CN108229071A - Cutting performance degradation assessment method and system based on AR models Yu SVDD algorithms - Google Patents
Cutting performance degradation assessment method and system based on AR models Yu SVDD algorithms Download PDFInfo
- Publication number
- CN108229071A CN108229071A CN201810133641.2A CN201810133641A CN108229071A CN 108229071 A CN108229071 A CN 108229071A CN 201810133641 A CN201810133641 A CN 201810133641A CN 108229071 A CN108229071 A CN 108229071A
- Authority
- CN
- China
- Prior art keywords
- signal data
- residual signals
- sample
- models
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000005520 cutting process Methods 0.000 title claims abstract description 46
- 230000015556 catabolic process Effects 0.000 title claims abstract description 38
- 238000006731 degradation reaction Methods 0.000 title claims abstract description 38
- 238000000034 method Methods 0.000 title claims abstract description 29
- 238000009826 distribution Methods 0.000 claims abstract description 45
- 230000001186 cumulative effect Effects 0.000 claims abstract description 38
- 238000012545 processing Methods 0.000 claims abstract description 38
- 238000001914 filtration Methods 0.000 claims description 17
- 238000004590 computer program Methods 0.000 claims description 10
- 238000009825 accumulation Methods 0.000 claims description 9
- 230000002087 whitening effect Effects 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 6
- 238000003860 storage Methods 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 4
- 238000013480 data collection Methods 0.000 claims description 2
- 239000012141 concentrate Substances 0.000 claims 1
- 238000012423 maintenance Methods 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 28
- 238000010586 diagram Methods 0.000 description 13
- 238000004519 manufacturing process Methods 0.000 description 6
- 230000007850 degeneration Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 230000008439 repair process Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 239000013078 crystal Substances 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000007731 hot pressing Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000013450 outlier detection Methods 0.000 description 1
- 238000002203 pretreatment Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000003746 surface roughness Effects 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- 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)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Complex Calculations (AREA)
Abstract
The present invention is a kind of cutting performance degradation assessment method based on AR models Yu SVDD algorithms, is included the following steps：According to the signal data sample of collected normal condition bottom tool, AR filter models are established；Signal data sample is handled by the AR filter models, obtain residual signals data set, and calculate the cumulative probability distribution of each residual signals data in residual signals data set, the feature set corresponding to normal condition bottom tool signal data and processing are established, obtains the hypersphere for containing minimum volume；The cutter signal data under collected current state is handled using AR filter models, obtains the first residual signals data set；The cumulative probability distribution characteristics of the corresponding first residual signals data of current state bottom tool signal data is calculated, calculates fisrt feature to the hyperspherical distance, the degree degenerated by Distance Judgment cutting performance.The present invention can realize the assessment degenerated to cutting performance, help to realize numerically-controlled machine tool prospective maintenance.
Description
Technical field
The present invention relates to mechanical signal technical field of data processing more particularly to a kind of based on AR models and SVDD algorithms
Cutting performance degradation assessment method and system.
Background technology
At present, in numerically-controlled machine tool process, cutter by hot pressing due to being split, physical friction, plastic deformation, spreads mill
It damages and the reasons such as crystal grain comes off influences, can gradually wear out, performance degradation occur until failure.The performance degradation of cutter can not only drop
Low machine tooling quality, influences the surface roughness and dimensional accuracy of workpiece, while also seriously affects the stability of numerically-controlled machine tool
With production OEE (Overall Equipment Effectiveness) index.Particularly in flexible manufacturing system and computer collection
Into the numerically-controlled machine tool in manufacture system, if cutting performance degree of degeneration cannot be assessed accurately and in time, once tool failure is sent out
Raw failure, often generates immeasurable loss.Therefore, exploitation is directed to the performance degradation method of cutting tool for CNC machine, not only
Machine cut parameter can be optimized and ensure the promotion of processing performance, when more can effectively control the non-programmed halt of numerically-controlled machine tool
Between, improve whole economic efficiency.
It retrieves and finds by technology, the failure that existing cutting tool for CNC machine management related patents focus primarily upon cutter is examined
Disconnected method and system, concern purpose is quickly and effectively be identified when cutting tool for CNC machine breaks down, so as to subtract
Few losses such as shutdown, waste material caused by cutter failure.But in flexible manufacturing system now and computer integrated manufacturing system
In, correction maintenance often cannot meet enterprise practical demand, the only angle from cutting performance degradation assessment, implement
Repair, that is, prospective maintenance based on state, can just be inherently eliminated nonscheduled down time, and then greatly improve enterprise
Economic benefit.
Invention content
The present invention in the prior art the shortcomings that, provide and a kind of moved back based on AR models and the cutting performance of SVDD algorithms
Change appraisal procedure and system.
In order to solve the above-mentioned technical problem, the present invention is addressed by following technical proposals：
The present invention discloses：
A kind of cutting performance degradation assessment method based on AR models Yu SVDD algorithms, includes the following steps：
According to the signal data sample of collected normal condition bottom tool, AR filter models are established；
Signal data sample is handled by the AR filter models, obtains residual signals data set, and calculate
The cumulative probability distribution of each residual signals data in residual signals data set, it is right to establish normal condition bottom tool signal data institute
The feature set answered handles the feature set, obtains the hypersphere for containing minimum volume；
The cutter signal data under collected current state is handled using AR filter models, it is residual to obtain first
Difference signal data set；
Calculate the cumulative probability distribution characteristics of the corresponding first residual signals data of current state bottom tool signal data, note
Make fisrt feature, calculate fisrt feature to the hyperspherical distance, the degree degenerated by Distance Judgment cutting performance.SVDD
It is exactly Support Vector data description that algorithm, which is,.
As a kind of embodiment, the signal data sample according to collected normal condition bottom tool is established
AR filter models, specific step are：
It is signal data sample set by the signal data Sample Establishing of collected normal condition bottom tool；
Arbitrary signal data sample in number of winning the confidence sample set, basis signal data sample establish AR wave filter moulds
Type, and determine the order and filter factor of AR filter models.
It is described that signal data sample is handled by the AR filter models as a kind of embodiment, it obtains
To residual signals data set, and the cumulative probability distribution of each residual signals data in residual signals data set is calculated, established just
Feature set corresponding to normal state bottom tool signal data, handles the feature set, obtains containing the super of minimum volume
Spherical surface, the specific steps are：
By the AR filter models to signal data sample carry out whitening processing, by the sample number of acquisition it is believed that
In number input AR filter models, carry out convolution algorithm with the filter coefficient and obtain filtering signal dataHere, p represents filter order, aiRepresent filter model coefficients, Y (k) represents the numerical value of k points
Linear expression can be worth by first i+k of X, then the residual signals data of AR filter models are expressed as ε (n)=x (n)-Y (n),
Residual signals data set E=[ε are formed by the residual signals data1,ε2,...εm], wherein, x (n) be arbitrary sample, Y
(n) it is filtering signal data；
Residual signals data in the residual signals data set of AR filter models filtering gained are substituted into signal data to tire out
In product probability distribution formula, the how corresponding feature set of normal condition bottom tool signal data is formed, feature set is expressed as G=[F1,
F2,...,Fm], here, signal data accumulation probability distribution is expressed asWherein, f (x) is signal data
Probability density function obtains cumulative probability distribution F (ε)；
The feature set is handled using Support Vector data description algorithm, obtains the hypersphere for containing minimum volume
Face is optimized using gaussian kernel function K (x, y), obtains decision function：
Wherein, αi, αjIt is to correspond to x by what training obtainedi, xjCoefficient, when coefficient is zero, corresponding target sample
Referred to as supporting vector xs, then hypersphere radius surface be expressed asK is gaussian kernel function, and α is for decision function f's (z)
Number, it is corresponding with kernel function K (x, y).
It is described to calculate the corresponding first residual signals number of current state bottom tool signal data as a kind of embodiment
According to cumulative probability distribution characteristics, be denoted as fisrt feature, calculate fisrt feature to the hyperspherical distance, pass through Distance Judgment
Cutting performance degenerate degree the specific steps are：
Calculate the cumulative probability distribution characteristics of the corresponding first residual signals data of current state bottom tool signal data, note
Make fisrt feature；
The distance of the fisrt feature to the suprasphere isIt is assessed by the size of distance HI
Cutting performance, if HI=0 represents that cutter is normal, if HI ＞ 0 represent that performance degradation occurs in cutter；HI values are bigger, show cutter
It can degenerate more serious.
Invention further discloses：
A kind of cutting performance degradation assessment system based on AR models Yu SVDD algorithms, including model building module, processing
It establishes module, processing module and calculates judgment module；
The model building module for the signal data sample according to collected normal condition bottom tool, establishes AR
Filter model；
Module is established in the processing, for being handled by the AR filter models signal data sample, is obtained
Residual signals data set, and the cumulative probability distribution of each residual signals data in residual signals data set is calculated, it establishes normal
Feature set corresponding to state bottom tool signal data, handles the feature set, obtains the hypersphere for containing minimum volume
Face；
The processing module, for using AR filter models to the cutter signal data under collected current state into
Row processing, obtains the first residual signals data set；
The calculating judgment module, for calculating the corresponding first residual signals data of current state bottom tool signal data
Cumulative probability distribution characteristics, be denoted as fisrt feature, calculate fisrt feature to the hyperspherical distance, pass through Distance Judgment knife
Has the degree of performance degradation.
As a kind of embodiment, the model building module includes establishing sample set unit and establishes model unit；
It is described to establish sample set unit, for being letter by the signal data Sample Establishing of collected normal condition bottom tool
Number sample set；
It is described to establish model unit, for the arbitrary signal data sample in number sample set of winning the confidence, basis signal number
According to Sample Establishing AR filter models, and determine the order and filter factor of AR filter models.
As a kind of embodiment, the processing establishes that module establishes unit including set of residuals, feature set establishes unit
With feature set processing unit；
The set of residuals establishes unit, for carrying out whitening to signal data sample by the AR filter models
By in the sample data signal data input AR filter models of acquisition, convolution algorithm is carried out with the filter coefficient for processing
Obtain filtering signal dataHere, p represents filter order, aiRepresent filter model coefficients, Y
(k) linear expression can be worth by first i+k of X by representing the numerical value of k points, then the residual signals data of AR filter models are expressed as
ε (n)=x (n)-Y (n) forms residual signals data set E=[ε by the residual signals data1,ε2,...εm], wherein, x
(n) it is arbitrary sample, Y (n) is filtering signal data；
The feature set establishes unit, for AR filter models to be filtered to the residual error in the residual signals data set of gained
Signal data is substituted into signal data cumulative probability distribution formula, forms the how corresponding feature of normal condition bottom tool signal data
Collection, feature set are expressed as G=[F1,F2,...,Fm], here, signal data accumulation probability distribution is expressed asWherein, f (x) is signal data probability density function, obtains cumulative probability distribution F (ε)；
The feature set processing unit, for being handled using Support Vector data description algorithm the feature set,
The hypersphere for containing minimum volume is obtained, is optimized using gaussian kernel function K (x, y), obtains decision function：
Wherein, αi, αjIt is to correspond to x by what training obtainedi, xjCoefficient, when coefficient is zero, corresponding target sample
Referred to as supporting vector xs, then hypersphere radius surface be expressed asK is gaussian kernel function, and α is for decision function f's (z)
Number, it is corresponding with kernel function K (x, y).
As a kind of embodiment, the calculating judgment module includes fisrt feature computing unit and judging unit；
The fisrt feature computing unit, for calculating corresponding first residual signals of current state bottom tool signal data
The cumulative probability distribution characteristics of data, is denoted as fisrt feature；
The judging unit, the distance for the fisrt feature to the suprasphere areBy away from
Size from HI assesses cutting performance, if HI=0 represents that cutter is normal, if HI ＞ 0 represent that performance degradation occurs in cutter；HI values
It is bigger, it is more serious to show that cutting performance is degenerated.
Invention further discloses：
A kind of computer readable storage medium, is stored with computer program, which realizes base when being executed by processor
In the cutting performance degradation assessment method of AR models and SVDD algorithms the step of.
The present invention has significant technique effect as a result of above technical scheme：
Method using the present invention can effectively realize the assessment degenerated to cutting performance, it is pre- to help to realize numerically-controlled machine tool
The property surveyed repair；
AR filter model pretreatments are carried out to cutter signal data, by the data whitening under different conditions, will after
Continuous contrast standard is unified for white noise signal data, so as to enhance its performance degradation assessment effect；
Difference is detected with the common detection method based on cumulative probability distribution, such as K-S, it is not general by calculating comparison accumulation
The similarity of rate density function, but the optimal of higher dimensional space is obtained using SVDD algorithms (Support Vector data description algorithm)
Suprasphere, so as to significantly change the otherness of comparison；
Using the distance of sample point to suprasphere as cutting performance degeneration index, realize to being usually used in outlier detection
The promotion of SVDD algorithms.
Description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, to embodiment or will show below
There is attached drawing needed in technology description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention, for those of ordinary skill in the art, without having to pay creative labor, may be used also
To obtain other attached drawings according to these attached drawings.
Fig. 1 is the overall flow schematic diagram of the present invention；
Fig. 2 is the detailed process schematic diagram of the wherein step of the present invention；
Fig. 3 is the cutting performance degradation assessment method flow based on AR models Yu SVDD algorithms；
Fig. 4 carries out whitening schematic diagram for vibration signal data by AR linear prediction models；
Fig. 5 is the schematic diagram of Support Vector data description algorithm；
Fig. 6 is respectively the cumulative probability distributional difference schematic diagram under cutter normal condition and cutting-tool wear state；
Fig. 7 is respectively that the HI indexs under cutter normal condition and cutting-tool wear state distinguish result；
Fig. 8 is the overall structure diagram of the present invention.
Label declaration in attached drawing：100th, model building module；200th, module is established in processing；300th, processing module；400、
Calculate judgment module；110th, sample set unit is established；120th, model unit is established；210th, set of residuals establishes unit；220th, feature
Collection establishes unit；230th, feature set processing unit；410th, fisrt feature computing unit；420th, judging unit；
Specific embodiment
With reference to embodiment, the present invention is described in further detail, following embodiment be explanation of the invention and
The invention is not limited in following embodiments.
The invention discloses：
A kind of cutting performance degradation assessment method based on AR models Yu SVDD algorithms, as shown in Figure 1, including following step
Suddenly：
S100, the signal data sample according to collected normal condition bottom tool, establish AR filter models；
S200, signal data sample is handled by the AR filter models, obtains residual signals data set, and
The cumulative probability distribution of each residual signals data in residual signals data set is calculated, establishes normal condition bottom tool signal data
Corresponding feature set handles the feature set, obtains the hypersphere for containing minimum volume；
S300, the cutter signal data under collected current state is handled using AR filter models, obtained
First residual signals data set；
S400, the cumulative probability distribution spy for calculating the corresponding first residual signals data of current state bottom tool signal data
Sign is denoted as fisrt feature, calculates fisrt feature to the hyperspherical distance, the journey degenerated by Distance Judgment cutting performance
Degree.
