CN110243605A - Multi-source time-frequency crestal line extracting method - Google Patents
Multi-source time-frequency crestal line extracting method Download PDFInfo
- Publication number
- CN110243605A CN110243605A CN201910674416.4A CN201910674416A CN110243605A CN 110243605 A CN110243605 A CN 110243605A CN 201910674416 A CN201910674416 A CN 201910674416A CN 110243605 A CN110243605 A CN 110243605A
- Authority
- CN
- China
- Prior art keywords
- frequency
- time
- crestal line
- fast path
- extracted
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
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
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
-
- 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
- G06F2218/16—Classification; Matching by matching signal segments
Landscapes
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- General Physics & Mathematics (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The multi-source time-frequency crestal line extracting method based on fast path optimum search and dynamic base angle degree that the invention discloses a kind of.The multi-source time-frequency crestal line extracting method based on fast path optimum search and dynamic base angle degree that the present invention provides a kind of, comprising: step 1: the selection of Short Time Fourier Transform and frequency band.To Short Time Fourier Analysis is used in vibration signal, bearing signal is divided into low-frequency range and resonance bands.Beneficial effects of the present invention: 1, fast path Optimal Searching is applied in the invention first, obtain continuous and accurate instantaneous frequency crestal line, then on the basis of obtained crestal line, calculate the corner cut of corresponding moment point, the basic function that frequency and echo signal frequency match is obtained, suitable for handling the bearing vibration signal under variable speed unstable condition.
Description
Technical field
The present invention relates to mechanical equipment field of signal processing, and in particular to one kind is based on fast path optimum search and dynamic
The multi-source time-frequency crestal line extracting method of base angle degree.
Background technique
Key components and parts one of of the bearing as rotating machinery, state directly influence the operating status of rotating machinery,
Once bearing breaks down, it will be likely to bring huge economic loss, even result in casualties.Signal disposal and analysis is
Realize the effective way of bearing failure diagnosis.In terms of mechanical breakdown feature extraction, earlier research be based on time-domain analysis and
The signal processing method of frequency-domain analysis is suitable for analyzing linear, stable Dynamic Signal mostly.These methods are simple and efficient,
It is used widely in actual machine fault diagnosis.
There are following technical problems for traditional technology:
But when there is local fault in the variation of rotating machinery operating condition or critical component, the vibration letter of acquisition
Number non-linear, non-stationary property is often presented, traditional time domain, frequency-domain analysis method often have significant limitation.
Summary of the invention
The technical problem to be solved in the present invention is to provide a kind of more based on fast path optimum search and dynamic base angle degree
Source time-frequency crestal line extracting method introduces dynamic base angle degree on the basis of being analyzed with traditional Fourier in short-term signal
Strategy optimizes the aggregation of time-frequency crestal line, so as to extract to obtain more accurate result.In addition, the invention is using quickly
Path optimal searching algorithm considers all amplitude maximum in entire start from, and establishes between all amplitude maximum of moment point of front and back
All possible path optimizing ensure that the continuity and accuracy of instantaneous time-frequency crestal line.Then, in the limit of taking advantage of a situation of extraction
On the basis of be calculated frequency and basic function that target matches, to improve time-frequency locality.
In order to solve the above-mentioned technical problems, the present invention provides one kind to be based on fast path optimum search and dynamic base angle degree
Multi-source time-frequency crestal line extracting method, comprising:
Step 1: the selection of Short Time Fourier Transform and frequency band.To Short Time Fourier Analysis is used in vibration signal, by axis
It holds signal and is divided into low-frequency range and resonance bands.
Step 2: calculating the local maximum of had amplitude and its corresponding frequency at each moment point.According to through street
Diameter optimal algorithm determines possible path optimizing between each extreme point and front and back extreme point.It is determined most according to the solution of optimization method
The corresponding extreme value of the frequency values that should be extracted at the latter moment point chooses corresponding that quickly most according to the extreme point
Shortest path, quickly optimal path is finally to extract obtained instantaneous frequency crestal line for this；
Step 3: a kind of dynamic base angle degree strategy is applied, after doing in conjunction with fast path optimum search result to original time-frequency figure
Processing, i.e., the instantaneous frequency crestal line extracted using fast path optimal algorithm are sought the corner cut of crestal line each point, obtain frequency
With the matched basic function of instantaneous frequency；Finally, due in low-frequency range and resonance bands any two revised time-frequency crestal lines it
Between ratio characterization failure type, carry out bearing failure diagnosis accordingly.
Step 2 specifically includes in one of the embodiments:
Step 2.1: calculating the local maximum of amplitude at each moment point on time-frequency figure, obtain had width at the moment point
It is worth the corresponding frequency values of maximum
fm(τn)=f, (d [X (τn, f)]/df=0 | | d2[X(τn,f)]/df2< 0), m=1,2 ..., Np (9)
Wherein, τn(n=1,2 ..., N) indicates at the time of point, N on time-frequency figurep(τn) indicate moment point τnLocate local maximum
The number of value, fm(τn) it is moment point τnLocate the corresponding frequency values of m-th of extreme value, amplitude is denoted as SPm(τn)。
Step 2.2: accurately to extract time-frequency crestal line, proposing a kind of path optimization's algorithm to determine at each moment point should be mentioned
The amplitude maximum taken, the extreme value is corresponding, fm(τn) it is exactly the frequency values that should be finally extracted, path optimization's algorithmic notation
It is as follows:
Wherein, mc(τn) determine moment point τnThe amplitude maximum that place is extracted, F [] indicates optimization method, according to this
Non trivial solution can extract optimal time-frequency crestal line.Optimization method is specifically expressed as follows:
M []=perc0.5[], IQR []=perc0.75[ ]-perc0.25[ ] (14)
Wherein, fd(τn-1) indicate moment point τn-1The candidate ridge frequency values at place, Δ fdIt is fdDerivative, percp[f (t)] table
Show the pth quantile of f (t).
Step 2.3: the solution of optimization method can be expressed as
Wherein, n=1,2 ..., N, m=1,2 ..., Np(τn), q (m, τn) to determine time instant τn(n=2 ..., N)
Path optimizing should be constituted with which extreme point at previous moment point by locating each amplitude maximum, plan all possible paths.U
(m,τn) indicate to obtain the intermediate vector of optimum results, that is, select time instant τNLocate U (m, τn) the amplitude maximum point pair that is maximized
The frequency values answered are as the instantaneous frequency that should be finally extracted, accordingly by q (m, τn) determine the extreme point under path make
It is optimal for fast path, so far extract instantaneous frequency crestal line.
The step 3 includes: in one of the embodiments,
Step 3.1: a kind of application dynamic base angle degree time-frequency figure enhancing strategy considers the angle of linear transformation base rather than window
It is long, corner cut is calculated using the instantaneous frequency crestal line extracted from original time-frequency figure by fast path optimal algorithm.It obtains
The basic function that frequency and echo signal frequency match, then post-processes time-frequency figure, to improve time-frequency figure aggregation.
Step 3.2: the ratio in low-frequency range and resonance bands between any two revised time-frequency crestal lines characterizes event
Hinder type, bearing failure diagnosis can be carried out accordingly.It is extracted from low-frequency range and corrects to obtain two time-frequency crestal line fr1And fr2, from
It is extracted in resonance bands and corrects to obtain two time-frequency crestal line fr3And fr4, ratio between any two has θ1=fr3/fr1, θ2=
fr3/fr2, θ3=fr4/fr1, θ4=fr4/fr2, they should be fault signature coefficient or its multiple.Diagnose different faults type
Bearing when calculating the mean value of its fault signature coefficient, copes with these ratios θ1, θ2, θ3And θ4Synchronization process is done, then seeks it
Mean value has θ=(η1θ1+η2θ2+η3θ3+η4θ4)/4；
Wherein, η1,η2,η3,η4Indicate the synchronization coefficient of corresponding ratio.So far, bearing fault type can be according to failure spy
Sign Coefficient Mean diagnosis obtains.
In one of the embodiments, in step 1, low-frequency range is [0,400] Hz.
A kind of computer equipment can be run on a memory and on a processor including memory, processor and storage
The step of computer program, the processor realizes any one the method when executing described program.
A kind of computer readable storage medium, is stored thereon with computer program, realization when which is executed by processor
The step of any one the method.
A kind of processor, the processor is for running program, wherein described program executes described in any item when running
Method.
Beneficial effects of the present invention:
1, fast path Optimal Searching is applied in the invention first, obtains continuous and accurate instantaneous frequency crestal line, so
Afterwards on the basis of obtained crestal line, the corner cut of corresponding moment point is calculated, obtains the base that frequency and echo signal frequency match
Function, suitable for handling the bearing vibration signal under variable speed unstable condition.
2, the multi-source time-frequency crestal line extraction side that the fast path optimum search and dynamic base angle degree that the invention is proposed combine
Method can obtain more accurate result compared with traditional method.It is searched for compared to traditional plots peak, fast path search
The calculation amount of signal analysis can be reduced and promote precision of analysis；Compared to traditional Short Time Fourier Transform, due to
Short Time Fourier Transform carries out time frequency analysis to signal using the long segmentation of fixed window, in conjunction with corresponding with actual frequency variation
Basic function can further enhance time-frequency locality.
Detailed description of the invention
Fig. 1 is the bearing test-bed for acquiring experimental signal.
Fig. 2 is proposed by the invention based on the extraction of the multi-source time-frequency crestal line of fast path optimum search and dynamic base angle degree
The flow chart of method.
Fig. 3 is the schematic diagram of fast path optimal searching algorithm, and (a) calculates local maximum；(b) planning path；(c) it selects
Take optimal path.
Fig. 4 is to extract result based on faulty bearings analog signal fast path optimum time frequency crestal line of the invention, wherein figure
It (a) is that low-frequency range is extracted as a result, figure (b) is that resonance bands extract result.
Fig. 5 obtains the faulty bearings being calculated after crestal line simulation based on fast path optimum search to be proposed by the present invention
Signal crestal line corner cut.
Fig. 6 be analog signal low-frequency range time-frequency figure comparison result of the present invention, wherein (a) window it is a length of 1200 when time-frequency figure；
(b) window it is a length of 1200 when crestal line extract result；(c) window it is a length of 2161 when time-frequency figure；(d) window it is a length of 2161 when crestal line mention
Take result.
Fig. 7 be analog signal resonance bands time-frequency figure comparison result of the present invention, wherein (a) window it is a length of 5200 when time-frequency
Figure；(b) window it is a length of 5200 when crestal line extract result；(c) window it is a length of 9295 when time-frequency figure；(d) window it is a length of 9295 when ridge
Line drawing result.
Fig. 8 is to extract result based on bearing fault experimental signal fast path optimum time frequency crestal line of the invention, wherein figure
It (a) is that low-frequency range is extracted as a result, figure (b) is that resonance bands extract result.
Fig. 9 obtains the faulty bearings being calculated after crestal line reality based on fast path optimum search to be proposed by the present invention
Signal crestal line corner cut.
Figure 10 be analog signal low-frequency range time-frequency figure comparison result of the present invention, wherein (a) window it is a length of 4590 when time-frequency
Figure；(b) window it is a length of 4590 when crestal line extract result；(c) window it is a length of 5818 when time-frequency figure；(d) window it is a length of 5818 when ridge
Line drawing result.
Figure 11 analog signal resonance band time-frequency figure comparison result of the present invention, wherein (a) window it is a length of 2500 when time-frequency
Figure；(b) window it is a length of 2500 when crestal line extract result；(c) window it is a length of 3377 when time-frequency figure；(d) window it is a length of 3377 when ridge
Line drawing result.
Specific embodiment
The present invention will be further explained below with reference to the attached drawings and specific examples, so that those skilled in the art can be with
It more fully understands the present invention and can be practiced, but illustrated embodiment is not as a limitation of the invention.
The construction for emulating signal model is as follows
X (t)=x1(t)+x2(t)+η(t) (17)
Wherein, x1(t) it indicates fault signature Related Component, is defined as
Wherein, [1+ α cos (2 π x2(t)) it] reflects by tach signal x2(t) covibration demodulated, α=0.11 are to adjust
Amplitude processed, M indicate signal length, AmIndicate the amplitude of m-th of impulse response, β=800 is indicated and damped related coefficient, ωr
Indicate the resonant frequency motivated, u (t) is a unit-step function, tmIt the time for indicating m-th of pulse generation, is defined as
Wherein, μmIt indicates slip rate, changes between 0.01~0.02, x2(t) the relevant ingredient of rotary shaft rotating speed, η are indicated
(t) indicate white Gaussian noise (signal-to-noise ratio is -12dB).x2(t) it is defined as
Wherein, AI=1,2,3Indicate amplitude, correspondingly value is 1.6,1.2 and 1, and the instantaneous axis frequency of analog signal is determined
Justice isFault characteristic frequency is set as 3.7.Sample frequency is 20kHz, sampling time 5.1s.
Experimental data is bearing outer ring failure.Bearing test-bed equipment is as shown in Figure 1.
Driving motor (SJ200-022NFU) links main shaft by shaft coupling, and major axis diameter is 1 inch.Two bearings
(ER16K) it is mounted in the bearing block on main shaft, left side rolling bearing is there are outer ring failure, the quality that a quality is 5.03kg
Disk is mounted between two bearings to provide load.The right end of drive shaft is linked on lesser pulley, and gearbox shaft link is larger
Pulley, the belt link drive shaft on two pulleys is to drive gearbox shaft, and the diameter of two pulleys is than being about 1:2.6.Tachometer
It is mounted on testing stand with acceleration transducer, respectively to measure axis revolving speed and vibration signal.For the vibration for ensuring gearbox
With amplitude advantage, accelerometer is mounted near gearbox.Then, vibration signal is collected by data collecting card and used
LABVIEW is sampled.Bearing design parameter is as shown in table 1:
1 outer ring faulty bearings parameter of table
As shown in Fig. 2, the present invention the following steps are included:
Step 1: the selection of Short Time Fourier Transform and frequency band.To Short Time Fourier Analysis is used in vibration signal, by axis
It holds signal and is divided into low-frequency range and resonance bands.Wherein, low-frequency range is about [0,400] Hz, it is sufficient to cover common axis revolving speed and its
Frequency multiplication information.
Step 2: calculating the local maximum of had amplitude and its corresponding frequency at each moment point.According to through street
Diameter optimal algorithm determines possible path optimizing between each extreme point and front and back extreme point.It is determined most according to the solution of optimization method
The corresponding extreme value of the frequency values that should be extracted at the latter moment point chooses corresponding that quickly most according to the extreme point
Shortest path, quickly optimal path is finally to extract obtained instantaneous frequency crestal line for this.
Step 3: to solve the time-frequency figure aggregation that obtains using Short Time Fourier Transform by linear transformation basic function frequency
The problem of being affected does original time-frequency figure in conjunction with fast path optimum search result using a kind of dynamic base angle degree strategy
Post-processing, i.e., the instantaneous frequency crestal line extracted using fast path optimal algorithm seek the corner cut of crestal line each point (i.e. and water
Acute angle folded by horizontal line is clockwise positive value, and equation counterclockwise is negative value), obtain frequency and the matched base of instantaneous frequency
Function.Finally, due to the ratio characterization failure class in low-frequency range and resonance bands between any two revised time-frequency crestal lines
Type carries out bearing failure diagnosis accordingly.
Further, the step specifically includes the following steps:
Step 1: to Short Time Fourier Analysis is used in vibration signal, bearing signal being divided into low-frequency range and resonance bands.
The step 2 includes:
Step 2.1: calculating the local maximum of amplitude at each moment point on time-frequency figure, obtain had width at the moment point
It is worth the corresponding frequency values of maximum
fm(τn)=f, (d [X (τn, f)]/df=0 | | d2[X(τn,f)]/df2< 0), m=1,2 ..., Np (21)
Wherein, τn(n=1,2 ..., N) indicates at the time of point, N on time-frequency figurep(τn) indicate moment point τnLocate local maximum
The number of value, fm(τn) it is moment point τnLocate the corresponding frequency values of m-th of extreme value, amplitude is denoted as SPm(τn)。
Step 2.2: accurately to extract time-frequency crestal line, proposing a kind of path optimization's algorithm to determine at each moment point should be mentioned
The amplitude maximum taken, the extreme value is corresponding, fm(τn) it is exactly the frequency values that should be finally extracted, path optimization's algorithmic notation
It is as follows
Wherein, mc(τn) determine moment point τnThe amplitude maximum that place is extracted, F [] indicates optimization method, according to the party
The solution of journey can extract optimal time-frequency crestal line.Optimization method is specifically expressed as follows
M []=perc0.5[], IQR []=perc0.75[ ]-perc0.25[ ] (26)
Wherein, fd(τn-1) indicate moment point τn-1The candidate ridge frequency values at place, Δ fdIt is fdDerivative, percp[f (t)] table
Show the pth quantile of f (t).
Step 2.3: the solution of optimization method can be expressed as
Wherein, n=1,2 ..., N, m=1,2 ..., Np(τn), q (m, τn) to determine time instant τn(n=2 ..., N)
Path optimizing should be constituted with which extreme point at previous moment point by locating each amplitude maximum, plan all possible paths.U
(m,τn) indicate to obtain the intermediate vector of optimum results, that is, select time instant τNLocate U (m, τn) the amplitude maximum point pair that is maximized
The frequency values answered are as the instantaneous frequency that should be finally extracted, accordingly by q (m, τn) determine the extreme point under path make
It is optimal for fast path, so far extract instantaneous frequency crestal line.
Generally speaking, step 2 considers all amplitude maximum in entire time domain by fast path optimal searching algorithm,
All possible path optimizing between all amplitude maximum of front and back moment point is established, ensure that the continuous of instantaneous time-frequency crestal line
Property and accuracy.
The step 3 includes:
Step 3.1: a kind of application dynamic base angle degree time-frequency figure enhancing strategy considers the angle of linear transformation base rather than window
It is long, corner cut is calculated using the instantaneous frequency crestal line extracted from original time-frequency figure by fast path optimal algorithm.It obtains
The basic function that frequency and echo signal frequency match, then post-processes time-frequency figure, to improve time-frequency figure aggregation.
Step 3.2: the ratio in low-frequency range and resonance bands between any two revised time-frequency crestal lines characterizes event
Hinder type, bearing failure diagnosis can be carried out accordingly.It is extracted from low-frequency range and corrects to obtain two time-frequency crestal line fr1And fr2, from
It is extracted in resonance bands and corrects to obtain two time-frequency crestal line fr3And fr4, ratio between any two has θ1=fr3/fr1, θ2=
fr3/fr2, θ3=fr4/fr1, θ4=fr4/fr2, they should be fault signature coefficient or its multiple.Diagnose different faults type
Bearing when calculating the mean value of its fault signature coefficient, copes with these ratios θ1, θ2, θ3And θ4Synchronization process is done, then seeks it
Mean value has θ=(η1θ1+η2θ2+η3θ3+η4θ4)/4。
Wherein, η1,η2,η3,η4Indicate the synchronization coefficient of corresponding ratio.So far, bearing fault type can be according to failure spy
Sign Coefficient Mean diagnosis obtains.
It is readable to also reduce time-frequency figure since basic function frequency is matched with target ridge frequency to a certain extent for the strategy
Property the sensibility long to window, efficiently solve basic function frequency and echo signal frequency under traditional Short Time Fourier Transform and mismatch
With the long problem for being difficult to selection of optimal window.
The schematic diagram of fast path optimal searching algorithm is as shown in Figure 3.
By analysis, the fault simulation signal results of bearing are as shown in figure 4, Fig. 4 (a) indicates that low-frequency range is extracted as a result, Fig. 4
(b) indicate that resonance bands extract result.Then average relative error is introduced, the instantaneous frequency error of two kinds of crestal line extraction algorithms is such as
Shown in table 2.
2 analog signal local peaking of table and fast path optimum search result compare
Compared with extracting result with local peaking's searching algorithm, it is apparent that the fluctuation and ridge of instantaneous frequency crestal line entirety
The mutation of line latter half makes moderate progress.List data is analyzed it is found that no matter for low-frequency range or resonance bands, fast path is most
It is higher that excellent searching algorithm compares local peaking's searching algorithm accuracy, and highest can make instantaneous frequency crestal line accuracy improve 74%.
For the validity for verifying dynamic base angle degree strategy, the wink extracted using fast path optimal searching algorithm is calculated
When frequency crestal line each moment point corner cut, as shown in figure 5, simultaneously accordingly choose obtain analog signal low-frequency range and resonance bands
Time-frequency figure.To verify the long influence to time-frequency figure aggregation of window under the strategy, this chapter will convert window length and utilize fast path most
Excellent searching algorithm extracts instantaneous frequency crestal line from the long lower time-frequency figure of different window, by the average relative error of analysis crestal line come
Verifying.
For low-frequency range, Fig. 6 (a) and 6 (c) is obtained when respectively indicating window a length of 1200 and 2161 using dynamic base angle degree strategy
The time-frequency figure arrived, Fig. 6 (c) and 6 (d) respectively indicate the instantaneous frequency therefrom extracted using fast path optimal searching algorithm
Crestal line, with shown in Fig. 4 (a) from the time-frequency figure that Short Time Fourier Transform obtains using fast path optimal searching algorithm extract
Obtained instantaneous frequency crestal line is compared, it can be seen that crestal line latter half is without mutation.Utilize fast path optimal searching algorithm
The average relative error for extracting obtained instantaneous frequency crestal line is all reduced, and highest can reduce by 52% error, demonstrate this
The validity of crestal line extraction algorithm.
Then, using the Fault Diagnosis Strategy based on fault signature Coefficient Mean, fast path optimal searching algorithm is utilized
It is as shown in table 3 from the ratio two-by-two between the instantaneous frequency crestal line extracted in the time-frequency figure under dynamic base angle degree strategy.It asks
Relative error of its mean value between 3.684, with setting value 3.7 is 0.4%, and diagnostic result further demonstrates fast path most
The validity of excellent searching algorithm and dynamic base angle degree strategy.
The ratio between instantaneous frequency crestal line two-by-two in 3 analog signal low-frequency range of table and resonance bands
Multi-source instantaneous frequency crestal line is extracted from low-frequency range and resonance bands using fast path optimal searching algorithm, is extracted
As a result respectively as shown in Fig. 8 (a) and 8 (b).Curve entirety fluctuation is reduced compared with local peaking's searching algorithm extracts result,
The catastrophe of the first vallate line front end is greatly improved especially in low-frequency range.Average relative error is introduced, table 4 indicates two kinds
The average relative error of instantaneous frequency crestal line under ridge search algorithm.Data in analytical table using fast path is optimal it is found that search
The average relative error for the instantaneous frequency crestal line that rope algorithm extracts all is reduced, and highest can reduce by 52% error, is tested
The validity of the crestal line extraction algorithm is demonstrate,proved.
4 bearing outer ring malfunction test signal local peaking of table and fast path optimum search result compare
Fig. 9 indicates the instantaneous frequency crestal line extracted using fast path optimal searching algorithm at each moment point
Corner cut obtains outer ring faulty bearings signal low-frequency range and resonance bands time-frequency figure accordingly.Using dynamic when window a length of 4590 and 5815
State base angle degree strategy obtains low-frequency range time-frequency figure, and such as Figure 10 (a) and 10 (c) is shown respectively, utilizes fast path optimal searching algorithm
The instantaneous frequency crestal line therefrom extracted as shown in Figure 10 (c) and 10 (d), extracts result with local peaking's searching algorithm respectively
It compares, crestal line front end, which is mutated, to be improved.Crestal line extracts accuracy and improves 35% or more, demonstrates dynamic base angle degree strategy phase
Time-frequency figure aggregation can be effectively improved compared with traditional Short Time Fourier Transform and instantaneous frequency crestal line extracts accuracy.
Seek the instantaneous frequency crestal line extracted under dynamic base angle degree strategy by fast path optimal searching algorithm two-by-two
Ratio, the results are shown in Table 5.Using the bearing failure diagnosis strategy set forth above based on fault signature Coefficient Mean, obtain
Fault signature coefficient is 3.533, is 0.9% with true value relative error.Further analysis table is it is found that each coefficient is closer
It is more accurate to show that crestal line extracts, effectively demonstrates fast path optimal searching algorithm and dynamic base angle degree strategy for true value
Feasibility.
The ratio between instantaneous frequency crestal line two-by-two in 5 bearing outer ring fault-signal low-frequency range of table and resonance bands
Local peaking is chosen to the dependence of initial time point and by experience for local peaking's searching algorithm and " one-step method "
The confinement problems of searching algorithm search range, present invention application fast path optimal searching algorithm are optimal according to fast path
Possible fast path is optimal between all amplitude maximum points of moment point before and after algorithmic rule, then therefrom chooses optimal solution conduct
Final extracting ridges ensure that accuracy and continuity that instantaneous frequency crestal line extracts.For Fourier becomes in short-term under variable speed
The frequency for changing basic function is difficult to match with the instantaneous frequency of time-varying and window long the problem of being affected to time-frequency figure aggregation, this
Dynamic base angle degree strategy is applied in invention, and it is each to seek crestal line for the instantaneous frequency crestal line extracted using fast path optimal algorithm
The corner cut of point obtains the basic function that frequency and echo signal frequency match, then carries out time-frequency conversion, and then obtain entire time domain
On time-frequency figure.On this basis building analog signal carry out analysis verifying, the results showed that fast path optimal searching algorithm and
Dynamic base angle degree strategy can effectively improve the accuracy of instantaneous frequency crestal line extraction.Utilize examining based on fault signature Coefficient Mean
Disconnected strategy carries out fault diagnosis to bearing based on obtained multi-source instantaneous frequency crestal line is extracted, and diagnostic result further demonstrates
The validity of fast path optimal searching algorithm and dynamic base angle degree strategy.The angle for considering linear transformation base, utilizes through street
The corner cut guidance of diameter optimum search result obtains analog signal low-frequency range and resonance bands time-frequency figure after optimization, again using fast
The instantaneous frequency crestal line error that fast path optimal searching algorithm therefrom extracts can reduce by 79%.Fast path optimal searching algorithm
There is centainly effective in terms of improving instantaneous frequency crestal line and extracting with bearing failure diagnosis accuracy with dynamic base angle degree strategy
Property.
Embodiment described above is only to absolutely prove preferred embodiment that is of the invention and being lifted, protection model of the invention
It encloses without being limited thereto.Those skilled in the art's made equivalent substitute or transformation on the basis of the present invention, in the present invention
Protection scope within.Protection scope of the present invention is subject to claims.
Claims (7)
1. a kind of multi-source time-frequency crestal line extracting method based on fast path optimum search and dynamic base angle degree, which is characterized in that
Include:
Step 1: the selection of Short Time Fourier Transform and frequency band.To Short Time Fourier Analysis is used in vibration signal, bearing is believed
Number it is divided into low-frequency range and resonance bands.
Step 2: calculating the local maximum of had amplitude and its corresponding frequency at each moment point.Most according to fast path
Excellent algorithm determines possible path optimizing between each extreme point and front and back extreme point.Last is determined according to the solution of optimization method
The corresponding extreme value of the frequency values that should be extracted at a moment point chooses that corresponding quick optimal road according to the extreme point
Diameter, quickly optimal path is finally to extract obtained instantaneous frequency crestal line for this；
Step 3: a kind of dynamic base angle degree strategy is applied, after doing in conjunction with fast path optimum search result to original time-frequency figure
Reason, i.e., the instantaneous frequency crestal line extracted using fast path optimal algorithm seeks the corner cut of crestal line each point, obtain frequency with
The matched basic function of instantaneous frequency；Finally, due in low-frequency range and resonance bands between any two revised time-frequency crestal lines
Ratio characterization failure type, carry out bearing failure diagnosis accordingly.
2. the multi-source time-frequency crestal line extraction side based on fast path optimum search and dynamic base angle degree as described in claim 1
Method, which is characterized in that step 2 specifically includes:
Step 2.1: calculating the local maximum of amplitude at each moment point on time-frequency figure, obtain had amplitude pole at the moment point
It is worth corresponding frequency values greatly
fm(τn)=f, (d [X (τn, f)]/df=0 | | d2[X(τn,f)]/df2< 0), m=1,2 ..., Np (1)
Wherein, τn(n=1,2 ..., N) indicates at the time of point, N on time-frequency figurep(τn) indicate moment point τnLocate local maximum
Number, fm(τn) it is moment point τnLocate the corresponding frequency values of m-th of extreme value, amplitude is denoted as SPm(τn)。
Step 2.2: accurately to extract time-frequency crestal line, proposing that a kind of path optimization's algorithm determines should be extracted at each moment point
Amplitude maximum, the extreme value is corresponding, fm(τn) it is exactly the frequency values that should be finally extracted, path optimization's algorithmic notation is as follows:
Wherein, mc(τn) determine moment point τnThe amplitude maximum that place is extracted, F [] indicates optimization method, according to the equation
Solution can extract optimal time-frequency crestal line.Optimization method is specifically expressed as follows:
M []=perc0.5[], IQR []=perc0.75[]-perc0.25[] (6)
Wherein, fd(τn-1) indicate moment point τn-1The candidate ridge frequency values at place, Δ fdIt is fdDerivative, percp[f (t)] indicates f
(t) pth quantile.
Step 2.3: the solution of optimization method can be expressed as
Wherein, n=1,2 ..., N, m=1,2 ..., Np(τn), q (m, τn) to determine time instant τnIt is each at (n=2 ..., N)
Amplitude maximum should constitute path optimizing with which extreme point at previous moment point, plan all possible paths.U(m,τn) table
Show the intermediate vector for obtaining optimum results, i.e. selection time instant τNLocate U (m, τn) the corresponding frequency of amplitude maximum point that is maximized
Value is as the instantaneous frequency that should be finally extracted, accordingly by q (m, τn) determine the extreme point under path as through street
Diameter is optimal, so far extracts instantaneous frequency crestal line.
3. the multi-source time-frequency crestal line extraction side based on fast path optimum search and dynamic base angle degree as described in claim 1
Method, which is characterized in that the step 3 includes:
Step 3.1: application a kind of dynamic base angle degree time-frequency figure enhancing strategy, consider linear transformation base angle rather than window is long, benefit
Corner cut is calculated with the instantaneous frequency crestal line extracted from original time-frequency figure by fast path optimal algorithm.Obtain frequency with
The basic function that echo signal frequency matches then post-processes time-frequency figure, to improve time-frequency figure aggregation.
Step 3.2: the ratio in low-frequency range and resonance bands between any two revised time-frequency crestal lines characterizes failure classes
Type can carry out bearing failure diagnosis accordingly.It is extracted from low-frequency range and corrects to obtain two time-frequency crestal line fr1And fr2, from resonance
It is extracted in frequency band and corrects to obtain two time-frequency crestal line fr3And fr4, ratio between any two has θ1=fr3/fr1, θ2=fr3/
fr2, θ3=fr4/fr1, θ4=fr4/fr2, they should be fault signature coefficient or its multiple.Diagnose the axis of different faults type
It holds, when calculating the mean value of its fault signature coefficient, copes with these ratios θ1, θ2, θ3And θ4Synchronization process is done, then seeks it
Value, there is θ=(η1θ1+η2θ2+η3θ3+η4θ4)/4；
Wherein, η1,η2,η3,η4Indicate the synchronization coefficient of corresponding ratio.So far, bearing fault type can be according to fault signature system
Number mean value diagnosis obtains.
4. the multi-source time-frequency crestal line extraction side based on fast path optimum search and dynamic base angle degree as described in claim 1
Method, which is characterized in that in step 1, low-frequency range is [0,400] Hz.
5. a kind of computer equipment including memory, processor and stores the meter that can be run on a memory and on a processor
Calculation machine program, which is characterized in that the processor realizes any one of claims 1 to 4 the method when executing described program
Step.
6. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the program is held by processor
The step of any one of claims 1 to 4 the method is realized when row.
7. a kind of processor, which is characterized in that the processor is for running program, wherein right of execution when described program is run
Benefit requires 1 to 4 described in any item methods.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2019104965868 | 2019-06-10 | ||
| CN201910496586.8A CN110174270A (en) | 2019-06-10 | 2019-06-10 | Multi-source time-frequency crestal line extracting method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110243605A true CN110243605A (en) | 2019-09-17 |
| CN110243605B CN110243605B (en) | 2020-04-10 |
Family
ID=67698030
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910496586.8A Pending CN110174270A (en) | 2019-06-10 | 2019-06-10 | Multi-source time-frequency crestal line extracting method |
| CN201910674416.4A Active CN110243605B (en) | 2019-06-10 | 2019-07-25 | Multi-source time-frequency ridge line extraction method |
Family Applications Before (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910496586.8A Pending CN110174270A (en) | 2019-06-10 | 2019-06-10 | Multi-source time-frequency crestal line extracting method |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN110174270A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111665469A (en) * | 2020-06-11 | 2020-09-15 | 浙江大学 | Underwater multipath signal parameter estimation method based on space time-frequency distribution |
| CN111797789A (en) * | 2020-07-10 | 2020-10-20 | 合肥工业大学 | Time-frequency ridge line extraction method and device for fault feature interpretation |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110907162B (en) * | 2019-12-13 | 2021-10-15 | 北京天泽智云科技有限公司 | Rotating machinery fault feature extraction method without tachometer under variable rotating speed |
| CN111458122B (en) * | 2020-04-08 | 2022-03-29 | 苏州大学 | Rotary machine fault diagnosis method based on matching enhancement time-frequency representation |
| CN111879508B (en) * | 2020-07-28 | 2022-06-10 | 无锡迈斯德智能测控技术有限公司 | Method and device for estimating instantaneous rotating speed of rotating machine based on time-frequency transformation and storage medium |
| CN113271161B (en) * | 2021-05-17 | 2022-06-03 | 湖南坤雷科技有限公司 | Method and device for screening electromagnetic signals |
| CN114492539B (en) * | 2022-02-21 | 2023-04-28 | 西南交通大学 | Bearing fault detection method and device, electronic equipment and storage medium |
| CN116380467B (en) * | 2023-05-24 | 2024-01-23 | 成都工业职业技术学院 | Rolling bearing fault diagnosis method based on multi-time-frequency ridge line extraction |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101666677A (en) * | 2009-09-25 | 2010-03-10 | 北京工业大学 | Method for extracting feature information of weak faults of low-speed heavy-duty equipment |
| CN102607845A (en) * | 2012-03-05 | 2012-07-25 | 北京工业大学 | Bearing fault characteristic extracting method for redundantly lifting wavelet transform based on self-adaptive fitting |
| CN105241666A (en) * | 2015-09-21 | 2016-01-13 | 华南理工大学 | Rolling bearing fault feature extraction method based on signal sparse representation theory |
| CN105547698A (en) * | 2015-12-31 | 2016-05-04 | 新疆金风科技股份有限公司 | Fault diagnosis method and apparatus for rolling bearing |
| CN107525674A (en) * | 2017-05-27 | 2017-12-29 | 苏州大学 | Frequency method of estimation and detection means are turned based on crestal line probability distribution and localised waving |
-
2019
- 2019-06-10 CN CN201910496586.8A patent/CN110174270A/en active Pending
- 2019-07-25 CN CN201910674416.4A patent/CN110243605B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101666677A (en) * | 2009-09-25 | 2010-03-10 | 北京工业大学 | Method for extracting feature information of weak faults of low-speed heavy-duty equipment |
| CN102607845A (en) * | 2012-03-05 | 2012-07-25 | 北京工业大学 | Bearing fault characteristic extracting method for redundantly lifting wavelet transform based on self-adaptive fitting |
| CN105241666A (en) * | 2015-09-21 | 2016-01-13 | 华南理工大学 | Rolling bearing fault feature extraction method based on signal sparse representation theory |
| CN105547698A (en) * | 2015-12-31 | 2016-05-04 | 新疆金风科技股份有限公司 | Fault diagnosis method and apparatus for rolling bearing |
| CN107525674A (en) * | 2017-05-27 | 2017-12-29 | 苏州大学 | Frequency method of estimation and detection means are turned based on crestal line probability distribution and localised waving |
Non-Patent Citations (1)
| Title |
|---|
| 姜万录 等: "基于小波脊线的滚动轴承故障诊断方法", 《振动与冲击》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111665469A (en) * | 2020-06-11 | 2020-09-15 | 浙江大学 | Underwater multipath signal parameter estimation method based on space time-frequency distribution |
| CN111665469B (en) * | 2020-06-11 | 2022-08-23 | 浙江大学 | Underwater multipath signal parameter estimation method based on spatial time-frequency distribution |
| CN111797789A (en) * | 2020-07-10 | 2020-10-20 | 合肥工业大学 | Time-frequency ridge line extraction method and device for fault feature interpretation |
| CN111797789B (en) * | 2020-07-10 | 2024-02-06 | 合肥工业大学 | Time-frequency ridge line extraction method and device for fault feature interpretation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110174270A (en) | 2019-08-27 |
| CN110243605B (en) | 2020-04-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110243605A (en) | Multi-source time-frequency crestal line extracting method | |
| Antoni et al. | Feedback on the Surveillance 8 challenge: Vibration-based diagnosis of a Safran aircraft engine | |
| CN110567574B (en) | Method and system for identifying timing vibration parameters of blade end of rotating blade | |
| Shi et al. | Bearing fault diagnosis under variable rotational speed via the joint application of windowed fractal dimension transform and generalized demodulation: A method free from prefiltering and resampling | |
| He et al. | Fault feature extraction of rolling element bearings using sparse representation | |
| Huang et al. | A method for tachometer-free and resampling-free bearing fault diagnostics under time-varying speed conditions | |
| Cong et al. | Research of singular value decomposition based on slip matrix for rolling bearing fault diagnosis | |
| CN109855874B (en) | Random resonance filter for enhancing detection of weak signals in vibration assisted by sound | |
| Wang et al. | Sparse and low-rank decomposition of the time–frequency representation for bearing fault diagnosis under variable speed conditions | |
| JP6906985B2 (en) | Vibration diagnosis system, vibration diagnosis method and parameter setting method | |
| CN109932179A (en) | A kind of rolling bearing fault testing method based on the reconstruct of DS Adaptive spectra | |
| KR20200075148A (en) | AI system and pre-conditioning method in use with noise data for detecting noise source | |
| Xu et al. | Generalized S-synchroextracting transform for fault diagnosis in rolling bearing | |
| Lin et al. | A review and strategy for the diagnosis of speed-varying machinery | |
| Dziedziech et al. | A novel method for speed recovery from vibration signal under highly non-stationary conditions | |
| Alsalaet et al. | Bearing fault diagnosis using normalized diagnostic feature-gram and convolutional neural network | |
| CN111487318B (en) | Time-varying structure instantaneous frequency extraction method | |
| CN115931318A (en) | Fault intelligent diagnosis method, device, equipment and storage medium | |
| CN114689298A (en) | Fault detection method and device for walking part of coal mining machine | |
| Ong et al. | An inconsistent phase selection assessment for harmonic peaks elimination in operational modal testing | |
| CN105987809A (en) | Centrifugal-compressor semi-open-type impeller crack detection method based on random resonance | |
| CN111307426A (en) | Rotating machinery fault feature extraction method based on FrFT-EWT principle | |
| Song et al. | Multispectral balanced automatic fault diagnosis for rolling bearings under variable speed conditions | |
| CN116358873B (en) | Rolling bearing fault diagnosis method, device, equipment and medium | |
| CN112049787B (en) | Non-contact water pump fault detection method and 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 |