CN103324147A - Cigarette quality evaluation method and system based on principal component analysis - Google Patents

Abstract

The invention discloses a cigarette quality evaluation method and system based on principal component analysis. The cigarette quality evaluation method and system solves the problem that an existing single-variable evaluation method cannot carry out procedure quality evaluation on multiple-variable procedures. The cigarette quality evaluation method and system uses a principal component analysis method to evaluate processing quality of the multiple-variable procedures and can provide basis for comprehensively grasping procedure quality processing states and procedure quality improvement. The method comprises the steps of firstly, calculating process capability indexes of all processing parameters, then using the principal component analyzing method to analyze all processing parameter data, confirming the number of principal components according to the principle that the accumulative total contribution rate is larger than or equal to 90%, confirming weights of all components according to the explanation variances of all principal components, and finally using an arithmetic weighting method to obtain procedure comprehensive process capability indexes. The cigarette quality evaluation method and system can provide more comprehensive and objective results and provides an effectively tool for an enterprise to know procedure total processing quality situations.

Description

Cigarette working procedure quality evaluation method and system thereof based on principal component analysis (PCA)

Technical field

The invention discloses a kind of cigarette working procedure quality evaluation method and system thereof based on principal component analysis (PCA), belong to tobacco working procedure quality technical field.

Background technology

Chinese patent discloses one " based on cigarette working procedure quality overall evaluation system and the method thereof of grey correlation analysis " (patent No. 200810237698.3), the technical essential of above-mentioned patent is: described system and method is the combination of single working procedure quality and Grey Incidence Analysis, at first calculate the process capability index of each technological parameter, then with grey correlation methods the Cpk value is carried out comprehensive evaluation, the comprehensive evaluation result that obtains at last is the good and bad rank of the process quality of each evaluation object.By Cpk and the good and bad ranking result of process quality are analyzed.The shortcoming of above-mentioned patent is as follows:

(1) result that only grey correlation analysis drawn of the method has carried out a quality of simply judging the comprehensive process quality of evaluation object, and the poorest operation of evaluation result improved, this evaluation result only is several operations results relatively, some the time can not truly reflect the true crudy level of operation, as at the Measure of Process Capability of several operations all more than or equal to 1.67 o'clock, the real process crudy all is in higher level, and do not have improved meaning for relatively relatively poor operation this moment yet; All less than 0.67 o'clock, the real process crudy all is in reduced levels simultaneously as at the Measure of Process Capability of several operations, this moment for evaluation result relatively preferably operation also must improve.

(2) a certain technological parameter that only provided for some operations in the production of cigarettes process of the method utilizes the method for grey correlation analysis to carry out the result of comprehensive evaluation, such as " comprehensive evaluation of the cigarette primary processing technique " moisture that provides of part for example, and in the technological parameter of each operation of fiber tow production, moisture only is the technological parameter of an outbalance wherein, also should comprise outlet temperature, hot blast temperature, flow etc., difference with concrete operation, the quantity of technological parameter and project be to some extent difference also, various process parameters all has very important decisive role to the quality of Cigarette processing quality, also directly affect the quality of the overall process quality of technology for making tobacco threds, therefore, some is unilateral undoubtedly only to use the quality of the overall process quality of a certain technological parameter evaluation of some operations, therefore can not reflect really the quality of overall process quality comprehensively, not have general representativeness.

Summary of the invention

Technical matters to be solved by this invention provides a kind of cigarette working procedure quality evaluation method and system thereof.

The technical solution adopted for the present invention to solve the technical problems:

Technical scheme one:

A kind of cigarette working procedure quality evaluation method based on principal component analysis (PCA), its step is as follows:

One, calculates the Measure of Process Capability C of single argument operation _{Pk, 1}:

(1) at first when production run is in steady state (SS), gathers the technological parameter data x that each operation on-line instrument instrument detects automatically _i, each production batch gathers N group data, N 〉=50; The data that gather are stored, and reject wrong or special sample in the described technological parameter data, make final data be in state in cont rol;

(2) standard value, specification center M and the specification tolerance T with above-mentioned technological parameter stores;

(3) calculate sample average x and the sample standard deviation σ of described technological parameter by following formula:

$x=\frac{\mathrm{\Σ}{x}_i}{N}---\left(1\right)$

$\mathrm{\σ}=\sqrt{\frac{\mathrm{\Σ}{(x_i-x)}^2}{N-1}}---\left(2\right)$

(4) calculate the Measure of Process Capability C of described single argument operation by following formula _{Pk, 1}:

$C_{\mathrm{pk},1}=(1-K)\frac{\mathrm{<figure-callout\; id="6"\; label="T"\; filenames="HSA00000687020100011.png"\; state="\{\{state\}\}">T</figure-callout>}}{6\mathrm{\&sigma;}}---\left(3\right)$

In the formula, T=|T _U-T _L| (4)

$K=\frac{2(M-x)}{T}---\left(5\right)$

Wherein, x is sample average, and σ is sample standard deviation, T _UBe upper specification limit, T _LBe the specification lower limit, T is the specification tolerance, and M is the specification center;

Two, calculate the Measure of Process Capability of a certain multivariate operation

(1) the major component number of at first definite described multivariate operation and corresponding major component load matrix L _Pc(n * l):

(1) chooses n technological parameter p of described multivariate operation ₁, p ₂P _n, collect m group data for each technological parameter synchronously by on-line automatic acquisition mode, the capable n column matrix of structure m X;

(2) adopt the z-score mode on column direction, matrix X to be carried out standardization and obtain X _StMatrix

Concrete steps are as follows: obtain average and the standard deviation of the every column data of X matrix, and structure n dimension mean vector μ and n dimension standard deviation vector ε, after matrix X each row of data deducted mean vector μ, point obtained X except standard deviation vector ε _StMatrix;

(3) utilize principal component analysis (PCA) to X _StMatrix carries out principal component analysis (PCA), draws X _StThe score matrix T of matrix _Pc(m * l), load matrix L _Pc(the explained variance λ of n * l) and each major component ₁, λ ₂λ _nThen obtain major component number l according to following accumulative total variance contribution ratio formula:

$\frac{\underset{i=1}{\overset{l}{\mathrm{\&Sigma;}}}{\mathrm{\&lambda;}}_i}{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&lambda;}}_i}\&GreaterEqual;90\%---\left(6\right)$

Thereby determine that corresponding major component load matrix is L _Pc(n * l), score matrix is T _Pc(m * l), and the matrix T that counts the score _Pc(the variance vectors σ of the column direction of the i row of m * l) _i

(2) determine the upper specification limit in major component space

The specification lower limit

With the specification center

Concrete steps are as follows:

Described n technological parameter p ₁, p ₂P _nUpper specification limit be

Be limited under the specification

The specification center

With P ^Max, P ^MinAnd P ^StDeduct mean vector μ, after point removes standard deviation vector ε, with major component loading matrix L _Pc(n * l) quadrature gets the upper specification limit in major component space

The specification lower limit

With the specification center $p_{\mathrm{pc}}^{\mathrm{st}}=[{\mathrm{<figure-callout\; id="1"\; label="p\; pc"\; filenames="HSA00000687020100011.png"\; state="\{\{state\}\}">p</figure-callout>}}_{\mathrm{pc}}^1,{\mathrm{<figure-callout\; id="2"\; label="p\; pc"\; filenames="HSA00000687020100011.png"\; state="\{\{state\}\}">p</figure-callout>}}_{\mathrm{pc}}^2...p_{\mathrm{pcl}}^{\mathrm{st}}];$

(3) calculate respectively the Measure of Process Capability C of each major component according to following formula _Pki:

$C_{\mathrm{pki}}=\min (\frac{P_{\mathrm{pci}}^{\max }-{\mathrm{<figure-callout\; id="3"\; label="P\; pci\; st"\; filenames="HSA00000687020100011.png"\; state="\{\{state\}\}">P</figure-callout>}}_{\mathrm{pci}\mathrm{st}}^{}}{{3\mathrm{\&sigma;}}_i},\frac{P_{\mathrm{pci}}^{\mathrm{st}}-{\mathrm{<figure-callout\; id="3"\; label="P\; pci\; min"\; filenames="HSA00000687020100011.png"\; state="\{\{state\}\}">P</figure-callout>}}_{\mathrm{pci}\min }^{}}{{3\mathrm{\&sigma;}}_i})---\left(7\right)$

Wherein,

It is the upper specification limit of i major component;

It is the specification lower limit of i major component;

It is the specification center of i major component;

σ _iBe score matrix T _Pc(the variance vectors of the column direction of the i row of m * l);

I is 1,2 ... l;

(4) calculate the Measure of Process Capability C of each major component _PkiWeight coefficient ω _i:

${\mathrm{\&omega;}}_i=\frac{{\mathrm{\&lambda;}}_i}{\underset{j=1}{\overset{l}{\mathrm{\&Sigma;}}}{\mathrm{\&lambda;}}_j}---\left(8\right)$

In the formula, λ _iIt is the explained variance of i major component;

(5) Measure of Process Capability of the described a certain multivariate operation of calculating

$C_{\mathrm{pk}}^r=\underset{i=1}{\overset{l}{\mathrm{\&Sigma;}}}{\mathrm{\&omega;}}_i{C}_{\mathrm{pki}}---\left(9\right)$

Three, can calculate the Measure of Process Capability of each multivariate operation according to above-mentioned second step

Four, with reference to " quality Specialized Theory and practice " (middle rank; In February, 2008 first published; The China Personnel Press publishes) in the book to the evaluation criterion of Measure of Process Capability, each operation of production of cigarettes is carried out working procedure quality one by one.

Technical scheme two:

Described cigarette working procedure quality evaluation system comprises authority management module, data importing module, data selection and pretreatment module, operation classification selection module, technological standards administration module, working procedure quality module, evaluation result output module, rights database, process parameter value database, operation category database, technological standards database, output device and evaluation result database;

Described authority management module is connected with rights database, described data importing module respectively with the process parameter value database, authority management module connects with data selection and pretreatment module, the operation classification selects module to be connected with data selection and pretreatment module with the operation category database respectively, the technological standards database is connected with the working procedure quality module with the technological standards administration module respectively, data selection and pretreatment module are connected with the working procedure quality module, and the working procedure quality module is successively through the evaluation result output module, output device is connected with the evaluation result database.

Described data importing module will be carried out computing in the described system of the data importing in the process parameter value database according to user's instruction.

Described data selection and pretreatment module are selected the importing data according to user's requirement and are processed, and reject mistake and special sample in the described importing data.

Described operation classification selects module according to the operation that will estimate, and the user selects corresponding operation from the operation category database; Described operation classification selects module also to comprise the selection of evaluation method, and described evaluation method comprises monthly to be estimated and estimate per year.

Described technological standards administration module is specification center M and specification tolerance T according to user's requirement typing or the process parameter value standard that gathers batch products from the technological standards database from the working procedure quality module.

Described working procedure quality module utilize principal component analysis (PCA) to gather the various process parameters Value Data add up, analyze, calculate the Measure of Process Capability of certain operation.

Described evaluation result output module sends to output device with evaluation result, and evaluation result is saved in the evaluation result database.

The principle of working procedure quality method of the present invention is as follows:

The present invention adopts Measure of Process Capability Cpk as the evaluation index of cigarette working procedure quality, mainly be because Measure of Process Capability is to weigh a kind of not only convenient but also handy appraisal tool that process capability satisfies product requirement (process quality), be mainly used at present in the evaluation of single technological parameter.Because cigarette is made each operation and comprised multinomial technological parameter, and exists correlativity in various degree between each technological parameter, adopt the Measure of Process Capability of a certain technological parameter to estimate the process quality situation, can not obtain comprehensive evaluation result.Therefore, when carrying out working procedure quality, need to carry out comprehensive evaluation to the multinomial technological parameter of operation.This method is the characteristics according to each operation of production of cigarettes, adopt principal component analysis (PCA) that the various process parameters data of multivariate operation are carried out dimension-reduction treatment, large little science according to the explained variance of each major component is determined its weight, on average draw at last the combined process Capability index of this operation by arithmetic weight, thereby realize the overall assessment to cigarette working procedure quality.Therefore this evaluation method is compared with other evaluation methods, seem more comprehensive and objective, it can realize the evaluation to each operation procedure quality control level of production of cigarettes, and the actual crudy situation of better understanding each operation of production of cigarettes for enterprise provides an effective instrument.

Principal component analysis (PCA) also claims principal component analysis, is intended to utilize the thought of dimensionality reduction, and many indexs are converted into a few overall target.In statistics, principal component analysis (PCA) (principal components analysis, PCA) is a kind of technology of reduced data collection.It is a linear transformation.This conversion transforms the data in the new coordinate system so that the first variance of any data projection is on first coordinate (being called first principal component), second largest variance on second coordinate (Second principal component), the like.Why the Selective principal component analysis method is estimated cigarette working procedure quality, mainly be because principal component analysis (PCA) can utilize dimensionality reduction technology to replace original a plurality of variable with a few generalized variable, these generalized variables have been concentrated the most information of original variable, when reducing the data set dimension, can keep the feature to the variance contribution maximum of data set; Secondly it can realize the scientific evaluation to objective economic phenomenon by calculating comprehensive major component function score; It more lays particular emphasis on contribute information influence power comprehensive evaluation on using again.

Beneficial effect of the present invention is as follows:

(1) the present invention utilizes the method for principal component analysis (PCA) that the multinomial process parameter value of operation is carried out synthetic study, by the weight that conversion and the computational science of data are determined each major component, then carry out arithmetic weight and on average draw the combined process Capability index of operation, thereby the process quality to cigarette stable state production run is estimated, the evaluation result that the method draws can comparatively comprehensively reflect in the process each main technologic parameters to the impact of the overall crudy of operation, and can embody each main technologic parameters to the difference of the overall crudy influence degree of operation.

(2) the present invention can reflect the actual crudy level of operation comparatively really, enterprise can be according to the result of each working procedure quality, to the relatively poor operation of the overall crudy of operation, can set about from aspects such as people, machine, material, method, rings, take as early as possible pointed measure to improve, promote the continuous lifting of production of cigarettes procedure quality control level.

(3) the present invention is by carrying out comprehensive objective appraisal to the process quality of cigarette stable state production run, help the innovation of enterprise's processing quality management, the process management of enterprise and quality control level, improve the stability of Cigarette processing procedure quality, the stable cigarette finished product quality that improves is had irreplaceable effect.

Description of drawings

Fig. 1 is the theory diagram of system of the present invention;

Fig. 2 is the operating process schematic diagram of the method for the invention.

Embodiment

The production of cigarettes process comprises cut tobacco processed, stem processed, expansion thread processed, sheet and shreds processed, mixes with addition of perfume, filter stick formation, seven main workshop sections of wraparound.Each workshop section comprises again operation separately, each operation comprises again the technological parameter that quantity does not wait, if certain operation includes only a main technologic parameters, this operation then is called as the single argument operation, for example the washing of stem processed workshop section stalk operation includes only main technologic parameters, i.e. a water temperature; If certain operation comprises the main technologic parameters more than 2, this operation is called as the multivariate operation, and for example the loosening and gaining moisture operation of cut tobacco processed workshop section comprises water percentage, temperature, flow, hot blast temperature and 5 main technologic parameters of steam injection amount.

Embodiment 1 (cigarette working procedure quality evaluation method):

Get final product according to technical scheme one operation in the foregoing invention content part.

Working procedure quality example below in conjunction with the loosening and gaining moisture operation in the cut tobacco processed workshop section further specifies this method:

5 main technologic parameters to the loosening and gaining moisture operation respectively gather 50 groups of data such as following table 1.

Table 1 loosening and gaining moisture procedure technology parameter monitoring table

The specification requirement of this operation various process parameters is respectively: water percentage (19.0 ± 1.0) %, temperature (53.0 ± 3.0) ℃, flow (6000.0 ± 100.0) Kg/h, hot blast temperature (70.0 ± 3.0) ℃, steam injection amount (1000.0 ± 50.0) Kg/h.

His-and-hers watches 1 data are carried out described standardization, and the data after the standardization see Table 2, each column mean and standard deviation see Table 3.

Table 2 loosening and gaining moisture procedure technology parameter monitoring result standard data

Average and the standard deviation of each technological parameter data of table 3

Project	Water percentage	Temperature	Flow	Hot blast temperature	The steam injection amount
						Average	19.214	53.192	5994.138	70.216	990.642
Standard deviation	0.353	0.656	41.659	0.285	8.749

Data after the standardization are carried out principal component analysis (PCA), draw loading matrix L _Pc(n * l), score matrix T _Pc(m * l) and each Principal Component Explanation variance λ ₁, λ ₂, λ ₃, λ ₄, λ ₅, result of calculation sees Table respectively 4, table 5, table 6.

Table 4 loading matrix L _Pc(n * l)

0.680	-0.285	0.093	-0.045	0.667
					0.546	0.336	0.374	0.514	-0.431
-0.109	-0.192	0.844	-0.474	-0.121
					-0.320	0.666	0.293	0.178	0.581
-0.354	-0.571	0.231	0.691	0.132

Table 5 score matrix T _Pc(m * l)

The explained variance of each major component of table 6

Project	λ 1	λ 2	λ 3	λ 4	λ 5
						Average	1.343	1.228	1.066	0.761	0.602

According to cumulative proportion in ANOVA

Principle determine that major component number l is 5.

To project to new space after the upper specification limit of water percentage, temperature, flow, hot blast temperature, 5 technological parameters of steam injection amount and specification lower limit and the distribution center's standardization, obtain the variance vectors of major component upper specification limit, major component specification lower limit, specification center, distribution center, column direction, its major component specification lower limit and major component upper specification limit are as follows respectively:

${\mathrm{<figure-callout\; id="1"\; label="P\; pc"\; filenames="HSA00000687020100011.png"\; state="\{\{state\}\}">P</figure-callout>}}_{\mathrm{pc}}^1=0.706;{\mathrm{<figure-callout\; id="2"\; label="P\; pc"\; filenames="HSA00000687020100011.png"\; state="\{\{state\}\}">P</figure-callout>}}_{\mathrm{pc}}^2=-4.752,{\mathrm{<figure-callout\; id="2"\; label="P\; pc"\; filenames="HSA00000687020100011.png"\; state="\{\{state\}\}">P</figure-callout>}}_{\mathrm{pc}}^2=2.639;{\mathrm{<figure-callout\; id="3"\; label="P\; pc"\; filenames="HSA00000687020100011.png"\; state="\{\{state\}\}">P</figure-callout>}}_{\mathrm{pc}}^3=-9.920,{\mathrm{<figure-callout\; id="3"\; label="P\; pc"\; filenames="HSA00000687020100011.png"\; state="\{\{state\}\}">P</figure-callout>}}_{\mathrm{pc}}^3=9.767;{\mathrm{<figure-callout\; id="4"\; label="P\; pc"\; filenames="HSA00000687020100011.png"\; state="\{\{state\}\}">P</figure-callout>}}_{\mathrm{pc}}^4=-5.658,$

-0.716 ,-1.057,0.077 ,-0.445 ,-0.587 the specification center is respectively:; Because raw data is all through standardization, distribution center is 0; The variance vectors of column direction is respectively:

${\mathrm{\&sigma;}}_i=\left(\begin{array}{c}1.159\\ 1.108\\ 1.032\\ 0.873\\ 0.776\end{array}\right).$

Calculate each major component Cpk value, C _Pk1, C _Pk2C _Pk5The result be respectively 0.203,0.794,3.154,2.161,2.455.Calculate each major component weights omega _i, ω ₁, ω ₂... ω ₅The result be respectively 0.269,0.246,0.213,0.152,0.120.Calculate this operation Measure of Process Capability

The result is 1.547.

According to this operation combined process Capability index that calculates, can judge that then the overall operation crudy of this operation belongs to the II level, illustrate that this operation process capability is abundant, the process quality comprehensive condition is good, and the technical management ability is fine, should continue to keep.Other each operation all can be according to the Measure of Process Capability Cpk of the method calculation process, then its process quality comprehensive condition is carried out overall assessment, for process quality comprehensive condition relatively poor operation or parameter, can set about from aspects such as people, machine, material, method, rings, take as early as possible targetedly measure to improve, promote the continuous lifting of manufacture process quality control level.

Embodiment 2 (cigarette working procedure quality evaluation system):

Implement to get final product according to the technical scheme two in the foregoing invention content part.

Operation steps of the present invention is as follows:

(1) user inputs username and password, and authority management module 1 is according to the data search user right in the rights database 8.Satisfy condition, then allow the user to operate use, responsible force users logs off;

(2) user by data importing module 2 with the described system of the data importing in the process parameter value database 9;

(3) user selects module 4 to select the operation classification that will estimate from operation category database 10 by the operation classification, and sends the data to data selection and pretreatment module 3;

(4) data selection and pretreatment module 3 are carried out pre-service according to the operation classification to the data that import described system; And rejecting imports mistake and special sample in the data;

(5) data in the working procedure quality module 6 combined process standard databases 11 calculate the Measure of Process Capability of this operation to pretreated data, then carry out the cigarette working procedure quality evaluation;

(6) evaluation result output module 7 is final appraisal results, sends to output device 12 and is saved in the evaluation result database 13.

The present invention can estimate the process crudy of each operation of production of cigarettes process, and fluctuation that can comparatively scientific comprehensive each operation main technologic parameters of reflection is on the impact of end product quality, and then provides foundation to impact, technology optimization and the production quality control of product quality for clear and definite operation.Simultaneously, by the problem that exists is is effectively rectified and improved, can improve constantly the online processing quality control level of enterprise, thereby improve the processing quality supportability of enterprise, guarantee the stable of product quality.By enforcement of the present invention, enterprise's production of cigarettes process quality control level is improved significantly, the Measure of Process Capability of 46 main technologic parameters of master operation all has raising in various degree, by 26 before implementing 39 of bringing up to after the enforcement, operation combined process Capability index also is significantly increased Measure of Process Capability more than or equal to 1.00 technological parameter.The cigarette working procedure quality evaluation method that the present invention sets up goes for the product of the different production lines of different tobacco productive corporations and different brands, different size, by to same tobacco productive corporation, same production line, same brand, same specification product, evaluation result at different time, different batches product compares, and can be clear and definite operation and provides foundation to impact, working process parameter optimization and the production quality control of product quality.

Claims (8)

1. cigarette working procedure quality evaluation method based on principal component analysis (PCA), its step is as follows:

One, calculates the Measure of Process Capability C of single argument operation _{Pk, 1}:

(1) at first when production run is in steady state (SS), gathers the technological parameter data x that each operation on-line instrument instrument detects automatically _i, each production batch gathers N group data, N 〉=50; The data that gather are stored, and reject wrong or special sample in the described technological parameter data, make final data be in state in cont rol;

(2) standard value, specification center M and the specification tolerance T with above-mentioned technological parameter stores;

(3) calculate sample average x and the sample standard deviation σ of described technological parameter by following formula:

$x=\frac{\mathrm{\&Sigma;}{x}_i}{N}---\left(1\right)$

$\mathrm{\&sigma;}=\sqrt{\frac{\mathrm{\&Sigma;}{(x_i-x)}^2}{N-1}}---\left(2\right)$

(4) calculate described single argument operation Measure of Process Capability C by following formula _{Pk, 1}:

$C_{\mathrm{pk},1}=(1-K)\frac{T}{6\mathrm{\&sigma;}}---\left(3\right)$

In the formula, T=|T _U-T _L| (4)

$K=\frac{2(M-x)}{T}---\left(5\right)$

Wherein, x is sample average, and σ is sample standard deviation, T _UBe upper specification limit, T _LBe the specification lower limit, T is the specification tolerance, and M is the specification center;

Two, calculate the Measure of Process Capability of a certain multivariate operation

(1) the major component number of at first definite described multivariate operation and corresponding major component load matrix L _Pc(n * l):

(1) chooses n technological parameter p of described multivariate operation ₁, p ₂P _n, n 〉=2; Collect m group data, m 〉=50 for each technological parameter synchronously by on-line automatic acquisition mode; The capable n column matrix of structure m X;

(2) adopt the z-score mode on column direction, matrix X to be carried out standardization and obtain X _StMatrix:

Concrete steps are as follows: obtain average and the standard deviation of the every column data of X matrix, and structure n dimension mean vector μ and n dimension standard deviation vector ε, after matrix X each row of data deducted mean vector μ, point obtained X except standard deviation vector ε _StMatrix;

(3) utilize principal component analysis (PCA) to X _StMatrix carries out principal component analysis (PCA), draws X _StThe score matrix T of matrix _Pc(m * l), load matrix L _Pc(the explained variance λ of n * l) and each major component ₁, λ ₂λ _nThen obtain major component number l according to following accumulative total variance contribution ratio formula:

$\frac{\underset{i=1}{\overset{l}{\mathrm{\&Sigma;}}}{\mathrm{\&lambda;}}_i}{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&lambda;}}_i}\&GreaterEqual;90\%---\left(6\right)$

Thereby determine that corresponding major component load matrix is L _Pc(n * l), score matrix is T _Pc(m * l), and the matrix T that counts the score _Pc(the variance vectors σ of the column direction of the i row of m * l) _i

(2) determine the upper specification limit in major component space The specification lower limit

With the specification center

Concrete steps are as follows:

Described n technological parameter p ₁, p ₂P _nUpper specification limit be

Be limited under the specification

The specification center With P ^Max, P ^MinAnd P ^StDeduct mean vector μ, after point removes standard deviation vector ε, with major component loading matrix L _Pc(n * l) quadrature gets the upper specification limit in major component space The specification lower limit

With the specification center

(3) calculate respectively the Measure of Process Capability C of each major component according to following formula _Pki:

$C_{\mathrm{pki}}=\min (\frac{P_{\mathrm{pci}}^{\max }-P_{\mathrm{pci}}^{\mathrm{st}}}{3{\mathrm{\&sigma;}}_i},\frac{P_{\mathrm{pci}}^{\mathrm{st}}-P_{\mathrm{pci}}^{\min }}{3{\mathrm{\&sigma;}}_i})---\left(7\right)$

Wherein,

It is the upper specification limit of i major component;

It is the specification lower limit of i major component;

It is the specification center of i major component;

σ _iBe score matrix T _Pc(the variance vectors of the column direction of the i row of m * l);

I is 1,2 ... l;

(4) calculate the Measure of Process Capability C of each major component _PkiWeight coefficient ω _i:

${\mathrm{\&omega;}}_i=\frac{{\mathrm{\&lambda;}}_i}{\underset{j=1}{\overset{l}{\mathrm{\&Sigma;}}}{\mathrm{\&lambda;}}_j}---\left(8\right)$

In the formula, λ _iIt is the explained variance of i major component;

(5) Measure of Process Capability of the described a certain multivariate operation of calculating

$C_{\mathrm{pk}}^r=\underset{i=1}{\overset{l}{\mathrm{\&Sigma;}}}{\mathrm{\&omega;}}_i{C}_{\mathrm{pki}}---\left(9\right)$

Three, calculate the Measure of Process Capability of each multivariate operation according to above-mentioned second step

Four, with reference to " quality Specialized Theory and practice " (middle rank; In February, 2008 first published; The China Personnel Press publishes) in the book to the evaluation criterion of Measure of Process Capability, each operation of production of cigarettes is carried out working procedure quality one by one.

2. a system that implements the described method of claim 1 is characterized in that comprising authority management module (1), data importing module (2), data selection and pretreatment module (3), the operation classification is selected module (4), technological standards administration module (5), working procedure quality module (6), evaluation result output module (7), rights database (8), process parameter value database (9), operation category database (10), technological standards database (11), output device (12) and evaluation result database (13);

Described authority management module (1) is connected with rights database (8), described data importing module (2) respectively with process parameter value database (9), authority management module (1) connects with data selection and pretreatment module (3), the operation classification selects module (4) to be connected with data selection and pretreatment module (3) with operation category database (10) respectively, technological standards database (11) is connected 6 with technological standards administration module (5) with the working procedure quality module respectively) be connected, data selection and pretreatment module (3) are connected with working procedure quality module (6), and working procedure quality module (6) is successively through evaluation result output module (7), output device (12) is connected 13 with the evaluation result database) connect.

3. described system according to claim 2 is characterized in that described data importing module (2) will carry out computing in the described system of data importing in the process parameter value database (9) according to user's instruction.

4. described system according to claim 3, it is characterized in that described data selection and pretreatment module (3) according to user's requirement to importing data and select and processing, reject mistake and special sample in the described importing data.

5. described system according to claim 4 is characterized in that described operation classification selects module (4) according to the operation that will estimate, and the user selects corresponding operation from operation category database (10); Described operation classification selects module (4) also to comprise the selection of evaluation method, and described evaluation method comprises laterally to be estimated and vertically estimate by the time (year, season, the moon) by the product brand specification.

6. described system according to claim 5 is characterized in that described technological standards administration module (5) is specification center M and specification tolerance T according to user's requirement typing or the various process parameters value standard that gathers batch products from technological standards database (11) from working procedure quality module (6).

7. described system according to claim 6 is characterized in that described working procedure quality module (6) utilizes principal component analysis (PCA) that the various process parameters Value Data that gathers is added up, analyzed, and calculates certain operation combined process Capability index.

8. described system according to claim 7 is characterized in that described evaluation result output module (7) sends to output device (12) with evaluation result, and evaluation result is saved in evaluation result counts in the storehouse (13).

Images