Further, in the step s 100, as shown in Fig. 2, the letter according to collected normal condition bottom tool
Number sample, establishes AR filter models, and specific step is：
S110, by the signal data Sample Establishing of collected normal condition bottom tool be signal data sample set X=
[x1,x2,...,xm], wherein, x1,x2,…,xmRepresent signal data sample；
Arbitrary signal data sample in S120, number of winning the confidence sample set, basis signal data sample establish AR filtering
Device model, and determine the order p and filter factor a of AR filter modelsq。
In the present embodiment, in step s 200, it is described that signal data sample is carried out by the AR filter models
Processing, obtains residual signals data set, and calculates the cumulative probability distribution of each residual signals data in residual signals data set,
Establish the feature set G=[F corresponding to normal condition bottom tool signal data1,F2,...,Fm], wherein, F1,F2,…FmIt represents
It is cumulative probability distribution, to the feature set G=[F1,F2,...,Fm] handled, the hypersphere for containing minimum volume is obtained,
The specific steps are：
S210, whitening processing is carried out to signal data sample by the AR filter models, by the sample of acquisition
In data signal data input AR filter models, carry out convolution algorithm with the filter coefficient and obtain filtering signal dataHere, p represents filter order, aiRepresent filter model coefficients, Y (k) represents the numerical value of k points
Linear expression can be worth by first i+k of X, then the residual signals data of AR filter models are expressed as ε (n)=x (n)-Y (n),
Residual signals data set E=[ε are formed by the residual signals data1,ε2,...εm], wherein, x (n) be arbitrary sample, Y
(n) it is filtering signal data；
It is prior, here select AR filter models the cutter signal data of acquisition is pre-processed, be in order to
Signal data under different cutting tool states is unified to original bench mark, so as to increase the reliability of later stage performance degradation assessment, ginseng
As shown in attached drawing 4.AR filter models can describe the linear regression relation of the recursion inside data sequence, and normal condition is shaken
Dynamic data signal data is handled by AR wave filters, i.e. signal data whitening process.If cutter is in normal condition,
The form of expression of its consequential signal data should be white noise signal data or close to white noise, if there is performance degradation in cutter,
Then filter result and white noise will appear deviation, and cutting performance can be assessed by the size of the deviation；
S220, the residual signals data in the residual signals data set of AR filter models filtering gained are substituted into signal number
According to the how corresponding feature set of normal condition bottom tool signal data in cumulative probability distribution formula, is formed, feature set is expressed as G=
[F1,F2,...,Fm], here, signal data accumulation probability distribution is expressed asWherein, f (x) is signal
Data probability density function obtains cumulative probability distribution F (ε)；
S230, the feature set is handled using Support Vector data description algorithm, obtains containing minimum volume
Hypersphere is optimized using gaussian kernel function K (x, y), obtains decision function：
Wherein, αi, αjIt is to correspond to x by what training obtainedi, xjCoefficient, when coefficient is zero, corresponding target sample
Referred to as supporting vector xs, then hypersphere radius surface be expressed asK is gaussian kernel function, and α is for decision function f's (z)
Number, it is corresponding with kernel function K (x, y), in this step, to build effective cutting performance degeneration index, calculated herein using SVDD
Method, that is, Support Vector data description algorithm, the algorithm are a kind of important data description method, can seek one comprising all
Or the suprasphere or domain of almost all of target sample and volume minimum, realize the hyperspherical description to target data, Fig. 5 is
Algorithm schematic diagram.If accumulated probability distribution is fallen into optimal suprasphere, then the sample is considered as one by Nonlinear Mapping
Normal point；Otherwise, picture of the sample in feature space is dropped into outside optimal suprasphere, then the sample is considered as an abnormal point,
And its performance degradation degree can be by determining to the distance of optimal suprasphere.
It is described to calculate the tired of the corresponding first residual signals data of current state bottom tool signal data in step S400
Product Probability Characteristics, are denoted as fisrt feature, calculate fisrt feature to the hyperspherical distance, pass through Distance Judgment cutter
Can degenerate degree the specific steps are：
S410, the accumulation for calculating corresponding first residual signals data ε ' (n) of current state bottom tool signal data x ' (n)
Probability Characteristics F ' (ε), is denoted as fisrt feature；
S420, the fisrt feature to the suprasphere distance beBy the size of distance HI come
Cutting performance is assessed, if HI=0 represents that cutter is normal, if HI ＞ 0 represent that performance degradation occurs in cutter；HI values are bigger, show knife
It is more serious to have performance degradation.
Fig. 6 is the disparity map with the cumulative probability distribution of signal data under state of wear under cutter normal condition.From figure
It can be seen that under different conditions cumulative probability of the signal data Jing Guo AR wave filter whitening handling results be distributed show it is bright
Aobvious ground otherness, can assess the state of cutter by the otherness.
To quantify otherness between the two, calculated using Support Vector data description algorithm, as a result such as Fig. 7 institutes
Show.From the figure, it can be seen that the sample point under normal condition is substantially at the position of HI=0 in figure, and there is performance degradation
The sample then position far from HI=0.The examination example the result shows that, the cutter proposed by the present invention based on AR models Yu SVDD algorithms
Performance degradation assessment method can effectively distinguish cutting tool state, have broad application prospects.
The invention also discloses：
A kind of cutting performance degradation assessment system based on AR models Yu SVDD algorithms, as shown in figure 8, including model foundation
Module 100, processing establish module 200, processing module 300 and calculate judgment module 400；
The model building module 100 for the signal data sample according to collected normal condition bottom tool, is established
AR filter models；
Module 200 is established in the processing, for being handled by the AR filter models signal data sample, is obtained
To residual signals data set, and the cumulative probability distribution of each residual signals data in residual signals data set is calculated, established just
Feature set corresponding to normal state bottom tool signal data, handles the feature set, obtains containing the super of minimum volume
Spherical surface；
The processing module 300, for using AR filter models to the cutter signal number under collected current state
According to being handled, the first residual signals data set is obtained；
The calculating judgment module 400, for calculating corresponding first residual signals of current state bottom tool signal data
The cumulative probability distribution characteristics of data is denoted as fisrt feature, calculates fisrt feature to the hyperspherical distance, is sentenced by distance
Breaking has the degree of performance degradation.
The model building module 100 includes establishing sample set unit 110 and establishes model unit 120；
It is described to establish sample set unit 110, for by the signal data Sample Establishing of collected normal condition bottom tool
For signal data sample set；
It is described to establish model unit 120, for the arbitrary signal data sample in number sample set of winning the confidence, basis signal
Data sample establishes AR filter models, and determines the order and filter factor of AR filter models.
The processing establishes that module 200 establishes unit 210 including set of residuals, feature set is established at unit 220 and feature set
Manage unit 230；
The set of residuals establishes unit 210, for carrying out white noise to signal data sample by the AR filter models
By in the sample data signal data input AR filter models of acquisition, convolution is carried out with the filter coefficient for soundization processing
Operation obtains filtering signal dataHere, p represents filter order, aiRepresent filter model system
Number, Y (k) represent that the numerical value of k points can be by the residual signals tables of data of preceding i+k the value linear expression, then AR filter models of X
ε (n)=x (n)-Y (n) is shown as, residual signals data set E=[ε are formed by the residual signals data1,ε2,...εm],
In, x (n) is arbitrary sample, and Y (n) is filtering signal data；
The feature set establishes unit 220, for AR filter models to be filtered in the residual signals data set of gained
Residual signals data are substituted into signal data cumulative probability distribution formula, how corresponding form normal condition bottom tool signal data
Feature set, feature set are expressed as G=[F1,F2,...,Fm], here, signal data accumulation probability distribution is expressed asWherein, f (x) is signal data probability density function, obtains cumulative probability distribution F (ε)；
The feature set processing unit 230, at using Support Vector data description algorithm to the feature set
Reason is obtained the hypersphere for containing minimum volume, is optimized using gaussian kernel function K (x, y), obtain decision function：
Wherein, αi, αjIt is to correspond to x by what training obtainedi, xjCoefficient, when coefficient is zero, corresponding target sample
Referred to as supporting vector xs, then hypersphere radius surface be expressed asK is gaussian kernel function, and α is for decision function f's (z)
Number, it is corresponding with kernel function K (x, y).
The calculating judgment module 400 includes fisrt feature computing unit 410 and judging unit 420；
The fisrt feature computing unit 410, for calculating corresponding first residual error of current state bottom tool signal data
The cumulative probability distribution characteristics of signal data, is denoted as fisrt feature；
The judging unit 420, the distance for the fisrt feature to the suprasphere areIt is logical
The size of distance HI is crossed to assess cutting performance, if HI=0 represents that cutter is normal, if HI ＞ 0 represent that performance degradation occurs in cutter；
HI values are bigger, and it is more serious to show that cutting performance is degenerated.
The invention also discloses：
A kind of computer readable storage medium, is stored with computer program, which realizes base when being executed by processor
In the cutting performance degradation assessment method of AR models and SVDD algorithms the step of.
For device embodiment, since it is basicly similar to embodiment of the method, so description is fairly simple, it is related
Part illustrates referring to the part of embodiment of the method.
Each embodiment in this specification is described by the way of progressive, the highlights of each of the examples are with
The difference of other embodiment, just to refer each other for identical similar part between each embodiment.
It should be understood by those skilled in the art that, the embodiment of the present invention can be provided as method, apparatus or computer program
Product.Therefore, the reality in terms of complete hardware embodiment, complete software embodiment or combination software and hardware can be used in the present invention
Apply the form of example.Moreover, the computer for wherein including computer usable program code in one or more can be used in the present invention
The computer program production that usable storage medium is implemented on (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.)
The form of product.
The present invention be with reference to according to the method for the present invention, the flow chart of terminal device (system) and computer program product
And/or block diagram describes.It should be understood that each flow in flowchart and/or the block diagram can be realized by computer program instructions
And/or the flow in box and flowchart and/or the block diagram and/or the combination of box.These computer programs can be provided to refer to
Enable the processor of all-purpose computer, special purpose computer, Embedded Processor or other programmable data processing terminal equipments with
Generate a machine so that the instruction performed by computer or the processor of other programmable data processing terminal equipments generates
It is used to implement the function specified in one flow of flow chart or multiple flows and/or one box of block diagram or multiple boxes
Device.
These computer program instructions, which may also be stored in, can guide computer or other programmable data processing terminal equipments
In the computer-readable memory to work in a specific way so that the instruction being stored in the computer-readable memory generates packet
The manufacture of command device is included, which realizes in one flow of flow chart or multiple flows and/or one side of block diagram
The function of being specified in frame or multiple boxes.
These computer program instructions can be also loaded into computer or other programmable data processing terminal equipments so that
Series of operation steps are performed on computer or other programmable terminal equipments to generate computer implemented processing, thus
The instruction offer performed on computer or other programmable terminal equipments is used to implement in one flow of flow chart or multiple flows
And/or specified in one box of block diagram or multiple boxes function the step of.
It should be noted that：
" one embodiment " or " embodiment " mentioned in specification means the special characteristic described in conjunction with the embodiments, structure
Or characteristic is included at least one embodiment of the present invention.Therefore, the phrase " reality that specification various places throughout occurs
Apply example " or " embodiment " same embodiment might not be referred both to.
Furthermore, it is necessary to illustrate, the specific embodiment described in this specification, the shape of parts and components is named
Title etc. can be different.The equivalent or simple change that all construction, feature and principles according to described in inventional idea of the present invention are done, is wrapped
It includes in the protection domain of patent of the present invention.Those skilled in the art can be to described specific implementation
Example is done various modifications or additions or is substituted in a similar way, without departing from structure of the invention or surmounts this
Range as defined in the claims, is within the scope of protection of the invention.
Claims (9)
- A kind of 1. cutting performance degradation assessment method based on AR models Yu SVDD algorithms, it is characterised in that include the following steps：According to the signal data sample of collected normal condition bottom tool, AR filter models are established；Signal data sample is handled by the AR filter models, obtains residual signals data set, and calculate residual error Signal data concentrates the cumulative probability distribution of each residual signals data, establishes corresponding to normal condition bottom tool signal data Feature set handles the feature set, obtains the hypersphere for containing minimum volume；The cutter signal data under collected current state is handled using AR filter models, obtains the first residual error letter Number collection；The cumulative probability distribution characteristics of the corresponding first residual signals data of current state bottom tool signal data is calculated, is denoted as the One feature calculates fisrt feature to the hyperspherical distance, the degree degenerated by Distance Judgment cutting performance.
- 2. the cutting performance degradation assessment method according to claim 1 based on AR models Yu SVDD algorithms, feature exist In the signal data sample according to collected normal condition bottom tool establishes AR filter models, specific step For：It is signal data sample set by the signal data Sample Establishing of collected normal condition bottom tool；Arbitrary signal data sample in number of winning the confidence sample set, basis signal data sample establish AR filter models, and Determine the order and filter factor of AR filter models.
- 3. the cutting performance degradation assessment method according to claim 2 based on AR models Yu SVDD algorithms, feature exist In, it is described that signal data sample is handled by the AR filter models, residual signals data set is obtained, and calculate residual The cumulative probability distribution of each residual signals data, is established corresponding to normal condition bottom tool signal data in difference signal data set Feature set, the feature set is handled, obtain contain minimum volume hypersphere, the specific steps are：Whitening processing is carried out to signal data sample by the AR filter models, by the sample data signal number of acquisition According in input AR filter models, carry out convolution algorithm with the filter coefficient and obtain filtering signal dataHere, p represents filter order, aiRepresent filter model coefficients, Y (k) represents the numerical value of k points Linear expression can be worth by first i+k of X, then the residual signals data of AR filter models are expressed as ε (n)=x (n)-Y (n), Residual signals data set E=[ε are formed by the residual signals data1,ε2,...εm], wherein, x (n) be arbitrary sample, Y (n) it is filtering signal data；It is general that residual signals data in the residual signals data set of AR filter models filtering gained are substituted into signal data accumulation In rate distribution formula, the how corresponding feature set of normal condition bottom tool signal data is formed, feature set is expressed as G=[F1, F2,...,Fm], here, signal data accumulation probability distribution is expressed asWherein, f (x) is signal data Probability density function obtains cumulative probability distribution F (ε)；The feature set is handled using Support Vector data description algorithm, the hypersphere for containing minimum volume is obtained, adopts It is optimized with gaussian kernel function K (x, y), obtains decision function：Wherein, αi, αjIt is to correspond to x by what training obtainedi, xjCoefficient, when coefficient is zero, corresponding target sample be known as branch Hold vector xs, then hypersphere radius surface be expressed asK is gaussian kernel function, and α is the coefficient of decision function f (z), with core Function K (x, y) is corresponding.
- 4. the cutting performance degradation assessment method according to claim 3 based on AR models Yu SVDD algorithms, feature exist In, the cumulative probability distribution characteristics for calculating the corresponding first residual signals data of current state bottom tool signal data, note Make fisrt feature, calculate fisrt feature to the hyperspherical distance, pass through the tool for the degree that Distance Judgment cutting performance is degenerated Body step is：The cumulative probability distribution characteristics of the corresponding first residual signals data of current state bottom tool signal data is calculated, is denoted as the One feature；The distance of the fisrt feature to the suprasphere isCutter is assessed by the size of distance HI Performance, if HI=0 represents that cutter is normal, if HI ＞ 0 represent that performance degradation occurs in cutter；HI values are bigger, show that cutting performance moves back Change more serious.
- 5. a kind of cutting performance degradation assessment system based on AR models Yu SVDD algorithms, which is characterized in that including model foundation Module, processing establish module, processing module and calculate judgment module；The model building module for the signal data sample according to collected normal condition bottom tool, establishes AR filtering Device model；Module is established in the processing, for being handled by the AR filter models signal data sample, obtains residual error Signal data collection, and the cumulative probability distribution of each residual signals data in residual signals data set is calculated, establish normal condition Feature set corresponding to bottom tool signal data handles the feature set, obtains the hypersphere for containing minimum volume；The processing module, at using AR filter models to the cutter signal data under collected current state Reason, obtains the first residual signals data set；The calculating judgment module, for calculating the tired of the corresponding first residual signals data of current state bottom tool signal data Product Probability Characteristics, are denoted as fisrt feature, calculate fisrt feature to the hyperspherical distance, pass through Distance Judgment cutter The degree that can be degenerated.
- 6. the cutting performance degradation assessment system based on AR models Yu SVDD algorithms according to claim 5, which is characterized in that The model building module includes establishing sample set unit and establishes model unit；It is described to establish sample set unit, for being signal number by the signal data Sample Establishing of collected normal condition bottom tool According to sample set；It is described to establish model unit, for the arbitrary signal data sample in number sample set of winning the confidence, basis signal data sample This establishes AR filter models, and determines the order and filter factor of AR filter models.
- 7. the cutting performance degradation assessment system based on AR models Yu SVDD algorithms according to claim 6, which is characterized in that The processing establishes that module establishes unit including set of residuals, feature set establishes unit and feature set processing unit；The set of residuals establishes unit, for being carried out at whitening to signal data sample by the AR filter models Reason by the sample data signal data input AR filter models of acquisition, carries out convolution algorithm with the filter coefficient and obtains To filtering signal dataHere, p represents filter order, aiRepresent filter model coefficients, Y (k) linear expression can be worth by first i+k of X by representing the numerical value of k points, then the residual signals data of AR filter models are expressed as ε (n)=x (n)-Y (n) forms residual signals data set E=[ε by the residual signals data1,ε2,...εm], wherein, x (n) it is arbitrary sample, Y (n) is filtering signal data；The feature set establishes unit, for AR filter models to be filtered to the residual signals in the residual signals data set of gained Data are substituted into signal data cumulative probability distribution formula, form the how corresponding feature set of normal condition bottom tool signal data, Feature set is expressed as G=[F1,F2,...,Fm], here, signal data accumulation probability distribution is expressed as Wherein, f (x) is signal data probability density function, obtains cumulative probability distribution F (ε)；The feature set processing unit for being handled using Support Vector data description algorithm the feature set, is obtained Contain the hypersphere of minimum volume, optimized using gaussian kernel function K (x, y), obtain decision function：Wherein, αi, αjIt is to correspond to x by what training obtainedi, xjCoefficient, when coefficient is zero, corresponding target sample be known as branch Hold vector xs, then hypersphere radius surface be expressed asK is gaussian kernel function, and α is the coefficient of decision function f (z), with core Function K (x, y) is corresponding.
- 8. the cutting performance degradation assessment system based on AR models Yu SVDD algorithms according to claim 7, which is characterized in that The calculating judgment module includes fisrt feature computing unit and judging unit；The fisrt feature computing unit, for calculating the corresponding first residual signals data of current state bottom tool signal data Cumulative probability distribution characteristics, be denoted as fisrt feature；The judging unit, the distance for the fisrt feature to the suprasphere arePass through distance HI Size assess cutting performance, if HI=0 represents that cutter is normal, if HI ＞ 0 represent that performance degradation occurs in cutter；HI values are got over Greatly, it is more serious to show that cutting performance is degenerated.
- 9. a kind of computer readable storage medium, is stored with computer program, which is characterized in that the program is executed by processor The step of Shi Shixian claim 1-4 any one the methods.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810133641.2A CN108229071B (en) | 2018-02-09 | 2018-02-09 | Cutter performance degradation evaluation method and system based on AR model and SVDD algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810133641.2A CN108229071B (en) | 2018-02-09 | 2018-02-09 | Cutter performance degradation evaluation method and system based on AR model and SVDD algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108229071A true CN108229071A (en) | 2018-06-29 |
| CN108229071B CN108229071B (en) | 2021-06-22 |
Family
ID=62661324
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810133641.2A Active CN108229071B (en) | 2018-02-09 | 2018-02-09 | Cutter performance degradation evaluation method and system based on AR model and SVDD algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108229071B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112085121A (en) * | 2020-09-18 | 2020-12-15 | 郑州大学 | A Nonlinear Profile Data Monitoring Method Based on LLE-SVDD |
| CN114095222A (en) * | 2021-11-10 | 2022-02-25 | 湖南大学 | LDoS attack detection method based on perceptual linear prediction and SVDD |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100205042A1 (en) * | 2009-02-11 | 2010-08-12 | Mun Johnathan C | Integrated risk management process |
| CN102072144A (en) * | 2011-03-11 | 2011-05-25 | 大连大学 | Vibration and noise online monitoring and fault diagnosis system of scroll compressor |
| CN103545832A (en) * | 2013-09-22 | 2014-01-29 | 国家电网公司 | A photovoltaic system energy storage capacity configuration method based on power generation prediction error |
| CN103822793A (en) * | 2014-01-20 | 2014-05-28 | 北京邮电大学 | Complicated equipment acoustic fault recognition and location method |
| WO2015163121A1 (en) * | 2014-04-25 | 2015-10-29 | 株式会社日立製作所 | Voltage stability-monitoring device and method |
| CN107092582A (en) * | 2017-03-31 | 2017-08-25 | 江苏方天电力技术有限公司 | One kind is based on the posterior exceptional value on-line checking of residual error and method for evaluating confidence |
| CN107144430A (en) * | 2017-06-27 | 2017-09-08 | 电子科技大学 | A kind of Method for Bearing Fault Diagnosis based on incremental learning |
| CN107563008A (en) * | 2017-08-08 | 2018-01-09 | 三峡大学 | Estimating operation reliability of tool method based on SVD conversion and supporting vector space |
-
2018
- 2018-02-09 CN CN201810133641.2A patent/CN108229071B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100205042A1 (en) * | 2009-02-11 | 2010-08-12 | Mun Johnathan C | Integrated risk management process |
| CN102072144A (en) * | 2011-03-11 | 2011-05-25 | 大连大学 | Vibration and noise online monitoring and fault diagnosis system of scroll compressor |
| CN103545832A (en) * | 2013-09-22 | 2014-01-29 | 国家电网公司 | A photovoltaic system energy storage capacity configuration method based on power generation prediction error |
| CN103822793A (en) * | 2014-01-20 | 2014-05-28 | 北京邮电大学 | Complicated equipment acoustic fault recognition and location method |
| WO2015163121A1 (en) * | 2014-04-25 | 2015-10-29 | 株式会社日立製作所 | Voltage stability-monitoring device and method |
| CN107092582A (en) * | 2017-03-31 | 2017-08-25 | 江苏方天电力技术有限公司 | One kind is based on the posterior exceptional value on-line checking of residual error and method for evaluating confidence |
| CN107144430A (en) * | 2017-06-27 | 2017-09-08 | 电子科技大学 | A kind of Method for Bearing Fault Diagnosis based on incremental learning |
| CN107563008A (en) * | 2017-08-08 | 2018-01-09 | 三峡大学 | Estimating operation reliability of tool method based on SVD conversion and supporting vector space |
Non-Patent Citations (7)
| Title |
|---|
| S. MA等: "Uncertainty Reduced Novelty Detection Approach Applied to Rotating Machinery for Condition Monitoring", 《SHOCK AND VIBRATION》 * |
| S.POYHONEN等: "Signal processing of vibrations for condition monitoring of an induction", 《FIRST INTERNATIONAL SYMPOSIUM ON CONTROL, COMMUNICATIONS AND SIGNAL PROCESSING,2004》 * |
| YASENJIANG JIARULA等: "A fault detection method for complex electromechanical system using improved integrated approach", 《2017 IEEE 3RD INFORMATION TECHNOLOGY AND MECHATRONICS ENGINEERING CONFERENCE (ITOEC)》 * |
| 从飞云等: "基于AR模型的Kolmogorov-Smirnov检验性能退化及预测研究", 《振动与冲击》 * |
| 刘雨: "基于支持向量数据描述与信息融合的设备性能退化评估研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
| 潘玉娜等: "小波包-支持向量数据描述在轴承性能退化评估中的应用研究", 《振动与冲击》 * |
| 陈斌等: "旋转设备声学故障特征提取与优化方法", 《北京邮电大学学报》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112085121A (en) * | 2020-09-18 | 2020-12-15 | 郑州大学 | A Nonlinear Profile Data Monitoring Method Based on LLE-SVDD |
| CN114095222A (en) * | 2021-11-10 | 2022-02-25 | 湖南大学 | LDoS attack detection method based on perceptual linear prediction and SVDD |
| CN114095222B (en) * | 2021-11-10 | 2022-11-11 | 湖南大学 | LDoS attack detection method based on perceptual linear prediction and SVDD |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108229071B (en) | 2021-06-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114078473B (en) | Tool detection method, electronic device and storage medium | |
| KR102229859B1 (en) | Operation management method, apparatus and system using machine learning and transfer learning | |
| CN114493057B (en) | Production process parameter recommendation method based on abnormal working conditions and related equipment | |
| CN108205580A (en) | A kind of image search method, device and computer readable storage medium | |
| CN106384197A (en) | Service quality evaluation method and device based on big data | |
| CN114918581B (en) | Welding parameter processing method and device, storage medium and processor | |
| CN108229071A (en) | Cutting performance degradation assessment method and system based on AR models Yu SVDD algorithms | |
| US10639765B2 (en) | System and method for compensating deviations in an optical lens manufacturing process | |
| EP2916164A1 (en) | Method for designing spectacle lens, and system for designing spectacle lens | |
| US20150293377A1 (en) | Method for providing an optical lens | |
| CN118819069B (en) | Intelligent path optimization method and system for CNC machining tools | |
| CN118427539B (en) | Waste water management control system and method based on data processing | |
| CN110781340A (en) | Offline evaluation method, system and device for recall strategy of recommendation system and storage medium | |
| CN112131467A (en) | Data prediction method, system, medium and device | |
| CN108491274A (en) | Optimization method, device, storage medium and the equipment of distributed data management | |
| CN110969483B (en) | Method and device for identifying merchant position and electronic equipment | |
| CN118428564A (en) | Sea chart data risk monitoring method and system based on multi-model linkage | |
| CN116307946B (en) | Processing environment-based adaptive energy consumption assessment method, device, equipment and medium | |
| CN115345237B (en) | Dimension reduction method, device and equipment for laser point cloud data and storage medium | |
| CN117348564A (en) | Process set recommendation method based on comprehensive error judgment and related equipment thereof | |
| CN112269879B (en) | Method and equipment for analyzing middle station log based on k-means algorithm | |
| CN115880696A (en) | Internet of things card management method and device based on deep learning and related media | |
| CN115690287A (en) | Diversion strip rendering method and device based on road surface elevation data | |
| CN115905373A (en) | Data query and analysis method, device, equipment and storage medium | |
| CN115841663A (en) | Wafer image identification method, device, equipment and application |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |