CN110146156A - A kind of denoising method of aircraft engine rotor system fault vibration signal - Google Patents
A kind of denoising method of aircraft engine rotor system fault vibration signal Download PDFInfo
- Publication number
- CN110146156A CN110146156A CN201910566648.8A CN201910566648A CN110146156A CN 110146156 A CN110146156 A CN 110146156A CN 201910566648 A CN201910566648 A CN 201910566648A CN 110146156 A CN110146156 A CN 110146156A
- Authority
- CN
- China
- Prior art keywords
- solution
- ultiwavelet
- algorithm
- coefficient
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 68
- 239000002245 particle Substances 0.000 claims abstract description 45
- 238000005457 optimization Methods 0.000 claims abstract description 24
- 238000010845 search algorithm Methods 0.000 claims abstract description 14
- RZVAJINKPMORJF-UHFFFAOYSA-N Acetaminophen Chemical compound CC(=O)NC1=CC=C(O)C=C1 RZVAJINKPMORJF-UHFFFAOYSA-N 0.000 claims abstract description 10
- 238000012545 processing Methods 0.000 claims abstract description 10
- 238000001914 filtration Methods 0.000 claims abstract description 6
- 230000008569 process Effects 0.000 claims description 13
- 238000000354 decomposition reaction Methods 0.000 claims description 5
- 230000001133 acceleration Effects 0.000 claims description 3
- 238000006073 displacement reaction Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 230000008929 regeneration Effects 0.000 claims description 2
- 238000011069 regeneration method Methods 0.000 claims description 2
- 230000017105 transposition Effects 0.000 claims description 2
- 230000007423 decrease Effects 0.000 abstract description 2
- 238000001228 spectrum Methods 0.000 description 17
- 230000000694 effects Effects 0.000 description 15
- 230000035939 shock Effects 0.000 description 15
- 230000009467 reduction Effects 0.000 description 14
- 230000006870 function Effects 0.000 description 13
- 230000004044 response Effects 0.000 description 12
- 239000000306 component Substances 0.000 description 9
- 238000010586 diagram Methods 0.000 description 6
- 230000007257 malfunction Effects 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 238000003745 diagnosis Methods 0.000 description 4
- 238000000605 extraction Methods 0.000 description 4
- 238000011946 reduction process Methods 0.000 description 4
- 230000007812 deficiency Effects 0.000 description 3
- 230000005764 inhibitory process Effects 0.000 description 3
- 238000012423 maintenance Methods 0.000 description 3
- 239000000203 mixture Substances 0.000 description 3
- 230000000737 periodic effect Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000015556 catabolic process Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000012804 iterative process Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000012805 post-processing Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- SNKNNAPOEZWYCH-UHFFFAOYSA-N (2-hydroxy-3-piperidin-1-ylpropyl) n-phenylcarbamate;hydrochloride Chemical compound Cl.C1CCCCN1CC(O)COC(=O)NC1=CC=CC=C1 SNKNNAPOEZWYCH-UHFFFAOYSA-N 0.000 description 1
- 206010000369 Accident Diseases 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013475 authorization Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 239000008358 core component Substances 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001035 drying Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 239000010437 gem Substances 0.000 description 1
- 229910001751 gemstone Inorganic materials 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000004615 ingredient Substances 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 239000011664 nicotinic acid Substances 0.000 description 1
- 230000002028 premature Effects 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01H—MEASUREMENT OF MECHANICAL VIBRATIONS OR ULTRASONIC, SONIC OR INFRASONIC WAVES
- G01H11/00—Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves by detecting changes in electric or magnetic properties
- G01H11/06—Measuring mechanical vibrations or ultrasonic, sonic or infrasonic waves by detecting changes in electric or magnetic properties by electric means
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M15/00—Testing of engines
- G01M15/02—Details or accessories of testing apparatus
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention discloses a kind of denoising methods of aircraft engine rotor system fault vibration signal, what this method first acquired aeroengine rotor touches the vibration signal that rubs, and signal is carried out to repeat row filtering processing and optimal threshold, initial solution particle swarm algorithm being assigned to by the suboptimal solution that iteration optimizing is searched out again in tabu search algorithm, then fitness value is calculated, by by the continuous search of the neighborhood of initial value, simultaneously flexibly using taboo list and special pardon criterion, tabu search algorithm is enabled to receive some low-quality solutions inferior to a certain extent, prevent algorithm from falling into local extremum, enhance the global optimizing ability of algorithm, effective solution particle swarm algorithm later period receives the problem of speed slows down, in comparison, although the convergence rate of Hybrid Particle Swarm phase has certain decline, but it can effectively avoid group's mistake Early falls into locally optimal solution, increase population seek globally optimal solution a possibility that, generally improve the optimization performance of algorithm.
Description
Technical field
The present invention relates to a kind of extracting method of signal characteristic information more particularly to a kind of events of aircraft engine rotor system
Hinder the denoising method of vibration signal.
Background technique
With the continuous development and progress of modern science and technology, on the one hand rotating machinery is sent out towards high speed, efficient direction
Exhibition, while it being faced with the challenge of harsher work and running environment again, the potential risk occurred so as to cause failure is corresponding
It increases, core component rotor-support-foundation system is even more one of highest position of rate of breakdown.Especially the aero-engine of aircraft turns
Subsystem is failure rate highest in Airborne Equipment, adjusts most complicated, the maximum system of maintenance workload, work shape
The quality of state directly influences the safe and reliable of aircraft and operation;Once if there is failure in operation in engine, so that it may
Can cause aircraft can not normal flight, jeopardize personal safety, cause great society, economic loss.
According to statistics, the rate of breakdown of aero-engine accounts for about the 30% of entire airplane fault, and aircraft is sent out because of mechanical reason
40% or so is as caused by engine failure in raw great aircraft accident, and engine is different from generally due to own characteristic
Machinery, even if very slight some mechanical defects or damage can all cause the vibration of rotor-support-foundation system, including from almost micro-ly small
It is inappreciaple to vibrate to the vibration large enough to lead to engine demolition.It is past due to the complexity of aircraft engine rotor system
Identical vibration can be shown as toward various faults, and same failure can show as different vibrations, thus will lead to fail to report with
Wrong report, has seriously affected the reliability of aircraft engine rotor system fault diagnosis, therefore, in order to improve aeroengine rotor
System fault diagnosis is horizontal, needs to develop practical, accurately and efficiently characteristics information extraction method.
Currently, wavelet analysis is because can carry out the mutation component and noise component(s) of signal from time domain and the angle of frequency domain
Analysis, so as to effectively eliminate the ambient noise in signal;Therefore, wavelet analysis has obtained extensively in field of signal processing
General application, but there are the deficiencies that wavelet basis function is difficult to choose in wavelet noise.Multi-wavelet transformation be wavelet theory into
One step development, needed for m ultiwavelet can have in the signal processings such as orthogonality, compact sup-port, symmetry and high degree of approximation simultaneously because of it
The key property wanted, the shortcomings that single wavelet can be made up, only one scaling function of single wavelet, and m ultiwavelet have simultaneously it is multiple
Scaling function.Therefore, symmetry, orthogonality, compact sup-port and the high-order that m ultiwavelet can have single wavelet and cannot be provided simultaneously with disappear
Lose the important features such as square, it is possible to obtain better effect in field of signal processing.
The patent application that the patent and application publication number that Authorization Notice No. is CN101968379B are CN107506709A point
The extracting method of two kinds of aircraft engine rotor system fault-signal characteristic informations is not disclosed, and both methods is all to utilize electricity
Eddy displacement sensor measures vibration signal, and the different Wave Cluster of design is respectively adopted and is analyzed, from a large amount of vibration positions
Excavate implicit characteristic information in shifting signal, realize the extraction of operating mode feature, former approach on extraction accuracy and speed still
So there is certain deficiency, latter extracting method has carried out breakthrough improvement on the basis of former, but its with the former still
So all there is a common problem, i.e., noise background frequency spectrum cannot be filtered out effectively in extracting result, and weak fault signature is believed
Breath and fault characteristic information frequency spectrum impact amplitude are close, affect rationally mentioning for rotor fault feature to a certain extent
It takes, so, in order to remove noise jamming and more reasonably carry out diagnosis and signal spy to aircraft engine rotor system failure
Sign is extracted, and also needs to be improved by the way that other directions are further;It is ground so carrying out noise reduction to rotor-support-foundation system fault vibration signal
Study carefully, it is significant to the generation for preventing sudden major accident.
Summary of the invention
(1) technical problems to be solved
In view of the deficiencies of the prior art, the purpose of the present invention is to provide a kind of aircraft engine rotor system fault vibrations
The denoising method of signal
(2) technical solution,
In order to solve the above-mentioned technical problems, the present invention provides such a aircraft engine rotor system fault vibration letters
Number denoising method, this method be specially the m ultiwavelet adjacent coefficient denoising method based on Hybrid particle swarm optimization, pass through the party
Method carries out the substantially step of noise reduction process to aircraft engine rotor system fault-signal are as follows: first to the signal after pretreatment
Multi-wavelet transformation is carried out, parameter required for Hybrid Particle Swarm Optimization is then calculated by decomposition coefficient, further according to adaptation
The minimization principle of degree function seeks optimal threshold by Hybrid Particle Swarm Optimization, and carries out to decomposition coefficient corresponding
Threshold process carries out m ultiwavelet reconstruct and post-processing later, obtains denoised signal.
Hybrid Particle Swarm Optimization is also referred to as PSO algorithm, first has to according to independent variable in optimization problem to be solved
Number determines the dimension of algorithm, assigns one initial position of all particles and initial velocity in population at random again later, passes through
The iterative search of algorithm, in iteration searching process each time, particles mainly instruct oneself by two parameters of tracking
Flight position and flying speed, one of parameter refer to that particle individual itself is experienced best in an iterative process
Position, referred to as " individual extreme value ", another parameter refer to the desired positions that entire group is lived through, also referred to as " global pole
Value ".
Particle swarm optimization algorithm is emerging bionic intelligence optimisation technique.The algorithm can converge on the overall situation most with greater probability
Excellent solution, the experimental results showed that, it is suitble to the optimizing in the environment such as single object optimization, constrained objective optimization, with other intelligent optimizations
Algorithm is compared, and particle swarm optimization algorithm realizes that step is relatively simple, and optimizing convergence capabilities are strong.
Tabu search algorithm is a kind of simulation to human mind's process, i.e., people is right when searching for a piece of unknown region
Binary search can be automatically avoided in the region searched for, and then the unknown sky of complete slice can be searched within the shortest time
Between, if do not searched, then the region gone is re-searched for, tabu search algorithm is based on this thought, from one
Initial solution is set out, and determines several specific directions of search (or to move) as souning out, then selection makes fitness function value
Reduce most directions to scan for, algorithm falls into local extremum in order to prevent, and tabu search algorithm uses a kind of flexible
" memory " technology instructs the next step direction of search of algorithm, i.e. establishment taboo with this to record to improved movement
Table saves the movement having been carried out in iterative process by taboo list, algorithm is avoided to fall into circulation, and causes the drop of search efficiency
It is low.Allow to lift a ban certain states using " special pardon criterion " simultaneously, so that algorithm, which has, receives some low-quality solutions inferior
Possibility prevents algorithm from falling into local extremum.
Tabu search algorithm is that some repetitions are avoided by the taboo list of single-input single-output based on neighborhood iterative search
Search, and absolved using special pardon criterion (aspiration criterion) it is some before the excellent conditions avoided by taboo list, protect whereby
Demonstrate,prove the diversity of search.Wherein the selection of neighbour structure, the acquisition of initial solution, the determination of Tabu Length, aspiration criterion, termination are quasi-
Formulating then all can generate critical impact to the performance of tabu search algorithm.Neighborhood function inherits the spy of local neighborhood search
Point, for execute to the neighborhood search currently solved;Flexible use of taboo list and special pardon criterion not only can be avoided algorithm
Repeat search, and ensure that effective search of diversity algorithm, so that algorithm is finally realized global optimizing.
The specific step of the denoising method of the aircraft engine rotor system fault vibration signal are as follows:
Step 1: being measured by determining time interval or sample frequency a certain number of by eddy current displacement sensor
Aeroengine rotor touches the vibration signal that rubs some sampling periods;
It carries out repeating row filtering processing Step 2: the aeroengine rotor collected to be touched to the vibration signal that rubs, and right
Signal after repeating row filtering processing carries out m ultiwavelet decomposition and reconstruct, obtains m ultiwavelet coefficient, and m ultiwavelet is decomposed and is reconstructed into
It is carried out by following formula,
In formula, cj,k=(c1,j,k,···,cr,j,k)TFor the low frequency component of r dimension, dj,k=(d1,j,k,···,dr,j,k)TFor
The high fdrequency component of r dimension, * are conjugate transposition operation;
Step 3: the invariant in setting Hybrid Particle Swarm Optimization, wherein setting Studying factors c1=c2=
1.4995, the quantity m=40 of population, maximum number of iterations T=200, the maximum value ω of weight coefficientmax=0.9, weight system
Several minimum value ωmin=0.4, the maximum speed v of particlemax=0.2 λmax, the maximum value λ of population location parameter λmaxWith
Minimum value λminIt is carried out by following formula,
λj=(2lognj)/q
In formula, njλ can be respectively obtained when q takes 0.1 and 1 respectively for the length of m ultiwavelet coefficientmaxAnd λmin;
Step 4: obtain optimal threshold by the function based on GCV criterion fitness, the suitable of each particle in group is obtained
Angle value is answered, the function based on GCV criterion fitness is following formula,
In formula, D and DλM ultiwavelet coefficient after respectively noisy m ultiwavelet coefficient and threshold value λ rule process, N is m ultiwavelet
The total number of coefficient, N00 number is set in thresholding process for m ultiwavelet coefficient;
Step 5: the desired positions p for living through its fitness value with it to each particleiFitness value make comparisons,
If preferably, by xiAs current desired positions pi, and then its fitness value lived through with all particles best
The fitness value of position is made comparisons;If preferably, as the desired positions p of current all particlesg;
Step 6: obtaining individual extreme value and global extremum by step 5 and then respectively by following formula Population Regeneration
The speed of particle and position,
vij(t+1)=wvij(t)+c1r1(t)[pij(t)-xij(t)]+c2r2(t)[pgj(t)-xij(t)]
xij(t+1)=xij(t)+vij(t+1)
In formula, w indicates inertia weight coefficient, c1And c2It represents Studying factors (also referred to as acceleration constant), i=1,
2, D, vijIndicate the speed of particle, vij∈[-vmax, vmax], vmaxIt is constant;By being manually set to particle rapidity
It is limited, parameter r1And r2It is uniform random number of the range between 0 to 1, for enhancing the randomness of particle search;
Step 7: whether the algorithm of judgment step six has reached maximum number of iterations, step 4 is back to if not up to
It continues to run, if having reached, exports optimal value λbest;
Step 8: enabling tabu search algorithm, and maximum number of iterations is set, and empties taboo list, by hybrid particle swarm
The optimal solution λ that optimization algorithm generatesbestCurrent solution is assigned to as the initial solution of tabu search algorithm, and initial solution;
Step 9: generating several neighborhood solutions using current solution neighborhood function, and therefrom determine several candidate solutions;
Step 10: obtaining each grain in group by obtaining the optimal threshold of the algorithm based on GCV criterion fitness function
The fitness value of son, and judge whether current candidate solution meets special pardon criterion with this, with satisfaction special pardon rule if meeting
Optimal candidate solution replaces current solution, and enters the object of taboo list earliest using optimal candidate solution as the replacement of taboo object, simultaneously
Current optimal solution is replaced with the best candidate solution for meeting special pardon rule, then jumps directly to step 12 operation;If discontented
It is sufficient then then downwards execute step 11;Function based on GCV criterion fitness is following formula,
In formula, D and DλM ultiwavelet coefficient after respectively noisy m ultiwavelet coefficient and threshold value λ rule process, N is m ultiwavelet
The total number of coefficient, N00 number is set in thresholding process for m ultiwavelet coefficient;
Step 11: the taboo attribute of candidate solution is determined when candidate solution is inferior to current optimal solution better than current solution, if
Candidate solution is then assigned to the current solution of next iteration not in taboo list by candidate solution, while being replaced to enter earliest with it and be prohibited
Avoid the object of table;It is inferior to currently to solve in candidate solution and when also inferior to current optimal solution, new neighborhood solution is regenerated with current solution;
Step 12: judging whether algorithm has reached termination condition, optimal solution and termination algorithm are exported if reaching;Such as
Fruit does not reach, and going to step nine continues to run;
Step 13: handling according to threshold rule m ultiwavelet coefficient, and coefficient after treatment is carried out more
Wavelet reconstruction obtains the denoised signal of aircraft engine rotor system fault vibration.
Pass through comparative experiments, the results showed that m ultiwavelet adjacent coefficient denoising method and biography based on Hybrid particle swarm optimization
The m ultiwavelet adjacent coefficient denoising method of system is compared, and is significantly improved in terms of ambient noise denoising effect, and more mono- than Db2 small
Wave, GHM m ultiwavelet method have more advantage in denoising effect, provide technology branch for the feature extraction in later period and fault diagnosis
It holds.
Wherein, common Research on threshold selection has DJ uniform threshold, based on zero-mean normal distribution under the prior art
Confidence interval threshold, desired threshold and GCV threshold value.In numerous wavelet threshold denoising algorithms, majority is the system using noise
Characteristic is counted to realize denoising.But in many practical applications, the priori knowledge in relation to noise be it is unknown, need to its into
Row estimation.And using GCV function come threshold value, he, which only depends on, outputs and inputs data and noise energy and its true number
According to unrelated.And by current test it was demonstrated that being a kind of minimum equal using generalized crossover confirmation (GCV) obtained threshold value
Miss the asymptotic optimality solution in meaning in side.Therefore, threshold value is sought using GCV principle, without obtaining any information of noise in advance, and
And the minutia of original image also can be preferably kept while denoising.
By actually drying method being gone to compare, and combine signal-to-noise ratio parameter (SNR) and mean square deviation (MSE) joint that can see
Out, noise reduction effect most preferably noise-reduction method proposed by the present invention, followed by m ultiwavelet adjacent coefficient Denoising Algorithm, and Db2 is mono- small
Wave threshold deniosing effect is worst;And from the denoising effect obtained to same denoising method using different threshold function table come
It sees, the soft-threshold denoising effect in each denoising method is obviously better than hard-threshold denoising.In summary it is found that it is proposed by the present invention
Noise-reduction method has maximum signal-to-noise ratio parameter (SNR), and mean square error (MSE) value is minimum simultaneously, illustrates method noise of the invention
Than highest, synchronous signal distortion level is low, and noise reduction effect is best.
By comparison, it is apparent that Hybrid Particle Swarm Optimization is easy to operate, there is stronger global group to have energy
Power, but can be gradually decreased in the late convergence of algorithm and precision, and also easily fall into precocity, so that algorithm can only
Obtain suboptimal solution;And tabu search algorithm has stronger dependence to initial solution in the initial stage, the quality of initial solution is often
Directly influence algorithm can converge to it is global most solve, therefore the present invention is breakthroughly by TABU search and Particle Swarm Optimization
Method blends, i.e. Hybrid Particle Swarm, and particle swarm algorithm is passed through the suboptimum that iteration optimizing is searched out by method of the invention
Solution is assigned to the initial solution in tabu search algorithm, then calculates fitness value, by by the continuous search of the neighborhood of initial value,
Flexibly tabu search algorithm is received to a certain extent some low-quality bad using taboo list and special pardon criterion simultaneously
Matter solution, prevents algorithm from falling into local extremum, enhances the global optimizing ability of algorithm, and the effective solution particle swarm algorithm later period receives
The problem of speed slows down in comparison, can although the convergence rate of Hybrid Particle Swarm phase has certain decline
Effectively avoid group is premature from falling into locally optimal solution, increase population seek globally optimal solution a possibility that, generally improve
The optimization performance of algorithm.
(3) beneficial effect
Compared with prior art, the beneficial effects of the present invention are: denoising methods of the invention by by hybrid particle swarm
Optimization algorithm is combined with m ultiwavelet adjacent coefficient denoising method, by the m ultiwavelet decomposition coefficient after vector quantization, passes through mangcorn
Subgroup optimization algorithm carries out threshold value optimizing, available threshold value more accurate than traditional uniform threshold method, and by should
Threshold process, is reconstructed using m ultiwavelet and post-processing reverts to one-dimensional de-noising signal, can be significantly reduced the dry of ambient noise
It disturbs, therefore noise reduction filtering excellent.
Detailed description of the invention
Illustrate the specific embodiment of the invention or technical solution in the prior art in order to clearer, it below will be to specific
Embodiment describes required attached drawing in the prior art and is briefly described, it should be apparent that, it is described below
Attached drawing is only one embodiment of the present invention, to those skilled in the art, is not being made the creative labor
Under the premise of, it is also possible to obtain other drawings based on these drawings.
Fig. 1 is rotor unbalanced signal time-domain diagram and its envelope spectrum in the specific embodiment of the invention.
Fig. 2 is rotor misalignment signal time-domain diagram and its envelope spectrum in the specific embodiment of the invention.
Fig. 3 is rotor rubbing signal time-domain diagram and its envelope spectrum in the specific embodiment of the invention.
Specific embodiment
To be easy to understand the technical means, the creative features, the aims and the efficiencies achieved by the present invention, below to this
Technical solution in invention specific embodiment carries out clear, complete description, with the present invention is further explained, it is clear that retouched
The specific embodiment stated is only a part of embodiment of the invention, rather than whole patterns.
Aero-engine is the key components on aircraft, is known as one " jewel " on aircraft industry at present
The big state of all aviations is all classified as aircraft engine maintenance technology and needs most conservative one of secret.And aero-engine due to
The precision and complexity of its components composition, so the probability to break down is higher, fault type is also to make a lot of variety, specific event
Hinder the aero-engine most common failure type and characteristic frequency of the visible such as the following table 1 of type, related Research statistics show aeroplane engine
Machine rotor system failure incidence is much higher than the failure of other types, and rotor-support-foundation system is the key that aero-engine composition portion
Point, therefore the maintenance of rotor-support-foundation system directly influences the service life of aircraft, is flight safety and lives and properties peace
Full important guarantee.Rotor-support-foundation system failure is the main reason for causing aero-engine to be shut down, according to aeroengine rotor system
The difference that system breaks down, it is also different to show characteristic frequency.
1 aero-engine most common failure type of table and characteristic frequency
The implementation case is the extracting method case study on implementation of aircraft engine rotor system fault-signal characteristic information, including
Following process:
A. to that there may be aeroengine rotors is uneven, misalign and the system of impact-rub malfunction in, rotor is turned
The acquisition that data sample is carried out when velocity modulation is to 2700r/min obtains the data that revolving speed is 2700r/min, sampling by sensor
Set of frequency is 10240Hz, and signal is that the sensor at rotor fault exerciser bearings at both ends in horizontal direction is collected
Discrete vibration acceleration signal;In view of signal, there may be errors at head and the tail endpoint, so to three kinds of fault-signals
When reason and calculating, only with 4096 data in data middle section, that is, the data more stable using rotor comprehensive state.
B. the collected rotor-support-foundation system imbalance fault vibration signal of institute is subjected to time frequency analysis, obtains Fig. 1 institute as shown in figure 1
The rotor unbalance signal time-domain diagram and its envelope spectrum shown;For the event in more clear convenient identification unbalanced signal
Hinder a large amount of interference components such as characteristic frequency and noise there are features, Envelope Analysis is carried out to signal, obtains imbalance fault
The envelope spectrum of signal, as shown in figure 1 shown in (b);By aircraft engine rotor system imbalance fault mechanism it is found that rotor not
The vibration performance of balance shows themselves in that fundamental frequency (turning frequency) vibration is obvious, without other apparent characteristic frequencies, so believing again imbalance
Low frequency part in number envelope spectrum has carried out partial enlargement processing, as shown in figure 1 shown in (c).
Place, time domain can be seen by the time domain waveform of the aircraft engine rotor system imbalance fault signal in the following figure
Waveform is disorderly and unsystematic, not the apparent periodic shock as caused by rotor-support-foundation system imbalance fault, by strong background
Noise is covered, and (c) can see in the Fig. 1 for passing through envelope spectrum, in the low-frequency range of imbalance fault vibration signal
(frequency range) there are the fundamental frequency shock responses of speed-frequency, the i.e. characteristic frequency of imbalance fault, but can see simultaneously,
In characteristic frequency with there are a large amount of noise jamming frequency contents around cluster, therefore imbalance fault signal need to be carried out at noise reduction
Reason.
Hereafter, then Db2 single wavelet, GHM m ultiwavelet, m ultiwavelet adjacent coefficient is respectively adopted and proposed by the present invention is based on
The m ultiwavelet adjacent coefficient noise-reduction method of Hybrid particle swarm optimization carries out noise reduction point to rotor-support-foundation system imbalance fault vibration signal
Analysis;As can be seen that by above-mentioned four kinds of noise-reduction methods to rotor-support-foundation system imbalance fault signal carry out noise reduction process after when
Domain signal is all more clear than the time domain waveform of original imbalance fault vibration signal, illustrates that four kinds of methods all have certain go
Make an uproar effect, thus compared and analyzed by noise reduction effect of the envelope spectrum to four kinds of methods it can be seen that, uneven therefore
Hindering in the low-frequency range of vibration signal (frequency range), four kinds of noise-reduction methods are all clearly present the fundamental frequency shock response of speed-frequency,
And the noise background signal around fundamental frequency has all obtained inhibition to a certain extent compared with original imbalance fault signal.But
It is by four kinds of denoising methods of comparison it is found that the m ultiwavelet adjacent coefficient denoising side proposed in this paper based on Hybrid particle swarm optimization
Although around the fundamental frequency of method there are still some small noise signal components, no matter from the quantity and wave of noise subharmonic
On peak size, denoising method proposed in this paper is compared with Db2 single wavelet, GHM m ultiwavelet and m ultiwavelet adjacent coefficient denoising method
More significant reduction has been obtained, has illustrated that noise-reduction method proposed in this paper can accomplish effective filter of rotor unbalance fault-signal
Wave, and noise reduction effect has more advantage compared with other three kinds of noise-reduction methods.
Equally, it carries out misaligning malfunction test using above-mentioned aircraft engine rotor system, by the collected rotor of institute
System misaligns vibration signal and carries out time frequency analysis, the rotor misalignment signal time-domain diagram and its envelope being illustrated in fig. 2 shown below
Demodulation spectra;Failure mechanism is misaligned by aircraft engine rotor system it is found that the main table of the vibration performance of rotor misalignment failure
It is existing are as follows: the evens frequency multiplication such as 2 frequencys multiplication and 4 frequencys multiplication is obvious, and present case carries out the high frequency section in signal envelope demodulation spectra thus
Partial enlargement processing, in Fig. 2 shown in (c).
It is compared and analyzed by noise reduction effect of the envelope spectrum to four kinds of denoising methods, by Fig. 2 of envelope spectrum
In (c) can be seen that in the high band for misaligning fault vibration signal (frequency range) that four kinds of noise-reduction methods can hold very much
The easy fundamental frequency shock response for finding speed-frequency, frequency multiplication shock response and frequency multiplication shock response containing certain impact amplitude,
By the suppressed degree of noise background signal around frequency multiplication shock response being made comparisons it is found that proposed by the present invention based on mixing
How small noise background signal around the m ultiwavelet adjacent coefficient denoising method frequency multiplication shock response of particle group optimizing is compared with original
Wave adjacent coefficient denoising method has obtained a degree of reduction, illustrates that the method for the present invention is misaligning failure letter to rotor-support-foundation system
Number noise reduction effect on promoted compared with original m ultiwavelet adjacent coefficient noise-reduction method, and in the inhibition of noise subharmonic
Also advantage is had more compared with other two kinds of noise-reduction methods.
Finally, carrying out time frequency analysis to by the collected vibration signal that rubs that touches of investigation of rotor rubbing faults experiment institute, obtain
Rotor rubbing signal time-domain diagram as shown in Figure 3 and its envelope spectrum.Periodic shock in original impact-rub malfunction signal, quilt
Strong ambient noise is covered and can not be identified, and (c) can be seen that in impact-rub malfunction in the Fig. 3 for passing through envelope spectrum
(frequency range) is not especially bright there are the fundamental frequency shock response of speed-frequency and impact amplitude in the low-frequency range of vibration signal
Aobvious frequency multiplication shock response, this is because the low-frequency range of rotor rubbing signal has strong noise jamming.And pass through envelope solution
(d) in spectrogram 3 is adjusted to can be seen that in the high band of impact-rub malfunction vibration signal (frequency range) it can be seen that four groups of interval phases
Deng frequency band cluster, spacing frequency is speed-frequency 45Hz, but is existed around four groups of equally spaced frequency band clusters a large amount of
Noise jamming frequency content.
The time domain waveform of time-domain signal ratio GHM m ultiwavelet denoising after having above-mentioned analysis to can be seen that noise reduction process is more
It is clear to add, and periodic shock is obvious, illustrates there is certain denoising effect, and can be seen that by each envelope spectrum region
In de-noising signal low-frequency range (frequency range) there are the fundamental frequency shock responses and shock response of speed-frequency, and in fundamental frequency
Around significantly interfere with ingredient there is no wave crest, illustrate that the noise background signal in low-frequency range is effectively inhibited, and from
Signal all unobvious, the nothing that can be seen that frequency multiplication shock response after Db2 single wavelet, GHM m ultiwavelet, m ultiwavelet adjacent coefficient noise reduction
Method effectively identifies, meanwhile, in de-noising signal high band (frequency range), therefore deduce that it is final remove dryness interpretation of result,
I.e. to noise-containing rotor-support-foundation system it is uneven, misalign and impact-rub malfunction signal, respectively by after noise reduction process, four kinds
Noise-reduction method can efficiently reduce the influence of the ambient noise interference component in signal, improve the signal-to-noise ratio of fault-signal.
But noise-reduction method proposed by the present invention better than other denoising methods in three in inhibition of the different frequency ranges to noise subharmonic,
Therefore the present invention proposes that the denoising effect of the m ultiwavelet adjacent coefficient noise-reduction method based on Hybrid particle swarm optimization is more preferable, can accomplish
Effective noise reduction of rotor fault signal.
The foregoing describe technical characteristics of the invention and basic principle and associated advantages, for those skilled in the art
For, it is clear that the present invention is not limited to the details of above-mentioned exemplary embodiment, and without departing substantially from design of the invention or
In the case where essential characteristic, the present invention can be realized in other specific forms.Therefore, in all respects, should all incite somebody to action
Above-mentioned specific embodiment regards exemplary as, and is non-limiting, the scope of the present invention by appended claims and
It is not that above description limits, it is intended that all changes that come within the meaning and range of equivalency of the claims are included
In the present invention.
Although not each embodiment is only in addition, it should be understood that this specification is described according to each embodiment
It contains an independent technical solution, this description of the specification is merely for the sake of clarity, and those skilled in the art answer
When considering the specification as a whole, the technical solution in each embodiment may also be suitably combined to form art technology
The other embodiments that personnel are understood that.
Claims (1)
1. a kind of denoising method of aircraft engine rotor system fault vibration signal, which is characterized in that specific step are as follows:
Step 1: measuring a certain number of aviations by determining time interval or sample frequency by eddy current displacement sensor
Engine rotor touches the vibration signal that rubs some sampling periods;
It carries out repeating row filtering processing Step 2: the aeroengine rotor collected to be touched to the vibration signal that rubs, and to repetition
Signal after row filtering processing carries out m ultiwavelet decomposition and reconstruct, obtains m ultiwavelet coefficient, and m ultiwavelet is decomposed and is reconstructed under
Formula carries out,
In formula, cj,k=(c1,j,k,···,cr,j,k)TFor the low frequency component of r dimension, dj,k=(d1,j,k,···,dr,j,k)TFor r
The high fdrequency component of dimension, * are conjugate transposition operation;
Step 3: the invariant in setting Hybrid Particle Swarm Optimization, wherein setting Studying factors c1=c2=1.4995,
The quantity m=40 of population, maximum number of iterations T=200, the maximum value ω of weight coefficientmax=0.9, the minimum of weight coefficient
Value ωmin=0.4, the maximum speed v of particlemax=0.2 λmax, the maximum value λ of population location parameter λmaxWith minimum value λmin
It is carried out by following formula,
λj=(2log nj)/q
In formula, njλ can be respectively obtained when q takes 0.1 and 1 respectively for the length of m ultiwavelet coefficientmaxAnd λmin;
Step 4: obtaining optimal threshold by the function based on GCV criterion fitness, the fitness of each particle in group is obtained
Value, the function based on GCV criterion fitness are following formula,
In formula, D and DλM ultiwavelet coefficient after respectively noisy m ultiwavelet coefficient and threshold value λ rule process, N are m ultiwavelet coefficient
Total number, N00 number is set in thresholding process for m ultiwavelet coefficient;
Step 5: the desired positions p for living through its fitness value with it to each particleiFitness value make comparisons, if compared with
It is good, then by xiAs current desired positions pi, and then desired positions that its fitness value and all particles are lived through
Fitness value is made comparisons;If preferably, as the desired positions p of current all particlesg;
Step 6: obtaining individual extreme value and global extremum by step 5 and then respectively by particle in following formula Population Regeneration
Speed and position,
vij(t+1)=wvij(t)+c1r1(t)[pij(t)-xij(t)]+c2r2(t)[pgj(t)-xij(t)]
xij(t+1)=xij(t)+vij(t+1)
In formula, w indicates inertia weight coefficient, c1And c2It represents Studying factors (also referred to as acceleration constant), i=1,2,
D, vijIndicate the speed of particle, vij∈[-vmax, vmax], vmaxIt is constant;
Step 7: whether the algorithm of judgment step six has reached maximum number of iterations, it is back to step 4 if not up to and continues
Operation, if having reached, exports optimal value λbest;
Step 8: enabling tabu search algorithm, and maximum number of iterations is set, and empties taboo list, by Hybrid particle swarm optimization
The optimal solution λ that algorithm generatesbestCurrent solution is assigned to as the initial solution of tabu search algorithm, and initial solution;
Step 9: generating several neighborhood solutions using current solution neighborhood function, and therefrom determine several candidate solutions;
Step 10: obtaining each particle in group by obtaining the optimal threshold of the algorithm based on GCV criterion fitness function
Fitness value, and judge whether current candidate solution meets special pardon criterion with this, the best of rule is specially pardoned with satisfaction if meeting
Candidate solution, which replaces, currently to be solved, and using optimal candidate solution as the replacement of taboo object earliest into the object of taboo list, while with completely
The best candidate solution of foot special pardon rule replaces current optimal solution, then jumps directly to step 12 operation;If being unsatisfactory for
Then step 11 is executed downwards;Function based on GCV criterion fitness is following formula,
In formula, D and DλM ultiwavelet coefficient after respectively noisy m ultiwavelet coefficient and threshold value λ rule process, N are m ultiwavelet coefficient
Total number, N00 number is set in thresholding process for m ultiwavelet coefficient;
Step 11: the taboo attribute of candidate solution is determined, if candidate when candidate solution is inferior to current optimal solution better than current solution
Candidate solution is then assigned to the current solution of next iteration not in taboo list by solution, while entering taboo list earliest with its replacement
Object;It is inferior to currently to solve in candidate solution and when also inferior to current optimal solution, new neighborhood solution is regenerated with current solution;
Step 12: judging whether algorithm has reached termination condition, optimal solution and termination algorithm are exported if reaching;If not yet
Have and reach, going to step nine continues to run;
Step 13: handling according to threshold rule m ultiwavelet coefficient, and m ultiwavelet is carried out to coefficient after treatment
Reconstruct, obtains the denoised signal of aircraft engine rotor system fault vibration.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910566648.8A CN110146156B (en) | 2019-06-27 | 2019-06-27 | Denoising method for fault vibration signal of aircraft engine rotor system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910566648.8A CN110146156B (en) | 2019-06-27 | 2019-06-27 | Denoising method for fault vibration signal of aircraft engine rotor system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110146156A true CN110146156A (en) | 2019-08-20 |
CN110146156B CN110146156B (en) | 2021-02-09 |
Family
ID=67596552
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910566648.8A Expired - Fee Related CN110146156B (en) | 2019-06-27 | 2019-06-27 | Denoising method for fault vibration signal of aircraft engine rotor system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110146156B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110470462A (en) * | 2019-08-22 | 2019-11-19 | 苏州旋械感知信息科技有限公司 | One kind being based on C0The reconstructing method of the mechanical system fault features of complexity |
CN111323227A (en) * | 2020-01-03 | 2020-06-23 | 南昌航空大学 | Method for extracting fault features of aeroengine rotor |
CN112345260A (en) * | 2020-10-16 | 2021-02-09 | 南昌航空大学 | Micro-heating-section vibration testing method based on micro-particle distribution |
CN117092206A (en) * | 2023-08-09 | 2023-11-21 | 国网四川省电力公司电力科学研究院 | Defect detection method for cable lead sealing area, computer equipment and storage medium |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103279813A (en) * | 2013-06-21 | 2013-09-04 | 哈尔滨工业大学(威海) | Steam load prediction method |
KR101549983B1 (en) * | 2014-02-28 | 2015-09-03 | 김종원 | The self-adaptive method for inserting/detecting a forensic mark using particle swarm optimization in the multiple transform domain and the apparatus thereof |
CN105651504A (en) * | 2016-01-15 | 2016-06-08 | 北京工业大学 | Rotary machinery fault feature extraction method based on self-adaptive wavelet energy |
CN107063427A (en) * | 2016-02-10 | 2017-08-18 | 株式会社神户制钢所 | The abnormality detection system and the method for detecting abnormal of whirler of whirler |
CN107506709A (en) * | 2017-08-15 | 2017-12-22 | 南昌航空大学 | Aircraft engine rotor system fault-signal characteristics information extraction method |
CN108550116A (en) * | 2018-03-12 | 2018-09-18 | 西安理工大学 | The self-adapting random resonant denoising method of silicon monocrystal growth image under low signal-to-noise ratio |
CN108731945A (en) * | 2018-08-02 | 2018-11-02 | 南昌航空大学 | A kind of extracting method of aircraft engine rotor system fault-signal characteristic information |
CN108760327A (en) * | 2018-08-02 | 2018-11-06 | 南昌航空大学 | A kind of diagnostic method of aeroengine rotor failure |
WO2019061006A1 (en) * | 2017-09-26 | 2019-04-04 | Schaeffler Technologies AG & Co. KG | Bearing failure diagnosis method and device, readable storage medium, and electronic device |
-
2019
- 2019-06-27 CN CN201910566648.8A patent/CN110146156B/en not_active Expired - Fee Related
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103279813A (en) * | 2013-06-21 | 2013-09-04 | 哈尔滨工业大学(威海) | Steam load prediction method |
KR101549983B1 (en) * | 2014-02-28 | 2015-09-03 | 김종원 | The self-adaptive method for inserting/detecting a forensic mark using particle swarm optimization in the multiple transform domain and the apparatus thereof |
CN105651504A (en) * | 2016-01-15 | 2016-06-08 | 北京工业大学 | Rotary machinery fault feature extraction method based on self-adaptive wavelet energy |
CN107063427A (en) * | 2016-02-10 | 2017-08-18 | 株式会社神户制钢所 | The abnormality detection system and the method for detecting abnormal of whirler of whirler |
CN107506709A (en) * | 2017-08-15 | 2017-12-22 | 南昌航空大学 | Aircraft engine rotor system fault-signal characteristics information extraction method |
WO2019061006A1 (en) * | 2017-09-26 | 2019-04-04 | Schaeffler Technologies AG & Co. KG | Bearing failure diagnosis method and device, readable storage medium, and electronic device |
CN108550116A (en) * | 2018-03-12 | 2018-09-18 | 西安理工大学 | The self-adapting random resonant denoising method of silicon monocrystal growth image under low signal-to-noise ratio |
CN108731945A (en) * | 2018-08-02 | 2018-11-02 | 南昌航空大学 | A kind of extracting method of aircraft engine rotor system fault-signal characteristic information |
CN108760327A (en) * | 2018-08-02 | 2018-11-06 | 南昌航空大学 | A kind of diagnostic method of aeroengine rotor failure |
Non-Patent Citations (4)
Title |
---|
HONG LIANHUAN, ET AL: "Compound fault diagnosis of rotating machinery based on adaptive maximum correlated kurtosis deconvolution and customized multiwavelet transform", 《MEASUREMENT SCIENCE AND TECHNOLOGY》 * |
LU ZHENYU,ET AL: "An improved FCM method for Image segmentation based on Wavelet transform and Particle swarm", 《IEEE》 * |
刘晓波 等: "基于广度优先搜索的小波聚类算法", 《振动与冲击》 * |
申江江 等: "基于改进粒子群算法的直升机振动信号盲分离", 《电子测量与仪器学报》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110470462A (en) * | 2019-08-22 | 2019-11-19 | 苏州旋械感知信息科技有限公司 | One kind being based on C0The reconstructing method of the mechanical system fault features of complexity |
CN110470462B (en) * | 2019-08-22 | 2021-09-28 | 苏州旋械感知信息科技有限公司 | Based on C0Reconstruction method of early fault characteristics of mechanical system with complexity |
CN111323227A (en) * | 2020-01-03 | 2020-06-23 | 南昌航空大学 | Method for extracting fault features of aeroengine rotor |
CN112345260A (en) * | 2020-10-16 | 2021-02-09 | 南昌航空大学 | Micro-heating-section vibration testing method based on micro-particle distribution |
CN112345260B (en) * | 2020-10-16 | 2022-06-24 | 南昌航空大学 | Micro-heating-section vibration testing method based on micro-particle distribution |
CN117092206A (en) * | 2023-08-09 | 2023-11-21 | 国网四川省电力公司电力科学研究院 | Defect detection method for cable lead sealing area, computer equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN110146156B (en) | 2021-02-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110146156A (en) | A kind of denoising method of aircraft engine rotor system fault vibration signal | |
Miao et al. | Improvement of kurtosis-guided-grams via Gini index for bearing fault feature identification | |
CN108760327B (en) | Diagnosis method for rotor fault of aircraft engine | |
Yu et al. | Weak fault feature extraction of rolling bearings using local mean decomposition-based multilayer hybrid denoising | |
Liao et al. | Extraction of repetitive transients with frequency domain multipoint kurtosis for bearing fault diagnosis | |
Wang et al. | Weak fault diagnosis of rolling bearing under variable speed condition using IEWT-based enhanced envelope order spectrum | |
CN103294579A (en) | Method for testing high-performance computing cluster application performance | |
Sheng et al. | Applications in bearing fault diagnosis of an improved Kurtogram algorithm based on flexible frequency slice wavelet transform filter bank | |
CN108801634B (en) | The method and its application of bearing fault characteristics frequency are extracted based on singular value decomposition and the frequency band entropy of optimization | |
Wang et al. | Fault diagnosis of wind turbine bearing based on optimized adaptive chirp mode decomposition | |
CN108731945A (en) | A kind of extracting method of aircraft engine rotor system fault-signal characteristic information | |
Xu et al. | An adaptive spectrum segmentation method to optimize empirical wavelet transform for rolling bearings fault diagnosis | |
CN111881594A (en) | Non-stationary signal state monitoring method and system for nuclear power equipment | |
Chi et al. | Bearing fault diagnosis through stochastic resonance by full-wave signal construction with half-cycle delay | |
Tan et al. | Rolling bearing incipient fault detection via optimized VMD using mode mutual information | |
Sun et al. | Application of wavelet soft threshold denoising algorithm based on EMD decomposition in vibration signals | |
CN109900447A (en) | A kind of circulation impact method for detecting vibration and system based on harmonic signal decomposition | |
Zhang et al. | An adaptive demodulation approach for bearing fault detection based on adaptive wavelet filtering and spectral subtraction | |
Zheng et al. | Zero-phase filter-based adaptive Fourier decomposition and its application to fault diagnosis of rolling bearing | |
Wang et al. | Fault early recognition and health monitoring on aeroengine rotor system | |
CN111323227A (en) | Method for extracting fault features of aeroengine rotor | |
CN111380680A (en) | Check valve fault feature extraction method based on improved permutation entropy | |
Shen et al. | Informative singular value decomposition and its application in fault detection of planetary gearbox | |
CN110186684A (en) | A kind of aero-engine mechanical oscillation fault-signal feature extracting method | |
Zhou et al. | A signal-filtering and feature-enhancement method based on ensemble local mean decomposition and adaptive morphological filtering |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20210209 |