CN106709192A - Power distribution network three-dimensional simulation training credibility evaluation method based on cloud matter-element model - Google Patents

Abstract

The invention relates to a power distribution network three-dimensional simulation training credibility evaluation method based on a cloud matter-element model. A correlation function of a matter-element model is expressed with a normal cloud model with randomness and fuzziness considered, the relation between various simulation training reliability evaluation indexes and the evaluation intervals of various evaluation result levels is quantified with the correlation function, conversion between quantitative calculation and the qualitative concept is achieved, and finally a power distribution network three-dimensional simulation training credibility evaluation result is worked out according to the maximum membership degree principle. Finally, with a power outage cable replacement subsystem as the embodiment, the credibility of the power outage cable replacement subsystem is evaluated, which shows the superiority and feasibility of the cloud matter-element algorithm. The cloud matter-element algorithm with the cloud model and the matter-element model combined is adopted to evaluate the credibility of power distribution network training simulation, wherein the correlation function of the matter-element model is expressed with the normal cloud model, expert scoring uncertainty is considered, and qualitative index quantification is achieved.

Description

A kind of power distribution network three-dimensional artificial training credibility evaluation method based on cloud matter-element model

Technical field

The present invention relates to power distribution network three-dimensional artificial training reliability assessment, more particularly, to a kind of based on cloud matter-element model Power distribution network three-dimensional artificial training credibility evaluation method.

Background technology

Power distribution network Simulated training is a newer field of training on electric power simulation study over the past two years, and power distribution network training is imitated The research of genuine confidence level is at present also also in the starting stage.Power distribution network Simulated training is different from numerical simulation, is not only to see imitative The error of true value and actual value, but also include that animation performance gives people visual experience and the next sense of hearing sense of voice strip for bringing Response speed is influenceed by, running environment, whether the setting of operating procedure standardizes, whether analogue system can be more comprehensive The operation for embodying various equipment etc..These fuzzyyer factors increased the difficulty to the research of its credibility quantification.

The method of reliability assessment has a lot, using the confidence level of theory of similarity research analogue system.Principle is：Will be credible Degree evaluation index is quantized into a similar elements, if evaluation process with ambiguity, it is necessary to fuzzy wheel by reliability assessment Index is converted into Fuzzy similarity unit.But the fuzzy similar participation for needing expert, it is inevitable with very big composition it is subjective because Element, and for power distribution network Simulated training, its corresponding real system evaluation index is too many, causes similar elements too many, To certain difficulty is brought in calculating, the distortion of assessment result is caused.

During reliability assessment, the selection of evaluation index, the division of state grade etc. is all by the special of the field Family's selection, formulates, and different people due to understanding, experience level it is different, assessment result is also different, inevitable to introduce Artificial subjective factor, with very strong uncertainty.Fuzzy wheel can reflect this uncertainty, and the key of fuzzy theory is The calculating of weight and the selection of degree of membership model, analytic hierarchy process (AHP) calculate the weight of evaluation index, excessive to introduce artificial subjective Factor, the correctness of impact evaluation result, and fuzzy comprehensive evaluation method typically separately discuss ambiguity and randomness, actual It is to influence each other to connect each other both upper, it is indivisible.

The content of the invention

In consideration of it, this patent is used first for problem present in power distribution network Simulated training confidence level is based on expert consulting Modified Delphi method set up power distribution network Simulated training credibility index system, establish one have recursive hierarchy structure can Reliability evaluation index system；Secondly power distribution network emulation training confidence level is estimated using cloud matter-element model, by cloud model and Matter-element theory is combined, it is proposed that one kind assesses power distribution network Simulated training confidence level based on cloud matter-element model, has taken into full account and has commented Estimate the randomness and ambiguity of index class boundaries value, qualitativing concept and the quantitative conversion for calculating are realized, finally using embodiment Show that the reliability assessment based on cloud matter-element model compares for grey Cluster Evaluation method, it is more objective rationally.

Above-mentioned technical problem of the invention is mainly what is be addressed by following technical proposals：

A kind of power distribution network three-dimensional artificial training credibility evaluation method based on cloud matter-element model, it is characterised in that including Following steps：

Step 1, the foundation of power distribution network Simulated training credibility index system, including following sub-step：

Step 1.1, selection and determination group member, choose the people of panel of expert 10；

Step 1.2, asks each expert according to knowledge and power distribution network production run simulation training system, and carrying out the first round comments Point, standards of grading are divided into of the utmost importance, critically important, important, general, inessential five grades, by result be entered as 5 points, 4 points, 3 Point, 2 points, 1 point；

Step 1.3, obtains the wheel appraisal result of expert second；Consulting result to the first round carries out statistical analysis treatment, will The average value and distribution situation of evaluation result feed back to every expert, carry out judging again for the second wheel, and ask expert to be given at sentencing Confidence level during each index of breaking；Confidence grade " very high ", " height ", " general ", " low ", " very low ", score value be 5 points, 4 points, 3 Point, 2 points, 1 point；The average value GCR (group confidence rating means) of the Confidence grade that all experts are given During not less than 3, show that group member has reached uniformity higher；

Step 1.4, the consulting result taken turns according to last is deleted and selects index；The first round expert meaning of modified Delphi method consulting Seeing to disperse, and the feedback that follow-up each wheel passes through information, expert opinion can be concentrated；

Step 1.5, for preferably analyze data result, using following parameter：

Parameter one：Desired value, usesRepresent：

Wherein, E_jIt is the score value of index significance level, n_ijIt is expert's number of j grade score values to judge i indexs, M is index Number, thenReflect the P desired value of expert judging；

Parameter two：Standard deviation, uses σ_iRepresent：

Reflect dispersion degree of the brainstrust to i index judge values, σ_i>63 need carry out the second wheel consulting；

Parameter three：Coordination degree, uses coefficient of variation V_iCharacterized with cooperation index W：

V_i=σ_i/E_i (3)

Expert is to i-th coordination degree of index for reflection；

Reflect coordination degree of the expert to whole evaluation index system, wherein

V_iSmaller, W is more big, represents and more coordinate；

Step 2, the calculating based on Normal Cloud reliability assessment index degree of membership；

Step 3, the power distribution network Simulated training confidence level comprehensive assessment based on Normal Cloud matter-element model.

In a kind of above-mentioned power distribution network three-dimensional artificial training reliability assessment based on cloud matter-element model, described step 2 Including following sub-step：

Step 2.1,3 numerical characteristics for calculating Normal Cloud：Desired value E_x, entropy E_nWith super entropy H_e, i.e.,：

Step 2.2, the degree of membership for calculating reliability assessment index：

Y (x)=exp [- (x-E_x)²/(2E_n ²)] (7)。

In a kind of above-mentioned power distribution network three-dimensional artificial training reliability assessment based on cloud matter-element model, described step 3 Comprising following sub-step：

Step 3.1, determines standard cloud：According to assessment needs, s evaluation rank is marked off, had for j-th grade

Wherein, R_0jIt is evaluation rank, c_i(i=1,2....i) is evaluation index, (Ex_i,En_i,He_i) it is R_0jOn index c_iThree numerical characteristics；

Step 3.2, determines matter-element to be assessed：For things q to be assessed, if it can be obtained on index c_iReally it is quantitative Value, then can use common matter-element to represent：

Wherein, q is things to be assessed, x_iThe specific data obtained when being assessment；

If the feature of things to be assessed can only be described using natural language value, available cloud matter-element is represented：

Wherein, (Ex_i,En_i,He_i) be things to be assessed numerical characteristic value；

Step 3.3, determines the degree of association between each evaluation rank of matter-element to be assessed and each index：Based on normal cloud model The calculating of the degree of association is different from the calculation of relationship degree in general Matter Analysis in Matter Element Analysis Method；According to things to be assessed not Same characteristic manner, including three below judges situation：

The degree of association between judgement situation one, matter-element that certainty numerical value is represented and cloud matter-element：The degree of association is defined as one Individual water dust, the problem for solving the degree of association is changed into the degree of certainty for seeking the water dust to this cloud；For Normal Cloud (E_x,E_n,H_e) table For the matter-element for showing, with reference to Normal Cloud generating algorithm, for evaluation index x, being subordinate to for its Normal Cloud is calculated with reference to formula (21) Degree is that the degree of association y of cloud matter-element model is

Judgement situation two, the degree of association between cloud matter-element and cloud matter-element：3En rules based on cloud model, by interval (Ex- 3En, Ex+3En) regard a set as, then and N and M is used in the total part and non-shared part of Normal Cloud 1 and Normal Cloud 2 respectively It is expressed as

Then degree of association y is

The degree of association between judgement situation three, matter-element that interval numerical value is represented and cloud matter-element：Power distribution network Simulated training is credible The class boundaries for spending assessment are considered as a double constraint space [c_min,c_max], then with the meter of the degree of association between above-mentioned cloud and cloud Calculation method is calculated, and calculates the parameter of cloud model with formula (13)-(15)：

He=m (15)

Wherein, m is a constant, can be adjusted according to the uncertain and actual conditions of comment；

Step 3.4 calculates the degree of association between matter-element to be assessed and each evaluation rank

If index c_iWeight be η_i, then the degrees of association of the things q on grade j to be assessed be calculated as

Wherein, y_j(x_i) be matter-element to be assessed the degrees of association of the index i on grade j；

Step 3.5 ranking

IfThen can determine that q belongs to jth₀Individual grade.

Advantages of the present invention：According to the characteristics of power distribution network Simulated training, power distribution network training is set up using modified Delphi method The index system of Simulation Credibility Evaluation, for the uncertain factor of reliability assessment process, set forth herein using cloud model The confidence level of the cloud matter-element algorithm evaluation power distribution network Simulated training being combined with matter-element model, the wherein correlation function of matter-element model Using normal cloud model, it is contemplated that uncertainty during expert estimation, and realize qualitative index quantification.

Brief description of the drawings

Accompanying drawing 1 is the determination flow of modified Delphi method in the present invention.

Accompanying drawing 2 is reliability assessment index of the present invention and consulting result.

Accompanying drawing 3 is index system of credibility evaluation of the present invention.

Accompanying drawing 4 is the standard cloud numerical characteristic of evaluation approach in the embodiment of the present invention.

Accompanying drawing 5 is the marking value that expert changes cables simulation subsystem reliability assessment to having a power failure in the embodiment of the present invention.

Accompanying drawing 6 is the assessment cloud numerical characteristic of each index in the embodiment of the present invention.

Accompanying drawing 7 is each degree of association assessed between cloud and standard cloud in the embodiment of the present invention.

Specific embodiment

Below by embodiment, and with reference to accompanying drawing, technical scheme is described in further detail.

Embodiment：

The step of here is using specific embodiment of the invention, it is as follows：

Evaluation rank is divided into " fabulous " by the first, requirement according to Simulation Credibility Evaluation, " fine ", " good ", " general ", " poor ", " very poor " six grades.The Linguistic Value of evaluation rank is separately converted into cloud model to represent.

2nd, expert group is asked for the emulation of " have a power failure and change cable " task, to the items in index system of credibility evaluation Content is scored, and makes to give a mark 100 points divide according to full marks.It is assumed that the weight of each expert is equal, all experts are commented The average value for dividing is used as index implementation value.

3rd, the score value of expert group is input into following backward cloud generator, obtains each of subsystem confidence level to be assessed The cloud of index is represented.

4th, the numerical characteristic assessed using backward cloud generator reduction expert, cloud is produced secondly by Normal Cloud Generator Drop, the substantially distribution of the assessed value of each index is shown with cloud atlas.

5th, the degree of association and weight between cloud matter-element to be assessed and standard cloud matter-element are sought.

Specific method is comprised the following steps：

Step 1, the foundation of power distribution network Simulated training credibility index system；Specific operating procedure is as follows：

Corresponding method is taken according to different assessment objects and information characteristics.Because power distribution network Simulated training confidence level is commented The majority parameters estimated all are qualitative indexes, and a kind of modified Delphi method (Modified based on expert group is taken herein Delphi, MD) set up index system of credibility evaluation.

1. group member is selected and determined, the people of panel of expert 10 is chosen.

2. designing and consuftng table, carries out first round expert consulting.Please each expert according to the experience and knowledge of oneself, to each Index is judged to the influence degree of simulation Credibility, and standards of grading are divided into " of the utmost importance ", " critically important ", " important ", " one As ", " inessential " five grades, for quantitative evaluation concept, result is entered as 5 points, 4 points, 3 points, 2 points, 1 point.

3. the appraisal result of the first round is analyzed, and carries out the consulting of the second wheel.Consulting result to the first round is counted Analyzing and processing, every expert is fed back to by the average value and distribution situation of evaluation result, carries out judging again for the second wheel, and please be special Family is given at judging confidence level during each index.Confidence level " very high ", " height ", " general ", " low ", " very low ", score value is 5 Point, 4 points, 3 points, 2 points, 1 point.Average value GCR (the group confidence of the Confidence grade that all experts are given Rating means) not less than 3 when, show that group member has reached uniformity higher.

4. the consulting result taken turns according to last is deleted and selects index.In general, the first round of modified Delphi method consulting is special Family's opinion meeting relative distribution, follow-up each wheel passes through the feedback of information, and expert opinion can be concentrated progressively.

Statistical analysis is with reference to several parameters once：

(1) desired value, usesRepresent：

Wherein, E_jIt is the score value of index significance level, n_ijIt is expert's number of j grade score values to judge i indexs, M is index Number, thenReflect the P desired value of expert judging.

(2) standard deviation, uses σ_iRepresent：

Reflect dispersion degree of the brainstrust to i index judge values, σ_i>63 need carry out the second wheel consulting.

(3) coordination degree, uses coefficient of variation V_iCharacterized with cooperation index W：

V_i=σ_i/E_i (19)

Expert is to i-th coordination degree of index for reflection.

Reflect coordination degree of the expert to whole evaluation index system, wherein

V_iSmaller, W is more big, represents and more coordinate.

Step 2, the calculating based on Normal Cloud reliability assessment index degree of membership；Specific operating procedure is as follows：

If the critical interval of reliability assessment each evaluation index state grade is (ab), then Normal Cloud degree of membership u (x) table It is up to formula Normal Cloud expression formula

Y (x)=exp [- (x-E_x)²/(2E_n ²)] (21)

Wherein

Step 3, the power distribution network Simulated training confidence level comprehensive assessment based on Normal Cloud matter-element model, specific operating procedure It is as follows：

The calculation procedure of the power distribution network three-dimensional artificial training reliability assessment based on cloud matter-element model is：

(1) standard cloud is determined

According to assessment needs, s evaluation rank is marked off, had for j-th grade

Wherein, R_0jIt is evaluation rank, c_i(i=1,2...i) is evaluation index, (Ex_i,En_i,He_i) it is R_0jOn index c_i Three numerical characteristics.

(2) matter-element to be assessed is determined

For things q to be assessed, if it can be obtained on index c_iReally quantitative values, then can use common matter-element to represent：

Wherein, q is things to be assessed, x_iThe specific data obtained when being assessment.

If the feature of things to be assessed can only be described using natural language value, available cloud matter-element is represented：

Wherein, (Ex_i,En_i,He_i) be things to be assessed numerical characteristic value.

(3) degree of association between each evaluation rank of matter-element to be assessed and each index is determined

The calculating of the degree of association is different from the association in general Matter Analysis in Matter Element Analysis Method based on normal cloud model Degree is calculated.According to the different characteristic manners of things to be assessed, following three kinds of situation discussion can be divided into.

1) degree of association between certainty numerical value is represented matter-element and cloud matter-element

In this case, the numerical value is regarded as a water dust, the problem for solving the degree of association is changed into and seeks the water dust pair The degree of certainty of this cloud.For Normal Cloud (E_x,E_n,H_e) for the matter-element that represents, with reference to Normal Cloud generating algorithm, for assessment Index x, the degree of membership i.e. degree of association y of cloud matter-element model for calculating its Normal Cloud with reference to formula (21) are

2) degree of association between cloud matter-element and cloud matter-element

With reference to the 3En rules of cloud model, interval (Ex-3En, Ex+3En) is regarded into as a set, then Normal Cloud 1 and just The total part and non-shared part of state cloud 2 are expressed as with N and M respectively

Then degree of association y is

3) degree of association between interval numerical value is represented matter-element and cloud matter-element

The class boundaries of power distribution network Simulated training reliability assessment are considered as a double constraint space [c_min,c_max], then Calculated with the computational methods of the degree of association between above-mentioned cloud and cloud, the ginseng of cloud model is calculated with formula (13)-(15) Number：

He=m (31)

Wherein, m is a constant, can be adjusted according to the uncertain and actual conditions of comment.

4) degree of association between matter-element to be assessed and each evaluation rank is calculated

If index c_iWeight be η_i, then the degrees of association of the things q on grade j to be assessed be calculated as

Wherein, y_j(x_i) be matter-element to be assessed the degrees of association of the index i on grade j.

5) ranking

IfThen can determine that q belongs to jth₀Individual grade.

Specific embodiment described in the present invention is only to the spiritual explanation for example of the present invention.Technology belonging to of the invention The technical staff in field can make various modifications or supplement to described specific embodiment or use similar mode Substitute, but without departing from spirit of the invention or surmount scope defined in appended claims.

Claims (3)

1. a kind of power distribution network three-dimensional artificial training credibility evaluation method based on cloud matter-element model, it is characterised in that including with Lower step：

Step 1, the foundation of power distribution network Simulated training credibility index system, including following sub-step：

Step 1.1, selection and determination group member, choose the people of panel of expert 10；

Step 1.2, asks each expert according to knowledge and power distribution network production run simulation training system, carries out first round scoring, Standards of grading are divided into of the utmost importance, critically important, important, general, inessential five grades, by result be entered as 5 points, 4 points, 3 points, 2 Point, 1 point；

Step 1.3, obtains the wheel appraisal result of expert second；Consulting result to the first round carries out statistical analysis treatment, will judge The average value and distribution situation of result feed back to every expert, carry out judging again for the second wheel, and ask expert to be given at judging each Confidence level during index；Confidence grade " very high ", " height ", " general ", " low ", " very low ", score value be 5 points, 4 points, 3 points, 2 Point, 1 point；The average value GCR (group confidence rating means) of the Confidence grade that all experts are given is not small When 3, show that group member has reached uniformity higher；

Step 1.4, the consulting result taken turns according to last is deleted and selects index；The first round expert opinion meeting of modified Delphi method consulting Dispersion, the feedback that follow-up each wheel passes through information, expert opinion can be concentrated；

Step 1.5, for preferably analyze data result, using following parameter：

Parameter one：Desired value, usesRepresent：

${\stackrel{\‾}{E}}_i=\underset{j=1}{\overset{5}{\mathrm{\Σ}}}{E}_j{n}_{ij}/P,\left(i=1,2,\mathrm{...},M\right)---\left(1\right)$

Wherein, E_jIt is the score value of index significance level, n_ijIt is expert's number of j grade score values to judge i indexs, M is index number, ThenReflect the P desired value of expert judging；

Parameter two：Standard deviation, uses σ_iRepresent：

${\mathrm{\σ}}_i={\[\underset{j=1}{\overset{5}{\mathrm{\Σ}}}{n}_{ij}{(E_j-{\stackrel{\‾}{E}}_j)}^2/(P-1)\]}^{\frac{1}{2}}---\left(2\right)$

Reflect dispersion degree of the brainstrust to i index judge values, σ_i>63 need carry out the second wheel consulting；

Parameter three：Coordination degree, uses coefficient of variation V_iCharacterized with cooperation index W：

V_i=σ_i/E_i (3)

Expert is to i-th coordination degree of index for reflection；

$W=\frac{12}{P^2(M^3-M)}\underset{i=1}{\overset{M}{\Σ}}{({\stackrel{\‾}{E}}_i-\stackrel{\‾}{E})}^2---\left(4\right)$

Reflect coordination degree of the expert to whole evaluation index system, wherein

V_iSmaller, W is more big, represents and more coordinate；

Step 2, the calculating based on Normal Cloud reliability assessment index degree of membership；

Step 3, the power distribution network Simulated training confidence level comprehensive assessment based on Normal Cloud matter-element model.

2. a kind of power distribution network three-dimensional artificial training reliability assessment side based on cloud matter-element model according to claim 1 Method, it is characterised in that described step 2 includes following sub-step：

Step 2.1,3 numerical characteristics for calculating Normal Cloud：Desired value E_x, entropy E_nWith super entropy H_e, i.e.,：

$Ex=\frac{b+a}{2}---\left(5\right)$

$En=\frac{b-a}{6}---\left(6\right)$

Step 2.2, the degree of membership for calculating reliability assessment index：

Y (x)=exp [- (x-E_x)²/(2E_n ²)] (7)。

3. a kind of power distribution network three-dimensional artificial training reliability assessment based on cloud matter-element model according to claim 1, its It is characterised by, described step 3 includes following sub-step：

Step 3.1, determines standard cloud：According to assessment needs, s evaluation rank is marked off, had for j-th grade

$R_{0j}=\left[\begin{array}{ccc}{N}_j,& c_1,& {({\mathrm{Ex}}_1,{\mathrm{En}}_1,{\mathrm{He}}_1)}_j\\ & c_2,& {({\mathrm{Ex}}_2,{\mathrm{En}}_2,{\mathrm{He}}_2)}_j\\ \begin{array}{c}\\ \\ \end{array}& \begin{array}{c}.\\ .\\ .\end{array}& \begin{array}{c}.\\ .\\ .\end{array}\\ & c_n,& {({\mathrm{Ex}}_n,{\mathrm{En}}_n,{\mathrm{He}}_n)}_j\end{array}\right]---\left(8\right)$

Wherein, R_0jIt is evaluation rank, c_i(i=1,2 ... .i) are evaluation index, (Ex_i,En_i,He_i) it is R_0jOn index c_iThree Individual numerical characteristic；

Step 3.2, determines matter-element to be assessed：For things q to be assessed, if it can be obtained on index c_iReally quantitative values, then Can be represented with common matter-element：

$R_0=\left[\begin{array}{ccc}q,& c_1,& x_1\\ & c_2,& x_2\\ \begin{array}{c}\\ \\ \end{array}& \begin{array}{c}.\\ .\\ .\end{array}& \begin{array}{c}.\\ .\\ .\end{array}\\ & c_n,& x_n\end{array}\right]---\left(9\right)$

Wherein, q is things to be assessed, x_iThe specific data obtained when being assessment；

If the feature of things to be assessed can only be described using natural language value, available cloud matter-element is represented：

$R_0=\left[\begin{array}{ccc}q,& c_1,& ({\mathrm{Ex}}_1,{\mathrm{En}}_1,{\mathrm{He}}_1)\\ & c_2,& ({\mathrm{Ex}}_2,{\mathrm{En}}_2,{\mathrm{He}}_2)\\ \begin{array}{c}\\ \\ \end{array}& \begin{array}{c}.\\ .\\ .\end{array}& \begin{array}{c}.\\ .\\ .\end{array}\\ & c_n,& ({\mathrm{Ex}}_n,{\mathrm{En}}_n,{\mathrm{He}}_n)\end{array}\right]---\left(10\right)$

Wherein, (Ex_i,En_i,He_i) be things to be assessed numerical characteristic value；

Step 3.3, determines the degree of association between each evaluation rank of matter-element to be assessed and each index：Matter-element based on normal cloud model The calculating of the degree of association is different from the calculation of relationship degree in general Matter Analysis in analysis method；According to the different tables of things to be assessed Mode is levied, including three below judges situation：

The degree of association between judgement situation one, matter-element that certainty numerical value is represented and cloud matter-element：The degree of association is defined as a cloud Drop, the problem for solving the degree of association is changed into the degree of certainty for seeking the water dust to this cloud；For Normal Cloud (E_x,E_n,H_e) represent For matter-element, with reference to Normal Cloud generating algorithm, for evaluation index x, the degree of membership for calculating its Normal Cloud with reference to formula (21) is i.e. The degree of association y of cloud matter-element model is

$y=\exp \left(-\frac{{\left(x-Ex\right)}^2}{2{\mathrm{En}}^2}\right)---\left(11\right)$

Judgement situation two, the degree of association between cloud matter-element and cloud matter-element：3En rules based on cloud model, by interval (Ex-3En, Ex+ 3En) regard a set as, then the total part and non-shared part of Normal Cloud 1 and Normal Cloud 2 are expressed as with N and M respectively

$N=\left\{({\mathrm{Ex}}^1-3{\mathrm{En}}^1,{\mathrm{Ex}}^1+3{\mathrm{En}}^1)\right\}\∩\left\{\left(\begin{array}{c}{\mathrm{Ex}}^2-3{\mathrm{En}}^2,{\mathrm{Ex}}^2\\ +3{\mathrm{En}}^2\end{array}\right)\right\}$

Then degree of association y is

$y=\frac{\left|N\right|}{\left|M\right|}---\left(12\right)$

The degree of association between judgement situation three, matter-element that interval numerical value is represented and cloud matter-element：Power distribution network Simulated training confidence level is commented The class boundaries estimated are considered as a double constraint space [c_min,c_max], then with the calculating side of the degree of association between above-mentioned cloud and cloud Method is calculated, and calculates the parameter of cloud model with formula (13)-(15)：

$Ex=\frac{c_{\max }+c_{\min }}{2}---\left(13\right)$

$En=\frac{c_{\max }-c_{\min }}{6}---\left(14\right)$

He=m (15)

Wherein, m is a constant, can be adjusted according to the uncertain and actual conditions of comment；

Step 3.4 calculates the degree of association between matter-element to be assessed and each evaluation rank

If index c_iWeight be η_i, then the degrees of association of the things q on grade j to be assessed be calculated as

$y_j\left(q\right)=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{\mathrm{\η}}_i{y}_j\left(x_i\right)---\left(16\right)$

Wherein, y_j(x_i) be matter-element to be assessed the degrees of association of the index i on grade j；

Step 3.5 ranking

IfThen can determine that q belongs to jth₀Individual grade.