CN108229382A - Vibration signal characteristics extracting method, device, storage medium and computer equipment - Google Patents
Vibration signal characteristics extracting method, device, storage medium and computer equipment Download PDFInfo
- Publication number
- CN108229382A CN108229382A CN201711480593.6A CN201711480593A CN108229382A CN 108229382 A CN108229382 A CN 108229382A CN 201711480593 A CN201711480593 A CN 201711480593A CN 108229382 A CN108229382 A CN 108229382A
- Authority
- CN
- China
- Prior art keywords
- signal
- noising
- obtains
- intrinsic mode
- mode function
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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
- G06F2218/04—Denoising
-
- 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
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)
- Complex Calculations (AREA)
Abstract
The present invention relates to a kind of vibration signal characteristics extracting method, device, storage medium and computer equipments.Vibration signal characteristics extracting method includes：Acquire the vibration signal of target device；Noise reduction process is carried out to vibration signal, obtains de-noising signal；Empirical mode decomposition is carried out to de-noising signal, obtains the original intrinsic mode function corresponding to de-noising signal；Original intrinsic mode function according to corresponding to de-noising signal obtains superposed signal, and empirical mode decomposition and removal overlap-add procedure are carried out to superposed signal, obtains the final intrinsic mode function corresponding to de-noising signal；Hilbert-Huang transform is carried out to final intrinsic mode function, obtains the spectrum signature of vibration signal.This method can improve the accuracy of feature extraction, and the feature of extraction can more accurately reflect the operating status of target device.
Description
Technical field
The present invention relates to signal analysis technology fields, more particularly to a kind of vibration signal characteristics extracting method, device, deposit
Storage media and computer equipment.
Background technology
Vibration signal is signal caused by the mechanical oscillation of equipment, and the feature of vibration signal can be transported with the machinery of embodiment device
Emotionally condition, therefore the feature for extracting vibration signal carries out analysis and helps to understand the operating status of equipment, finds that failure is hidden as early as possible
Suffer from.For example, load ratio bridging switch, as movable member unique in power transformer, operating status is directly related to entire electricity
The safe and reliable operation of net；The feature of load ratio bridging switch vibration signal is extracted, can accurately analyze the fortune of load ratio bridging switch
Row state, so as to find the potential faults of power grid as early as possible.
The mode of traditional extraction vibration signal characteristics, is using EMD (Empirical Mode mostly
Decomposition empirical mode decompositions) algorithm, decomposes vibration signal and extracts spectrum signature, so as to carry out equipment
The judgement of operation conditions.However, there are more serious modal overlap phenomenons for traditional EMD algorithms, influence final feature and carry
It takes as a result, feature extraction accuracy is low.
Invention content
Based on this, it is necessary to extract the problem of accuracy is low for traditional vibration signal characteristics, providing one kind can improve
Vibration signal characteristics extracting method, device, storage medium and the computer equipment of feature extraction accuracy.
A kind of vibration signal characteristics extracting method, including：
Acquire the vibration signal of target device；
Noise reduction process is carried out to the vibration signal, obtains de-noising signal；
Empirical mode decomposition is carried out to the de-noising signal, obtains the original intrinsic mode letter corresponding to the de-noising signal
Number；
Original intrinsic mode function according to corresponding to the de-noising signal obtains superposed signal, to the superposed signal into
Row empirical mode decomposition and removal overlap-add procedure, obtain the final intrinsic mode function corresponding to the de-noising signal；
Hilbert-Huang transform is carried out to the final intrinsic mode function, obtains the spectrum signature of the vibration signal.
A kind of vibration signal characteristics extraction element, including：
Signal acquisition module, for acquiring the vibration signal of target device；
Signal denoising module for carrying out noise reduction process to the vibration signal, obtains de-noising signal；
For carrying out empirical mode decomposition to the de-noising signal, it is right to obtain the de-noising signal institute for initial decomposition module
The original intrinsic mode function answered；
Decomposing module is superimposed, superposition letter is obtained for the original intrinsic mode function according to corresponding to the de-noising signal
Number, empirical mode decomposition and removal overlap-add procedure are carried out to the superposed signal, is obtained final corresponding to the de-noising signal
Intrinsic mode function；
Characteristic extracting module for carrying out Hilbert-Huang transform to the final intrinsic mode function, obtains described shake
The spectrum signature of dynamic signal.
Above-mentioned vibration signal characteristics extracting method and device carry out noise reduction process by the vibration signal to acquisition and are dropped
Noise cancellation signal carries out empirical mode decomposition to de-noising signal and obtains the original intrinsic mode function corresponding to de-noising signal, further according to
Original intrinsic mode function corresponding to de-noising signal obtains superposed signal, and empirical mode decomposition and removal are carried out to superposed signal
Overlap-add procedure obtains the final intrinsic mode function corresponding to de-noising signal, to final intrinsic mode function carry out Hilbert-
Huang obtains the spectrum signature of vibration signal.On the one hand, noise reduction is carried out to vibration signal before empirical mode decomposition, it can
Avoid the effect of influence of noise feature extraction；On the other hand, superposed signal is obtained after obtaining original intrinsic mode function, further according to
Superposed signal carries out empirical mode decomposition and removal overlap-add procedure, can limit the frequency bandwidth of final intrinsic mode function, energy
The modal overlap phenomenon present in empirical mode decomposition is effectively solved, so as to improve the accuracy of feature extraction, the spy of extraction
Sign can more accurately reflect the operating status of target device.
A kind of storage medium, is stored with computer program, is realized when the computer program of storage is executed by processor above-mentioned
The step of vibration signal characteristics extracting method.
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 above-mentioned vibration signal characteristics extracting method when performing the computer program.
Above-mentioned storage medium and computer equipment, due to the step of realizing above-mentioned vibration signal characteristics extracting method, together
Reason can improve the accuracy of feature extraction.
Description of the drawings
Fig. 1 is the flow chart of vibration signal characteristics extracting method in one embodiment；
Fig. 2 is the flow chart of vibration signal characteristics extracting method in second embodiment；
Fig. 3 is the flow chart of vibration signal characteristics extracting method in third embodiment；
Fig. 4 is the flow chart of vibration signal characteristics extracting method in the 4th embodiment；
Fig. 5 is the structure chart of vibration signal characteristics extraction element in an embodiment；
Fig. 6 is the de-noising signal under normal condition in application examples；
De-noising signal when Fig. 7 is contact slap；
Fig. 8 is the Hilbert marginal spectrums under normal condition；
Hilbert marginal spectrums when Fig. 9 is contact slap.
Specific embodiment
With reference to figure 1, the vibration signal characteristics extracting method in an embodiment includes the following steps：
S110：Acquire the vibration signal of target device.
Target device is to generate vibration signal, need according to the equipment of analysis of vibration signal operating status.For example, target
Equipment can be the load ratio bridging switch in power transformer.
S130：Noise reduction process is carried out to vibration signal, obtains de-noising signal.
The vibration signal of collection in worksite usually contains noise, can effect characteristics extraction effect, by vibration signal into
Row noise reduction process, obtained de-noising signal are the signal after denoising.
S150：Empirical mode decomposition is carried out to de-noising signal, obtains the original intrinsic mode function corresponding to de-noising signal.
The vibration signal of acquisition typically exhibits out non-stationary.Empirical mode decomposition is self-adapting signal time frequency processing side
Method, suitable for the analyzing and processing of nonlinear and nonstationary signal.Empirical mode decomposition is carried out to signal, can be this by signal decomposition
Levy mode function (abbreviation IMF, Intrinsic Mode Function)；Original intrinsic mode function corresponding to de-noising signal
The obtained intrinsic mode function of empirical mode decomposition as is carried out to de-noising signal.Specifically, the quantity of intrinsic mode function
Can have multiple.
S170：Original intrinsic mode function according to corresponding to de-noising signal obtains superposed signal, and superposed signal is carried out
Empirical mode decomposition and removal overlap-add procedure, obtain the final intrinsic mode function corresponding to de-noising signal.
Superposed signal is to be superimposed with the signal of the original intrinsic mode function corresponding to de-noising signal.Overlap-add procedure is gone to remove
The processing of Signal averaging.
S190：Hilbert-Huang transform is carried out to final intrinsic mode function, obtains the spectrum signature of vibration signal.
The spectrum signature of vibration signal, frequency spectrum are obtained by carrying out Hilbert-Huang transform to final intrinsic mode function
Feature can be used for analyzing and determining the operating status of target device, to understand the potential faults of target device in time.
Above-mentioned vibration signal characteristics extracting method carries out noise reduction process by the vibration signal to acquisition and obtains noise reduction letter
Number, empirical mode decomposition is carried out to de-noising signal and obtains the original intrinsic mode function corresponding to de-noising signal, further according to noise reduction
Original intrinsic mode function corresponding to signal obtains superposed signal, and empirical mode decomposition is carried out to superposed signal and removal is superimposed
Processing, obtains the final intrinsic mode function corresponding to de-noising signal, and Hilbert-xanthochromia is carried out to final intrinsic mode function
It changes, obtains the spectrum signature of vibration signal.On the one hand, noise reduction is carried out to vibration signal before empirical mode decomposition, can avoided
The effect of influence of noise feature extraction；On the other hand, superposed signal is obtained after obtaining original intrinsic mode function, further according to superposition
Signal carries out empirical mode decomposition and removal overlap-add procedure, can limit the frequency bandwidth of final intrinsic mode function, can be effective
The modal overlap phenomenon present in empirical mode decomposition is solved, so as to improve the accuracy of feature extraction, the feature of extraction is more
The operating status of target device can be accurately reflected.
In one embodiment, target device is load ratio bridging switch.In the present embodiment, step S110 includes：Pass through peace
The vibration signal of load ratio bridging switch in normal state and each is acquired respectively mounted in the acceleration transducer of load ratio bridging switch
Vibration signal under malfunction.
The working condition of load ratio bridging switch includes normal condition (i.e. non-faulting state) and malfunction, and malfunction
Type there are many.Shaking under the vibration signal and each malfunction under normal condition by acquiring load ratio bridging switch respectively
Dynamic signal, step S130 to step S190 is performed based on each vibration signal respectively, extractable to obtain vibration signal under normal condition
Spectrum signature and each malfunction under vibration signal spectrum signature, the more effective different operations for distinguishing load ratio bridging switch
State.
In one embodiment, with reference to figure 2 or Fig. 4, step S130 includes S131 and step S133.
S131：Wavelet-packet noise reduction is carried out to vibration signal using the Optimum Wavelet Packet and Decomposition order deposited, is obtained small
Wave packet de-noising signal.
S133：Singular value decomposition is carried out to wavelet-packet noise reduction signal, obtains de-noising signal.
Wavelet-packet noise reduction and singular value decomposition (abbreviation SVD, Singularly Valuable Decomposition) are two
The different noise reduction process mode of kind.Vibration signal is unable to reach preferably using individual wavelet-packet noise reduction, singular value decomposition
Denoising effect and the feature for retaining vibration signal.By the way that wavelet-packet noise reduction and singular value decomposition are combined, first using wavelet packet
Noise reduction to vibration signal carry out a noise reduction, then the wavelet-packet noise reduction signal to being obtained after a noise reduction using singular value decomposition into
The secondary noise reduction of row, can reach preferable denoising effect, and can retain the feature of original vibration signal, so as to improve feature indirectly
The accuracy of extraction.
Optionally, Optimum Wavelet Packet preparation process and Decomposition order preparation process can also be included before step S131.
Wherein, Optimum Wavelet Packet preparation process includes：According to the entropy function for giving a sequence, in all alternative wavelet packet basis
Searching makes the wavelet packet basis of entropy function minimum, as Optimum Wavelet Packet and stores.Decomposition order preparation process includes：To given
Master sample carry out the wavelet decompositions of a variety of Decomposition orders and obtain decomposition result, more each decomposition result simultaneously chooses optimal point
Solve the Decomposition order as a result, corresponding to the decomposition result of optimal storage.Specifically, the comparison of decomposition result can be based on row
The good and bad judgment criteria of wavelet decomposition is carried out in the industry.
Specifically, step S131 includes：Vibration signal is carried out using the Optimum Wavelet Packet and Decomposition order deposited small
Wave Decomposition obtains the coefficient of multiple high fdrequency components and each high fdrequency component；Threshold is carried out to the coefficient of high fdrequency component using predetermined threshold value
It is worth quantification treatment；Wavelet package reconstruction is carried out to threshold value quantizing treated signal, obtains wavelet-packet noise reduction signal.
Vibration signal is carried out can obtain after wavelet decomposition the high fdrequency component divided using preset frequency for separation with
Low frequency component, each low frequency component and each high fdrequency component are corresponding, and there are one coefficients.Predetermined threshold value can be specific according to actual needs
Setting；Threshold value quantizing processing is carried out to the coefficient of high fdrequency component using predetermined threshold value, can be specifically each high fdrequency component of comparison
The size of coefficient and predetermined threshold value remains larger than the coefficient of predetermined threshold value using hard threshold method, will be less than or equal to predetermined threshold value
High fdrequency component coefficient zero setting；Threshold value quantizing processing or ratio are carried out to the coefficient of high fdrequency component using predetermined threshold value
The coefficient of more each high fdrequency component and the size of predetermined threshold value using Soft thresholding by less than the coefficient zero setting of predetermined threshold value, are incited somebody to action big
It does and shrinks to zero in the coefficient of the high fdrequency component of predetermined threshold value.
Specifically, step S133 includes：Phase space reconfiguration is carried out to wavelet-packet noise reduction signal and obtains the rail of characterization attractor
Road matrix；Singular value decomposition is carried out to track matrix and obtains de-noising signal.
Continuous wavelet transform can reflect kinetic characteristics, and Smooth Systems signal, random noise signal are to continuous wavelet transform track
Singular values of a matrix has Different Effects.By carrying out singular value decomposition to track matrix, reduced using the characteristic of singular spectrum small
Noise in wave packet de-noising signal is so as to obtain the de-noising signal of secondary noise reduction.
In one embodiment, step S150 includes step (A) to step (D).
Step (A)：Calculate all Local Extremums of de-noising signal x (t).Wherein, Local Extremum includes maximum
Point and minimum point.
Step (B)：Coenvelope line and all minimum point structures that all maximum points are formed are obtained according to Local Extremum
Into lower envelope line, be denoted as u respectively0(t) and v0(t)。
Step (C)：The mean value m of coenvelope line and lower envelope line is obtained0, and remember that signal difference is h0：
m0=[u0(t)+v0(t)]/2；
h0=x (t)-m0；
Step (D)：Judge signal difference h0Whether first condition and second condition are met.First condition：Signal difference h0In zero
Points are equal with pole number or at most differ 1；Second condition：Signal difference h0Upper any point, is determined by Local modulus maxima
Envelope and the mean value of envelope determined by local minizing point are 0, i.e. signal difference h0About time shaft Local Symmetric.If
Meet first condition and second condition, then by signal difference h0As the original intrinsic mode function (IMF) corresponding to de-noising signal；
Otherwise, note signal difference h0For x (t), i.e., by signal difference h0As new de-noising signal, step (A) to step (D) is repeated, until
To the original intrinsic mode function corresponding to a de-noising signal, it is denoted as F1(t)。
In one embodiment, with reference to figure 3 or Fig. 4, step S170 includes step S171 to step S177.
S171：Original intrinsic mode function according to corresponding to default mask coefficient and de-noising signal obtains corresponding mask
Signal.
Default mask coefficient is according to the pre-set value of actual needs.Specifically, it presets mask coefficient and is more than 1.
S173：De-noising signal is added to obtain superposed signal with corresponding mask signal.
S175：Empirical mode decomposition is carried out to superposed signal, obtains the intrinsic mode function corresponding to superposed signal.
Intrinsic mode function corresponding to superposed signal is that the eigen mode that empirical mode decomposition obtains is carried out to superposed signal
State function.Specifically, the detailed step of empirical mode decomposition is carried out to superposed signal with carrying out empirical modal point to de-noising signal
The detailed step of solution is identical, and the object of empirical mode decomposition only is changed to superposed signal.
S177：The mask signal included in the intrinsic mode function corresponding to superposed signal is removed, obtains de-noising signal institute
Corresponding final intrinsic mode function.
Wherein, the mask signal included in the intrinsic mode function corresponding to superposed signal, i.e., according to this superposition letter
The mask signal that de-noising signal corresponding to number obtains.For example, the mask signal that meter step S171 is obtained is s (t), superposed signal
The mask signal included in corresponding intrinsic mode function is similarly s (t).
Empirical mode decomposition is carried out to signal to can be understood as signal passing through EMD wave filters.Believed by obtaining mask
Number and mask signal is added with de-noising signal to obtain superposed signal and empirical mode decomposition is carried out to superposed signal, can be changed
The centre frequency of EMD wave filters inhibits the ingredient of low frequency IMF components to be mixed into high-frequency I MF components, realizes the frequency to IMF
Bandwidth is limited, and can effectively solve the problems, such as modal overlap.
In one embodiment, step S171 includes step (a1) and step (a2).
Step (a1)：Hilbert transform is carried out to the original intrinsic mode function corresponding to de-noising signal, is obtained instantaneous
Frequency and instantaneous amplitude.
Step (a2)：Mask signal frequency, and base are calculated according to instantaneous frequency, instantaneous amplitude and default mask coefficient
Mask signal is generated in mask signal frequency.
Specifically, step (a2) includes：
S (t)=A0sin(2πft)；
In formula, K is to preset mask coefficient, and K ＞ 1；a1(i) and f1(i) it is respectively that i-th of instantaneous amplitude and i-th are instantaneous
Frequency；F is mask signal frequency；S (t) is mask signal；A0For intermediate parameters, can be taken as original corresponding to de-noising signal
1.6 times of the average amplitude of intrinsic mode function.
In one embodiment, step S177 includes：
Wherein,For the intrinsic mode function corresponding to superposed signal, s (t) is mask signal, Fi' (t) believes for noise reduction
Final intrinsic mode function corresponding to number.
In one embodiment, with reference to figure 4, step S177 further includes step S178 and step S179 later.
S178：Judge whether the quantity of final intrinsic mode function reaches preset quantity.
If it is not, representing that the quantity of final intrinsic mode function is less than preset quantity, then step S179 is performed；If so, it holds
Row step S190.
S179：De-noising signal is subtracted into corresponding final intrinsic mode function and obtains next de-noising signal and as new
De-noising signal, return to step S150.
Return to step S150 after new de-noising signal is obtained, specifically performs step：Experience is carried out to new de-noising signal
Mode decomposition obtains the original intrinsic mode function corresponding to new de-noising signal；According to the original corresponding to new de-noising signal
Beginning intrinsic mode function obtains new superposed signal, and empirical mode decomposition and removal overlap-add procedure are carried out to new superposed signal,
Obtain the final intrinsic mode function corresponding to new de-noising signal.
When being less than preset quantity by the quantity in final intrinsic mode function, new de-noising signal and return to step are obtained
S150 repeats the quantity for obtaining final intrinsic mode function until final intrinsic mode function equal to preset quantity, so as to walk
Rapid S190 is according to the spectrum signature of multiple final intrinsic mode function extraction vibration signals, accuracy height.
In the present embodiment, note preset quantity is n, finally can obtain n final intrinsic mode functions；First new noise reduction
Signal r1(t) it is represented by：
r1(t)=x (t)-F1＇ (t)；
Original de-noising signal x (t) can be expressed as：
Wherein, rn(t) it is；F1' (t) is the final intrinsic mode function corresponding to first de-noising signal.
In the present embodiment, spectrum signature includes Hilbert spectrums and Hilbert marginal spectrums.Step S190 includes：Respectively to more
A final intrinsic mode function carries out function after Hilbert-Huang transform is converted；According to function after each transformation and corresponding
Final intrinsic mode function is calculated magnitude function, phase function, instantaneous frequency and constructs the parsing for including real and imaginary parts
Signal；The real part of analytic signal is unfolded to obtain Hilbert spectrums；Hilbert marginal spectrums are calculated according to Hilbert spectrums.
Hilbert marginal spectrums can reflect changing rule of the amplitude with frequency.It is analyzed according to marginal spectrum, can accurately analyze target and set
Standby operating status.
Specifically, it is c to remember i-th of final intrinsic mode functioni(t).Step S190 includes：
Wherein, H (ci(t)) for i-th, final intrinsic mode function is ci(t) carry out what Hilbert-Huang transform obtained
Function after transformation；ai(t) it is magnitude function；For phase function；wi(t) it is instantaneous frequency；zi(t) it is analytic signal；H2
(w, t) is Hilbert energy spectrums, and H (w, t) is composed for Hilbert；H (w) is Hilbert marginal spectrums.Specifically, Hilbert energy
Spectrum can reflect the time-frequency changing rule of vibration signal energy.
With reference to figure 5, in one embodiment, a kind of vibration signal characteristics extraction element is provided, including signal acquisition module
110th, signal denoising module 130, initial decomposition module 150, superposition decomposing module 170 and characteristic extracting module 190.
Signal acquisition module 110 is used to acquire the vibration signal of target device.
Signal denoising module 130 is used to carry out noise reduction process to vibration signal, obtains de-noising signal.
Initial decomposition module 150 is used to carry out empirical mode decomposition to de-noising signal, obtains the original corresponding to de-noising signal
Beginning intrinsic mode function.
Original intrinsic mode function corresponding to de-noising signal is to obtained by de-noising signal progress empirical mode decomposition
Intrinsic mode function.Specifically, the quantity of intrinsic mode function can have multiple.
It is superimposed decomposing module 170 and obtains superposed signal for the original intrinsic mode function according to corresponding to de-noising signal,
Empirical mode decomposition and removal overlap-add procedure are carried out to superposed signal, obtains the final intrinsic mode letter corresponding to de-noising signal
Number.
Characteristic extracting module 190 is used to carry out Hilbert-Huang transform to final intrinsic mode function, obtains vibration signal
Spectrum signature.
Above-mentioned vibration signal characteristics extraction element shakes to 110 acquisition of signal acquisition module by signal denoising module 130
Dynamic signal carries out noise reduction process and obtains de-noising signal, and initial decomposition module 150 carries out empirical mode decomposition to de-noising signal and obtains
Original intrinsic mode function corresponding to de-noising signal, superposition decomposing module 170 are original intrinsic according to corresponding to de-noising signal
Mode function obtains superposed signal, and empirical mode decomposition and removal overlap-add procedure are carried out to superposed signal, obtains de-noising signal institute
Corresponding final intrinsic mode function, characteristic extracting module 190 carry out Hilbert-Huang transform to final intrinsic mode function,
Obtain the spectrum signature of vibration signal.On the one hand, noise reduction is carried out to vibration signal before empirical mode decomposition, noise can be avoided
The effect of effect characteristics extraction；On the other hand, superposed signal is obtained after obtaining original intrinsic mode function, further according to superposed signal
Empirical mode decomposition and removal overlap-add procedure are carried out, the frequency bandwidth of final intrinsic mode function can be limited, can effectively be solved
Modal overlap phenomenon present in empirical mode decomposition, so as to improve the accuracy of feature extraction, the feature of extraction more can be accurate
Really reflect the operating status of target device.
In one embodiment, target device is load ratio bridging switch.Signal acquisition module 110 is divided by being mounted on to carry
The acceleration transducer for connecing switch is acquired respectively under the vibration signal and each malfunction of load ratio bridging switch in normal state
Vibration signal.The vibration under the vibration signal and each malfunction under normal condition by acquiring load ratio bridging switch respectively
Signal, the extractable frequency spectrum for obtaining vibration signal under the spectrum signature of vibration signal under normal condition and each malfunction are special
Sign, the more effective different operating statuses for distinguishing load ratio bridging switch.
In one embodiment, signal denoising module 130 includes the first denoising unit (not shown) and the second denoising unit
(not shown).First denoising unit carries out wavelet packet drop using the Optimum Wavelet Packet and Decomposition order deposited to vibration signal
It makes an uproar, obtains wavelet-packet noise reduction signal；Second denoising unit carries out singular value decomposition to wavelet-packet noise reduction signal, obtains noise reduction letter
Number.By the way that wavelet-packet noise reduction and singular value decomposition are combined, a noise reduction is first carried out to vibration signal using wavelet-packet noise reduction,
Secondary noise reduction is carried out using singular value decomposition to the wavelet-packet noise reduction signal obtained after a noise reduction again, can reach preferable denoising
Effect, and the feature of original vibration signal can be retained, so as to improve the accuracy of feature extraction indirectly.
Optionally, signal denoising module 130 can also include preparatory unit (not shown), for being held in the first denoising unit
Before row corresponding function, according to the entropy function for giving a sequence, being searched in all alternative wavelet packet basis makes entropy function most
Small wavelet packet basis as Optimum Wavelet Packet and stores；The small wavelength-division of a variety of Decomposition orders is carried out to given master sample
Solution obtains decomposition result, and more each decomposition result simultaneously chooses optimal decomposition result, corresponding to the decomposition result of optimal storage
Decomposition order.
In one embodiment, the first denoising unit uses the Optimum Wavelet Packet deposited and Decomposition order to vibration signal
Wavelet decomposition is carried out, obtains the coefficient of multiple high fdrequency components and each high fdrequency component；Using predetermined threshold value to the coefficient of high fdrequency component
Carry out threshold value quantizing processing；Wavelet package reconstruction is carried out to threshold value quantizing treated signal, obtains wavelet-packet noise reduction signal.
In one embodiment, the second denoising unit carries out wavelet-packet noise reduction signal phase space reconfiguration and obtains characterization attraction
The track matrix of son；Singular value decomposition is carried out to track matrix and obtains de-noising signal.
In one embodiment, initial decomposition module 150 is specifically used for：Calculate all local extremums of de-noising signal
Point, wherein, Local Extremum includes maximum point and minimum point；It is obtained what all maximum points were formed according to Local Extremum
The lower envelope line that coenvelope line and all minimum points are formed；The mean value of coenvelope line and lower envelope line is obtained and is obtained according to mean value
Take signal difference；Judge whether signal difference meets first condition and second condition；First condition：Zero number and pole in signal difference
Number is equal or at most differs 1；Second condition：Any point in signal difference, the envelope determined by Local modulus maxima and by office
The mean value for the envelope that portion's minimum point determines is 0, i.e., signal difference is about time shaft Local Symmetric；If meet first condition and
Second condition, then using signal difference as the original intrinsic mode function corresponding to de-noising signal；Otherwise, using signal difference as new
De-noising signal, return recalculates all Local Extremums of de-noising signal, until obtaining corresponding to a de-noising signal
Original intrinsic mode function.
In one embodiment, superposition decomposing module 170 includes mask signal acquiring unit (not shown), superposed signal obtains
Take unit (not shown), superposed signal resolving cell (not shown) and mask signal removal unit (not shown).
Mask signal acquiring unit is used for according to the original intrinsic mode letter corresponding to default mask coefficient and de-noising signal
Number obtains corresponding mask signal.Wherein, it presets mask coefficient and is more than 1.Superposed signal acquiring unit be used for de-noising signal and
Corresponding mask signal is added to obtain superposed signal.Superposed signal resolving cell is used to carry out empirical modal point to superposed signal
Solution, obtains the intrinsic mode function corresponding to superposed signal.Wherein, the intrinsic mode function corresponding to superposed signal is to superposition
Signal carries out the intrinsic mode function that empirical mode decomposition obtains.Mask signal removal unit is used to remove corresponding to superposed signal
Intrinsic mode function in the mask signal that includes, obtain the final intrinsic mode function corresponding to de-noising signal.Wherein, it is superimposed
The mask signal included in intrinsic mode function corresponding to signal, i.e., according to de-noising signal corresponding to this superposed signal
The mask signal of acquisition.
Empirical mode decomposition is carried out to signal to can be understood as signal passing through EMD wave filters.Believed by obtaining mask
Number and mask signal is added with de-noising signal to obtain superposed signal and empirical mode decomposition is carried out to superposed signal, can be changed
The centre frequency of EMD wave filters inhibits the ingredient of low frequency IMF components to be mixed into high-frequency I MF components, realizes the frequency to IMF
Bandwidth is limited, and can effectively solve the problems, such as modal overlap.
In one embodiment, mask signal acquiring unit carries out the original intrinsic mode function corresponding to de-noising signal
Hilbert transform obtains instantaneous frequency and instantaneous amplitude；It is calculated according to instantaneous frequency, instantaneous amplitude and default mask coefficient
Mask signal is generated to mask signal frequency, and based on mask signal frequency.
Specifically, mask signal frequency is：
Mask signal is：
S (t)=A0sin(2πft)；
In formula, K is to preset mask coefficient, and K ＞ 1；a1(i) and f1(i) it is respectively that i-th of instantaneous amplitude and i-th are instantaneous
Frequency；F is mask signal frequency；S (t) is mask signal；A0For intermediate parameters, can be taken as original corresponding to de-noising signal
1.6 times of the average amplitude of intrinsic mode function.
In one embodiment, mask signal removal unit according to：
Remove mask signal；Wherein,For the intrinsic mode function corresponding to superposed signal, s (t) is mask signal,
Fi' (t) is the final intrinsic mode function corresponding to de-noising signal.
In one embodiment, superposition decomposing module 170 is including cycling element (not shown), in final eigen mode
When the quantity of state function is less than preset quantity, de-noising signal is subtracted into corresponding final intrinsic mode function and obtains next drop
Noise cancellation signal and as new de-noising signal, return repeats 150 corresponding function of initial decomposition module, until final intrinsic mode
The quantity of function is equal to preset quantity.In this way, characteristic extracting module 190 can be according to multiple final intrinsic mode function extraction vibrations
The spectrum signature of signal, accuracy are high.
In the present embodiment, spectrum signature includes Hilbert spectrums and Hilbert marginal spectrums.Characteristic extracting module 190 is specifically used
In：Function after Hilbert-Huang transform is converted is carried out to multiple final intrinsic mode functions respectively；According to letter after each transformation
Number and corresponding final intrinsic mode function be calculated magnitude function, phase function, instantaneous frequency and construct include real part and
The analytic signal of imaginary part；The real part of analytic signal is unfolded to obtain Hilbert spectrums；Hilbert is calculated according to Hilbert spectrums
Marginal spectrum.Hilbert marginal spectrums can reflect changing rule of the amplitude with frequency.It is analyzed, can accurately be divided according to marginal spectrum
Analyse the operating status of target device.
Specifically, it is c to remember i-th of final intrinsic mode functioni(t).Step S190 includes：
Wherein, H (ci(t)) for i-th, final intrinsic mode function is ci(t) carry out what Hilbert-Huang transform obtained
Function after transformation；ai(t) it is magnitude function；For phase function；wi(t) it is instantaneous frequency；zi(t) it is analytic signal；H2
(w, t) is Hilbert energy spectrums, and H (w, t) is composed for Hilbert；H (w) is Hilbert marginal spectrums.Specifically, Hilbert energy
Spectrum can reflect the time-frequency changing rule of vibration signal energy.
It is illustrated with a specific application examples, above-mentioned vibration signal characteristics extracting method/device is applied to there is load
The vibration signal of tap switch carries out feature extraction, acquisition load ratio bridging switch vibration signal in normal state and contact pine
Vibration signal under dynamic malfunction.Fig. 6 be normal condition under de-noising signal, Fig. 7 be contact slap when de-noising signal；
Hilbert marginal spectrums under normal condition and contact slap situation are distinguished shown in Fig. 8 and Fig. 9.To Hilbert marginal spectrums carry out
Analysis it is found that vibration signal for load ratio bridging switch normal condition and contact slap state, frequency be in 300Hz~
At 500Hz, the Hilbert limits spectral amplitude ratio of vibration signal is almost without variation, and frequency contact slap at 800Hz~1200Hz
State is remarkably reinforced compared with the signal amplitude of normal condition.
In one embodiment, a kind of storage medium is provided, is stored with computer program, the computer program of storage is located
The step of reason device realizes above-mentioned vibration signal characteristics extracting method when performing.Specifically, storage medium is computer-readable storage
Medium.
In one embodiment, a kind of computer equipment is provided, including memory, processor and is stored on a memory simultaneously
The computer program that can be run on a processor, processor realize above-mentioned vibration signal characteristics extraction side when performing computer program
The step of method.
Above-mentioned storage medium and computer equipment, due to the step of realizing above-mentioned vibration signal characteristics extracting method, together
Reason can improve the accuracy of feature extraction.
Each technical characteristic of embodiment described above can be combined arbitrarily, to make description succinct, not to above-mentioned reality
It applies all possible combination of each technical characteristic in example to be all described, as long as however, the combination of these technical characteristics is not deposited
In contradiction, it is all considered to be the range of this specification record.
Embodiment described above only expresses the several embodiments of the present invention, and description is more specific and detailed, but simultaneously
It cannot therefore be construed as limiting the scope of the patent.It should be pointed out that those of ordinary skill in the art are come
It says, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to the protection of the present invention
Range.Therefore, the protection domain of patent of the present invention should be determined by the appended claims.
Claims (10)
1. a kind of vibration signal characteristics extracting method, which is characterized in that including：
Acquire the vibration signal of target device；
Noise reduction process is carried out to the vibration signal, obtains de-noising signal；
Empirical mode decomposition is carried out to the de-noising signal, obtains the original intrinsic mode function corresponding to the de-noising signal；
Original intrinsic mode function according to corresponding to the de-noising signal obtains superposed signal, to the superposed signal into passing through
Mode decomposition and removal overlap-add procedure are tested, obtains the final intrinsic mode function corresponding to the de-noising signal；
Hilbert-Huang transform is carried out to the final intrinsic mode function, obtains the spectrum signature of the vibration signal.
2. vibration signal characteristics extracting method according to claim 1, which is characterized in that the target device is has load point
Connect switch, the vibration signal of the acquisition target device, including：
The load ratio bridging switch is acquired in normal shape by the acceleration transducer for being mounted on the load ratio bridging switch respectively
The vibration signal under vibration signal and each malfunction under state.
3. vibration signal characteristics extracting method according to claim 1, which is characterized in that it is described to the vibration signal into
Row noise reduction process, obtains de-noising signal, including：
Wavelet-packet noise reduction is carried out to the vibration signal using the Optimum Wavelet Packet and Decomposition order deposited, obtains wavelet packet drop
Noise cancellation signal；
Singular value decomposition is carried out to the wavelet-packet noise reduction signal, obtains the de-noising signal.
4. vibration signal characteristics extracting method according to any one of claim 1-3, which is characterized in that described according to institute
It states the original intrinsic mode function corresponding to de-noising signal and obtains superposed signal, empirical mode decomposition is carried out to the superposed signal
And removal overlap-add procedure, the final intrinsic mode function corresponding to the de-noising signal is obtained, including：
Original intrinsic mode function according to corresponding to default mask coefficient and the de-noising signal obtains corresponding mask signal；
The de-noising signal is added to obtain superposed signal with corresponding mask signal；
Empirical mode decomposition is carried out to the superposed signal, obtains the intrinsic mode function corresponding to the superposed signal；
The mask signal included in the intrinsic mode function corresponding to the superposed signal is removed, obtains the de-noising signal
Corresponding final intrinsic mode function.
5. vibration signal characteristics extracting method according to claim 4, which is characterized in that the basis presets mask coefficient
Original intrinsic mode function corresponding to the de-noising signal obtains corresponding mask signal, including：
Hilbert transform is carried out to the original intrinsic mode function corresponding to the de-noising signal, obtains instantaneous frequency and instantaneous
Amplitude；
Mask signal frequency, and base are calculated according to the instantaneous frequency, the instantaneous amplitude and the default mask coefficient
Mask signal is generated in the mask signal frequency.
6. vibration signal characteristics extracting method according to claim 4, which is characterized in that described to remove the superposed signal
The mask signal included in corresponding intrinsic mode function obtains the final intrinsic mode corresponding to the de-noising signal
Function, including：
Wherein,For the intrinsic mode function corresponding to the superposed signal, s (t) is the mask signal, Fi' (t) is institute
State the final intrinsic mode function corresponding to de-noising signal.
7. vibration signal characteristics extracting method according to claim 4, which is characterized in that described to remove the superposed signal
The mask signal included in corresponding intrinsic mode function obtains the final intrinsic mode corresponding to the de-noising signal
After function, further include：
If the quantity of the final intrinsic mode function is less than preset quantity, the de-noising signal is subtracted corresponding final
Intrinsic mode function obtains next de-noising signal and as new de-noising signal, return it is described to the de-noising signal into passing through
The step of testing mode decomposition, obtaining the original intrinsic mode function corresponding to the de-noising signal, until the final eigen mode
The quantity of state function is equal to the preset quantity.
8. a kind of vibration signal characteristics extraction element, which is characterized in that including：
Signal acquisition module, for acquiring the vibration signal of target device；
Signal denoising module for carrying out noise reduction process to the vibration signal, obtains de-noising signal；
Initial decomposition module for carrying out empirical mode decomposition to the de-noising signal, is obtained corresponding to the de-noising signal
Original intrinsic mode function；
Decomposing module is superimposed, superposed signal is obtained for the original intrinsic mode function according to corresponding to the de-noising signal, it is right
The superposed signal carries out empirical mode decomposition and removal overlap-add procedure, obtains the final eigen mode corresponding to the de-noising signal
State function；
Characteristic extracting module for carrying out Hilbert-Huang transform to the final intrinsic mode function, obtains the vibration letter
Number spectrum signature.
9. a kind of storage medium, is stored with computer program, which is characterized in that when the computer program of storage is executed by processor
It realizes such as the step of any one of claim 1-7 the methods.
10. 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 is realized when performing the computer program as described in claim any one of 1-7
The step of method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711480593.6A CN108229382A (en) | 2017-12-29 | 2017-12-29 | Vibration signal characteristics extracting method, device, storage medium and computer equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711480593.6A CN108229382A (en) | 2017-12-29 | 2017-12-29 | Vibration signal characteristics extracting method, device, storage medium and computer equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108229382A true CN108229382A (en) | 2018-06-29 |
Family
ID=62647186
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711480593.6A Pending CN108229382A (en) | 2017-12-29 | 2017-12-29 | Vibration signal characteristics extracting method, device, storage medium and computer equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108229382A (en) |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109100103A (en) * | 2018-07-06 | 2018-12-28 | 哈尔滨工业大学（深圳） | Based on blower 1p signal recognition method, device, terminal and the computer readable storage medium continuously monitored |
| CN109507579A (en) * | 2018-12-05 | 2019-03-22 | 中国电力科学研究院有限公司 | Load ratio bridging switch changeover program on-line checking diagnostic method |
| CN109813417A (en) * | 2019-01-18 | 2019-05-28 | 国网江苏省电力有限公司检修分公司 | A kind of shunt reactor method for diagnosing faults based on improvement EMD |
| CN109871758A (en) * | 2019-01-14 | 2019-06-11 | 南京航空航天大学 | Fault-signal SVD noise-reduction method based on Multiscale Morphological optimization |
| CN109900465A (en) * | 2019-03-12 | 2019-06-18 | 西安交通大学 | A kind of load ratio bridging switch method for diagnosing faults based on voice signal |
| CN110672327A (en) * | 2019-10-09 | 2020-01-10 | 西南交通大学 | Asynchronous motor bearing fault diagnosis method based on multilayer noise reduction technology |
| CN110763445A (en) * | 2019-10-12 | 2020-02-07 | 西人马(西安)测控科技有限公司 | Brake monitoring method, device, equipment and storage medium |
| CN112014107A (en) * | 2020-03-12 | 2020-12-01 | 岭南师范学院 | Improved empirical mode decomposition bearing vibration analysis method and system |
| CN112101089A (en) * | 2020-07-27 | 2020-12-18 | 北京建筑大学 | Signal noise reduction method and device, electronic equipment and storage medium |
| CN112595514A (en) * | 2020-11-26 | 2021-04-02 | 上海航天控制技术研究所 | High-speed bearing assembly vibration signal noise reduction processing method |
| CN114234361A (en) * | 2021-12-14 | 2022-03-25 | 北京工业大学 | Central air-conditioning sensor fault detection method based on double noise reduction and fuzzy indexes |
| CN114235405A (en) * | 2021-11-24 | 2022-03-25 | 阿里巴巴（中国）有限公司 | Feature extraction method and device of vibration signal, and equipment analysis method and device |
| CN114462458A (en) * | 2022-04-11 | 2022-05-10 | 自然资源部第一海洋研究所 | Ship underwater signal noise reduction and target enhancement method |
| CN114888634A (en) * | 2022-03-23 | 2022-08-12 | 北京工业大学 | Milling cutter wear monitoring method and device |
| CN115659128A (en) * | 2022-12-12 | 2023-01-31 | 浙江工业大学 | Signal noise reduction method based on ensemble empirical mode decomposition method and power spectrum |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101900708A (en) * | 2010-08-18 | 2010-12-01 | 哈尔滨工业大学 | Vibration and audio signal-based high-speed train track defect detecting method |
| CN103837334A (en) * | 2012-11-20 | 2014-06-04 | 中国石油大学(北京) | Detection method of mechanical failure |
| US20160011268A1 (en) * | 2014-07-10 | 2016-01-14 | National Cheng Kung University | Motor fault detecting method and motor fault detecting system |
| CN105510711A (en) * | 2015-12-24 | 2016-04-20 | 合肥工业大学 | Empirical mode decomposition-based improved harmonic analysis method |
| CN105547627A (en) * | 2016-01-28 | 2016-05-04 | 南京信息工程大学 | Rotating machinery feature extraction method on the basis of WPT-CEEMD |
| CN106840637A (en) * | 2017-03-24 | 2017-06-13 | 国网山东省电力公司莱芜供电公司 | Based on the GIS mechanical oscillation signal Time-Frequency Analysis Methods for improving HHT algorithms |
-
2017
- 2017-12-29 CN CN201711480593.6A patent/CN108229382A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101900708A (en) * | 2010-08-18 | 2010-12-01 | 哈尔滨工业大学 | Vibration and audio signal-based high-speed train track defect detecting method |
| CN103837334A (en) * | 2012-11-20 | 2014-06-04 | 中国石油大学(北京) | Detection method of mechanical failure |
| US20160011268A1 (en) * | 2014-07-10 | 2016-01-14 | National Cheng Kung University | Motor fault detecting method and motor fault detecting system |
| CN105510711A (en) * | 2015-12-24 | 2016-04-20 | 合肥工业大学 | Empirical mode decomposition-based improved harmonic analysis method |
| CN105547627A (en) * | 2016-01-28 | 2016-05-04 | 南京信息工程大学 | Rotating machinery feature extraction method on the basis of WPT-CEEMD |
| CN106840637A (en) * | 2017-03-24 | 2017-06-13 | 国网山东省电力公司莱芜供电公司 | Based on the GIS mechanical oscillation signal Time-Frequency Analysis Methods for improving HHT algorithms |
Non-Patent Citations (2)
| Title |
|---|
| 段礼祥等: ""柴油机振动信号的小波包奇异值降噪"", 《中国石油大学学报（自然科学版）》 * |
| 段若晨等: ""基于优化HHT算法与洛伦兹信息量度的换流变用有载分接开关机械特征提取"", 《中国电机工程学报》 * |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020007375A1 (en) * | 2018-07-06 | 2020-01-09 | 哈尔滨工业大学（深圳） | Continuous monitoring-based method and device for identifying 1p signal of wind turbine, terminal, and computer readable storage medium |
| CN109100103B (en) * | 2018-07-06 | 2020-04-14 | 哈尔滨工业大学（深圳） | Fan 1p signal identification method, device, terminal and computer readable storage medium based on continuous monitoring |
| CN109100103A (en) * | 2018-07-06 | 2018-12-28 | 哈尔滨工业大学（深圳） | Based on blower 1p signal recognition method, device, terminal and the computer readable storage medium continuously monitored |
| CN109507579B (en) * | 2018-12-05 | 2022-09-20 | 中国电力科学研究院有限公司 | On-load tap-changer switching program on-line detection and diagnosis method |
| CN109507579A (en) * | 2018-12-05 | 2019-03-22 | 中国电力科学研究院有限公司 | Load ratio bridging switch changeover program on-line checking diagnostic method |
| CN109871758A (en) * | 2019-01-14 | 2019-06-11 | 南京航空航天大学 | Fault-signal SVD noise-reduction method based on Multiscale Morphological optimization |
| CN109813417A (en) * | 2019-01-18 | 2019-05-28 | 国网江苏省电力有限公司检修分公司 | A kind of shunt reactor method for diagnosing faults based on improvement EMD |
| CN109900465A (en) * | 2019-03-12 | 2019-06-18 | 西安交通大学 | A kind of load ratio bridging switch method for diagnosing faults based on voice signal |
| CN109900465B (en) * | 2019-03-12 | 2020-07-28 | 西安交通大学 | On-load tap-changer fault diagnosis method based on sound signals |
| CN110672327A (en) * | 2019-10-09 | 2020-01-10 | 西南交通大学 | Asynchronous motor bearing fault diagnosis method based on multilayer noise reduction technology |
| CN110763445A (en) * | 2019-10-12 | 2020-02-07 | 西人马(西安)测控科技有限公司 | Brake monitoring method, device, equipment and storage medium |
| CN112014107A (en) * | 2020-03-12 | 2020-12-01 | 岭南师范学院 | Improved empirical mode decomposition bearing vibration analysis method and system |
| CN112101089A (en) * | 2020-07-27 | 2020-12-18 | 北京建筑大学 | Signal noise reduction method and device, electronic equipment and storage medium |
| CN112101089B (en) * | 2020-07-27 | 2023-10-10 | 北京建筑大学 | Signal noise reduction method and device, electronic equipment and storage medium |
| CN112595514A (en) * | 2020-11-26 | 2021-04-02 | 上海航天控制技术研究所 | High-speed bearing assembly vibration signal noise reduction processing method |
| CN114235405A (en) * | 2021-11-24 | 2022-03-25 | 阿里巴巴（中国）有限公司 | Feature extraction method and device of vibration signal, and equipment analysis method and device |
| CN114234361A (en) * | 2021-12-14 | 2022-03-25 | 北京工业大学 | Central air-conditioning sensor fault detection method based on double noise reduction and fuzzy indexes |
| CN114888634A (en) * | 2022-03-23 | 2022-08-12 | 北京工业大学 | Milling cutter wear monitoring method and device |
| CN114888634B (en) * | 2022-03-23 | 2023-09-01 | 北京工业大学 | Milling cutter abrasion monitoring method and device |
| CN114462458A (en) * | 2022-04-11 | 2022-05-10 | 自然资源部第一海洋研究所 | Ship underwater signal noise reduction and target enhancement method |
| CN115659128A (en) * | 2022-12-12 | 2023-01-31 | 浙江工业大学 | Signal noise reduction method based on ensemble empirical mode decomposition method and power spectrum |
| CN115659128B (en) * | 2022-12-12 | 2023-04-07 | 浙江工业大学 | Signal noise reduction method based on ensemble empirical mode decomposition method and power spectrum |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108229382A (en) | Vibration signal characteristics extracting method, device, storage medium and computer equipment | |
| CN108845250B (en) | On-load tap-changer fault identification method based on vibration signal feature extraction | |
| CN107451557B (en) | Power transmission line short-circuit fault diagnosis method based on empirical wavelet transform and local energy | |
| Yu et al. | Weak fault feature extraction of rolling bearings using local mean decomposition-based multilayer hybrid denoising | |
| Konar et al. | Multi-class fault diagnosis of induction motor using Hilbert and Wavelet Transform | |
| Carvalho et al. | Identification of partial discharges immersed in noise in large hydro-generators based on improved wavelet selection methods | |
| CN105547465A (en) | Transformer vibration signal winding state feature extraction method | |
| CN107179486A (en) | A kind of GIS device monitors ultrahigh-frequency signal noise-reduction method on-line | |
| CN106441547A (en) | Transformer vibration monitoring method and apparatus | |
| CN108734192A (en) | A kind of support vector machines mechanical failure diagnostic method based on voting mechanism | |
| Koley et al. | Wavelet-aided SVM tool for impulse fault identification in transformers | |
| CN109506938A (en) | Based on the failure extracting method for improving wavelet packet threshold denoising and local mean value decomposition | |
| CN109541455A (en) | A kind of OLTC impact characteristics extracting method based on S-transformation time-frequency spectrum SVD noise reduction | |
| Varanis et al. | Statements on wavelet packet energy–entropy signatures and filter influence in fault diagnosis of induction motor in non-stationary operations | |
| Hosseini et al. | Image sharpness metric based on maxpol convolution kernels | |
| CN117473414A (en) | Bearing fault position identification method based on low-noise time-frequency image | |
| CN113268924B (en) | Time-frequency characteristic-based fault identification method for on-load tap-changer of transformer | |
| Ambika et al. | Vibration signal based condition monitoring of mechanical equipment with scattering transform | |
| Chilukuri et al. | Time-frequency based pattern recognition technique for detection and classification of power quality disturbances | |
| Mitiche et al. | Deep residual neural network for EMI event classification using bispectrum representations | |
| Nayak et al. | Advanced signal processing techniques for feature extraction in data mining | |
| CN110160778A (en) | Gearbox fault state identification method based on sequential hypothesis testing | |
| CN114609515A (en) | GIS ultrahigh frequency partial discharge detection interference suppression method based on sequential hierarchical signal processing | |
| Shicheng et al. | An effective S-transform feature extraction method for classification of power quality disturbance Signals | |
| Zhu et al. | Fault diagnosis of rolling bearing based on WT-VMD and random forest |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20200925 Address after: 510620 Tianhe District, Guangzhou, Tianhe South Road, No. two, No. 2, No. Applicant after: Guangzhou Power Supply Bureau of Guangdong Power Grid Co.,Ltd. Address before: 510620 Tianhe District, Guangzhou, Tianhe South Road, No. two, No. 2, No. Applicant before: GUANGZHOU POWER SUPPLY Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180629 |
|
| RJ01 | Rejection of invention patent application after publication |