm
The time-frequency power density spectrum SPWVDA of rate signalU(n, m) and SPWVDAC(n, m), in time domain lower boundary ND, time domain coboundary NU、
Frequency domain lower boundary MDWith frequency domain coboundary MUIn the rectangular area surrounded, sum to time-frequency power density spectrum, acquired results
FU、FCRespectively as speed signal time-frequency energy failure feature and control signal time-frequency energy failure feature.
Embodiment:
As shown in figure 4, underwater robot longitudinal direction target velocity 0.3m/s, start operation since 0 moment, longitudinal velocity by
Cumulative big, after 100 clap, underwater robot starts to run with the steady state speed of 0.3m/s, opens from the 250th time beat
Begin, thrust loss failure occurs for propeller, and fault degree is respectively 0%, 10%, 20%, 30%, 40%, until off-test.
Propeller fault degree is that 10%, 20%, 30%, 40% corresponding control signal is gradually increased since clapping the 250th, is then existed
One stationary value nearby fluctuates.Fault degree is that 10%, 20%, 30%, 40% corresponding speed signal is clapped the 250th to the 350th
The singular behavior risen after falling before is formed in clapping, as shown in circle oval in Fig. 4.
The length of one group of experimental data of underwater robot is 600 in the present embodiment, wherein preceding 100 clap, and underwater robot is also
Do not reach stable state, cause this segment data that cannot use, so 500 data clapped can be used for experimental analysis after only.To obtain 50
A fault sample, time window slides 50 times, by time window length L1It is set to 400.When initial, using length L1=400 time window
The experimental data that interception 101~500 is clapped is used for experimental analysis, the time window is then slid to the right a time beat, works as cunning
When moving 50 time beats, the experimental data of time window interception 151~550 is used for experimental analysis.
Using length L1=400 time window function intercepts the speed signal that the 101st bat is clapped to the 500th in Fig. 4, counts respectively
The instantaneous spectrum entropy of propeller different faults degree corresponding speed signal time-frequency power density spectrum is calculated, as a result as shown in Figure 5.Fig. 5
The corresponding speed signal instantaneous spectrum entropy curve minimum point 3.805 of middle propeller fault degree 40% is in the 279th time beat
Place, then thus position extends to neighbouring maximum point to both sides simultaneously, and left side extends to the maximum point 5.371 of the 192nd bat,
Right side extends to the maximum point 5.352 of 387 bats, and as shown in corresponding vertical line, then the distortion section of instantaneous spectrum entropy curve is
[192 387] time beat, i.e., the time domain boundary in fault energy region in the corresponding time-frequency power density spectrum of fault degree 40%
For [192 387] time beat.Similarly, the corresponding time-frequency power density spectrum of propeller fault degree 0%, 10%, 20%, 30%
Segmentum intercalaris when the time domain boundary in middle fault energy region is respectively [100 239], [226 332], [174 314], [211 337]
It claps.
As shown in fig. 6, time domain boundary according to figure 5, divided in speed signal time-frequency power density spectrum shown in Fig. 6
Fault energy region time domain boundary, from the result of time domain boundary demarcation as can be seen that the time domain boundary divided according to Fig. 5 with
The practical time domain boundary in area of energy concentration domain is consistent, illustrates the fault energy region time domain of the invention based on instantaneous spectrum entropy
Boundary Recognition method is effective.
As shown in fig. 7, the maximum value of the corresponding noise energy difference of propeller fault degree 40% is 0.0199, it is corresponding
Wavelet decomposition scales are 5, similarly, the corresponding noise energy difference maximum value difference of propeller fault degree 0%, 10%, 20%, 30%
It is 0.0001,0.0006,0.0020,0.0111, corresponding wavelet decomposition scales are respectively 6,5,5,5.
As shown in figure 8, the corresponding small echo approximation component of the wavelet decomposition scales according to determined by Fig. 7, according to small echo approximation
Frequency band corresponding to component divides fault energy region frequency domain boundary, as can be seen that basis from the result of frequency domain boundary demarcation
The frequency domain boundary that Fig. 7 is divided is consistent with the actual frequency domain boundary in area of energy concentration domain, illustrates of the invention based on noise energy
The fault energy region frequency domain Boundary Recognition method of amount difference is effective.
As shown in Figure 9 and Figure 10, close in the corresponding speed signal time-frequency power of propeller a certain kind fault degree in Fig. 9
It in degree spectrum, is not provided with any boundary, directly all values in time-frequency power density spectrum being summed, acquired results are as propulsion
The mapping relations of device time-frequency energy failure feature, this kind of method gained fault signature and fault degree are not unique, i.e. a failure
Characteristic value may correspond to various faults degree, as corresponding four kinds of fault degrees of fault signature 0.060 4.9%, 13.7%,
26.6%, 32.7%.In Figure 10, in the corresponding speed signal time-frequency power density spectrum of propeller a certain kind fault degree, root
Fault energy region time domain boundary and frequency domain boundary are identified according to this patent method, then carry out the time-frequency power density in boundary
Summation, acquired results are as propeller time-frequency energy failure feature, the mapping of this kind of method gained fault signature and fault degree
Relationship is unique, i.e., a fault eigenvalue only corresponds to a kind of fault degree, such as a kind of 0.010 fault degree of correspondence of fault signature
27.2%.
Length is used to intercept the 101st bat is clapped to the 500th in Fig. 4 speed signal data and control for 400 time window function
Signal data processed corresponds to each fault degree of propeller, therefrom extracts time-frequency using this patent characteristic boundary recognition methods
Energy failure feature constructs a fault sample, and then time window moves right a time beat, and it is special to extract failure again
Sign constructs a fault sample, repeats this process, by mobile 50 beats of time window, constructs 50 fault samples, gained failure
Sample distribution is as shown in figure 11.In Figure 11, the fault sample of same fault degree flocks together, and different fault degrees are corresponding
Fault sample between distance it is larger, this be beneficial to propeller fault degree classification.
In Figure 11, correspond to each fault degree, randomly selects 25 samples as training sample, remaining 25
Sample is as test sample.Conventional support vector domain description method is based on using training sample and establishes the classification of propeller fault degree
Model brings test sample into disaggregated model, the relative distance of test sample and each hyper-sphere model is calculated, as a result such as Figure 12 institute
Show, wherein hypersphere 1, hypersphere 2, hypersphere 3, hypersphere 4, the corresponding fault degree of hypersphere 5 be respectively 0%, 10%, 20%, 30%,
40%.
In Figure 12, with fault degree corresponding to the smallest hyper-sphere model of test sample relative distance as the test specimens
This fault degree classification results, for example, in Figure 12 (a), the corresponding test sample of fault degree 0% and 1 relative distance of hypersphere
Minimum, then the corresponding fault degree 0% of hypersphere 1 is by the fault degree classification results as the test sample.Institute in statistical chart 12
There is the fault degree nicety of grading of test sample, result 100% illustrates of the invention based on Entropy of Instantaneous Frequency and noise energy
The fault signature that the fault energy zone boundary identification of difference is obtained with feature extracting method, is conducive to propeller fault degree point
Class, and nicety of grading is 100%.
Claims (3)
1. a kind of fault energy region time domain Boundary Recognition method based on instantaneous spectrum entropy, it is characterised in that: including following step
It is rapid:
The first step, acquisition length are L1Underwater robot speed signal；
Second step, calculating speed signal time-frequency power density spectrum；Speed is sought using conventional smooth puppet Wigner-Ville Distribution Algorithm
Smoothing Pseudo Eugene Wigner-Willie spectrum SPWVD (n, m) of signal is spent, n is the serial number on time shaft, n=1,2,3 ..., L1, m is frequency
Serial number on rate axis, m=1,2,3 ..., N3, N3For frequency axis demarcation interval number；And by the conventional absolute of SPWVD (n, m)
Value operation obtains speed signal time-frequency power spectral density spectrum SPWVDA (n, m) in turn；
Third step calculates the distribution of instantaneous spectrum entropy；In time-frequency power density spectrum SPWVDA (n, m), in time shaft serial number n=1
Place constructs the probability density function p (1, m) at the time point by formula (1), and then calculates the time point by formula (2)
The Shannon entropy at place, calculated result H (1), as the instantaneous spectrum entropy at time shaft serial number n=1；
This step content is repeated, calculating time shaft serial number n is respectively 2,3 ..., L1When instantaneous spectrum entropy H (n), obtain instantaneous frequency
Compose entropy distribution；
In formula, SPWVDA (1, m) is in time-frequency power density spectrum SPWVDA (n, m), on time shaft serial number n=1, m-th of frequency band
Energy, p (1, m) be probability density function, N3For frequency axis demarcation interval number, H (1) is the wink at time shaft serial number n=1
Time-frequency spectrum entropy；
4th step determines fault signature region time domain boundary, i.e., determines distortion section in the distribution of instantaneous spectrum entropy；
Determine minimum value position in the distribution of instantaneous spectrum entropy, then thus position extends to neighbouring maximum to both sides simultaneously
Point continues to extend to both sides if this maximum is less than the minimum in other regions, until the maximum encountered is greater than it
Until the minimum in his region, then the region between two final maximum points is section of distorting；Then with left side maximum
Point position is the time domain lower boundary N in fault energy regionD, using right side maximum point position as fault energy region
Time domain coboundary NU。
2. a kind of fault energy region frequency domain Boundary Recognition method based on noise energy difference, it is characterised in that: including following step
It is rapid:
The first step, acquisition length are L1Underwater robot speed signal data；
Second step carries out the wavelet function feedback of multiple scales using conventional wavelet transform method to speed signal, obtains more
A small echo approximation component ujA(n), j is wavelet decomposition scales, j=0,1,2 ..., 8, wherein j=0 indicates that speed signal does not carry out
Wavelet decomposition；
Third step calculates time-frequency power density spectrum；Using conventional smooth puppet Wigner-Ville Distribution Algorithm and absolute value algorithm meter
Calculate j-th of multi-scale wavelet approximation component ujA(n) time-frequency power density spectrum SPWVDAj(n,m)；
4th step calculates the distribution of noise energy difference；
The time domain lower boundary N in fault energy region is determined based on instantaneous spectrum entropyDWith time domain coboundary NU；
In time-frequency power density spectrum SPWVDAjIn (n, m), by time domain boundary NDWith NUBetween time-frequency power density spectrum SPWVDAj
(n, m) sums, acquired results ESjAs fault energy, by time domain boundary NDWith NUExcept time-frequency power density spectrum
SPWVDAj(n, m) sums, acquired results ENjAs noise energy, by E△SNj=ESj-ENjAs noise energy difference；It calculates
J=0,1,2 ..., 8 noise energy difference E△SNj, obtain the distribution of noise energy difference；
5th step determines fault energy region frequency domain boundary；In noise energy difference E△SNjIn distribution, determine that noise energy difference is maximum
Value, determines the corresponding wavelet decomposition scales of the maximum value, determines the corresponding small echo approximation component of the decomposition scale, then the small echo is close
Frequency band [the M like corresponding to componentD MU] it is fault energy region frequency domain boundary, wherein MDFor frequency domain lower boundary, MUFor frequency domain
Coboundary.
3. a kind of identification of fault energy zone boundary and feature extracting method based on instantaneous spectrum entropy and noise energy difference, special
Sign is: the following steps are included:
The first step, obtaining length respectively is L1Underwater robot speed signal and propeller control voltage change ratio signal data；
Second step determines fault energy region time domain lower boundary N based on instantaneous spectrum entropyDAnd time domain coboundary NU；
Third step determines fault energy region frequency domain lower boundary M based on noise energy differenceDAnd frequency domain coboundary MU；
4th step extracts propeller time-frequency energy failure feature；
Speed signal, control voltage change ratio letter are calculated separately using smooth and pseudo Wigner-Ville distribution algorithm, absolute value algorithm
Number time-frequency power density spectrum SPWVDAU(n, m) and SPWVDAC(n, m), in time domain lower boundary ND, time domain coboundary NU, frequency domain
Lower boundary MDWith frequency domain coboundary MUIn the rectangular area surrounded, sum to time-frequency power density spectrum respectively, acquired results
FU、FCRespectively as speed signal time-frequency energy failure feature and control signal time-frequency energy failure feature.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910001146.0A CN109633270B (en) | 2019-01-02 | 2019-01-02 | Fault energy region boundary identification and feature extraction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910001146.0A CN109633270B (en) | 2019-01-02 | 2019-01-02 | Fault energy region boundary identification and feature extraction method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109633270A true CN109633270A (en) | 2019-04-16 |
| CN109633270B CN109633270B (en) | 2020-10-02 |
Family
ID=66056466
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910001146.0A Active CN109633270B (en) | 2019-01-02 | 2019-01-02 | Fault energy region boundary identification and feature extraction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109633270B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110739984A (en) * | 2019-11-08 | 2020-01-31 | 江苏科技大学 | camouflage communication method based on wavelet analysis |
| CN112001314A (en) * | 2020-08-25 | 2020-11-27 | 江苏师范大学 | Early fault detection method for variable speed hoist |
| CN112051836A (en) * | 2020-09-11 | 2020-12-08 | 江苏科技大学 | Underwater robot propeller state monitoring method based on multi-core model |
| CN113655778A (en) * | 2021-08-16 | 2021-11-16 | 江苏科技大学 | Underwater propeller fault diagnosis system and method based on time-frequency energy |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101672873A (en) * | 2009-10-20 | 2010-03-17 | 哈尔滨工业大学 | Detection method of transient harmonic signals of power system based on combination of Tsallis wavelet singular entropy and FFT computation |
| JP4743914B2 (en) * | 2009-06-17 | 2011-08-10 | 有限会社諏訪トラスト | Time series data analysis system and time series data analysis program |
| CN103344875A (en) * | 2013-07-02 | 2013-10-09 | 福州大学 | Classification line selection method for single-phase earth fault of resonance earthing system |
| CN109033965A (en) * | 2018-06-22 | 2018-12-18 | 江苏科技大学 | A kind of underwater robot propeller failure time-frequency characteristics Enhancement Method |
-
2019
- 2019-01-02 CN CN201910001146.0A patent/CN109633270B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4743914B2 (en) * | 2009-06-17 | 2011-08-10 | 有限会社諏訪トラスト | Time series data analysis system and time series data analysis program |
| CN101672873A (en) * | 2009-10-20 | 2010-03-17 | 哈尔滨工业大学 | Detection method of transient harmonic signals of power system based on combination of Tsallis wavelet singular entropy and FFT computation |
| CN103344875A (en) * | 2013-07-02 | 2013-10-09 | 福州大学 | Classification line selection method for single-phase earth fault of resonance earthing system |
| CN109033965A (en) * | 2018-06-22 | 2018-12-18 | 江苏科技大学 | A kind of underwater robot propeller failure time-frequency characteristics Enhancement Method |
Non-Patent Citations (1)
| Title |
|---|
| 张铭钧等: "自主式水下机器人故障特征增强方法", 《哈尔滨工程大学学报》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110739984A (en) * | 2019-11-08 | 2020-01-31 | 江苏科技大学 | camouflage communication method based on wavelet analysis |
| CN112001314A (en) * | 2020-08-25 | 2020-11-27 | 江苏师范大学 | Early fault detection method for variable speed hoist |
| CN112051836A (en) * | 2020-09-11 | 2020-12-08 | 江苏科技大学 | Underwater robot propeller state monitoring method based on multi-core model |
| CN113655778A (en) * | 2021-08-16 | 2021-11-16 | 江苏科技大学 | Underwater propeller fault diagnosis system and method based on time-frequency energy |
| CN113655778B (en) * | 2021-08-16 | 2024-03-26 | 江苏科技大学 | Underwater propeller fault diagnosis system and method based on time-frequency energy |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109633270B (en) | 2020-10-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109633270A (en) | The identification of fault energy zone boundary and feature extracting method based on instantaneous spectrum entropy and noise energy difference | |
| CN111651937B (en) | Method for diagnosing faults of in-class self-adaptive bearing under variable working conditions | |
| CN103308804B (en) | Based on quick K-S converting electric power quality disturbance signal time and frequency parameter extracting method | |
| CN103208808B (en) | Power system sub-synchronous oscillation mode identification method | |
| CN112629863A (en) | Bearing fault diagnosis method for dynamic joint distribution alignment network under variable working conditions | |
| CN111898447B (en) | Xin Jihe modal decomposition-based wind turbine generator fault feature extraction method | |
| CN109253882A (en) | A kind of rotor crack fault diagnostic method based on variation mode decomposition and gray level co-occurrence matrixes | |
| Wei et al. | An optimal variational mode decomposition for rolling bearing fault feature extraction | |
| CN110866448A (en) | Flutter signal analysis method based on convolutional neural network and short-time Fourier transform | |
| CN112147601B (en) | Sea surface small target detection method based on random forest | |
| CN108051189A (en) | A kind of rotary machine fault characteristic extraction method and device | |
| CN111191740A (en) | Fault diagnosis method for rolling bearing | |
| CN109696906B (en) | Underwater robot propeller fault diagnosis method based on wavelet correction Bayes convolution energy | |
| CN109683591A (en) | Underwater propeller fault degree discrimination method based on fusion signal time domain energy and time-frequency entropy | |
| CN108334704A (en) | Structure damage monitoring method based on density self-adapting peak value-mixing probabilistic Modeling | |
| CN108376233A (en) | A kind of the separation sparse representation method and inaction interval detection method of fault detect | |
| CN107367552A (en) | Damage propatagtion method of real-time based on FCM algorithms under the influence of time varying temperature | |
| CN108181059A (en) | Multiphase flow pipeline leakage acoustic signals recognition methods based on small echo signal | |
| CN109973251A (en) | A kind of blind source separation method for solid propellant rocket vibration signal noise reduction | |
| CN103383413A (en) | Real-time harmonic detection method based on direct weight determination method | |
| CN111025100A (en) | Transformer ultrahigh frequency partial discharge signal mode identification method and device | |
| CN109633269A (en) | Based on the wavelet decomposition scales determination of time-frequency power spectral density maximum and fault signature Enhancement Method | |
| CN114764113A (en) | Diagnosis method suitable for faults of different metal particles in GIL | |
| CN114065813A (en) | Transformer winding fault diagnosis method for improving G-means vector elements | |
| Wang et al. | Identification of engine foreign object impact based on acoustic emission and radical basis function neural network |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |