CN108694417A - A kind of sliding bearing-rotor system amount of unbalance recognition methods - Google Patents
A kind of sliding bearing-rotor system amount of unbalance recognition methods Download PDFInfo
- Publication number
- CN108694417A CN108694417A CN201810934874.2A CN201810934874A CN108694417A CN 108694417 A CN108694417 A CN 108694417A CN 201810934874 A CN201810934874 A CN 201810934874A CN 108694417 A CN108694417 A CN 108694417A
- Authority
- CN
- China
- Prior art keywords
- unbalance
- amount
- sliding bearing
- rotor system
- sample
- 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
- 238000000034 method Methods 0.000 title claims abstract description 58
- 238000009826 distribution Methods 0.000 claims abstract description 65
- 230000004044 response Effects 0.000 claims abstract description 43
- 238000005070 sampling Methods 0.000 claims abstract description 25
- 238000012360 testing method Methods 0.000 claims abstract description 23
- 230000002068 genetic effect Effects 0.000 claims description 16
- 238000005516 engineering process Methods 0.000 claims description 11
- 238000004364 calculation method Methods 0.000 claims description 9
- 238000013401 experimental design Methods 0.000 claims description 8
- 238000004458 analytical method Methods 0.000 claims description 7
- 238000004088 simulation Methods 0.000 claims description 5
- 238000013461 design Methods 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 4
- 238000005457 optimization Methods 0.000 claims description 4
- 238000012546 transfer Methods 0.000 claims description 3
- 230000003044 adaptive effect Effects 0.000 claims 1
- 230000006870 function Effects 0.000 description 23
- 239000012141 concentrate Substances 0.000 description 6
- 238000006073 displacement reaction Methods 0.000 description 5
- 238000002474 experimental method Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000005259 measurement Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 239000004744 fabric Substances 0.000 description 3
- 238000012804 iterative process Methods 0.000 description 3
- 230000001052 transient effect Effects 0.000 description 3
- 230000005012 migration Effects 0.000 description 2
- 238000013508 migration Methods 0.000 description 2
- 230000010355 oscillation Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 238000005299 abrasion Methods 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000006880 cross-coupling reaction Methods 0.000 description 1
- 238000013016 damping Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000003050 experimental design method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000011900 installation process Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000010181 polygamy Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000002922 simulated annealing Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/29—Graphical models, e.g. Bayesian networks
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M1/00—Testing static or dynamic balance of machines or structures
- G01M1/14—Determining imbalance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Testing Of Balance (AREA)
Abstract
The present invention relates to a kind of sliding bearing-rotor system amount of unbalance recognition methods, belong to uncertain indirect problem technical field, a large amount of sample point will be generated in sampling by solving bayesian theory and MCMC methods, and solve time-consuming problem serious, efficiency is low when direct problem.Including:Obtain amount of unbalance prior distribution space, initial sample and test unbalance response;Cost function is solved, and obtains the distribution of posteriority joint probability density;Calculate the minimum value of cost function, when minimum value is more than convergence precision, update prior distribution space, and utilize hereditary intellegent sampling technical limit spacing new samples, and then calculate new cost function, otherwise, the approximate edge posterior probability Density Distribution for obtaining amount of unbalance is identified using MCMC methodology, and then determines mean value, the confidence interval of amount of unbalance.This method can obtain higher computational efficiency under the premise of not sacrificing computational accuracy, can that accurately and quickly identify the information such as mean value and the confidence interval of amount of unbalance.
Description
Technical field
The present invention relates to uncertain indirect problem technical field more particularly to a kind of sliding bearing-rotor system are uneven
Measure recognition methods.
Background technology
In the caused deformation of inhomogeneities, manufacture and installation process and work of sliding bearing-rotor system material
Abrasion etc. can all cause the unbalance vibration of bearing rotor system.Shafting dynamic balance is carried out using amount of unbalance identification technology to carry
The product quality of high rotor and its composition reduces noise and vibration, improves the service life on probation of bearing, to ensure the length of shafting operation
Phase property and stability are a kind of calibrating modes that bearing rotor system is commonly used.But due to the diversity of influence factor and again
Polygamy, there are certain errors for amount of unbalance recognition result.For sliding bearing-rotor system, rotor geometric properties, sliding axle
Even holding oil film characteristic coefficient and these uncertain factors of randomness for measuring response in the case that smaller, it is also likely to lead
Amount of unbalance recognition result is caused to generate larger deviation.And engineering is in practice, engineer can be rule of thumb with knowledge to imbalance
Amount parameter has certain pre-estimate before not yet obtaining experiment metrical information.How to be reduced using these prior informations not true
Determine factor and treats the influence of identification parameter as the research hotspot in uncertain indirect problem field.
