CN104700200B - A kind of product multivariate quality monitoring method towards digitlization workshop - Google Patents
A kind of product multivariate quality monitoring method towards digitlization workshop Download PDFInfo
- Publication number
- CN104700200B CN104700200B CN201410797712.0A CN201410797712A CN104700200B CN 104700200 B CN104700200 B CN 104700200B CN 201410797712 A CN201410797712 A CN 201410797712A CN 104700200 B CN104700200 B CN 104700200B
- Authority
- CN
- China
- Prior art keywords
- msub
- mrow
- control
- principal component
- quality
- 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.)
- Active
Links
Abstract
A kind of product multivariate quality monitoring method towards digitlization workshop, first construct the quality information acquisition scheme in a set of suitable digitlization workshop；The quality control towards digitlization workshop and improved model are constructed again, and the specially multimass characteristic for critical product critical process between machine extra bus has been carried out based on the complex process capability analysis for improving principal component analytical method；It is difficult to carry out diagnosis orientation problem to process exception for multivariate statistics control control figure, proposes to realize multivariate process quality control and diagnosis using principal component analysis technology；The present invention can make the assessment that workshop is quantified for the assurance ability of the critical process of critical product, and the source caused by timely alignment quality problem when there are quality problems, so as to avoid further cost allowance.
Description
Technical field
The invention belongs to Mechanical Product's Machining manufacturing process Quality Control Technology field, more particularly to it is a kind of towards digitlization
The product multivariate quality monitoring method in workshop, it is a kind of process for adding product manufacturing quality to machine based on principal component analytical method
Capability analysis and the technology of quality problems diagnosis.
Background technology
Statistical Process Control (SPC) refer to using Mathematical Statistics Analysis theory to production process carry out product quality monitoring and
The method of control, it is the important technology in quality control, is to obtain qualified products quality instrument is effectively ensured, simultaneously and
Process performance monitors and the basis of process exception diagnosis.Traditional SPC technologies utilize standard Shewhart control chart, being capable of monitoring process
Stable state whether is in, and early warning is carried out to process exception.This method of quality control is widely used, and is taken
Obtained good economic benefit.SPC technologies are quite ripe, but traditional SPC technologies still suffer from some in actual applications asks
Topic：
First, China's manufacturing industry is horizontal improves constantly, and a large amount of advanced production equipments constantly come into operation, and production efficiency obtains
Surprising raising is arrived, traditional working method that quality characteristics data is recorded by collected by hand can not meet enterprise
Need.At present, manufacturing business largely uses advanced numerically controlled processing equipment, improves Workshop Production rate, and workshop needs to measure
Qualitative data amount explosive growth is presented, the efficiency of workshop quality inspection, which has become, restricts the bottleneck that production efficiency improves.
In addition, workshop, in order to reduce the unnecessary circulation between process, using pipeline work, this working method is to mass number
According to collection propose higher requirement, it is desirable to qualitative data can obtain timely processing in time, to prevent a upper procedure
Quality problems enter next process, cause the unnecessary wasting of resources.Finally, Workshop production becoming with flexibility
Gesture, the product category in workshop is varied, and this also proposes new requirement to the collection of qualitative data, it is desirable to for different type
Qualitative data can timely collection analysis.Based on the newest feature in digitlization workshop, traditional data acquisition modes are not
Requirement of the statistical Process Control to data acquisition in digitlization workshop can be met.
Secondly, traditional statistical Process Control often can be only done to monotropic due to limitations such as e measurement technology, analytical technologies
The statistical Process Control of amount.Take in actual applications and the Critical to quality of critical process is individually monitored, this side
Method is widely applied in the past few decades, and it is horizontal to substantially increase workshop crudy to a certain extent.But with
The horizontal continuous improvement of manufacturing industry, and continuous improvement of the people to quality requirement, it is this that single argument is individually controlled
Mode exposes some problems.In actual workshop manufacturing process, between each process and between each mass property not
It is separate entirely, is often related between them, their simple monitoring respectively are got up inherently to give workshop quality
Control brings certain error.And bigger economic loss will be brought to enterprise in Digital manufacturing workshop, these errors.In order to
Dependency relation between variable is taken into account, K.S.Chen, the pass between W.L.Pearn and P.C.Lin et al. proposition variable
Connection figure does further monitoring, in this way the shortcomings that be the number of associated diagram when the variable quantity for needing to monitor is more
Amount increase is too fast, and workload is big and operability is little.Under this background we there is an urgent need to by multivariatestatistical process control
(Multivariate Statistical Process Control, MSPC) is applied to mill count process matter with diagnostic techniques
In amount control.Quality management is carried out to workshop manufacturing process using MSPC technologies, can be effectively to each in manufacturing process
Variable carries out unified monitoring, finds hidden danger present in whole manufacturing process and timely processing in time, improves Workshop Production quality
It is horizontal.
