CN106092578A - A kind of machine tool mainshaft bearing confined state online test method based on wavelet packet and support vector machine - Google Patents
A kind of machine tool mainshaft bearing confined state online test method based on wavelet packet and support vector machine Download PDFInfo
- Publication number
- CN106092578A CN106092578A CN201610561304.4A CN201610561304A CN106092578A CN 106092578 A CN106092578 A CN 106092578A CN 201610561304 A CN201610561304 A CN 201610561304A CN 106092578 A CN106092578 A CN 106092578A
- Authority
- CN
- China
- Prior art keywords
- signal
- bearing
- wavelet packet
- confined state
- vibration
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
Abstract
A kind of machine tool mainshaft bearing confined state online test method based on wavelet packet and support vector machine of the present invention, comprises the steps, step 1, utilizes vibration acceleration sensor to gather rolling bearing and works well the vibration signal under state and abnormal condition；Step 2, carries out denoising Processing, and the vibration signal after de-noising carries out the wavelet package reconstruction of signal the vibration signal collected；Step 3, characterizes reconstruction signal vibration information in this frequency range by the vibration signal after reconstruct through m layer WAVELET PACKET DECOMPOSITION, the decomposed signal in each frequency band；Step 4, is normalized each frequency band energy value characteristic of correspondence vector, builds new characteristic vector；Step 5, based on new characteristic vector, is input in the grader that utilized support vector machine is set up be trained as sample using this new characteristic vector, obtains training SVM diagnostic cast, it is achieved on-line checking and the diagnosis to main shaft bearing confined state.
Description
Technical field
The present invention relates to machine tool mainshaft bearing assembling quality determining method, be specially one and based on wavelet packet and support vector
The machine tool mainshaft bearing confined state online test method of machine.
Background technology
Owing to machine tool chief axis is to be assembled the exquisite system combined, the assembling matter of each parts by numerous parts
Amount directly influences the military service performance of axis system, and in main shaft, main shaft performance all will be had by the slight change of parts confined state
Great impact.Its middle (center) bearing is the strength member in axis system, and its confined state is very big to main shaft performance impact.Such as when
Bearing is installed when there is deflection or dislocation, the most then cause main shaft to produce noise, abnormal vibration, and severe one causes main shaft precision to lose
Even cause bearing local pyrexia excessive and burn out；Additionally, high speed and precision main shaft bearing is often at high temperature, high rotating speed, high load capacity
Run Deng under complex working condition, easily cause bearing and bearing block deflection on axially and radially direction, cause Bearing inner to swim
Gap presents non-uniform Distribution, and then causes main shaft running precision to be lost, therefore the assembling quality testing of research main shaft bearing and operation
State estimation has important engineering significance.
Main shaft bearing vibrates the power with noise and the frequency content comprised thereof and main shaft produced by operation process
Running status has close contact, and the diagnosis of existing machine tool chief axis focuses mostly on to bearing diagnosing malfunction, detects with this
The early stage faint damage defect of main shaft bearing existence or the service life of assessment bearing.But, existing method is based on machine spindle
Hold inside and have under certain defect premise, found the fault type of bearing by different signal processing method, this method for
One new assembling main shaft, for only the bearing that assembling tilts occurring and improper because newly assembling bearing the most do not exist therefore
Barrier feature.When deflection occurs in bearing assembling, deflected condition the vibration signal caused is the faintest, and this type of vibration signal
Usually it is hidden in and is difficult to be found among bearing normal vibration signal, how the main shaft bearing only existing assembly problem is carried out
Identify the key being to ensure that spindle mounted quality.
But in the past is studied, for the analysis and research of spindle vibration signal, often as a kind of main shaft-bearing fault
The method that diagnosis is conventional, detects early stage faint damage defect present in main shaft or bearing, and rare people considers by means of advanced person
Signal processing method assess main shaft bearing installment state.In actual monitoring diagnoses, due to by multiple on main shaft and axle
The impact of parts vibration, the interference of bearing vibration signal is many, and comparison of ingredients is complicated, the most traditional single time frequency analysis
Method is more difficult therefrom extracts difference.And wavelet packet analysis can signal be mapped to one group flexible by small echo, translation
On basic function, it is achieved signal is in different frequency bands, reasonable separation the most in the same time, for the non-stationary description, faint of Dynamic Signal
Extracting of signal provides the strongest instrument realizing the identification of bearing confined state.Relevant scholar's research rolling bearing
Break down and correlation dimension under normal condition at inner ring, outer ring and rolling element, utilize the difference of its dimension to identify rolling
The running status of bearing.Also have and utilize the several characteristic quantities in monitoring of rolling bearings (in peak value, virtual value, kurtosis value and bearing
The characteristic frequency etc. of outer ring) as the input parameter of neutral net, the running status of bearing is monitored.But these methods
Only being adapted to process obvious bearing fault signal, the spindle vibration signal caused due to bearing deflection or dislocation can't draw
Play obvious fault signature, thus limit its application.
Additionally, main shaft bearing vibrational feature extracting does not has clear and definite standard, lead to meet the requirement of fault diagnosis precision
Often need to extract multiple fault signature, which increase amount of calculation, simultaneously will fault restriction diagnosis along with the increase of fault signature
Precision.It is difficult to obtain mass data sample, identify knowledge for small-signal present in main shaft bearing confined state identification
Obtaining the problems such as difficulty, inferential capability is weak, feature extraction is difficult, how rapid extraction bearing assembles the bad Weak characteristic caused
Change, sets up and readily appreciates and confined state detection model that computation complexity is relatively low quickly judge the assembling of main shaft bearing
State, is the emphasis of bearing confined state detection research.
Summary of the invention
For problems of the prior art, the present invention provides a kind of lathe master based on wavelet packet and support vector machine
Axle bearing confined state online test method, it is possible to extract strong noise background lower bearing and assemble the frequency of the bad small-signal caused
Band energy feature, sets up simple efficient bearing confined state detection model, and on-line checking is quick and precisely.
The present invention is to be achieved through the following technical solutions:
A kind of machine tool mainshaft bearing confined state online test method based on wavelet packet and support vector machine, including as follows
Step,
Step 1, utilizes vibration acceleration sensor to gather rolling bearing and works well shaking under state and abnormal condition
Dynamic signal；Wherein, abnormal condition refer to main shaft bearing after the inner and/or outer circle deflection that causes being installed or running one section due to main shaft
The bearing that each parts thermal deformation inequality of system causes tilts；
Step 2, carries out denoising Processing to the vibration signal collected；Determine best wavelet packet basis, determine optimal decomposition layer
Number m, quantifies WAVELET PACKET DECOMPOSITION coefficient threshold, is finely divided vibration signal in low frequency part and HFS simultaneously, obtains m
The WAVELET PACKET DECOMPOSITION coefficient of layer 2m sub-band from low to high, is expressed as:After de-noising
Vibration signal carries out the wavelet package reconstruction of signal；
Step 3, is broken down into 2 by the vibration signal after reconstruct through m layer WAVELET PACKET DECOMPOSITION, the energy of reconstruction signalmIndividual just
Hand on frequency band, obtain each frequency band energy value and construct characteristic of correspondence vector, with Sm0Represent Xm0Reconstruction signal, Sm1Represent Xm1
Reconstruction signal, the like, obtain the signal of each sub-band scope:After WAVELET PACKET DECOMPOSITION
Signal is 2mEnergy value summation on individual frequency band is consistent with the energy value of reconstruction signal, and the decomposed signal in each frequency band characterizes weight
Structure signal vibration information in this frequency range；
Step 4, is normalized each frequency band energy value characteristic of correspondence vector solved in step 3, structure
Build new characteristic vector；
Step 5, based on being tried to achieve new characteristic vector, inputs this new characteristic vector as sample in step 4
It is trained in the grader that utilized support vector machine is set up, from new characteristic vector, chooses 2n group data, wherein n group
Being used for training SVM diagnostic cast, n group is used for testing the confined state judging main shaft bearing；When corresponding data in SVM diagnostic cast
State normal time, bearing state is normal；When in SVM diagnostic cast, the state of corresponding data is abnormal, bearing is deflection shape
State；Thus realize the on-line checking to main shaft bearing confined state and diagnosis.
Preferably, in step 2, the vibration signal collected is as follows,
y(ti)=f (ti)+n(ti) i=1 ..., N；
Wherein, f (ti) it is primary signal, n (ti) for being desired for 0, variance be σ2Independent identically distributed white Gaussian noise,
tiRepresent time variable corresponding under different operating mode.
Preferably, during signal noise silencing removes pseudo-component in step 2, use the threshold denoising method improved, the most such as
Under,
Introduce alpha factor, it is proposed that soft and hard threshold compromise method, for object function, alpha factor and Decomposition order are entered with signal to noise ratio
Row optimizes, and obtains optimum α value, replaces original hard-threshold or soft-threshold threshold denoising method；
Wherein, calculate optimum signal-noise ratio SNR after de-noising by following formula,
In formula, f (t) is primary signal；For the signal after de-noising.
Preferably, in step 3, the noise cancellation signal that vibration signal obtains through denoising Processing is carried out WAVELET PACKET DECOMPOSITION, its
In, wavelet packet basis determines according to bearing type, and Decomposition order determines according to the complexity of signal, calculates in each frequency band range
Energy value, extract the signal of each frequency band range；IfCorresponding energy is Emj, obtain each frequency
The total capacity that in band, signal is corresponding is EmjAs follows:
Wherein, xikRepresent reconstruction signal SmjThe amplitude of discrete point, j=0,1 ..., 8；K=1,2 ..., 2m-1)。
Preferably, in step 4, construct a stack features vector T, T=with frequency band different-energy element each in reconstruction signal
[E1,E1,E2,…,Ek], wherein k=2m-1。
Preferably, in step 5, when the characteristic vector characterizing bearing confined state is carried out classifier training, first will instruction
Certain class sample in white silk sample set is as a class, and its classification logotype is 1, and remaining sample is another kind of, is designated-1；Training aids divides
Class is I, builds categorised decision function.
Preferably, in step 5, new characteristic vector is inputted the grader utilizing support vector machine to set up, specifically includes
Following steps:
Step 5.1, every kind of bearing state choose 2n group data, and wherein n group sample data is used for training SVM to diagnose mould
Type, n group is used for testing；
Step 5.2, according to the one-to-many strategy in svm classifier, sets up 2 two classification SVM classifier, respectively correspondences two kinds
Bearing confined state types of bearings assembling deflection and normal condition, and main shaft bearing assembling running status is carried out classification based training and
Test；
Step 5.3, chooses the RBF kernel function as SVM diagnostic cast；
Step 5.4, according to categorical attribute, determines bearing assembling deflection or normal condition, and grader I respective shaft takes up joins partially
Ramp-like state, grader II respective shaft takes up joins normal condition.
Compared with prior art, the present invention has a following useful technique effect:
The present invention is that a kind of machine tool mainshaft bearing based on wavelet packet combination supporting vector machine assembles quality determining method, solves
The bearing caused due to each parts thermal deformation inequality of axis system after main shaft bearing of having determined installation deflection or operation one section inclines
The bearing fault characteristics weak output signal caused under the conditions of Xie, extraction difficult problem；Bearing is extracted by analysis method of wavelet packet
Assemble bad middle Weak fault characteristic signal, and be characterized vector with the energy value after the normalized on different frequency bands, build
Found state recognition model based on different frequency bands energy eigenvalue, it is possible to fast and accurately identify bearing installation quality detection and
Condition monitoring and diagnosis, be used in the on-line checking of machine tool mainshaft bearing confined state.
Accompanying drawing explanation
Fig. 1 is the overall flow block diagram of method described in present example.
Fig. 2 is WAVELET PACKET DECOMPOSITION described in present example and the basic flow sheet of subsequent processes thereof.
Fig. 3 is wavelet packet threshold denoising flow chart described in present example.
Fig. 4 a is that present example middle (center) bearing Internal and external cycle is installed or normal operation state vibration signal.
Fig. 4 b is that present example middle (center) bearing Internal and external cycle is installed or runs abnormal condition vibration signal.
Vibration signal before Fig. 5 is present example middle (center) bearing de-noising and after de-noising.
Fig. 6 a is assembling normal condition wavelet pack energy feature rectangular histogram in present example.
Fig. 6 b is assembling defective mode wavelet pack energy feature rectangular histogram in present example.
Fig. 7 is present example middle (center) bearing confined state quality testing classifying quality figure.
In figure, 1 and 2 is vibration acceleration sensor, and 3 is harvester, and 4 is the mastery routine mould processing spindle vibration signal
Block.
Detailed description of the invention
Below in conjunction with specific embodiment, the present invention is described in further detail, described in be explanation of the invention and
It not to limit.
A kind of machine tool mainshaft bearing confined state online test method based on wavelet packet and support vector machine of the present invention, profit
The vibration signal under machine tool mainshaft bearing normal operating condition and bearing deflected condition is gathered with vibration acceleration sensor；
Use improved wavelet threshold Denoising Method, introduce alpha factor, vibration signal is carried out wavelet decomposition, by vibration signal
It is mapped on one group of basic function that, translation flexible by small echo is formed, chooses suitable decomposition scale and make signal at different frequency bands, no
Separate in the same time, it is achieved vibration signal denoising Processing.
Signal after de-noising is carried out wavelet package reconstruction, makes the vibration signal under different confined state believe in some frequency band
Number energy reduces, and strengthens at some inband signal energy, improves the identification of signal；
Use Wavelet Packet Technique irredundant, independent without slipping, decomposing orthogonally to components different in reconstruction signal
In each frequency band, the energy correspondence different conditions identification information of signal in each frequency band；
Extract the energy feature at each frequency band of vibration signal under different operating mode as bearing confined state characteristic of division；Profit
Small sample, non-linear, higher-dimension can be preferably solved by support vector machine (Support Vector Machines is called for short SVM)
The advantages such as pattern recognition, set up the mapping relations of frequency band energy characteristic vector and bearing confined state, it is achieved the dress of main shaft bearing
Join state recognition.
Solve that the bearing assembling machine tool chief axis vibratory impulse feature that causes of deflection is faint causes asking of the state that is difficult to
Topic, it is determined that new method based on the main shaft bearing confined state detection that wavelet packet and support vector machine combine.Such as Fig. 2 institute
Show, specifically include following steps:
Step 1, utilizes vibration acceleration sensor to gather rolling bearing and works well shaking under state and abnormal condition
Dynamic signal；Wherein, abnormal condition refer to main shaft bearing after the inner and/or outer circle deflection that causes being installed or running one section due to main shaft
The bearing that each parts thermal deformation inequality of system causes tilts.
Step 2, carries out denoising Processing to the vibration signal collected, if f is (ti) it is primary signal, n (ti) for being desired for 0,
Variance is σ2Independent identically distributed white Gaussian noise, then the vibration signal collected can model as follows；
y(ti)=f (ti)+n(ti) i=1 ..., N；
Determine best wavelet packet basis, use the denoising method improving threshold value, quantify WAVELET PACKET DECOMPOSITION coefficient threshold, determine
Good decomposition scale i.e. Decomposition order, be finely divided in low frequency part and HFS vibration signal, after de-noising simultaneously
Signal carries out the wavelet package reconstruction of signal；
Step 3, is broken down into 2 by the vibration signal after reconstruct through m layer WAVELET PACKET DECOMPOSITION, the energy of reconstruction signalmIndividual just
Hand on frequency band.Signal after WAVELET PACKET DECOMPOSITION is 2mEnergy summation on individual frequency band and the energy coincidence of reconstruction signal, each
Decomposed signal in frequency band characterizes reconstruction signal vibration information in this frequency range；
Step 4, is normalized each frequency band energy value solved in step 3, builds new characteristic vector；
Step 5, based on being tried to achieve new characteristic vector, is input to be utilized by the characteristic vector that this is new in step 4
Being trained in the grader that support vector machine is set up, choose 2n group data, wherein n group is used for training SVM diagnostic cast, n group
Being used for testing the confined state judging main shaft bearing, when SVM state is normal, bearing state is normal, when SVM state is abnormal
Time, bearing is deflected condition；Thus realize the on-line checking to main shaft bearing confined state and diagnosis, the present invention uses radially
Basic function is the kernel function of algorithm of support vector machine.
In step 1～3, the most key is how selected threshold and quantifies it, and this is directly connected to signal
De-noising quality.During in step 2, signal noise silencing removes pseudo-component, the softest, the different characteristics of hard-threshold Denoising Method, adopt
By the threshold denoising method improved, introduce alpha factor, for object function, alpha factor and decomposition scale are optimized with signal to noise ratio, obtain
Optimum α value, replaces original hard-threshold or soft-threshold Denoising Method；The optimum signal-noise ratio (SNR) after de-noising, letter is calculated by program
Make an uproar and be defined as follows than calculating:
In step 3, the noise cancellation signal that vibration signal obtains through denoising Processing is carried out WAVELET PACKET DECOMPOSITION, wherein, small echo
Bao Ji choose depending on bearing type, Decomposition order depending on the complexity of signal, calculate the energy in each frequency band range
Value, extracts the signal of each frequency band range.Wherein SijRepresent the reconstruction signal of the jth node of i-th layer.If corresponding total of signal
Ability is Ej, then have:
In step 3, construct a stack features vector T with signal different frequency bands energy element after reconstruct as follows,
T=[E1,E1,E2,…,Ek] wherein k=2m-1；
Each inband signal energy value may be bigger, data analysis for convenience, needs to return characteristic vector T
One change processes, and obtains new characteristic vector T ', reflect bearing confined state feature.
In step 5, certain class sample that training sample is concentrated when carrying out classifier training, is first made by the countershaft confined state that holds
Being a class, its classification logotype is 1, and remaining sample is another kind of, is designated-1；Training aids is categorized as I, builds categorised decision letter
Number；Thus realize machine tool mainshaft bearing confined state on-line checking and diagnosis.
This preferred embodiment uses the experimental data of NSK7210C bearing to verify, the deflection of this bearing internal external circle is by interior
The piezoelectric actuator being placed in main shaft realizes.
As it is shown in figure 1, comprise the following steps:
Step 1, utilizes vibration acceleration sensor 1 and 2 to measure, by harvester 3 collect rolling bearing install good,
The each 50 groups of signals of Internal and external cycle deflected condition, it is thus achieved that 100 groups of signals, sample frequency is 8192Hz, and rolling bearing two states vibrates
Signal is as shown in figures 4 a and 4b.
Step 2, carries out denoising to the vibration signal collected, if f is (ti) it is primary signal, n (ti) for being desired for 0, variance
For σ2Independent identically distributed white Gaussian noise, then the signal collected can model as follows；
y(ti)=f (ti)+n(ti) i=1 ..., N；
As it is shown on figure 3, by means of wavelet packet de-noise technology, the pseudo-component of cancelling noise signal, to the vibration letter collected
Number it is finely divided in low frequency part and HFS simultaneously；Wherein, signal i.e. noise cancellation signal such as Fig. 5 institute after primary signal and denoising
Show；
Step 3, determines best wavelet packet basis, quantifies WAVELET PACKET DECOMPOSITION coefficient threshold, the most again noise cancellation signal is carried out letter
Number wavelet package reconstruction；
Step 4, as in figure 2 it is shown, through the routine processes of step 3, select DaubechiesI (db1) to shaking after reconstruct
Dynamic signal in orthogonal WAVELET PACKET DECOMPOSITION, Decomposition order J=3, the energy of the most former vibration signal is broken down into 23On individual orthogonal frequency band.
Signal is 23Energy summation on individual frequency band and the energy coincidence of source signal, the vibration signal in each frequency band characterizes original signal and exists
Vibration information in this frequency range；
Step 5, is characterized vector with the energy on the different frequency bands tried to achieve in step 4, wherein, assembling normal condition and
Defective mode wavelet pack energy feature histogram is as shown in figures 6 a and 6b；The characteristic vector input profit again these set up
The grader set up by support vector machine, specifically includes following steps:
Step 5.1, every kind of bearing state choose 100 groups of data, and wherein 50 groups of sample datas are used for training SVM to diagnose mould
Type, 50 groups are used for testing；
Step 5.2, according to " one-to-many " strategy in svm classifier, sets up 2 two classification SVM classifier, to main shaft bearing
Assembling running status is classified；
Step 5.3, chooses the RBF kernel function as the SVM algorithm used in the present invention；
Step 5.4, according to categorical attribute, determines bearing assembling deflection or normal condition.
Diagnostic result shows, total rate of correct diagnosis is 92%, shown in part diagnostic result Fig. 7.
The conversion of any equivalence that technical solution of the present invention is taked, the claim being the present invention is contained.
Claims (7)
1. a machine tool mainshaft bearing confined state online test method based on wavelet packet and support vector machine, its feature exists
In, comprise the steps,
Step 1, utilizes vibration acceleration sensor to gather the vibration letter that rolling bearing works well under state and abnormal condition
Number；Wherein, abnormal condition refer to main shaft bearing after the inner and/or outer circle deflection that causes being installed or running one section due to axis system
The bearing that each parts thermal deformation inequality causes tilts；
Step 2, carries out denoising Processing to the vibration signal collected；Determine best wavelet packet basis, determine optimal Decomposition order m,
Quantify WAVELET PACKET DECOMPOSITION coefficient threshold, vibration signal is finely divided in low frequency part and HFS simultaneously, obtain m layer from
Low frequency, to the WAVELET PACKET DECOMPOSITION coefficient of 2m sub-band of high frequency, is expressed as:To the vibration after de-noising
Signal carries out the wavelet package reconstruction of signal；
Step 3, is broken down into 2 by the vibration signal after reconstruct through m layer WAVELET PACKET DECOMPOSITION, the energy of reconstruction signalmIndividual orthogonal frequency band
On, obtain each frequency band energy value and construct characteristic of correspondence vector, with Sm0Represent Xm0Reconstruction signal, Sm1Represent Xm1Reconstruct
Signal, the like, obtain the signal of each sub-band scope:Signal after WAVELET PACKET DECOMPOSITION exists
2mEnergy value summation on individual frequency band is consistent with the energy value of reconstruction signal, and the decomposed signal in each frequency band characterizes reconstruction signal
Vibration information in this frequency range；
Step 4, is normalized each frequency band energy value characteristic of correspondence vector solved in step 3, builds new
Characteristic vector；
Step 5, based on being tried to achieve new characteristic vector, is input to institute using this new characteristic vector as sample in step 4
Utilizing in the grader that support vector machine sets up and be trained, choose 2n group data from new characteristic vector, wherein n group is used for
Training SVM diagnostic cast, n group is used for testing the confined state judging main shaft bearing；When the shape of corresponding data in SVM diagnostic cast
When state is normal, bearing state is normal；When in SVM diagnostic cast, the state of corresponding data is abnormal, bearing is deflected condition；From
And realize the on-line checking to main shaft bearing confined state and diagnosis.
A kind of machine tool mainshaft bearing confined state based on wavelet packet and support vector machine the most according to claim 1 is online
Detection method, it is characterised in that in step 2, the vibration signal collected is as follows,
y(ti)=f (ti)+n(ti) i=1 ..., N；
Wherein, f (ti) it is primary signal, n (ti) for being desired for 0, variance be σ2Independent identically distributed white Gaussian noise, tiRepresent
Time variable corresponding under different operating modes.
A kind of machine tool mainshaft bearing confined state based on wavelet packet and support vector machine the most according to claim 1 is online
Detection method, it is characterised in that during signal noise silencing removes pseudo-component in step 2, uses the threshold denoising method improved, tool
Body is as follows,
Introduce alpha factor, it is proposed that soft and hard threshold compromise method, for object function, alpha factor and Decomposition order are carried out excellent with signal to noise ratio
Change, obtain optimum α value, replace original hard-threshold or soft-threshold threshold denoising method；
Wherein, calculate optimum signal-noise ratio SNR after de-noising by following formula,
In formula, f (t) is primary signal；For the signal after de-noising.
A kind of machine tool mainshaft bearing confined state based on wavelet packet and support vector machine the most according to claim 1 is online
Detection method, it is characterised in that in step 3, carries out wavelet packet by vibration signal through the noise cancellation signal that denoising Processing obtains and divides
Solving, wherein, wavelet packet basis determines according to bearing type, and Decomposition order determines according to the complexity of signal, calculates each frequency band
In the range of energy value, extract the signal of each frequency band range；If Smj(j=0,1 ..., 2m-1), corresponding energy is Emj, obtain
The total capacity that in each frequency band, signal is corresponding is EmjAs follows:
Wherein, xikRepresent reconstruction signal SmjThe amplitude of discrete point, j=0,1 ..., 8；K=1,2 ..., 2m-1)。
A kind of machine tool mainshaft bearing confined state based on wavelet packet and support vector machine the most according to claim 1 is online
Detection method, it is characterised in that in step 4, constructs a stack features vector T, T with frequency band different-energy element each in reconstruction signal
=[E1,E1,E2,…,Ek], wherein k=2m-1。
A kind of machine tool mainshaft bearing confined state based on wavelet packet and support vector machine the most according to claim 1 is online
Detection method, it is characterised in that in step 5, when the characteristic vector characterizing bearing confined state is carried out classifier training, first
Certain the class sample concentrated by training sample is as a class, and its classification logotype is 1, and remaining sample is another kind of, is designated-1；Training
Device is categorized as I, builds categorised decision function.
A kind of machine tool mainshaft bearing confined state based on wavelet packet and support vector machine the most according to claim 1 is online
Detection method, it is characterised in that in step 5, inputs the grader utilizing support vector machine to set up, specifically by new characteristic vector
Comprise the following steps:
Step 5.1, every kind of bearing state choose 2n group data, and wherein n group sample data is used for training SVM diagnostic cast, n group
It is used for testing；
Step 5.2, according to the one-to-many strategy in svm classifier, sets up 2 two classification SVM classifier, the most corresponding two kinds of bearings
Confined state types of bearings assembling deflection and normal condition, and main shaft bearing assembling running status is carried out classification based training and survey
Examination；
Step 5.3, chooses the RBF kernel function as SVM diagnostic cast；
Step 5.4, according to categorical attribute, determines bearing assembling deflection or normal condition, and grader I respective shaft takes up joins deflection shape
State, grader II respective shaft takes up joins normal condition.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610561304.4A CN106092578A (en) | 2016-07-15 | 2016-07-15 | A kind of machine tool mainshaft bearing confined state online test method based on wavelet packet and support vector machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610561304.4A CN106092578A (en) | 2016-07-15 | 2016-07-15 | A kind of machine tool mainshaft bearing confined state online test method based on wavelet packet and support vector machine |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106092578A true CN106092578A (en) | 2016-11-09 |
Family
ID=57220450
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610561304.4A Pending CN106092578A (en) | 2016-07-15 | 2016-07-15 | A kind of machine tool mainshaft bearing confined state online test method based on wavelet packet and support vector machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106092578A (en) |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106908232A (en) * | 2017-02-24 | 2017-06-30 | 伍婷婷 | A kind of vibration signal fault recognition method of rotating machinery |
| CN106901731A (en) * | 2017-03-07 | 2017-06-30 | 重庆邮电大学 | A kind of EMG Feature Extraction for merging wavelet packet and double-spectrum analysis |
| CN106989924A (en) * | 2017-04-05 | 2017-07-28 | 中车青岛四方机车车辆股份有限公司 | bearing abnormal sound detecting system |
| CN107271024A (en) * | 2017-05-26 | 2017-10-20 | 国网山东省电力公司青岛供电公司 | A kind of load ratio bridging switch diagnostic method and device |
| CN107560844A (en) * | 2017-07-25 | 2018-01-09 | 广东工业大学 | A kind of fault diagnosis method and system of gearbox of wind turbine |
| CN107688820A (en) * | 2017-07-11 | 2018-02-13 | 浙江新再灵科技股份有限公司 | A kind of Elevator Fault Diagnosis method based on BCSA Support Vector Machines Optimizeds |
| CN107909118A (en) * | 2017-12-11 | 2018-04-13 | 北京映翰通网络技术股份有限公司 | A kind of power distribution network operating mode recording sorting technique based on deep neural network |
| CN108398265A (en) * | 2018-01-15 | 2018-08-14 | 上海电力学院 | A kind of online fault detection method of rolling bearing |
| CN108710889A (en) * | 2018-04-02 | 2018-10-26 | 天津大学 | A kind of scarce cylinder method for diagnosing faults of automobile engine |
| CN110030187A (en) * | 2019-03-20 | 2019-07-19 | 杭州哲达智慧能源技术研究院有限公司 | A kind of pump Ankang diagnostic method based on support vector machines |
| CN110132596A (en) * | 2019-04-24 | 2019-08-16 | 昆明理工大学 | A method of the rolling bearing fault diagnosis based on wavelet packet and GWO-SVM |
| CN110161119A (en) * | 2019-06-07 | 2019-08-23 | 湘潭大学 | Wind electricity blade defect identification method |
| CN111220312A (en) * | 2020-02-28 | 2020-06-02 | 中国铁道科学研究院集团有限公司 | Bolt state diagnosis method and system |
| CN111458116A (en) * | 2020-03-10 | 2020-07-28 | 博世华域转向系统有限公司 | Method for detecting assembling quality of automobile steering column |
| CN113326774A (en) * | 2021-05-28 | 2021-08-31 | 武汉科技大学 | Machine tool energy consumption state identification method and system based on AlexNet network |
| CN114062995A (en) * | 2021-11-15 | 2022-02-18 | 通号（长沙）轨道交通控制技术有限公司 | Mutual inductor fault diagnosis method, equipment and medium based on electric quantity multi-feature fusion |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101634589A (en) * | 2009-08-21 | 2010-01-27 | 武汉钢铁(集团)公司 | Processing method of equipment vibration signal |
| CN102183366A (en) * | 2011-03-08 | 2011-09-14 | 上海大学 | Device and method for vibration measurement and failure analysis of rolling bearing |
| CN103076177A (en) * | 2013-01-16 | 2013-05-01 | 昆明理工大学 | Rolling bearing fault detection method based on vibration detection |
| CN105760839A (en) * | 2016-02-22 | 2016-07-13 | 重庆大学 | Bearing fault diagnosis method based on multi-feature manifold learning and support vector machine |
-
2016
- 2016-07-15 CN CN201610561304.4A patent/CN106092578A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101634589A (en) * | 2009-08-21 | 2010-01-27 | 武汉钢铁(集团)公司 | Processing method of equipment vibration signal |
| CN102183366A (en) * | 2011-03-08 | 2011-09-14 | 上海大学 | Device and method for vibration measurement and failure analysis of rolling bearing |
| CN103076177A (en) * | 2013-01-16 | 2013-05-01 | 昆明理工大学 | Rolling bearing fault detection method based on vibration detection |
| CN105760839A (en) * | 2016-02-22 | 2016-07-13 | 重庆大学 | Bearing fault diagnosis method based on multi-feature manifold learning and support vector machine |
Non-Patent Citations (5)
| Title |
|---|
| 中国通信学会青年工作委员会主编: "《第十三届全国青年通信学术会议论文集》", 31 December 2008, 国防工业出版社 * |
| 周传德: "《传感器与测试技术》", 30 June 2009, 重庆大学出版社 * |
| 徐光华: "《机械设备检修技术基础》", 31 August 2008, 中国铁道出版社 * |
| 沈阳阳: "基于振动信号分析的风力发电机轴承故障诊断", 《中国优秀硕士学位论文全文数据库 工程科技II辑》 * |
| 陈益: "改进的小波阈值消噪法应用于超声信号处理", 《北京航空航天大学学报》 * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106908232A (en) * | 2017-02-24 | 2017-06-30 | 伍婷婷 | A kind of vibration signal fault recognition method of rotating machinery |
| CN106901731B (en) * | 2017-03-07 | 2020-05-12 | 重庆邮电大学 | Electromyographic signal feature extraction method fusing wavelet packet and bispectrum analysis |
| CN106901731A (en) * | 2017-03-07 | 2017-06-30 | 重庆邮电大学 | A kind of EMG Feature Extraction for merging wavelet packet and double-spectrum analysis |
| CN106989924A (en) * | 2017-04-05 | 2017-07-28 | 中车青岛四方机车车辆股份有限公司 | bearing abnormal sound detecting system |
| CN107271024A (en) * | 2017-05-26 | 2017-10-20 | 国网山东省电力公司青岛供电公司 | A kind of load ratio bridging switch diagnostic method and device |
| CN107688820B (en) * | 2017-07-11 | 2020-09-08 | 浙江新再灵科技股份有限公司 | Elevator fault diagnosis method based on BCSA optimized support vector machine |
| CN107688820A (en) * | 2017-07-11 | 2018-02-13 | 浙江新再灵科技股份有限公司 | A kind of Elevator Fault Diagnosis method based on BCSA Support Vector Machines Optimizeds |
| CN107560844A (en) * | 2017-07-25 | 2018-01-09 | 广东工业大学 | A kind of fault diagnosis method and system of gearbox of wind turbine |
| CN107909118A (en) * | 2017-12-11 | 2018-04-13 | 北京映翰通网络技术股份有限公司 | A kind of power distribution network operating mode recording sorting technique based on deep neural network |
| CN107909118B (en) * | 2017-12-11 | 2022-02-22 | 北京映翰通网络技术股份有限公司 | Power distribution network working condition wave recording classification method based on deep neural network |
| CN108398265A (en) * | 2018-01-15 | 2018-08-14 | 上海电力学院 | A kind of online fault detection method of rolling bearing |
| CN108710889A (en) * | 2018-04-02 | 2018-10-26 | 天津大学 | A kind of scarce cylinder method for diagnosing faults of automobile engine |
| CN110030187A (en) * | 2019-03-20 | 2019-07-19 | 杭州哲达智慧能源技术研究院有限公司 | A kind of pump Ankang diagnostic method based on support vector machines |
| CN110030187B (en) * | 2019-03-20 | 2021-05-14 | 杭州哲达智慧能源技术研究院有限公司 | Pump health diagnosis method based on support vector machine |
| CN110132596A (en) * | 2019-04-24 | 2019-08-16 | 昆明理工大学 | A method of the rolling bearing fault diagnosis based on wavelet packet and GWO-SVM |
| CN110161119B (en) * | 2019-06-07 | 2021-11-26 | 湘潭大学 | Wind power blade defect identification method |
| CN110161119A (en) * | 2019-06-07 | 2019-08-23 | 湘潭大学 | Wind electricity blade defect identification method |
| CN111220312A (en) * | 2020-02-28 | 2020-06-02 | 中国铁道科学研究院集团有限公司 | Bolt state diagnosis method and system |
| CN111458116A (en) * | 2020-03-10 | 2020-07-28 | 博世华域转向系统有限公司 | Method for detecting assembling quality of automobile steering column |
| CN113326774A (en) * | 2021-05-28 | 2021-08-31 | 武汉科技大学 | Machine tool energy consumption state identification method and system based on AlexNet network |
| CN114062995A (en) * | 2021-11-15 | 2022-02-18 | 通号（长沙）轨道交通控制技术有限公司 | Mutual inductor fault diagnosis method, equipment and medium based on electric quantity multi-feature fusion |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106092578A (en) | A kind of machine tool mainshaft bearing confined state online test method based on wavelet packet and support vector machine | |
| CN104655425B (en) | Bearing fault classification diagnosis method based on sparse representation and LDM (large margin distribution machine) | |
| Cui et al. | Quantitative trend fault diagnosis of a rolling bearing based on Sparsogram and Lempel-Ziv | |
| CN108318249B (en) | Fault diagnosis method for rotary mechanical bearing | |
| CN112257530B (en) | Rolling bearing fault diagnosis method based on blind signal separation and support vector machine | |
| CN105760839A (en) | Bearing fault diagnosis method based on multi-feature manifold learning and support vector machine | |
| CN104155108B (en) | A kind of Fault Diagnosis of Roller Bearings based on vibration time frequency analysis | |
| Gao et al. | Rolling bearing compound fault diagnosis based on parameter optimization MCKD and convolutional neural network | |
| CN109827777B (en) | Rolling bearing fault prediction method based on partial least square method extreme learning machine | |
| CN104849050A (en) | Rolling bearing fault diagnosis method based on composite multi-scale permutation entropies | |
| CN103558042B (en) | A kind of unit Fault Quick Diagnosis method based on full state information | |
| CN105241680A (en) | Health state assessment method for rotary machine based on probability density function | |
| CN105678343A (en) | Adaptive-weighted-group-sparse-representation-based diagnosis method for noise abnormity of hydroelectric generating set | |
| CN107817106A (en) | Fault Diagnosis of Roller Bearings based on Bayes's residual transform singular value decomposition Gaussian Mixture HMM framework | |
| CN109858104A (en) | A kind of rolling bearing health evaluating and method for diagnosing faults and monitoring system | |
| CN111458142A (en) | Sliding bearing fault diagnosis method based on generation of countermeasure network and convolutional neural network | |
| CN106596149A (en) | Method for monitoring and diagnosing flat wheel state of rail vehicle based on VMD | |
| CN113375941A (en) | Open set fault diagnosis method for high-speed motor train unit bearing | |
| CN111707472B (en) | Rolling bearing fault diagnosis method based on FFT and full-connection layer-SVM | |
| CN109165632A (en) | A kind of equipment fault diagnosis method based on improvement D-S evidence theory | |
| CN111238843A (en) | Fan health evaluation method based on rapid spectrum kurtosis analysis | |
| CN105784340B (en) | Valve fault diagnosis method based on Hybrid Intelligent Technology | |
| CN114462480A (en) | Multi-source sensor rolling mill fault diagnosis method based on non-equilibrium data set | |
| CN114263621A (en) | Test method and system for diagnosing and simulating cavitation fault of centrifugal pump | |
| CN106339720A (en) | Automobile engine failure detection method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161109 |