Bayesian theory carries out reverse analysis to structural model parameter, while considering known parameters probability density sample letter
Breath and unknown parameter prior information.But what bayesian theory solution engineering uncertainty indirect problem often referred to take very much just asks
Topic calculates, it is difficult to meet requirement of the Practical Project to computational efficiency.Meanwhile MCMC (Markov Chain Monte Carlo, horse
Er Kefu chains Monte Carlo) method will generate a large amount of sample point when obtaining amount of unbalance edge posterior probability Density Distribution, and it wants
It is a large amount of that time-consuming direct problem is called to calculate, the low problem of efficiency.
Invention content
In view of above-mentioned analysis, the present invention is intended to provide a kind of sliding bearing-rotor system amount of unbalance recognition methods, is used
Serious, effect is taken when generating a large amount of sample point in sampling to solve bayesian theory and MCMC methods, and solve direct problem
The low problem of rate.
The purpose of the present invention is mainly achieved through the following technical solutions:
A kind of sliding bearing-rotor system amount of unbalance recognition methods is provided, is included the following steps:
Step S1, sliding bearing-rotor system amount of unbalance prior distribution space, initial sample and sliding bearing-are obtained
The test unbalance response of rotor-support-foundation system;
Step S2 solves cost function using bayesian theory, and obtains the posteriority joint probability density point of amount of unbalance
Cloth;
Step S3 calculates the minimum value of above-mentioned cost function, when the minimum value is more than convergence precision, updates the cunning
Dynamic bearing-rotor-support-foundation system amount of unbalance prior distribution space, and using hereditary intellegent sampling technical limit spacing new samples, enter step
Otherwise S2 enters step S4;The heredity intellegent sampling technology is that old sample is genetic in new samples;
Step S4, the approximate edge posteriority for identifying to obtain sliding bearing-rotor system amount of unbalance using MCMC methodology are general
Rate Density Distribution, and then determine mean value, the confidence interval of sliding bearing-rotor system amount of unbalance.
The present invention has the beneficial effect that:The present invention is based on the sliding bearing-rotor system of hereditary intellegent sampling technology injustice
Recognition methods is weighed, the calculation times of sample are greatly reduced, keeps collecting sample and priori spatial gradually close to joint posterior probability
The high-density region of degree distribution is concentrated, and is improved sampling efficiency, and hereditary effective sample point, is reduced direct problem call number.Effectively
Recognition result is evaluated on ground, reduces influence of the uncertain factor to recognition result.This method is not sacrificing computational accuracy
Under the premise of can obtain higher computational efficiency, can that accurately and quickly identify mean value and confidence interval of amount of unbalance etc.
Information.In turn, it carries out rotor dynamic balancing for engineer and aid decision foundation is provided.
On the basis of said program, the present invention has also done following improvement:
Further, described using hereditary intellegent sampling technical limit spacing new samples, to be set by the experiment of hereditary Latin hypercube
New samples in meter update prior distribution space, including:
According to minimax distance criterion, the sample to dropping into prior distribution of lower generation space screens;
Make new samples project the new samples uniformly and generated on each design variable using the fiery optimization solver of simulation to arrive
The distance of genetic sample is maximum.
Advantageous effect using above-mentioned further scheme is:Constantly more by hereditary Latin hypercube experimental design (ILHD)
The sample in area is trusted in new priori spatial and heredity, makes collecting sample and priori spatial gradually to joint posterior probability Density Distribution
High-density region concentrate, improve sampling efficiency.
Further, the solution cost function, including:
It is imitated by sliding bearing-rotor system unbalance response using amount of unbalance as input variable using transfer matrix method
True analysis, seeks sliding bearing-rotor system unbalance response;
Using above-mentioned sliding bearing-rotor system test the error between unbalance response and the unbalance response sought as
Cost function.
Further, the update prior distribution space, to obtain prior distribution of lower generation space according to trusted zones update method.
Further, the trusted zones update method is:Amount of unbalance prior distribution is adaptively adjusted by Fibonacci method
The up-and-down boundary X in spaceLAnd XR, and then obtain the prior distribution space of sampled point of lower generation.
Further, the acquisition amount of unbalance prior distribution includes:
Error constant ε is set, the equally distributed initial up-and-down boundary of sliding bearing-rotor system amount of unbalance priori is givenWithGive each iteration step sample point quantity Ns i。
Further, the posteriority joint probability density distribution p (X| of the amount of unbalance;Q) it is:
In formula, X is sliding bearing-rotor system amount of unbalance;Q is that the test imbalance of sliding bearing-rotor system is rung
It answers;A is constant;G (X) is cost function;σ2For the variance of sliding bearing-rotor system random noise.
Further, cost function calculation formula is:
In formula, X is sliding bearing-rotor system amount of unbalance, and h is measuring point number, and q is sliding bearing-rotor system
Test unbalance response, qj(X) it is that amount of unbalance is emulated as input variable by sliding bearing-rotor system unbalance response
Analyze the unbalance response obtained.
Further, the prior distribution space for obtaining sampled point of lower generation is:
In formula, 1- α are golden section point, α=0.382;WithThe upper and lower side in prior distribution space is sampled for the i-th generation
Boundary;WithFor the upper and lower boundary in i+1 generation sampling prior distribution space.
Further, in MCMC methodology, chain length takes 105Magnitude.
It in the present invention, can also be combined with each other between above-mentioned each technical solution, to realize more preferred assembled schemes.This
Other feature and advantage of invention will illustrate in the following description, also, certain advantages can become from specification it is aobvious and
It is clear to, or understand through the implementation of the invention.The purpose of the present invention and other advantages can by specification, claims with
And it realizes and obtains in specifically noted content in attached drawing.
Description of the drawings
Attached drawing is only used for showing the purpose of specific embodiment, and is not considered as limitation of the present invention, in entire attached drawing
In, identical reference mark indicates identical component.
Fig. 1 is sliding bearing-rotor system amount of unbalance recognition methods flow chart in the embodiment of the present invention 1;
Fig. 2 is rotor testbed and testing scheme in the embodiment of the present invention 2;
Fig. 3 is rotor testbed parameter model in the embodiment of the present invention 2;
Fig. 4 is rotor testbed transfer matrix method model in the embodiment of the present invention 2;
Fig. 5 is two measuring point x radial direction transient responses of rotor testbed in the embodiment of the present invention 2;
Fig. 6 is two measuring point y radial direction transient responses of rotor testbed in the embodiment of the present invention 2;
Fig. 7 is the hereditary intelligent updating iterative process (first step) of sample point in the embodiment of the present invention 2;
Fig. 8 is the hereditary intelligent updating iterative process (second step) of sample point in the embodiment of the present invention 2;
Fig. 9 is the hereditary intelligent updating iterative process (third step) of sample point in the embodiment of the present invention 2;
Figure 10 is the two-dimensional random migration figure sampled in the embodiment of the present invention 2;
Figure 11 is the one-dimensional random migration figure sampled in the embodiment of the present invention 2;
Figure 12 is that the X in the embodiment of the present invention 3 before and after two measuring point dynamic balancing compares figure to displacement time domain response;
Figure 13 is that figure is compared in the Y-direction displacement time domain response in the embodiment of the present invention 3 before and after two measuring point dynamic balancing.
Specific implementation mode
Specifically describing the preferred embodiment of the present invention below in conjunction with the accompanying drawings, wherein attached drawing constitutes the application part, and
It is used to illustrate the principle of the present invention together with embodiments of the present invention, be not intended to limit the scope of the present invention.
Embodiment 1
In order on the one hand be utilized by sliding bearing-rotor system Service Environment in sliding bearing-rotor dynamic balancing application
The sample information of uncertain factor present in variation, the limitation of test condition and sliding bearing-rotor system information, to reduce
Influence of the uncertain factor to amount of unbalance recognition result.Another aspect utilizing works teacher is rule of thumb with knowledge to amount of unbalance
Parameter pre-estimates information to improve the accuracy of identification of amount of unbalance before not yet obtaining experiment metrical information.The present embodiment
It is obtained not using the bayesian theory and MCMC methods of the prior information for the sample information and amount of unbalance that can handle uncertain factor
Mean value, the confidence interval of aequum.It is asked for computational efficiency of the bayesian theory in the identification of bearing rotor system amount of unbalance
Topic, it is proposed that a kind of sliding bearing-rotor system amount of unbalance recognition methods based on hereditary intellegent sampling technology, including it is following
Step:
Step S1, sliding bearing-rotor system amount of unbalance prior distribution space, initial sample and sliding bearing-are obtained
The test unbalance response of rotor-support-foundation system;
Step S2 solves cost function using bayesian theory, and obtains the posteriority joint probability density point of amount of unbalance
Cloth;
Step S3 calculates the minimum value of above-mentioned cost function, when the minimum value is more than convergence precision, updates the cunning
Dynamic bearing-rotor-support-foundation system amount of unbalance prior distribution space, and using hereditary intellegent sampling technical limit spacing new samples, enter step
Otherwise S2 enters step S4;The heredity intellegent sampling technology is that old sample is genetic in new samples;
Step S4, the approximate edge posteriority for identifying to obtain sliding bearing-rotor system amount of unbalance using MCMC methodology are general
Rate Density Distribution, and then determine mean value, the confidence interval of sliding bearing-rotor system amount of unbalance.
Compared with prior art, sliding bearing-rotor system amount of unbalance recognition methods provided in this embodiment, subtracts significantly
The small calculation times of sample make the high-density region of collecting sample and priori spatial gradually to joint posterior probability Density Distribution
It concentrates, improves computational efficiency.Effectively recognition result is evaluated, reduces influence of the uncertain factor to recognition result.
This method can obtain higher computational efficiency under the premise of not sacrificing computational accuracy, can that accurately and quickly identify injustice
The information such as the mean value and confidence interval of measurement.These information carry out rotor dynamic balancing for engineer and provide aid decision foundation.