For the problem of quality control aspect is present between traditional vehicle, build the workshop quality information towards digitlization workshop and adopt
Collection scheme simultaneously has urgent and real meaning to the expansion further investigation of workshop multivariate process quality control problem.
The content of the invention
The problem of existing for prior art, it is an object of the invention to provide a kind of product towards digitlization workshop is polynary
Quality control method, it can make what workshop was quantified for the assurance ability of the critical process of critical product by this method
Assess, and the source caused by timely alignment quality problem when there are quality problems, so as to avoid further cost allowance.
To achieve these goals, present invention employs following technological means.
A kind of product multivariate quality monitoring method towards digitlization workshop, comprise the following steps：
Step 1：Pass through the acquisition scheme of internal lan structure digitlization workshop qualitative data between machine extra bus；
Step 2：Qualitative data is obtained by the Critical to quality for gathering critical process, to the multivariate quality of critical process
Characteristic carries out complex process capability analysis, when complex process capability analysis result meets customer's requirement or quality management body
During the requirement of system, it is transferred to according to Critical to quality to determine the process of the type of sampling plan and control figure；Work as complex process
When capability analysis result is unsatisfactory for the requirement of customer's requirement or quality management system, quality problems need to be analyzed,
Find the measure solved the problems, such as and corrected, until reaching quality level；
Step 3：It is that variable control figure is preferential that control figure, which chooses rule, does not find suitable variable control really
In the case of drawing, consider further that and reckon by the piece or enumeration control figure；Qualitative data according to collecting is handled sample data,
Corresponding specification limit is calculated, and draws corresponding initial control figure and process sentence surely；
Step 4：When control figure is not at controlled state, searching system is different because if it is different because of supplement sampling not find system
And carry out the drafting of control figure；If it is different because eliminating different because after, control figure being repainted according to sampling plan to find system；Work as control
When drawing is in controlled state, the monitoring of production status is carried out；
Step 5：When monitoring process appearance exception, abnormal Producing reason is found；If find abnormal different because whether see needs
Change control figure；If desired control figure is modified, sampling prescription should be redefined, and carry out the drafting of control figure, instead
Use before control figure carry out production status monitoring；If monitoring process does not occur exception, process finishing, it is transferred to under
The process control of one sample.
The present invention has advantages below：
Digitlization workshop quality data collection scheme proposed by the present invention has a reliability, promptness, integrality and continuous
The characteristics of property.These data are to ensure the accuracy of qualitative data analysis result, the basis of reliability.
Present invention improves over traditional deficiency that dimensionality reduction is carried out for complex process capability analysis, improved principal component is proposed
Specification interval computation method, and propose a set of Multivariate Process Capability Indices computational methods so that analysis result is more accurate.
The present invention can only be monitored for multivariate process quality control figure to multivariate statistics amount, for being which variable draws actually
Play the problem of exception is difficult and makes accurate explanation and propose a kind of multivariate quality diagnostic method based on principal component analysis, it is fast to help
The positioning out-of-the way position of speed, reduce the workshop cost loss thereby resulted in.
Brief description of the drawings
Fig. 1 is present invention digitlization workshop data acquisition scheme.
Fig. 2 is the multivariate process quality control and improved model figure of the present invention.
Fig. 3 is the control figure selection flow chart of the present invention.
Fig. 4 is the multivariate process quality control diagnostic flow chart based on principal component analysis of the present invention.
Fig. 5 is example part drawing.
Fig. 6 is hole milling process T2Control figure.
Fig. 7 is hole milling process MEWMA control figures.
Fig. 8 is first principal component control figure.
Fig. 9 is Second principal component, control figure.
The principal component control figures of Figure 10 the 3rd.
Figure 11 is the mean chart of diameter variable.
Figure 12 is the mean chart of depth variable.
Figure 13 is the mean chart of the variable of distance 1.
Figure 14 is the mean variable value control figure of distance 2.
Table 1 is part quality characteristic requirements.
Table 2 is actual measurement data.
Table 3 is hole milling process sample data principal component analysis result.
Embodiment
Technical scheme is described further below in conjunction with the accompanying drawings.
Step 1：By the acquisition scheme of internal lan structure digitlization workshop qualitative data between machine extra bus, such as Fig. 1 institutes
Show.Manual data acquisition mode is primarily directed to some and only has just obtainable data by dedicated gauge and range estimation formula weight
And the data of some attributes.When RFID is mainly used for quality planning establishment, the product of RFID label tag and each model in workshop
Correspond, scanning RFID label tag by scanner in production scene can will examine item to include counting corresponding to the product
In process control system.Digital intelligent amount instrument is mainly connected by wire/wireless mode with computer, and quality inspection personnel is to parts
Characteristic can be transmitted measurement data automatically to Statistical Quality Control System to carry out the analysis of data when measuring.
Step 2：By the qualitative data collected, complex process ability is carried out to the Multivariate Quality Characteristics of critical process
Analysis, using the complex process capability analysis method based on principal component analysis.
Variable X represents the Critical to quality of critical process, if F1Represent first linear combination institute shape of original variable
Into principal component component, i.e. F1=a11X1+a21X2+...+ap1Xp,ai=(a1i,a2i,...,api)TIt is the covariance matrix of variable X
Σ ith feature root λiCharacteristic vector, it is assumed that variable X obey be p dimension normal distribution, meet workshop data collection reality
Border situation, in the case where p ties up normal distribution, covariance matrix Σ and the calculation formula of covariance matrix Σ characteristic roots and characteristic vector will be
It is illustrated below, first principal component includes original variable information size and can measured with its variance size, its variance yields Var (F1)
It is bigger, show that the information that first principal component component includes original variable is more, it is desirable to first principal component component F1Comprising original
Variable information amount is maximum, therefore the F chosen1Should be original variable X1,X2,...,XpAll linear combinations in variance it is maximum
, it can so construct variance yields is sequentially reduced the second, the three ..., p-th of principal component, as shown by the following formula：
Wherein principal component FiIt is exactly the ith feature root λ with data matrix X covariance matrix ΣiCharacteristic vector ai=
(a1i,a2i,...,api)TFor the linear combination of coefficient, and there are Var (Fi)=λi。
Represent be i-th of principal component contribution rate, Var (Fi) represent principal component FiSide
Poor size.The variance size sum of P principal component before expression；
In actual applications, for p original variable information, it is not necessary to build p principal component component and analyzed, work as m
≤ p, when m principal component component can reflect most information in original variable, it can not continue to build principal component component,
The accumulation of current m principal component component and contribution rateOne can consider that this m principal component component during more than 90%
Reflect most information that p original variable includes.
Assuming that variable X obeys p dimension normal distributions, i.e. X~Np(μ, Σ), wherein μ are population mean vector, and Σ is covariance
Matrix.
Due to Σp×pFor positive definite symmetric matrices, therefore positive definite matrix U be present so that
Wherein, λ1,λ2,...,λpIt is Σ characteristic value, and meets λ1≥λ2≥...≥λp＞ 0, U=(U1,U2,...,Up)
For unit orthogonal matrix,
Remember Y=(Y1,Y2,...,Yp)=UTX then Y1,Y2,...,YpX the 1st, 2 ..., p principal components are referred to as,
Because X obeys multivariate normal distributions, according to the linear combination of normal distribution still Normal Distribution, then Y obediences are more
First normal distribution.
Again
EY=E (UTX)=UTEX=UTμ
Cov(Z,YT)=Cov (UTX,XTU)=UTCov(X,XT) U=UTΣ U=Λ
The proportion of each principal component component synthesis original variable information after conversion is different, utilizes principal component component mistake
Journey Capability index needs to assign different weight coefficients to each principal component when constructing complex process ability, by principal component analysis
Principle understands it is rational, therefore multivariable process Capability index definable by the use of principal component component contribution rate as weight coefficient
For：
Wherein ri=ri/ tr (Λ), tr (Λ)=λ1+λ2+...+λp,For the single of the i-th principal component
Multivariable process Capability index,
If MCP>=0.9973, then the potential ability of process is to meet to require.If process meets MCP≥
0.9973, step 3 is transferred to as shown in Figure 2, otherwise carries out quality improvement, until it is that process capability index meets to require to reach.
Step 3：Selection control figure will be carried out according to the overall principle of economy and accuracy, due to value or the enumeration of reckoning by the piece
The sample size that Value Data control figure generally requires is larger, so cost of sampling is high, Check-Out Time length, so control figure is chosen typically
Principle is that variable control figure is preferential, in the case of not finding suitable variable control figure really, considers further that and reckons by the piece or count
Point control figure.It is as shown in Figure 3 that control figure chooses flow.Qualitative data according to collecting is handled sample data, meter
Corresponding specification limit is calculated, and draws corresponding initial control figure and process sentence surely.Why first to carry out sentencing surely, be due to
If the control figure established when process is in non-statistical controlled process state with sample point controls follow-up production process, not only rise
Less than good control effect, the forecast of mistake can be brought to enterprise on the contrary, is caused damage to enterprise.
Step 4：, can be to use statistical Process Control as shown in Fig. 2 if step 3 system, which sentences steady result, is in controlled state
Figure is monitored to production process, is carried out early warning to the unusual fluctuation in production in time, timely processing is answered when there is process exception, such as
There is not exception in fruit, then carries out described point control to next sample, until process terminates or occurs abnormal.
If workshop is answered establishment officer to find reason and eliminated when process is in non-statistical controlled process state, eliminate
System is different should to be sampled according to sampling plan again because after, and draws control figure, see whether system is in controlled state.This step
Suddenly should repeat until process is in controlled state.
Step 5：If previous step determine controlled state control figure produce abnormal cause will not go out in outside, can only be by
In caused by internal cause.If monitoring process does not occur exception, process finishing, the process control to next sample is transferred to.
When the process of monitoring occurs abnormal, multivariate process quality control and diagnosis are realized using principal component analysis technology.Its flow such as Fig. 4 institutes
Show.Key step is as follows：
(1) after occurring extremely, principal component analysis calculating is carried out to multivariate quality data sample data.
(2) by obtain accounting for variation influence after principal component analysis 90% or so preceding k principal component component.
(3) control figure (mean chart or EWMA control figures) of k main variables before drawing.
(4) once finding principal component PCiGeneration is abnormal, it is determined that to principal component PCiInfluence maximum original variable Xi, tentatively
It is due to original variable X to diagnose the exception of principal component control figure and the exception of multivariate process quality control figureiException it is caused.
(5) original variable X is drawniCorresponding monodrome control figure, to original variable XiBe monitored, judge whether be due to
Original variable XiThe caused multivariate process quality control figure of exception it is abnormal.
(6) according to original variable XiException take process correction corrective measure, to process carry out loop control improvement,
To realize the sustained improvement of quality.
Embodiment
As shown in figure 5, certain on-and-off enterprise process it is a certain switch parts during, it is necessary in the parts list facing cut one
Blind hole, the process have four specific parameter requests, as shown in table 1.Deploy multivariate quality control for four mass propertys of table 1
System and diagnosis research, the measurement data of four mass propertys are as shown in table 2.Based on plant site to diameter, depth, distance 1, away from
50 groups of sample data T that from 2 four mass propertys are gathered2Control figure and MEWMA control figures such as Fig. 6, shown in Fig. 7.
The T of paired observation hole milling process2Control figure and MEWMA control figures, we can easily have found, in T2Control
Figure, sample 37 will exceed T2Upper control limit, but without departing from；In MEWMA control figures, sample 37 is substantially beyond control figure
Upper control limit, occur abnormal.
While finding the exception in process in time using multivariate process quality control figure, for multivariate process quality control we
It is caused by the exception of which variable (diameter, depth, distance 1, distance 2) actually to also need to determine in time, so as to right in time
Exception is handled.Principal component analysis is carried out to process using principal component analysis technology, and entered according to the result of principal component analysis
One step searches error source.
The result of calculation that principal component analysis is carried out to hole milling process sample data is as shown in table 3.
By above principal component analysis result, it will be seen that the accumulation contribution rate of first three principal component is 85% or so,
Substantially, so we take first three principal component to be analyzed, can so it be achieved that polynary with the abnormal conditions of reflected sample
The dimension-reduction treatment (dimension has been reduced to three-dimensional from the four-dimension) of qualitative data, reduces analysis difficulty.Three principal components are made respectively
Monodrome control figure, as shown in Fig. 8, Fig. 9, Figure 10.
The 37th sample occurs different at Second principal component, it can be seen from three above principal component component monodrome control figure
Often, the calculating formula of Second principal component, is:PC2=0.992x1+0.01x2+0.074x3-0.1x4, wherein x1,x2,x3,x4Table respectively
Show the diameter in hole, depth, distance 1,2 four variate-values of distance.The diameter in hole is it can be seen from Second principal component, expression formula
The ratio accounted in two principal components is maximum, is the principal element that this principal component is deteriorated, and the influence of distance 2 is taken second place.In diagnosis hole milling work
During sequence error source, it should consider the diameter factor in hole first.The exception for tentatively judging the process is existed by the diameter variable in hole
The exception of sample point 37 is caused.Make the monodrome control figure of 4 variables as shown in Figure 11, Figure 12, Figure 13, Figure 14, diameter variable
Control figure there is exception in sample point 37, beyond upper control limit.The diagnostic result obtained by principal component analysis is carried out with us
With uniformity.
The inventive method can be used for digitlization workshop machine to add the abnormal alarm of product quality problem and process abnormal cause to examine
It is disconnected, the quality problems in workshop can be found in time, are prevented the quality problems of the last process in workshop from flowing into next procedure, are eliminated car
Between potential cost allowance, had broad application prospects in the actual production in workshop.
The requirement of the mass property of table 1
The actual measurement data of table 2
The hole milling process sample data principal component analysis result of table 3
It is last it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention and it is unrestricted, although Shen
Ask someone that the present invention is described in detail with reference to preferred embodiment, it will be understood by those within the art that, those are right
Technical scheme is modified or equivalent substitution, all should without departing from the objective and scope of technical solution of the present invention
Cover among scope of the presently claimed invention.
Claims (3)
1. a kind of product multivariate quality monitoring method towards digitlization workshop, it is characterised in that comprise the following steps：
Step 1：Pass through the acquisition scheme of internal lan structure digitlization workshop qualitative data between machine extra bus；
Step 2：Qualitative data is obtained by the Critical to quality for gathering critical process, to the Multivariate Quality Characteristics of critical process
Complex process capability analysis is carried out, when complex process capability analysis result meets customer's requirement or quality management system
It is required that when, it is transferred to according to Critical to quality to determine the process of the type of sampling plan and control figure；When complex process ability
When analysis result is unsatisfactory for the requirement of customer's requirement or quality management system, quality problems need to be analyzed, found
The measure that solves the problems, such as simultaneously is corrected, until reaching quality level；
Step 3：It is that variable control figure is preferential that control figure, which chooses rule, does not find suitable variable control figure really
In the case of, consider further that and reckon by the piece or enumeration control figure；Qualitative data according to collecting is handled sample data, is calculated
Corresponding specification limit, and draw corresponding initial control figure and process sentence surely；
Step 4：When control figure is not at controlled state, searching system is different because if it is different because supplement sampling is gone forward side by side not find system
The drafting of row control figure；If it is different because eliminating different because after, control figure being repainted according to sampling plan to find system；Work as control figure
During in controlled state, the monitoring of production status is carried out；
Step 5：When monitoring process occurs abnormal, abnormal Producing reason is found, if finding abnormal different because whether see needs to repair
Change control figure, if desired control figure is modified, sampling prescription should be redefined, and carry out the drafting of control figure, otherwise make
The monitoring of production status is carried out with control figure before, if monitoring process does not occur exception, process finishing, is transferred to next
The process control of sample；
Complex process capability analysis method described in step 2 is specially：
Variable X represents the Critical to quality of critical process, if F1Represent the master that first linear combination of original variable is formed
Composition component, i.e. F1=a11X1+a21X2+...+ap1Xp,ai=(a1i,a2i,...,api)TIt is the i-th of the covariance matrix Σ of variable X
Individual characteristic root λiCharacteristic vector, it is assumed that variable X obey be p dimension normal distribution, meet workshop data collection actual conditions,
First principal component includes original variable information size and can measured with its variance size, its variance yields Var (F1) bigger, show
The information that first principal component component includes original variable is more, it is desirable to first principal component component F1Include original variable information content
Maximum, therefore the F chosen1Should be original variable X1,X2,...,XpAll linear combinations in variance it is maximum, so can be with
Construct variance yields is sequentially reduced the second, the three ..., p-th of principal component, as shown by the following formula：
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>F</mi>
<mn>1</mn>
</msub>
<mo>=</mo>
<msub>
<mi>a</mi>
<mn>11</mn>
</msub>
<msub>
<mi>X</mi>
<mn>1</mn>
</msub>
<mo>+</mo>
<msub>
<mi>a</mi>
<mn>12</mn>
</msub>
<msub>
<mi>X</mi>
<mn>2</mn>
</msub>
<mo>+</mo>
<mn>...</mn>
<mo>+</mo>
<msub>
<mi>a</mi>
<mrow>
<mn>1</mn>
<mi>p</mi>
</mrow>
</msub>
<msub>
<mi>X</mi>
<mi>p</mi>
</msub>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>F</mi>
<mn>2</mn>
</msub>
<mo>=</mo>
<msub>
<mi>a</mi>
<mn>21</mn>
</msub>
<msub>
<mi>X</mi>
<mn>1</mn>
</msub>
<mo>+</mo>
<msub>
<mi>a</mi>
<mn>22</mn>
</msub>
<msub>
<mi>X</mi>
<mn>2</mn>
</msub>
<mo>+</mo>
<mn>...</mn>
<mo>+</mo>
<msub>
<mi>a</mi>
<mrow>
<mn>2</mn>
<mi>p</mi>
</mrow>
</msub>
<msub>
<mi>X</mi>
<mi>p</mi>
</msub>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mo>.</mo>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>F</mi>
<mi>p</mi>
</msub>
<mo>=</mo>
<msub>
<mi>a</mi>
<mrow>
<mi>p</mi>
<mn>1</mn>
</mrow>
</msub>
<msub>
<mi>X</mi>
<mn>1</mn>
</msub>
<mo>+</mo>
<msub>
<mi>a</mi>
<mrow>
<mi>p</mi>
<mn>2</mn>
</mrow>
</msub>
<msub>
<mi>X</mi>
<mn>2</mn>
</msub>
<mo>+</mo>
<mn>...</mn>
<mo>+</mo>
<msub>
<mi>a</mi>
<mrow>
<mi>p</mi>
<mi>p</mi>
</mrow>
</msub>
<msub>
<mi>X</mi>
<mi>p</mi>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
Wherein principal component FiIt is exactly the ith feature root λ with data matrix X covariance matrix ΣiCharacteristic vector ai=(a1i,
a2i,...,api)TFor the linear combination of coefficient, and there are Var (Fi)=λi；
Represent be i-th of principal component contribution rate, Var (Fi) represent principal component FiVariance it is big
It is small,The variance size sum of P principal component before expression；
In actual applications, for p original variable information, it is not necessary to build p principal component component and analyzed, as m≤p, m
When individual principal component component can reflect most information in original variable, it can not continue to build principal component component, current m
The accumulation of individual principal component component and contribution rateOne can consider that this m principal component component reflection during more than 90%
The most information that p original variable includes；
Assuming that variable X obeys p dimension normal distributions, i.e. X~Np(μ, Σ), wherein μ are population mean vector, and Σ is covariance matrix；
Due to Σp×pFor positive definite symmetric matrices, therefore positive definite matrix U be present so that
Wherein, λ1,λ2,...,λpIt is Σ characteristic value, and meets λ1≥λ2≥...≥λp＞ 0, U=(U1,U2,...,Up) it is single
Position orthogonal matrix,
Remember Y=(Y1,Y2,...,Yp)=UTX then Y1,Y2,...,YpX the 1st, 2 ..., p principal components are referred to as,
Because X obeys multivariate normal distributions, according to the linear combination of normal distribution still Normal Distribution, then Y obediences are polynary just
State is distributed；
Again
EY=E (UTX)=UTEX=UTμ
Cov(Z,YT)=Cov (UTX,XTU)=UTCov(X,XT) U=UTΣ U=Λ
The proportion of each principal component component synthesis original variable information after conversion is different, utilizes principal component component processes energy
Power index needs to assign different weight coefficients to each principal component when constructing complex process ability, by the principle of principal component analysis
Understand by the use of principal component component contribution rate as weight coefficient it is rational, therefore multivariable process Capability index may be defined as：
<mrow>
<msub>
<mi>MC</mi>
<mi>p</mi>
</msub>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>p</mi>
</munderover>
<msub>
<mi>r</mi>
<mi>i</mi>
</msub>
<msub>
<mi>C</mi>
<mrow>
<mi>p</mi>
<mo>,</mo>
<msub>
<mi>pc</mi>
<mi>i</mi>
</msub>
</mrow>
</msub>
</mrow>
<mrow>
<msub>
<mi>MC</mi>
<mrow>
<mi>p</mi>
<mi>k</mi>
</mrow>
</msub>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>p</mi>
</munderover>
<msub>
<mi>r</mi>
<mi>i</mi>
</msub>
<msub>
<mi>C</mi>
<mrow>
<mi>p</mi>
<mi>k</mi>
<mo>,</mo>
<msub>
<mi>pc</mi>
<mi>i</mi>
</msub>
</mrow>
</msub>
</mrow>
<mrow>
<msub>
<mi>MC</mi>
<mrow>
<mi>p</mi>
<mi>m</mi>
</mrow>
</msub>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>p</mi>
</munderover>
<msub>
<mi>r</mi>
<mi>i</mi>
</msub>
<msub>
<mi>C</mi>
<mrow>
<mi>p</mi>
<mi>m</mi>
<mo>,</mo>
<msub>
<mi>pc</mi>
<mi>i</mi>
</msub>
</mrow>
</msub>
</mrow>
Wherein ri=ri/ tr (Λ), tr (Λ)=λ1+λ2+...+λp,For the single variable of the i-th principal component
Measure of Process Capability,
If MCP>=0.9973, then the potential ability of process is to meet to require；If process meets MCP≥
0.9973, step 3 is transferred to, otherwise carries out quality improvement, until it is that process capability index meets to require to reach.
A kind of 2. product multivariate quality monitoring method towards digitlization workshop according to claim 1, it is characterised in that
Described in step 1 structure digitlization workshop qualitative data acquisition scheme be specifically：Manual data acquisition mode mainly for
It is some only by dedicated gauge and estimates formula weight tool just obtainable data and the data of some attributes；RFID is mainly
When being worked out for quality planning, the product of RFID label tag and each model in workshop corresponds, and passes through scanner in production scene
Scanning RFID label tag can will examine item to include in statistical process control system corresponding to the product；Digital intelligent amount instrument master
To be connected by wire/wireless mode with computer, can be by measurement data when quality inspection personnel measures to parts characteristic
Automatically transmit to Statistical Quality Control System to carry out the analysis of data.
A kind of 3. product multivariate quality monitoring method towards digitlization workshop according to claim 1, it is characterised in that
There is exception in monitoring process of working as described in step 5, and it is as follows to find abnormal Producing reason key step：
(1) after occurring extremely, principal component analysis calculating is carried out to multivariate quality data sample data；
(2) by obtain accounting for variation influence after principal component analysis 90% or so preceding k principal component component；
(3) control figure, mean chart or the EWMA control figures of k main variables before drawing；
(4) once finding principal component PCiGeneration is abnormal, it is determined that to principal component PCiInfluence maximum original variable Xi, tentative diagnosis
The exception of principal component control figure and the exception of multivariate process quality control figure are due to original variable XiException it is caused；
(5) original variable X is drawniCorresponding monodrome control figure, to original variable XiIt is monitored, judges whether to be due to original
Variable XiThe caused multivariate process quality control figure of exception it is abnormal；
(6) according to original variable XiException take process correction corrective measure, to process carry out loop control improvement, with reality
The sustained improvement of existing quality.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410797712.0A CN104700200B (en) | 2014-12-18 | 2014-12-18 | A kind of product multivariate quality monitoring method towards digitlization workshop |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410797712.0A CN104700200B (en) | 2014-12-18 | 2014-12-18 | A kind of product multivariate quality monitoring method towards digitlization workshop |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104700200A CN104700200A (en) | 2015-06-10 |
| CN104700200B true CN104700200B (en) | 2018-03-16 |
Family
ID=53347293
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410797712.0A Active CN104700200B (en) | 2014-12-18 | 2014-12-18 | A kind of product multivariate quality monitoring method towards digitlization workshop |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104700200B (en) |
Families Citing this family (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106845826B (en) * | 2017-01-18 | 2021-02-02 | 西安交通大学 | PCA-Cpk-based cold continuous rolling production line service quality state evaluation method |
| CN108122293A (en) * | 2017-03-27 | 2018-06-05 | 沈阳工业大学 | A kind of waste product remanufactures process quality control method |
| CN107203198A (en) * | 2017-05-27 | 2017-09-26 | 四川用联信息技术有限公司 | Improved manufacturing process multivariate quality diagnostic classification device |
| CN107346122A (en) * | 2017-05-27 | 2017-11-14 | 四川用联信息技术有限公司 | Improve the manufacturing process multivariate quality diagnostic classification device of fuzzy support vector machine |
| CN107256000A (en) * | 2017-05-27 | 2017-10-17 | 四川用联信息技术有限公司 | Algorithm of the improved Domain Volume than weighing manufacturing process multivariate quality ability |
| CN107256001A (en) * | 2017-05-27 | 2017-10-17 | 四川用联信息技术有限公司 | The improved algorithm for weighing manufacturing process multivariate quality ability |
| CN107357167A (en) * | 2017-05-27 | 2017-11-17 | 四川用联信息技术有限公司 | The feature extraction algorithm of manufacturing process abnormal quality based on principal component analytical method |
| CN107291065A (en) * | 2017-05-27 | 2017-10-24 | 四川用联信息技术有限公司 | The improved manufacturing process multivariate quality diagnostic classification device based on decision tree |
| CN107390667A (en) * | 2017-05-27 | 2017-11-24 | 四川用联信息技术有限公司 | Manufacturing process multivariate quality diagnostic classification device based on decision tree |
| CN107256002A (en) * | 2017-05-27 | 2017-10-17 | 四川用联信息技术有限公司 | The algorithm of new measurement manufacturing process multivariate quality ability |
| CN107256003A (en) * | 2017-05-27 | 2017-10-17 | 四川用联信息技术有限公司 | A kind of manufacturing process multivariate quality diagnostic classification device of fuzzy support vector machine |
| CN108090675A (en) * | 2017-12-18 | 2018-05-29 | 北京工业大学 | The quality control system that a kind of vehicle is manufactured |
| CN108255142B (en) * | 2018-01-19 | 2019-08-20 | 山东大陆计量科技有限公司 | Quality of production control method and device |
| CN110163479A (en) * | 2019-04-18 | 2019-08-23 | 红云红河烟草(集团)有限责任公司 | A kind of tobacco rolls up the method for building up of exceptional quality problem main cause model between hired car |
| CN110427011A (en) * | 2019-07-16 | 2019-11-08 | 珠海达明科技有限公司 | A kind of intelligent monitoring for automation equipment and real-time compensation system and method |
| CN112668015B (en) * | 2019-12-12 | 2022-02-01 | 华控清交信息科技(北京)有限公司 | Data processing method and device and data processing device |
| CN111192454B (en) * | 2020-01-07 | 2021-06-01 | 中山大学 | Traffic abnormity identification method and system based on travel time evolution and storage medium |
| CN111950785B (en) * | 2020-08-05 | 2021-09-28 | 上海微亿智造科技有限公司 | Event prediction and factor recognition dynamic simulation model modeling method and system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101038485A (en) * | 2006-12-22 | 2007-09-19 | 浙江大学 | System and method for detecting date and diagnosing failure of propylene polymerisation production |
| US7337124B2 (en) * | 2001-08-29 | 2008-02-26 | International Business Machines Corporation | Method and system for a quality software management process |
| CN103839128A (en) * | 2012-11-28 | 2014-06-04 | 沈阳铝镁设计研究院有限公司 | Control method for electrolytic aluminum factory production process quality management |
| CN104062968A (en) * | 2014-06-10 | 2014-09-24 | 华东理工大学 | Continuous chemical process fault detection method |
-
2014
- 2014-12-18 CN CN201410797712.0A patent/CN104700200B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7337124B2 (en) * | 2001-08-29 | 2008-02-26 | International Business Machines Corporation | Method and system for a quality software management process |
| CN101038485A (en) * | 2006-12-22 | 2007-09-19 | 浙江大学 | System and method for detecting date and diagnosing failure of propylene polymerisation production |
| CN103839128A (en) * | 2012-11-28 | 2014-06-04 | 沈阳铝镁设计研究院有限公司 | Control method for electrolytic aluminum factory production process quality management |
| CN104062968A (en) * | 2014-06-10 | 2014-09-24 | 华东理工大学 | Continuous chemical process fault detection method |
Non-Patent Citations (2)
| Title |
|---|
| 基于主成分分析和不合格品率的多元过程能力分析;赵凯;《西北工业大学学报》;20111015;第29卷(第5期);2-5 * |
| 复杂机械产品装配过程在线质量控制方法研究;赵志彪;《中国博士学位论文全文数据库 工程科技Ⅰ辑》;20140415(第4期);61-74 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104700200A (en) | 2015-06-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104700200B (en) | A kind of product multivariate quality monitoring method towards digitlization workshop | |
| CN104808587B (en) | A kind of mobility statistical method based on machining apparatus running status | |
| Woodall et al. | Using control charts to monitor process and product quality profiles | |
| CN109978403A (en) | A kind of quality management-control method, device and the equipment of Product Assembly process | |
| CN103423189B (en) | A kind of server fan power consumption measurement method | |
| WO2016004749A1 (en) | Method for recognizing tool abrasion degree of large numerical control milling machine | |
| CN107067100A (en) | Wind power anomalous data identification method and device for identifying | |
| Li et al. | Application of Big Data Technology in Evaluation of Operating State of High-Pressure Hydrogenation Heat Exchanger | |
| CN107655692A (en) | Engine pedestal automatic calibration method, device and system | |
| CN109840671A (en) | Operational development effect calculates equipment, operational development effect calculation method and recording medium | |
| CN102848266A (en) | Machine tool spindle accuracy prediction method | |
| CN115438726A (en) | Device life and fault type prediction method and system based on digital twin technology | |
| CN109901476A (en) | A kind of unusual service condition monitoring method and system based on multivariate data section similitude | |
| Liu et al. | Thermal error modeling for machine tools: Mechanistic analysis and solution for the pseudocorrelation of temperature-sensitive points | |
| Niu et al. | Application of AHP and EIE in reliability analysis of complex production lines systems | |
| CN104268350A (en) | Closed-loop quality control simulation method with simulation prediction and actual production integrated | |
| CN113128047A (en) | Aero-engine thrust calculation method, storage medium and calculation device | |
| CN109948746A (en) | A kind of sensor-based steel billet information tracing method | |
| WO2023241327A1 (en) | Power grid anomaly locating method based on maximum eigenvector | |
| CN116339260A (en) | Production time adjustment analysis method and system for flexible manufacturing of black lamp factory | |
| CN111443602A (en) | Hidden Markov-based valve health degree estimation and life prediction method | |
| CN111367255A (en) | Performance evaluation test system and method for multi-variable control system | |
| CN106651056B (en) | Energy prediction system based on equipment working condition combination | |
| Tsmots et al. | Selection of hardware features and synthesis of multilevel control system components of energy efficiency of production processes | |
| CN105701716A (en) | Energy efficiency analysis method based on user power consumption data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |