CN109033965A - A kind of underwater robot propeller failure time-frequency characteristics Enhancement Method - Google Patents
A kind of underwater robot propeller failure time-frequency characteristics Enhancement Method Download PDFInfo
- Publication number
- CN109033965A CN109033965A CN201810653360.XA CN201810653360A CN109033965A CN 109033965 A CN109033965 A CN 109033965A CN 201810653360 A CN201810653360 A CN 201810653360A CN 109033965 A CN109033965 A CN 109033965A
- Authority
- CN
- China
- Prior art keywords
- time
- underwater robot
- propeller
- dynamic signal
- frequency
- 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
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000009826 distribution Methods 0.000 claims abstract description 50
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 12
- 230000004888 barrier function Effects 0.000 claims description 2
- 235000013399 edible fruits Nutrition 0.000 claims description 2
- 230000035945 sensitivity Effects 0.000 abstract description 12
- 230000002708 enhancing effect Effects 0.000 description 12
- 238000009499 grossing Methods 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 7
- 238000013507 mapping Methods 0.000 description 5
- 239000000284 extract Substances 0.000 description 4
- 239000012141 concentrate Substances 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000001965 increasing effect Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 101150004636 SYM1 gene Proteins 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 101150065184 sym-2 gene Proteins 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Investigating Or Analyzing Materials By The Use Of Ultrasonic Waves (AREA)
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
Abstract
The present invention discloses a kind of underwater robot propeller failure time-frequency characteristics Enhancement Method, first acquire and record the Dynamic Signal of underwater robot, current time beat and before the underwater robot dynamic signal data of time beat are intercepted using time domain window function, multilevel wavelet decomposition is carried out to the data of time domain window function interception again and obtains small echo approximation component, small echo approximation component data are handled based on amendment bayes method, obtained result is handled based on smooth and pseudo Wigner-Ville distribution, obtain time-frequency distributions, then it first takes absolute value to time-frequency distributions, two-dimensional convolution operation is carried out again, obtain time-frequency distributions two-dimensional convolution operation result, operation result is converted into probability density function, Shannon entropy is finally calculated based on probability density function, using Shannon entropy as failure time-frequency characteristics value, the present invention, which can effectively enhance, to be pushed away The distance between fault signature is corresponded to the sensitivity of fault degree and different faults degree into device failure time-frequency characteristics.
Description
Technical field
The present invention relates to underwater robot fields, specifically underwater robot propeller method for diagnosing faults, in particular to
Underwater robot propeller failure time-frequency characteristics Enhancement Method.
Background technique
Propeller is the most heavy component of underwater robot load, is easy to happen failure.When propeller breaks down, underwater machine
The Dynamic Signals such as device people motion state signal, control signal would generally generate singular behavior, mention from dynamic signal singularity behavior
The fault signature for taking characterization propeller fault message is the necessary condition for carrying out propeller fault diagnosis.Due to underwater robot
Motion state signal, control signal have it is non-linear, non-stationary, so Time-Frequency Analysis Method is a kind of more applicable underwater
Robot propeller fault signature extracting method.In Time-Frequency Analysis Method, Smoothing Pseudo Eugene Wigner-Willie Shannon entropy method is a kind of
Typical time-frequency characteristics extracting method.This method calculates underwater robot dynamic by smooth and pseudo Wigner-Ville distribution and believes
Number time-frequency distributions, the Shannon entropy of time-frequency distributions is then calculated, and using Shannon entropy as fault eigenvalue.The principle of this method
Be: propeller failure causes underwater robot Dynamic Signal to generate singular behavior, and Dynamic Signal singular behavior will be in Dynamic Signal
Time-frequency distributions in cause energy to be concentrated, and fault degree is bigger, and phenomenon of energy concentration is more obvious, and Shannon entropy is bigger.In reality
In trampling, Smoothing Pseudo Eugene Wigner-Willie Shannon entropy method can extract and propeller failure from underwater robot Dynamic Signal
The unique fault signature of degree mapping relations, but fault signature is corresponding to the sensitivity of fault degree and different faults degree
The distance between fault signature is smaller.
Summary of the invention
The purpose of the present invention is above-mentioned known Smoothing Pseudo Eugene Wigner-Willie Shannon entropy method there are aiming at the problem that, provide
A kind of underwater robot propeller failure time-frequency characteristics Enhancement Method enhances underwater machine in terms of time domain, frequency domain, time-frequency domain three
Device people's Dynamic Signal singular behavior caused energy in Dynamic Signal time-frequency distributions is concentrated, and then increases fault signature to failure
The sensitivity of degree and different faults degree correspond to the distance between fault signature.
The technical solution adopted by the present invention is that according to the following steps sequentially:
Step 1: the Dynamic Signal of acquisition and record underwater robot;
Step 2: intercepting current time beat and before the underwater robot dynamic of time beat using time domain window function
Signal data gives up first data in window function and the data newly received is placed on time window when receiving new data
End remains the data length in window function;
Step 3: carrying out multilevel wavelet decomposition to the data of time domain window function interception, small echo approximation component is obtained;
Step 4: being handled based on amendment bayes method small echo approximation component data;
Step 5: being handled based on the result that smooth and pseudo Wigner-Ville distribution obtains the 4th step, frequency division when obtaining
Cloth;
Step 6: first taking absolute value to time-frequency distributions, then two-dimensional convolution operation is carried out, obtains time-frequency distributions two-dimensional convolution fortune
Calculate result;
Step 7: time-frequency distributions two-dimensional convolution operation result is converted to probability density function;
Step 8: calculating Shannon entropy based on probability density function;
Step 9: using Shannon entropy as underwater robot propeller failure time-frequency characteristics value.
The beneficial effect of the present invention by adopting the above technical scheme is: it is special that the present invention can effectively enhance propeller failure time-frequency
Sign corresponds to the distance between fault signature to the sensitivity of fault degree and different faults degree.
Detailed description of the invention
Fig. 1 is well known Smoothing Pseudo Eugene Wigner-Willie Shannon entropy method flow chart in background technique;
Fig. 2 is a kind of flow chart of underwater robot propeller failure time-frequency characteristics Enhancement Method of the present invention;
Fig. 3 is that underwater robot controls voltage signal and longitudinal speed signal time domain waveform;
Fig. 4 is known smooth and pseudo Wigner-Ville distribution treated underwater robot longitudinal speed signal time-frequency distributions
Figure;
Fig. 5 is the enhanced underwater robot longitudinal speed signal time frequency distribution map of frequency domain of the present invention;
Fig. 6 is the enhanced underwater robot longitudinal speed signal time frequency distribution map of time domain of the present invention;
Fig. 7 is the enhanced underwater robot longitudinal speed signal time frequency distribution map of time-frequency domain of the present invention;
Fig. 8 is that the time-frequency characteristics of longitudinal speed signal extract and enhancing result figure;
Fig. 9 is the time-frequency characteristics extraction and enhancing result figure for controlling voltage change ratio signal;
Figure 10 is to control voltage change ratio signal to extract and enhancing knot with the signal time-frequency characteristics that merge of longitudinal speed signal
Fruit figure;
Figure 11 is the comparing result figure of the present invention and known method in terms of sensitivity;
Figure 12 is the comparing result figure of the present invention and known method in fault signature apart from aspect.
Specific embodiment
Known Smoothing Pseudo Eugene Wigner-Willie Shannon entropy method flow diagram as shown in Figure 1, well known Smoothing Pseudo Eugene Wigner-
Willie Shannon entropy method is to acquire and record first underwater robot Dynamic Signal initial data, is intercepted using time domain window function dynamic
State signal initial data, then Dynamic Signal original number is calculated using smooth and pseudo Wigner-Ville distribution algorithm, it calculates
The absolute value of time domain distribution, then the Shannon entropy of probability density function and time-frequency distributions is calculated, finally using Shannon entropy as table
Levy the fault eigenvalue of propeller fault message.
Underwater robot propeller failure time-frequency characteristics Enhancement Method flow chart of the present invention as shown in Figure 2.1 He of comparison diagram
Fig. 2 it is found that the two the difference is that: the present invention be based on wavelet-decomposing method, amendment bayes method, two-dimensional convolution fortune
Calculation method realizes the concentrated to underwater robot Dynamic Signal time-frequency distributions energy in terms of frequency domain, time domain, time-frequency domain three
Once, second, third time enhancing.Specific implementation step is:
Step 1: the Dynamic Signals such as acquisition and record underwater robot motion state signal and control signal, and store.
Step 2: use length for L=400 time domain window function intercept current time beat and it is 399 first when segmentum intercalaris
The underwater robot dynamic signal data of bat is given up first data in window function and will newly be received when receiving new data
Data be placed on the end of window function, remain that the data length in window function is L=400.
Step 3: carrying out multilevel wavelet decomposition to the data that time domain window function in second step intercepts.Based on wavelet decomposition side
Method is filtered underwater robot Dynamic Signal, to reduce external random noise interference effect, passively highlights propulsion
Dynamic Signal singular behavior amplitude caused by device failure, and then realized from frequency domain angle and Dynamic Signal time-frequency distributions energy is concentrated
First time enhancing.
Wavelet basis function is " db4 ", and Decomposition order 3 obtains wavelet scale coefficient and wavelet details coefficient, casts out small echo
Detail coefficients, and the reconstruct of small echo list branch is carried out to wavelet scale coefficient, obtain small echo approximation component.
Wherein, multilevel wavelet decomposition formula is as follows:
Small echo list branch reconstruction formula is as follows:
In formula, u (n) be underwater robot Dynamic Signal, n be dynamic signal data time shaft on serial number, n=1,
2 ..., L, L=400,For db4 scaling function, j is the wavelet decomposition number of plies, and j=1,2,3, k are
Wavelet coefficient serial number, k=1,2 ..., it is not more than 2-jThe maximum integer of × L,For db4 small echo letter
Number, DjIt (k) is wavelet details coefficient, AjIt (k) is wavelet scale coefficient, uAIt (n) is Dynamic Signal small echo approximation component,<>is interior
Product operator.Wavelet scaling function and wavelet function are chosen from wavelet basis function library.In existing wavelet basis function library
The wavelet scaling function and wavelet function of plurality of classes has been defined, as Symlets small echo, Daubechies small echo,
Morlet small echo, Meyer small echo etc., the wavelet basis function of each classification can be divided into different models again, as Symlets is small
Wave has eight kinds of models, respectively sym1, sym2 ..., sym8, Daubechies small echo has ten kinds of models, be expressed as db1,
Db2 ..., db10 etc. suitable small echo letter can be directly selected from wavelet basis function library when using wavelet function
Number, in the present invention, db4 small echo are most suitable for the data in the processing present invention.
Step 4: at the Dynamic Signal small echo approximation component data obtained based on amendment bayesian algorithm to third step
Reason, thus it is abnormal based on dynamic signal parameter, actively enhance Dynamic Signal singular behavior amplitude caused by propeller failure, in turn
Second of the enhancing concentrated to Dynamic Signal time-frequency distributions energy is realized from time domain angle.Detailed process is as follows:
Calculate the desired value of underwater robot small echo approximation component when propeller works normallyAnd varianceFormula is such as
Under:
In formula, uA0Underwater robot Dynamic Signal small echo approximation component when being worked normally for propeller, N1Enough for one
Big positive integer, the present invention take N1=400,Respectively desired value and variance.
Then, it is based on varianceUsing underwater robot when following formula is normal to propeller and different faults degree
Dynamic Signal small echo approximation component is handled, to enhance Dynamic Signal singular behavior amplitude:
Wherein, uAIt (n) is underwater robot Dynamic Signal small echo when propeller is normal and different faults degree
Approximation component, i=1,2 ..., N2, N2For window function length, N in the present invention2=20, duAIt (n) is amendment bayesian algorithm processing
As a result.
Step 5: based on smooth and pseudo Wigner-Ville distribution algorithm to the amendment bayesian algorithm processing result of the 4th step
duA(n) data are handled, and obtain the time-frequency distributions of data, which can describe the Energy distribution of Dynamic Signal.It is flat
Sliding puppet Eugene Wigner-Willie distribution calculation formula is as follows:
In formula, SPWVD (n, m) be time-frequency distributions calculated result, n be dynamic signal data time shaft on serial number, n=1,
2 ..., L, L=400, m are the serial number on dynamic signal data frequency axis, m=1,2 ..., N3, N3For frequency axis demarcation interval number,
N3=512, h (k1) and g (l1) it is Gauss function, k1=-(K1- 1)~(K1- 1), l1=-(L1- 1)~(L1- 1), K1For not
Greater than (N3The maximum integer of)/4, L1For no more than (N3The maximum integer of)/5, z (n) are signal duA(n) analytical form, z*
(n) conjugate complex number for being z (n).
Step 6: the time-frequency distributions calculated result obtained to the 5th step takes absolute value, calculation formula is as follows:
SPWVDA (n, m)=| SPWVD (n, m) |,
SPWVDA represents the absolute value of SPWVD.
Step 7: the absolute value data obtained to the 6th step carries out two-dimensional convolution operation, thus based on two dimensional image from
Correlation, the opposite time-frequency distributions amplitude for enhancing area of energy concentration domain, and then realize from time-frequency domain angle to Dynamic Signal time-frequency
The third time enhancing that distribution energy is concentrated, calculation formula are as follows:
In formula, C (nc, mc) is time-frequency distributions two-dimensional convolution calculated result, and nc is a variable, value nc=1,
2 ..., (2*L-1), similarly, mc are another variable, value mc=1,2 ..., (2*N3-1)。
Step 8: the 7th step data is converted into probability density function, conversion formula is as follows:
P (nc, mc)=C (nc, mc)/∑ ∑ C (nc, mc).
Step 9: calculating Shannon entropy according to the 8th step data, calculation formula is as follows:
H=- ∑ ∑ p (nc, mc) log2p(nc,mc)。
Tenth step, the Shannon entropy H that the 9th step is calculated is as underwater robot propeller failure time-frequency characteristics value.
Fig. 3 is that underwater robot controls voltage signal and longitudinal speed signal time domain waveform.Underwater robot propeller
Failure normally behaves as output loss, and after propeller breaks down, underwater robot longitudinal velocity decline, closed loop controller is mentioned
Height control voltage is to compensate output loss caused by propeller failure, so that longitudinal velocity will gradually promote target value.Fig. 3
In, underwater robot starts since static, reaches target velocity 0.3m/s at the 100th time beat, when from the 150th
Between beat start to the 500th time beat to terminate, output loss failure occurs for left main thruster.Upper figure is underwater machine in Fig. 3
Device people or so main thruster controls voltage, and the following figure is underwater robot longitudinal velocity situation of change in Fig. 3.λ in upper figure in Fig. 3=
10% indicates that propeller fault degree is 10%, i.e. the practical power output of propeller output loss 10%, i.e. propeller is theoretical power output
90%, similarly, λ=0%, λ=20%, λ=30%, λ=40% respectively indicate propeller fault degree be 0%, 20%,
30%, 40%, propeller output loss is 0%, 20%, 30%, 40%, the practical power output of propeller is theoretical 100% to contribute,
80%, 70%, 60%.In Fig. 3 in the partial enlarged view of the following figure, from top to bottom, be corresponding in turn to propeller fault degree be 0%,
10%, 20%, 30%, 40% when, the situation of change of underwater robot longitudinal velocity.The present invention was with 20 seconds to 100 seconds in Fig. 3
For the experimental data of (the 100th time beat to the 500th time beat), the working principle of the invention and beneficial is shown
Effect.
As shown in figure 4, Fig. 4 is known smooth and pseudo Wigner-Ville distribution treated underwater robot longitudinal velocity letter
Number time frequency distribution map.In Fig. 4, fault degree 0%, 10%, 20%, 30%, 40%, expression propeller output loss 0%,
10%, 20%, 30%, 40%, i.e. the practical power output of propeller is 100%, 90%, 80%, 70%, the 60% of theoretical power output.This
In it should be strongly noted that when propeller fault degree be 0% when, indicate propeller work normally.In Fig. 4 (a), propeller
When normal work, as shown in dotted outline in FIG., there are multiple energy to concentrate in the time-frequency distributions of underwater robot longitudinal speed signal
Region, Energy distribution is relatively uniform, as shown in Fig. 4 (b)~4 (e), as shown in dotted outline in FIG., with propeller fault degree
Increase, the energy in time-frequency distributions gradually concentrates to a region, and phenomenon of energy concentration gradually increases.
The present invention as shown in Figure 5 is in the enhanced underwater robot longitudinal speed signal time frequency distribution map of frequency domain, Fig. 6 institute
The present invention shown is in the enhanced underwater robot longitudinal speed signal time frequency distribution map of time domain;It is shown in Fig. 7 the present invention when
The enhanced underwater robot longitudinal speed signal time frequency distribution map of frequency domain.Comparison diagram 4 and Fig. 5, Fig. 6, Fig. 7, especially comparison are empty
Content shown in wire frame, enhances in time-frequency distributions first time of the invention, second, third time enhancing to some extent
Energy concentrate.
Fig. 8 is that the time-frequency characteristics of longitudinal speed signal extract and enhancing result figure.According to Fig. 4~Fig. 7 data, calculate in figure
It is as shown in Figure 8 to obtain propeller failure time-frequency characteristics value for the Shannon entropy of time-frequency distributions.As can be seen from Figure 8, correspond to known
The mapping relations curve of method, present invention first time, second, third time Enhancement Method, fault eigenvalue and fault degree
Slope gradually increases, and the difference of mapping relations curve initial and end fault eigenvalue is also gradually increased.
As shown in Figure 9 and Figure 10, Fig. 9 is time-frequency characteristics extraction and enhancing result figure, the figure for controlling voltage change ratio signal
10 be to control voltage change ratio signal to extract and enhancing result figure, Fig. 9, figure with the signal time-frequency characteristics that merge of longitudinal speed signal
Data shown in data shown in 10 and Fig. 8 have same rule, that is, correspond to known method, the present invention for the first time, second, the
The mapping relations slope of a curve of Enhancement Method three times, fault eigenvalue and fault degree gradually increases, and mapping relations curve
The difference of initial and end fault eigenvalue is also gradually increased.
Figure 11 is the comparing result figure of the present invention and known method in terms of sensitivity.According to Fig. 8, Fig. 9, Figure 10 data,
Sensitivity of the fault eigenvalue to fault degree in figure is calculated, as a result as shown in figure 11.Calculation of Sensitivity formula is as follows:
In formula, S is sensitivity of the fault eigenvalue to fault degree, HjFor the corresponding Shannon entropy of jth kind fault degree,
H0Shannon entropy when being worked normally for propeller, λjFor jth kind fault degree value, λ0Failure when being worked normally for propeller
Degree value, i.e. λ0=0.
As shown in figure 11, correspond to known method, present invention first time, second, third time Enhancement Method, from longitudinal speed
The fault eigenvalue that extracts successively increases the sensitivity of fault degree in degree signal, in addition, from control voltage change ratio signal,
The fault signature extracted in fusion signal also has same rule, i.e. fault eigenvalue successively increases the sensitivity of fault degree
Add, illustrate the invention patent compared with known Smoothing Pseudo Eugene Wigner-Willie Shannon entropy method, is capable of increasing fault eigenvalue to event
The sensitivity of barrier degree.
Figure 12 is the comparing result figure of the present invention and known method in fault signature apart from aspect.According to Fig. 8, Fig. 9, Figure 10
Data, the different faults degree for calculating each Dynamic Signal in figure correspond to the distance of fault signature, as a result as shown in figure 12.Therefore
It is as follows to hinder characteristic distance calculation formula:
In formula, D is fault signature distance, and n is the number of fault degree type, Hp、HjRespectively P kind, jth kind failure
The corresponding Shannon entropy of degree.
As shown in figure 12, correspond to known method, present invention first time, second, third time Enhancement Method, from longitudinal speed
The fault eigenvalue extracted in degree signal, different degrees of failure corresponds to the distance between fault signature and successively increases, in addition, from control
The fault signature extracted in voltage change ratio signal processed, fusion signal also has same rule, i.e., different degrees of failure is corresponding
The distance between fault signature successively increases, and illustrates of the invention compared with known Smoothing Pseudo Eugene Wigner-Willie Shannon entropy method, energy
Enough increase different degrees of failure and corresponds to the distance between fault signature.
Claims (8)
1. a kind of underwater robot propeller failure time-frequency characteristics Enhancement Method, it is characterized in that according to the following steps sequentially:
Step 1: the Dynamic Signal of acquisition and record underwater robot;
Step 2: intercepting current time beat and before the underwater robot Dynamic Signal of time beat using time domain window function
Data give up first data in window function and the data newly received are placed on to the end of time window when receiving new data,
Remain the data length in window function;
Step 3: carrying out multilevel wavelet decomposition to the data of time domain window function interception, small echo approximation component is obtained;
Step 4: small echo approximation component data are handled based on amendment bayes method,
Step 5: handling based on the result that smooth and pseudo Wigner-Ville distribution obtains the 4th step, time-frequency distributions are obtained;
Step 6: first taking absolute value to time-frequency distributions, then two-dimensional convolution operation is carried out, obtains time-frequency distributions two-dimensional convolution operation knot
Fruit;
Step 7: time-frequency distributions two-dimensional convolution operation result is converted to probability density function;
Step 8: calculating Shannon entropy based on probability density function;
Step 9: using Shannon entropy as underwater robot propeller failure time-frequency characteristics value.
2. a kind of underwater robot propeller failure time-frequency characteristics Enhancement Method according to claim 1, it is characterized in that: the
In two steps, length is used to intercept the underwater of current time beat and preceding 399 time beats for the time domain window function of L=400
Robotic Dynamic signal data.
3. a kind of underwater robot propeller failure time-frequency characteristics Enhancement Method according to claim 1, it is characterized in that: the
In three steps, when multilevel wavelet decomposition, wavelet basis function db4, Decomposition order 3 obtains wavelet scale coefficient and wavelet details
Coefficient casts out wavelet details coefficient, and carries out the reconstruct of small echo list branch to wavelet scale coefficient, obtains small echo approximation component.
4. a kind of underwater robot propeller failure time-frequency characteristics Enhancement Method according to claim 3, it is characterized in that: the
In four steps, the desired value and variance of underwater robot small echo approximation component when propeller works normally first are calculated, then to propeller
Underwater robot Dynamic Signal small echo approximation component when normal and different faults degree is handled.
5. a kind of underwater robot propeller failure time-frequency characteristics Enhancement Method according to claim 4, it is characterized in that: the
In six steps, using formulaCarry out two-dimentional volume
Product operation, C (nc, mc) are time-frequency distributions two-dimensional convolution calculated result, and n=1,2 ..., L, L=400, m are dynamic signal data
Serial number on frequency axis, m=1,2 ..., N3, N3For frequency axis demarcation interval number, SPWVDA (n, m) is the absolute of time-frequency distributions
Value, nc are a variables, value nc=1,2 ..., (2*L-1), similarly, mc is another variable, value mc=1,
2 ..., (2*N3-1)。
6. a kind of underwater robot propeller failure time-frequency characteristics Enhancement Method according to claim 3, it is characterized in that: more
The formula of layer wavelet decomposition is:The public affairs of small echo list branch reconstruct
Formula is:U (n) is underwater robot Dynamic Signal, and n is the dynamic signal data time
Serial number on axis, n=1,2 ..., L, L=400,For db4 scaling function, j is wavelet decomposition layer
Number, j=1,2,3, k be wavelet coefficient serial number, k=1,2 ..., be not more than 2-jThe maximum integer of × L,For db4 wavelet function, DjIt (k) is wavelet details coefficient, AjIt (k) is wavelet scale coefficient, uA
It (n) is Dynamic Signal small echo approximation component,<>is inner product operation symbol.
7. a kind of underwater robot propeller failure time-frequency characteristics Enhancement Method according to claim 4, it is characterized in that: small
The desired value of wave approximation componentVarianceuA0For the normal work of propeller
Underwater robot Dynamic Signal small echo approximation component, N when making1=400.
8. a kind of underwater robot propeller failure time-frequency characteristics Enhancement Method according to claim 7, special
Sign is: using following formulaWhen and different faults degree normal to propeller
Underwater robot Dynamic Signal small echo approximation component is handled, uA(n) normally and different former for propeller
Underwater robot Dynamic Signal small echo approximation component when barrier degree, i=1,2 ..., N2, N2For window function length, N2=20, duA
It (n) is amendment bayesian algorithm processing result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810653360.XA CN109033965B (en) | 2018-06-22 | 2018-06-22 | Underwater robot propeller fault time-frequency feature enhancement method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810653360.XA CN109033965B (en) | 2018-06-22 | 2018-06-22 | Underwater robot propeller fault time-frequency feature enhancement method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109033965A true CN109033965A (en) | 2018-12-18 |
CN109033965B CN109033965B (en) | 2021-10-01 |
Family
ID=64610147
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810653360.XA Expired - Fee Related CN109033965B (en) | 2018-06-22 | 2018-06-22 | Underwater robot propeller fault time-frequency feature enhancement method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109033965B (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109633270A (en) * | 2019-01-02 | 2019-04-16 | 江苏科技大学 | The identification of fault energy zone boundary and feature extracting method based on instantaneous spectrum entropy and noise energy difference |
CN109633269A (en) * | 2019-01-02 | 2019-04-16 | 江苏科技大学 | Based on the wavelet decomposition scales determination of time-frequency power spectral density maximum and fault signature Enhancement Method |
CN109683591A (en) * | 2018-12-27 | 2019-04-26 | 江苏科技大学 | Underwater propeller fault degree discrimination method based on fusion signal time domain energy and time-frequency entropy |
CN111091159A (en) * | 2019-12-27 | 2020-05-01 | 哈尔滨工程大学 | Weak fault feature extraction method for autonomous underwater robot propeller |
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 |
CN113780355A (en) * | 2021-08-12 | 2021-12-10 | 上海理工大学 | Deep convolutional neural network learning method for deep sea submersible propeller fault identification |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107132760A (en) * | 2017-03-27 | 2017-09-05 | 哈尔滨工程大学 | The underwater robot state monitoring method described based on fuzzy support vector domain |
CN107845389A (en) * | 2017-12-21 | 2018-03-27 | 北京工业大学 | A kind of sound enhancement method based on multiresolution sense of hearing cepstrum coefficient and depth convolutional neural networks |
-
2018
- 2018-06-22 CN CN201810653360.XA patent/CN109033965B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107132760A (en) * | 2017-03-27 | 2017-09-05 | 哈尔滨工程大学 | The underwater robot state monitoring method described based on fuzzy support vector domain |
CN107845389A (en) * | 2017-12-21 | 2018-03-27 | 北京工业大学 | A kind of sound enhancement method based on multiresolution sense of hearing cepstrum coefficient and depth convolutional neural networks |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109683591A (en) * | 2018-12-27 | 2019-04-26 | 江苏科技大学 | Underwater propeller fault degree discrimination method based on fusion signal time domain energy and time-frequency entropy |
CN109683591B (en) * | 2018-12-27 | 2021-03-19 | 江苏科技大学 | Underwater propeller fault degree identification method based on fusion signal time domain energy and time-frequency entropy |
CN109633270A (en) * | 2019-01-02 | 2019-04-16 | 江苏科技大学 | The identification of fault energy zone boundary and feature extracting method based on instantaneous spectrum entropy and noise energy difference |
CN109633269A (en) * | 2019-01-02 | 2019-04-16 | 江苏科技大学 | Based on the wavelet decomposition scales determination of time-frequency power spectral density maximum and fault signature Enhancement Method |
CN109633270B (en) * | 2019-01-02 | 2020-10-02 | 江苏科技大学 | Fault energy region boundary identification and feature extraction method |
CN111091159A (en) * | 2019-12-27 | 2020-05-01 | 哈尔滨工程大学 | Weak fault feature extraction method for autonomous underwater robot propeller |
CN111091159B (en) * | 2019-12-27 | 2023-03-17 | 哈尔滨工程大学 | Weak fault feature extraction method for autonomous underwater robot propeller |
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 |
CN113780355A (en) * | 2021-08-12 | 2021-12-10 | 上海理工大学 | Deep convolutional neural network learning method for deep sea submersible propeller fault identification |
CN113780355B (en) * | 2021-08-12 | 2024-02-09 | 上海理工大学 | Deep convolution neural network learning method for fault identification of deep sea submersible propeller |
Also Published As
Publication number | Publication date |
---|---|
CN109033965B (en) | 2021-10-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109033965A (en) | A kind of underwater robot propeller failure time-frequency characteristics Enhancement Method | |
Guo et al. | A review of wavelet analysis and its applications: Challenges and opportunities | |
CN106127217A (en) | The method and system that neutral net detects is goed deep into for anatomical object for approximation | |
CN105374026B (en) | A kind of detection method of marine infrared small target suitable for coast defence monitoring | |
Yi et al. | Mechanical compound faults extraction based on improved frequency domain blind deconvolution algorithm | |
CN110151175A (en) | Surface electromyogram signal noise-eliminating method based on CEEMD and improvement wavelet threshold | |
Yu et al. | CWGAN: Conditional wasserstein generative adversarial nets for fault data generation | |
CN107292852A (en) | A kind of Image denoising algorithm theoretical based on low-rank | |
CN111899168A (en) | Remote sensing image super-resolution reconstruction method and system based on feature enhancement | |
CN107886078A (en) | A kind of Threshold Denoising method based on layered self-adapting threshold function table | |
CN109859131A (en) | A kind of image recovery method based on multi-scale self-similarity Yu conformal constraint | |
He et al. | GMC sparse enhancement diagnostic method based on the tunable Q-factor wavelet transform for detecting faults in rotating machines | |
Zhu et al. | A transformer model with enhanced feature learning and its application in rotating machinery diagnosis | |
CN107966287A (en) | A kind of adaptive dynamoelectric equipment Weak fault feature extracting method | |
Mukherjee et al. | Study of efficient methods of detection and reconstruction of gravitational waves from nonrotating 3D general relativistic core collapse supernovae explosion using multilayer signal estimation method | |
Jothi et al. | Fetal electrocardiogram extraction using adaptive neuro-fuzzy inference systems and undecimated wavelet transform | |
Meng et al. | Intelligent fault diagnosis of gearbox based on multiple synchrosqueezing S-Transform and convolutional neural networks | |
CN114781427A (en) | Wind generating set rolling bearing fault diagnosis system based on information fusion technology | |
Xie et al. | An iterative method with enhanced Laplacian-scaled thresholding for noise-robust compressive sensing magnetic resonance image reconstruction | |
CN110309689A (en) | The domain Gabor gesture identification detection method based on ULTRA-WIDEBAND RADAR | |
Jiang et al. | Fault diagnosis of three-phase inverter based on CEEMDAN and GWO-SVM | |
Cao et al. | Fault Diagnosis Based on Optimized Wavelet Packet Transform and Time Domain Convolution Network | |
CN113704688A (en) | Defect vibration signal recovery method based on variational Bayes parallel factor decomposition | |
Chen et al. | A new method for fault diagnosis of hydraulic system based on improved empirical wavelet transform and kernel extreme learning machine | |
CN108765509B (en) | Rapid image reconstruction method for linear imaging system |
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 | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20220815 Address after: 212021 room 1-413, high tech entrepreneurship service center, Zhenjiang high tech Zone, Zhenjiang City, Jiangsu Province Patentee after: Zhenjiang Haiyun Marine Technology Co.,Ltd. Address before: 212003, No. 2, Mengxi Road, Zhenjiang, Jiangsu Patentee before: JIANGSU University OF SCIENCE AND TECHNOLOGY |
|
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20211001 |