The identification of rotor unbalance value carries out on the basis of rotor dynamics is analyzed, i.e., according to rotor dynamics
Basic analyzing method establishes the equation of motion of rotor, and then the sliding bearing by that can predict-rotor oscillation response (ring by imbalance
Answer), in conjunction with the equation of motion, foundation includes the amount of unbalance identification equation of unknown system parameter.The angular speed of rotor-support-foundation system, damping
With stiffness matrix it is known that amount of unbalance (quality, eccentricity, phase) be unknown parameter as parameter to be identified, unbalance response
It can be obtained by testing to measure, amount of unbalance is identified by unbalance response.When unknown parameter or Structural Parameters of its Rotor have
When uncertain factor, unbalance response is also uncertain.In this case it is not to be based on unbalance response identification amount of unbalance
Certainty identifies problem.
In bayesian theory, when structure unknown parameter joint probability density obtains cost function, to obtain high-precision
Joint posterior probability Density Distribution space will largely be sampled, and a kind of effective way can be made in priori distribution space
What is generated concentrates on presenting the height of true joint posterior probability Density Distribution for building the finite sample point needed for cost function
Density area.It is the larger space in relatively true section by given parameter prior distribution in Practical Project, therefore, sampling
When constantly to adjust prior distribution space to concentrate limited sample point to reflect true posteriority space high density to the maximum extent
Region, and then ensure the precision of joint Posterior probability distribution.The present embodiment is based on hereditary intellegent sampling technology sampling, passes through trust
Domain model management method detects non-domination solution region, constantly updates prior distribution space, to ensure to obtain and really solve close
Section.Technology is layouted by intelligence and sample Genetic Strategies make the sample in each trusted zones be uniformly distributed, and passes through hereditary part
The point that sample falls into trusted zones of lower generation is layouted as intelligence, is reduced direct problem calculation times and is improved key area in trusted zones
Precision, to accelerate convergence rate.
Specifically, step S101 obtains amount of unbalance prior distribution;
Error constant (i.e. convergence precision) ε is set, sliding bearing-rotor system amount of unbalance priori is equally distributed initial
Up-and-down boundaryWithGive each iteration step sample point quantity Ns i。
It can be set as relatively small it should be noted that giving initial sample point quantity, sample point at this time is mainly used
In the subsequently encrypted direction of detection sample point, while being not easy to too small, smaller sample point and will be unable to reaction actual parameter joint
The main feature of posterior probability Density Distribution, this can increase the number of encryption iteration.
Bound is by reality to be indicated with being uniformly distributed for the prior distribution of sliding bearing-rotor system amount of unbalance
Experience or expertise are trampled to determine.
Step S102, samples sample;Preferably, sliding bearing-rotor system imbalance is obtained using LHD methods
The initial sample X of amount1;In often step iteration later, by hereditary Latin hypercube experimental design method, sliding axle is obtained
Hold-the new samples of rotor-support-foundation system amount of unbalance.
Step S103, measurement obtain the test unbalance response of sliding bearing-rotor system (such as:Immediate movement responds);
In step s 2, cost function is solved;It is rung by the test imbalance of the sliding bearing-rotor system of above-mentioned acquisition
It answers, and calculates and seek the imbalance sound that amount of unbalance is obtained as input variable by rotor-support-foundation system unbalance response simulation analysis
It answers, obtains cost function, and then obtain the posteriority joint probability density distribution of amount of unbalance;
Specifically, using following bayesian theory approximate formulas, the posteriority of sliding bearing-rotor system amount of unbalance is sought
Joint probability density is distributed:
X is sliding bearing-rotor system amount of unbalance (quality m, eccentricity e, phase in formula);Q turns for sliding bearing-
The test unbalance response of subsystem;p(X|Q) it is the posteriority joint probability density of sliding bearing-rotor system amount of unbalance point
Cloth;A is constant;G (X) is cost function;H is measuring point number;qj(X) be sliding bearing-rotor system amount of unbalance as defeated
Enter unbalance response (the i.e. unbalance response meter that variable is obtained by sliding bearing-rotor system unbalance response simulation analysis
Calculation value).
The uncertainty of test unbalance response is characterized with independent random noise, be mean value is 0, variance σ2Normal state
Distribution.
It is emphasized that cost function g (X), which is sliding bearing-rotor system, tests unbalance response and its imbalance
Error between response computation value, i.e. error under least square meaning use cost function as indicator in the present embodiment
To adjust the priori spatial of amount of unbalanceSo that it gradually covers high density posteriority space.
Wherein, the unbalance responses value of sliding bearing-rotor system is by transmitting moments method, in conjunction with sliding bearing-
What equation of rotor motion was sought, high-precision cost function is built with a small amount of effective sample, and then it is uneven to obtain unknown parameter
The posteriority joint probability density distribution p (X| of measurement;q).
In step s3, the minimum value of above-mentioned cost function is calculated, and by this minimum value and preset convergence precision
It is compared;When above-mentioned minimum value is more than convergence precision, priori spatial is updated, and new using hereditary intellegent sampling technical limit spacing
Sample enters step S2, and cost function is sought using new samples;Otherwise, S4 is entered step;
Its true posteriority space whether is contained since prior distribution is uncertain, can only reflect parameter posteriority space sometimes
Local message, sometimes even differ larger, it is limited to concentrate to the maximum extent therefore, it is necessary to adjust prior distribution space
Sample point reacts its true posteriority space high-density region, and then ensures the approximation quality of joint posterior probability Density Distribution.Tool
Body,
Step S301 calculates the minimum value of cost function;And by the convergence precision set in this minimum value and step S1 into
Row comparison;
For the minimum value of the i-th step cost function, obtaining in this step leads to cost function minimum value
Amount of unbalance vectorSince error is minimum at this time, illustrate that the high-density region of joint priori probability density distribution concentrates on
Sample pointAround.By this minimum value and convergence precision ε, if yminIt is high then to combine posterior probability Density Distribution by≤ε
Density area obtains, and iteration terminates, and enters step S5.When more than convergence precision, then entering step S402.
Step S302 determines trusted zones of lower generation region (i.e. the prior distribution space in lower generation) according to trusted zones update method;
When more than convergence precision, the equally distributed up-and-down boundary of amount of unbalance priori is adaptively adjusted with Fibonacci methodWithTo obtain the trusted zones region of sampled point of lower generation.
It is with the amount of unbalance priori spatial relationship currently walked in next step:
1- α are golden section point, α=0.382 in formula;WithThe up-and-down boundary of priori spatial is walked for current i;WithFor the up-and-down boundary of (i+1 steps) priori spatial in next step.
The equally distributed up-and-down boundary of amount of unbalance priori is updated according to trusted zones update methodWithEnsure to generate
Finite sample point can reflect unknown parameter amount of unbalance joint posterior probability Density Distribution high-density region.
Step S303 is sampled in the new priori spatial of above-mentioned acquisition based on hereditary intellegent sampling technology;
Lower generation trusted zones have the region overlapped with the trusted zones in the present age.The sample point in the present age may fall into the trust of lower generation
In domain.The present embodiment combines the new samples that the old sample of heredity and hereditary Latin hypercube experimental design (ILHD) generate
The sample point vertical as cost function of lower generation structure, can greatly reduce the total sample for needing to do unbalance responses in this way
Number improves computational efficiency.
In addition, if sample obtained in the previous step all entailed in next step, a part of area sample will produce excessively
Sample that is compact and being unfavorable for intelligently layouting is uniformly distributed.For this purpose, the present embodiment uses sample Genetic Strategies as far as possible fully
Latin hypercube experimental design is optimized according to minimax distance criterion using under the principle of the old sample of heredity, is screened
Part drops into the sample of trusted zones of lower generation, meanwhile, make new samples in each design variable using simulated annealing optimization solver
Upper projection is uniformly and the distance of the new samples of generation to genetic sample is maximum, and genetic sample and the new new samples that generate are in the trust of lower generation
Domain region keeps space to be evenly distributed with property, projection uniformity.
Wherein, according to minimax distance principle, sample set S is calculatedi+1Minimum range dmin(the i.e. most narrow spacing of adjacent sample
From), such as following formula:
In formula, NsFor trusted zones total sample number of lower generation;Respectively i+1 is for jth in sample set and k-th
Sample point.
Fall into the genetic sample of trusted zones of lower generationTo sample set S of lower generationi+1Maximum distance DmaxTo seek formula as follows
It is shown:
In formula,For the i-th generation genetic sampleIn m-th of sample point, m=1,2 ..., Np,NpFor genetic sample
Number.It is i+1 for sample set Si+1In n-th of sample point, n=1,2 ..., Ns。
It will be set with using the Latin hypercube experiment after optimization by the above-mentioned genetic sample dropped into new trusted zones
Count sample of the new samples generated as a new generation.
In step s 4, it is based on high-precision cost function, utilizes the posteriority joint probability of the above-mentioned amount of unbalance sought
Density Distribution identifies the approximate edge posterior probability Density Distribution for obtaining amount of unbalance using MCMC methodology, and then determines uneven
Mean value, the confidence interval of measurement.
Chain length (sample point number) N is setmax, it is preferred that chain length NmaxTake 105Magnitude.It is analyzed by many experiments, chain length
MCMC methods when this magnitude are selected to be likely to obtain the accurate approximate solution of Posterior distrbutionp.
Specifically, it is distributed according to the posteriority joint probability density of unknown parameter amount of unbalance, Markov Chain is from current sample
Candidate samples are obtained in this;By solve acceptance probability, judge candidate samples point whether this receiving;Candidate samples are constantly obtained,
Until meeting Markov Chain elongate member.
Embodiment 2
The present embodiment verifies the method in embodiment 1 by specifically testing,
First, rotor-bearing system model is established
INV1612 types rotor testbed is illustrated in figure 2 as method testing model in embodiment 1, disc thickness is
Artificially unbalance mass, is arranged in the screw in circumferentially distributed 16 holes at 15mm, diameter 60mm, configurable different weight.Rotor
Both ends by sliding supported condition, oil film characteristic coefficient such as table 1.Design of Rotation is 2000r/min.
1 oil film bearings rigidity of table and damped coefficient
The midpoint and section of 2 uncertain variables of table
Then, according to the method in embodiment 1, amount of unbalance identification is carried out;
In order to avoid error caused by the cross-couplings of magnetic field occurs for transducer magnetic head, so x, y sensor are not mounted on
In same plane.Arrange two measuring points, a measuring point is 73mm apart from left bearing, for surveying the radial displacement of the directions x, a measuring point
Apart from left bearing 334mm, for surveying the radial displacement of the directions y.The displacement transient response of two measuring points is measured by eddy current sensor such as
Shown in Fig. 5, Fig. 6.
Assuming that the initial prior information of amount of unbalance obey it is independent be uniformly distributed m~U (0,20) and
Error constant is taken as 1e-7, and Markov Chain a length of 50000 obtains 11 using optimal Latin hypercube experimental design (OLHD)
Initial step sample.Heredity falls the sample point in next step trust-region, using ILHD experimental designs genetic sample point and is trusting
Region generates new sample point.Fig. 7 provides the detailed process using identification amount of unbalance quality and phase.Fig. 7-Fig. 9 displays are used
For this method by the hereditary intelligent updating encryption iteration process of 3 step sample points, Fig. 7,8,9 show each newer priori of iteration step
The overall distribution situation that sample point is encrypted in space, with symbol " * ", " " and " ☆ " expression is corresponding respectively walk in the sample that generates
This updates the new samples in prior distribution space by hereditary Latin hypercube experimental design in embodiment 1, and the 1st step generates 11
A sample point, the 2nd step generate 12 new samples, 11 samples of heredity, the 3rd step generates 16 new samples, 23 samples of heredity
This.Sample point (this reality of the new samples that ILHD experimental designs generate the old sample of heredity and ILHD in combination as lower generation
Apply in example, genetic sample selected by brush, in preceding 3 step, the genetic sample of previous generation has all been genetic in the next generation), in this way
The number that total sample can be greatly reduced reaches and puies forward efficient purpose.3rd step, 23 samples of heredity, reduce 23 times and adopt
Sample, this is significant for the engineering problem that sample acquisition takes.
Figure 10 is provided in amount of unbalance two-dimensional space, and MCMC methodology combines Posterior probability distribution in unknown parameter amount of unbalance
The result of calculations of upper 50000 sampling, Figure 11 gives unbalance mass, and phase angle is respective markovian restrained
Journey shows that its corresponding recognition result has 95% possibility to fall in Xia Zhaiqujian [1.2319,1.2322]And [46.0354,
46.0358]In.The receptance of its sampling results is 42.6%, this acceptance probability ranging from 30%- suggested with Tarantola
50% comparison coincide.
Embodiment 3
It, can be each to solve directly to rotor into action balance after identifying rotor unbalance value by above-described embodiment
Rotor unbalance caused by kind factor.
In order to verify the accuracy and precision of amount of unbalance identification, the present embodiment is tested using the rotor of example 2 above
Device carries out the identification of degree of unbalancedness, and pair the rotor-support-foundation system after counterweight has been added to test.It the size of counterweight and identifies
Amount of unbalance size is identical, and phase angle i.e. in original angle on the contrary, plus 180 ° or subtract 180 °, test has added counterweight
System vibration response afterwards, and carry out the front and back response comparison of counterweight.Measuring point is can be seen that before and after dynamic balancing from Figure 12,13
Oscillation Amplitude has prodigious decline, dynamically balanced effect relatively good.
To sum up, the advantageous effect that technical solution of the present invention is brought:Present invention utilizes the priori of engineer's amount of unbalance letters
Breath, it is also considered that the environmental factor and test noise of rotor-support-foundation system meet Practical Project condition, there is practical implementation value.
Meanwhile being reduced in bayesian theory the present invention is based on hereditary intellegent sampling technology and calling time-consuming direct problem calculation times, it improves
Computational efficiency, and make limited sample concentrated reflection true posteriority space high-density region, and then ensure that amount of unbalance is known
Other precision.
It will be understood by those skilled in the art that realizing all or part of flow of above-described embodiment method, meter can be passed through
Calculation machine program instruction relevant hardware is completed, and the program can be stored in computer readable storage medium.Wherein, described
Computer readable storage medium is disk, CD, read-only memory or random access memory etc..
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
Any one skilled in the art in the technical scope disclosed by the present invention, the change or replacement that can be readily occurred in,
It should be covered by the protection scope of the present invention.
Claims (10)
1. a kind of sliding bearing-rotor system amount of unbalance recognition methods, which is characterized in that include the following steps:
Step S1, sliding bearing-rotor system amount of unbalance prior distribution space, initial sample and sliding bearing-rotor are obtained
The test unbalance response of system;
Step S2 solves cost function using bayesian theory, and obtains the posteriority joint probability density distribution of amount of unbalance;
Step S3 calculates the minimum value of above-mentioned cost function, when the minimum value is more than convergence precision, updates the sliding axle
- rotor-support-foundation system amount of unbalance prior distribution space is held, and using hereditary intellegent sampling technical limit spacing new samples, enters step S2,
Otherwise, S4 is entered step;The heredity intellegent sampling technology is that old sample is genetic in new samples;
Step S4, the approximate edge posterior probability for identifying to obtain sliding bearing-rotor system amount of unbalance using MCMC methodology are close
Degree distribution, and then determine mean value, the confidence interval of sliding bearing-rotor system amount of unbalance.
2. according to the method described in claim 1, it is characterized in that, described using hereditary intellegent sampling technical limit spacing new samples,
To update the new samples in prior distribution space by hereditary Latin hypercube experimental design, including:
According to minimax distance criterion, the sample to dropping into prior distribution of lower generation space screens;
New samples are made to project new samples that are uniform and generating on each design variable to heredity using the fiery optimization solver of simulation
The distance of sample is maximum.
3. method according to claim 1 or 2, which is characterized in that the solution cost function, including:
The emulation point of sliding bearing-rotor system unbalance response is passed through using amount of unbalance as input variable using transfer matrix method
Analysis, seeks sliding bearing-rotor system unbalance response;
Using the error between test unbalance response and the unbalance response sought of above-mentioned sliding bearing-rotor system as generation
Valence function.
4. according to the method described in claim 3, it is characterized in that, the update prior distribution space, for according to trusted zones more
New method obtains prior distribution of lower generation space.
5. according to the method described in claim 4, it is characterized in that, the trusted zones update method is:Pass through Fibonacci method
The up-and-down boundary X in adaptive adjustment amount of unbalance prior distribution spaceLAnd XR, and then the prior distribution for obtaining sampled point of lower generation is empty
Between.
6. according to the method described in claim 1, it is characterized in that, the acquisition amount of unbalance prior distribution includes:
Error constant ε is set, the equally distributed initial up-and-down boundary of sliding bearing-rotor system amount of unbalance priori is given
WithGive each iteration step sample point quantity Ns i。
7. according to the method described in claim 6, it is characterized in that, the posteriority joint probability density distribution p of the amount of unbalance
(X|Q) it is:
In formula, X is sliding bearing-rotor system amount of unbalance;Q is the test unbalance response of sliding bearing-rotor system;a
For constant;G (X) is cost function;σ2For the variance of sliding bearing-rotor system random noise.
8. the method according to the description of claim 7 is characterized in that cost function calculation formula is:
In formula, X is sliding bearing-rotor system amount of unbalance, and h is measuring point number, and q is the test of sliding bearing-rotor system
Unbalance response, qj(X) it is that amount of unbalance passes through sliding bearing-rotor system unbalance response simulation analysis as input variable
The unbalance response of acquisition.
9. according to the method described in claim 5, it is characterized in that, the prior distribution space for obtaining sampled point of lower generation is:
In formula, 1- α are golden section point, α=0.382;WithThe upper and lower boundary in prior distribution space is sampled for the i-th generation;WithFor the upper and lower boundary in i+1 generation sampling prior distribution space.
10. according to the method described in claim 9, it is characterized in that, in MCMC methodology, chain length takes 105Magnitude.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810934874.2A CN108694417A (en) | 2018-08-16 | 2018-08-16 | A kind of sliding bearing-rotor system amount of unbalance recognition methods |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810934874.2A CN108694417A (en) | 2018-08-16 | 2018-08-16 | A kind of sliding bearing-rotor system amount of unbalance recognition methods |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108694417A true CN108694417A (en) | 2018-10-23 |
Family
ID=63841685
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810934874.2A Pending CN108694417A (en) | 2018-08-16 | 2018-08-16 | A kind of sliding bearing-rotor system amount of unbalance recognition methods |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108694417A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110276164A (en) * | 2019-07-10 | 2019-09-24 | 湖南科技大学 | Annular components gusset welding magnitude of misalignment causes rotor structure amount of unbalance analytic approach |
| CN111113265A (en) * | 2019-12-12 | 2020-05-08 | 湖南工程学院 | Method for identifying eccentricity of motorized spindle-cutter system |
| CN115826407A (en) * | 2022-11-29 | 2023-03-21 | 中国航发沈阳发动机研究所 | Control method for reducing rotation inertia excitation of drum shaft |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102254184A (en) * | 2011-07-18 | 2011-11-23 | 华中科技大学 | Method for fusing multi-physical-domain feature information |
| WO2014107303A1 (en) * | 2013-01-04 | 2014-07-10 | Siemens Corporation | Probabilistic modeling and sizing of embedded flaws in nondestructive inspections for fatigue damage prognostics and structural integrity assessment |
| CN107255549A (en) * | 2017-06-14 | 2017-10-17 | 西安交通大学 | A kind of rotor multiple spot amount of unbalance recognition methods |
-
2018
- 2018-08-16 CN CN201810934874.2A patent/CN108694417A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102254184A (en) * | 2011-07-18 | 2011-11-23 | 华中科技大学 | Method for fusing multi-physical-domain feature information |
| WO2014107303A1 (en) * | 2013-01-04 | 2014-07-10 | Siemens Corporation | Probabilistic modeling and sizing of embedded flaws in nondestructive inspections for fatigue damage prognostics and structural integrity assessment |
| CN107255549A (en) * | 2017-06-14 | 2017-10-17 | 西安交通大学 | A kind of rotor multiple spot amount of unbalance recognition methods |
Non-Patent Citations (2)
| Title |
|---|
| W.ZHANG 等: "A fast Bayesian approach using adaptive densifying approximation technique accelerated MCMC", 《INVERSE PROBLEMS IN SCIENCE AND ENGINEERING, 2015》 * |
| 陈国栋: "基于代理模型的多目标优化方法及其在车身设计中的应用", 《中国博士学位论文全文数据库信息科技辑》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110276164A (en) * | 2019-07-10 | 2019-09-24 | 湖南科技大学 | Annular components gusset welding magnitude of misalignment causes rotor structure amount of unbalance analytic approach |
| CN111113265A (en) * | 2019-12-12 | 2020-05-08 | 湖南工程学院 | Method for identifying eccentricity of motorized spindle-cutter system |
| CN111113265B (en) * | 2019-12-12 | 2021-07-06 | 湖南工程学院 | Method for identifying eccentricity of motorized spindle-cutter system |
| CN115826407A (en) * | 2022-11-29 | 2023-03-21 | 中国航发沈阳发动机研究所 | Control method for reducing rotation inertia excitation of drum shaft |
| CN115826407B (en) * | 2022-11-29 | 2024-04-09 | 中国航发沈阳发动机研究所 | Control method for reducing drum shaft rotation inertia excitation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Compaction quality control of earth-rock dam construction using real-time field operation data | |
| Chang et al. | Optimal sensor placement for modal identification of bridge systems considering number of sensing nodes | |
| Ahmadizadeh et al. | Compensation of actuator delay and dynamics for real‐time hybrid structural simulation | |
| Song et al. | Computation of Rayleigh Damping Coefficients for the Seismic Analysis of a Hydro‐Powerhouse | |
| Cantero et al. | Bridge damage detection using weigh-in-motion technology | |
| Rosales et al. | Crack detection in beam-like structures | |
| CN108694417A (en) | A kind of sliding bearing-rotor system amount of unbalance recognition methods | |
| Shao et al. | Real-time hybrid simulation with online model updating: methodology and implementation | |
| Li et al. | Bayesian updating of fragility functions using hybrid simulation | |
| Aydin | Power of global test in deformation analysis | |
| Barbato et al. | Structural reliability applications of nonstationary spectral characteristics | |
| JP6001740B1 (en) | High precision evaluation of structure transfer functions, earthquake response prediction, deterioration diagnosis system and method | |
| Ma et al. | Assessment of intelligent compaction quality evaluation index and uniformity | |
| Wan et al. | Statistical framework for sensitivity analysis of structural dynamic characteristics | |
| Sumer et al. | Effect of uncertainty on water distribution system model design decisions | |
| CN112001067B (en) | Uncertainty measurement-based reliability assessment method for complex equipment simulation system | |
| Wang et al. | Crack parameters identification based on a kriging surrogate model for operating rotors | |
| Alipour et al. | Analytical approach to characterize tornado-induced loads on lattice structures | |
| Rostami et al. | Bayesian joint state-parameter-input estimation of flexible-base buildings from sparse measurements using Timoshenko beam models | |
| Casero et al. | Regularization methods applied to noisy response from beams under static loading | |
| Pourzeynali et al. | Reliability analysis of suspension bridges against fatigue failure from the gusting of wind | |
| Saha et al. | Monte–Carlo based method for predicting extreme value statistics of uncertain structures | |
| Zhang et al. | Bayesian updating model for structural vibration–induced hydrodynamic added mass of rectangular pile cap submerged in water | |
| Wen et al. | Application of Improved Combined Deterministic‐Stochastic Subspace Algorithm in Bridge Modal Parameter Identification | |
| Li et al. | Modal contribution coefficients in bridge condition evaluation |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181023 |
|
| RJ01 | Rejection of invention patent application after publication |