CN103886512A - Thermal power unit index evaluation unit based on gray level clustering - Google Patents

Abstract

The invention discloses a thermal power unit index evaluation unit based on gray level clustering, which constructs a relatively impersonal thermal power unit comprehensive evaluation system on the basis of establishing a comprehensive evaluation norm. The economical evaluation can be performed on the thermal power unit by combining four first level economical indexes which are boiler efficiency (%), generation unit thermal consumption (kJ/(kw*h)), station service power consumption rate (%) and coal quality, and coal-related secondary level indexes which are calorific value (MJ/kg), volatiles (%), sulfur content (%), ash content and price (Yuan/ton). The comprehensive economical index can be obtained through pairwise comparison and gray level clustering analysis. The analysis method is simple and easy to do, and reduces subjectivity.

Description

Fired power generating unit economic target evaluation method based on gray scale cluster

Technical field

The present invention relates to fired power generating unit economic target evaluation method, relate in particular to the fired power generating unit economic target evaluation method based on gray scale cluster, belong to machine learning modeling field.

Background technology

Cluster analysis refers to the set of physics or abstract object to be grouped into the analytic process of multiple classes of the similar object composition of serving as reasons.Belonging to a kind of multivariate statistical analysis, is also the important branch of non-supervised recognition.Cluster is a process that things is distinguished and sorted out, and in this process, not with the relevant priori of classification, the inherent similarity between things is unique criterion that generic is divided.The target of cluster analysis is exactly on similar basis, to collect data to classify.As the unsupervised sorting technique of one, it according to certain criterion a sample set divide into several classes who there is no a classification mark, making as far as possible similar sample assemble is a class, and dissimilar sample gathers inhomogeneity, is generally divided into traditional cluster and the large class of fuzzy clustering two.Traditional cluster analysis is a kind of hard division, and object to be identified is strictly divided into a certain class, has either-or character.But under actual conditions, the weight of sample in variety classes considered in fuzzy clustering, more has practical significance.Gray scale cluster belongs to a kind of fuzzy clustering.

The evaluating system of fired power generating unit overall operation situation accurately, scientifically comprehensive evaluation power plant units economy, the fired power generating unit appraisal model and method of formation scientific quantification.This model not only can and then be supervised at electric power enterprise and carry out energy-saving and emission-reduction, optimization operation work good and bad the economy of same type units sequence, can promote the exchange of technology between electric power enterprise and strengthen energy-saving and cost-reducing consciousness simultaneously.

Many experts study power plant's evaluation method, and Chen Jianhong etc. pass judgment on and set up power plant project Objective Comprehensive Evaluation Method model in conjunction with domain expert, and Judgement Method is passed judgment on for fired power generating unit contest.The people such as Dusmanta are the reliability evaluation for power plant for self-supply by fuzzy theory, estimates to carry out power plant's grading evaluation system construction according to operating standard and expertise.But take expert opinion process as main fuzzy overall evaluation, each expert gives a mark with subjectivity, practical operation complexity, therefore need to set up a set of succinct easily unit comprehensive evaluation model row, that reduce artificial subjectivity, for the contest contrast between the different units of different power plant, model also can be used as similar unit appraisal standards simultaneously.

Summary of the invention

Goal of the invention: in order to overcome the deficiencies in the prior art, the invention provides a kind of fired power generating unit economic target evaluation method based on gray scale cluster.

Technical scheme: for solving the problems of the technologies described above, the fired power generating unit economic target evaluation method based on gray scale cluster provided by the invention, comprises the following steps:

Step 1, utilizes tournament method to determine two-level index weight

Tournament method only compares two attributes in n attribute, and sets i and two factors of j are carried out to significance level contrast, has following rule: " very important " is designated as 9; " very important " is designated as 7; " important " is designated as 5; " slightly important " is designated as 3; " equally important " is designated as 1; " slightly less important " is designated as 1/3; " less important " is designated as 1/5; " very less important " is designated as 1/7; " very less important " is designated as 1/9; Carry out significance level comparison between factor between two, can obtain comparator matrix A=[a _ij] _{n × n}, wherein a _ijrepresent the significance level comparison of i item attribute and j item attribute;

Step 2, adopts and long-pending method is calculated weight coefficient

(1) contrast matrix A is by row standardization $\stackrel{\‾}{a_{\mathrm{ij}}}=a_{\mathrm{ij}}/\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{a}_{\mathrm{ij}},i,j=\mathrm{1,2},...,n$

(2) be added by row $\stackrel{\‾}{w_i}=\underset{j=1}{\overset{n}{\mathrm{\Σ}}}\stackrel{\‾}{a_{\mathrm{ij}}}$

(3) after standardization, obtain weight yarn number

Step 3, consistency check, checks the consistance of decision maker to multiattribute assessment, definition coincident indicator I is

I＝(λ _max-n)／(n-1)

${\mathrm{\λ}}_{\max }=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{\left[\mathrm{AW}\right]}_i/{\mathrm{nw}}_i$

The mean value of different n coincident indicator is R, meets Q=I/R<0.1, can assert that the judgement of comparator matrix can be accepted;

Step 4: utilize the comprehensive clustering methodology of gray scale to evaluate

Be provided with n clustering object, m cluster index, s different grey classes, according to i (i=1,2 ..., n) individual object about j (j=1,2 ..., m) the observed reading x of index _ij(i=1,2 ..., n, j=1,2 ..., m) i object is included into k (k ∈ (and 1,2 ..., s) individual grey class, is called gray scale cluster;

The comprehensive sorting procedure of gray scale is as follows

(1) require to divide grey class according to comprehensive evaluation and count s, provide the albefaction weight function of cluster index j about k ash class, adopt triangle Bai Quanhua function

$f_j^k\left(x\right)=\left\{\begin{array}{cc}0& x\&NotElement;[a_{k-1},a_{k+2}]\\ (x-a_{k-1})/({\mathrm{\&lambda;}}_k-a_{k-1})& x\&Element;[a_{k-1},{\mathrm{\&lambda;}}_k]\\ (a_{k+2}-x)/(a_{k+2}-{\mathrm{\&lambda;}}_k)& x\&Element;[{\mathrm{\&lambda;}}_k,a_{k+2}]\end{array}\right.,k=\mathrm{1,2},...,s,j=\mathrm{1,2,3}...m$

Wherein λ _k=(a _k+ a _k+1)/2;

(2) determine the cluster weight coefficient w of each index according to qualitative analysis conclusion _i, j=1,2 ... m;

(3) white function drawing according to step (1) and (2) and weight coefficient and object i are for the sample value x of j index _ij, i=1,2 ... n; J=1,2 ... m calculates the cluster coefficients of object i about k ash class

(4) calculating object i is about the normalization cluster coefficients of grey class k

(5) according to the normalization cluster coefficients vector of clustering object

with the weight vector η of cluster coefficients=(1,2 ... s-1, s) ^tthe comprehensive cluster coefficients ω of calculating object i _i=δ _iη (i=1,2 ... n);

(6) according to the s decile of comprehensive cluster coefficients, when

time, judge that object i belongs to k class.

Beneficial effect: the present invention is based on gray scale cluster and establish on the basis of comprehensive evaluation criterion, built comparatively objectively fired power generating unit System of Comprehensive Evaluation:

1, different from hard clustering, gray scale cluster can meet the be subordinate to situation of object in actual conditions better;

2, every two-level index of ature of coal is not adopted to equal weight, but tournament method calculates weight, consider that power plant cost is subject to the size of each two-level index degree of fluctuation, makes cluster result have more practical meaning in engineering;

3, quantize similar unit economic target, the economic target quality of the similar unit of visual evaluation, also, for machine group performance optimization and status monitoring provide new thinking and method, has certain engineering practical value;

4, be that providing can reference model for power plant's monitoring information system Premium Features module (as boiler, steam turbine, electrical operation optimization).

Accompanying drawing explanation

Fig. 1 is the process flow diagram of embodiments of the invention.

Embodiment

Embodiment:

Take same type units in certain branch office of electricity power group as example, illustrate the method flow process.Whole process mainly contains the nucleus modules such as coal data input pre-service, knowledge base information extraction, ature of coal gray scale cluster module, the typing of SIS one-level economic target, comprehensive economic index classification, working storage and monitoring management.Detailed process is as shown in Figure 1:

1, ature of coal analysis data via serial line interface the input data as ature of coal classification processing module;

2,10 kinds of typical ature of coal samples, as ature of coal knowledge base, fall into 5 types according to calorific value, volatile matter, sulphur content, ash content and price attribute.Calorific value, volatile matter, sulphur content, ash content and 5 two-level index of price are made as respectively x ₁, x ₂, x ₃, x ₄, x ₅, comparator matrix A between two

$A=\left|\begin{array}{ccccc}1& 3& 2& 5& 4\\ 1/3& 1& 1& 3& 3\\ 1/2& 1& 1& 3& 3\\ 1/5& 1/3& 1/3& 1& 2\\ 1/4& 1/3& 1/3& 1/2& 1\end{array}\right|$

Standardize by row according to formula (1) contrast matrix A $\stackrel{\&OverBar;}{a_{\mathrm{ij}}}=a_{\mathrm{ij}}/\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{a}_{\mathrm{ij}},i,j=\mathrm{1,2},...,n$

(2) be added by row $\stackrel{\&OverBar;}{w_i}=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}\stackrel{\&OverBar;}{a_{\mathrm{ij}}}$

(3) after standardization, obtain weight coefficient

5 two-level index weight coefficients that can obtain ature of coal are respectively w ₁=0.4207, w ₂=0.2015, w ₃=0.2161, w ₄=0.0903, w ₅=0.0713.

This weight coefficient is carried out to consistency check,

I＝(λ _max-n)／(n-1)

${\mathrm{\&lambda;}}_{\max }=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\left[\mathrm{AW}\right]}_i/{\mathrm{nw}}_i$

Q=I/R<0.1 meets condition for consistence, therefore can accept.

3, minimum value and the maximal value of selecting every index of this 10 kinds of typical coals, round the rear span as each index, then the span of each index is correspondingly divided into 5 grey classes, a ₁, a ₂..., a ₆value and left and right continuation value a ₀, a ₇form albefaction Effects of Parameters of Weight table.Fire coal is divided into 5 classes and is stored in 5 silos, the span of comprehensive cluster coefficients should be divided into 5 mutually disjoint isometric intervals, i.e. [1,1.8], [1.8,2.6], [2.6,3.4], [3.4,4.2], [4.2,5], with since judge the affiliated grey class of object.Choose minimum value and the maximal value of every index of 10 kinds of coals, round the rear span as each index, can try to achieve the albefaction Effects of Parameters of Weight of each index observed reading according to said method, as shown in table 1.

Table 1 albefaction Effects of Parameters of Weight value

4, input data enter ature of coal classification processing module in conjunction with albefaction Effects of Parameters of Weight table, carry out the gray scale cluster of ature of coal two-level index, and the comprehensive cluster coefficients and the classification that connect 10 kinds of coals that step 2 tries to achieve are as shown in table 2:

Table 2 ature of coal classification results

5, access the economic target from SIS system: boiler efficiency, steam turbine efficiency and station service power consumption rate, in conjunction with the comprehensive cluster index of ature of coal, equal weight enters comprehensive economic index categorizing system.

6, the comprehensive cluster index that step 5 obtains is as the quantitative measurement of similar unit economic target, and storage, the reference for related personnel as evaluation unit.

Claims (1)

1. the fired power generating unit economic target evaluation method based on gray scale cluster, is characterized in that comprising the following steps:

Step 1, utilizes tournament method to determine two-level index weight

Tournament method only compares two attributes in n attribute, and sets i and two factors of j are carried out to significance level contrast, has following rule: " very important " is designated as 9; " very important " is designated as 7; " important " is designated as 5; " slightly important " is designated as 3; " equally important " is designated as 1; " slightly less important " is designated as 1/3; " less important " is designated as 1/5; " very less important " is designated as 1/7; " very less important " is designated as 1/9; Carry out significance level comparison between factor between two, can obtain comparator matrix A=[a _ij] _{n × n}, wherein a _ijrepresent the significance level comparison of i item attribute and j item attribute;

Step 2, adopts and long-pending method is calculated weight coefficient

(1) contrast matrix A is by row standardization $\stackrel{\&OverBar;}{a_{\mathrm{ij}}}=a_{\mathrm{ij}}/\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{a}_{\mathrm{ij}},i,j=\mathrm{1,2},...,n$

(2) be added by row $\stackrel{\&OverBar;}{w_i}=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}\stackrel{\&OverBar;}{a_{\mathrm{ij}}}$

(3) after standardization, obtain weight yarn number

Step 3, consistency check, checks the consistance of decision maker to multiattribute assessment, definition coincident indicator I is

I＝(λ _max-n)／(n-1)

${\mathrm{\&lambda;}}_{\max }=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\left[\mathrm{AW}\right]}_i/{\mathrm{nw}}_i$

The mean value of different n coincident indicator is R, meets Q=I/R<0.1, can assert that the judgement of comparator matrix can be accepted;

Step 4: utilize the comprehensive clustering methodology of gray scale to evaluate

Be provided with n clustering object, m cluster index, s different grey classes, according to i (i=1,2 ..., n) individual object about j (j=1,2 ..., m) the observed reading x of index _ij(i=1,2 ..., n, j=1,2 ..., m) i object is included into k (k ∈ (and 1,2 ..., s) individual grey class, is called gray scale cluster;

The comprehensive sorting procedure of gray scale is as follows

(1) require to divide grey class according to comprehensive evaluation and count s, provide the albefaction weight function of cluster index j about k ash class, adopt triangle Bai Quanhua function

$f_j^k\left(x\right)=\left\{\begin{array}{cc}0& x\&NotElement;[a_{k-1},a_{k+2}]\\ (x-a_{k-1})/({\mathrm{\&lambda;}}_k-a_{k-1})& x\&Element;[a_{k-1},{\mathrm{\&lambda;}}_k]\\ (a_{k+2}-x)/(a_{k+2}-{\mathrm{\&lambda;}}_k)& x\&Element;[{\mathrm{\&lambda;}}_k,a_{k+2}]\end{array}\right.,k=\mathrm{1,2},...,s,j=\mathrm{1,2,3}...m$

Wherein λ _k=(a _k+ a _k+1)/2;

(2) determine the cluster weight coefficient w of each index according to qualitative analysis conclusion _j, j=1,2 ... m;

(3) white function drawing according to step (1) and (2) and weight coefficient and object i are for the sample value x of j index _ij, i=1,2 ... n; J=1,2 ... m calculates the cluster coefficients of object i about k ash class

(4) calculating object i is about the normalization cluster coefficients of grey class k

(5) according to the normalization cluster coefficients vector of clustering object

with the weight vector η of cluster coefficients=(1,2 ... s-1, s) ^tthe comprehensive cluster coefficients ω of calculating object i _i=δ _iη (i=1,2 ... n);

(6) according to the s decile of comprehensive cluster coefficients, when

time, judge that object i belongs to k class.

Images