CN106251047A - Consider the relay protection change power fuzzy synthesis method for evaluating state that degree of membership is variable - Google Patents

Abstract

The invention discloses a kind of relay protection change power fuzzy synthesis method for evaluating state considering that degree of membership is variable.The most simple Information Entropy normal weight coefficient of establishment that uses, transition dependence Historical Monitoring data, its linear model can only carry out simple evaluation simultaneously, and scoring item weight is fixed, and the accuracy of evaluation haves much room for improvement.The present invention uses the combination of Information Entropy and Fuzzy AHP often to weigh calculation, is simultaneously introduced variable-weight theory, the amplitude that constructs distinguishing punishment stimulable type variable weight function, establishes combination and becomes power phasor；Theoretical in conjunction with fuzzy discrimination, construct, the variable membership function model of normal distribution variable based on coefficient, and consider the dynamic corrections problem of degree of membership fuzzy discrimination matrix, propose the fuzzy matrix dynamic correcting method that effect becomes apparent from, finally choose weighted average type operator and calculate fuzzy evaluation result.The present invention can react the virtual condition of relay protection device exactly, improves the accuracy for relay protection state evaluation.

Description

Consider the relay protection change power fuzzy synthesis method for evaluating state that degree of membership is variable

Technical field

The present invention relates to relay protection field, a kind of relay protection change power fuzzy synthesis considering that degree of membership is variable Method for evaluating state.

Background technology

Along with the continuous growth of electrical network scale, traditional by equipment stop relay protection periodic inspection pattern premised on labour by Gradually expose some drawback: when Inspection cycle arrives, may be because of scheduled overhaul overlong time, scheduling cannot arrange equipment to have a power failure, Maintenance is caused to delay even to cancel；On the other hand, no matter virtual condition quality all accepts inspection equipment, and maintenance specific aim is the most not By force, blindly maintenance is also possible to bring security risk.And repair based on condition of component is by effectively monitoring and the science of equipment state objectively Evaluate, grasp the operation conditions of equipment timely and accurately, carry out service work on this basis, can precision customized maintenance content, Reasonable arrangement repair schedule, significantly shortens the repair time and improves repair quality.To this end, study relay protection accurate, reliable Method for evaluating state has positive effect.In recent years, Chinese scholars is in succession in relay protection repair based on condition of component and state evaluation Carry out many Beneficials, formulate condition grading standard from early days by expert discussion, to application model identification technology identification Relay protected operation state, then to utilizing fuzzy synthetic appraisement method and variable-weight theory to carry out relay protection state evaluation, state Evaluation theory has obtained increasing concern in relay protection field.

In the actual application of relay protection state evaluation, some power grid enterprises combine self-operating experience, successively promulgate About the correlation technique directive/guide of relay protection repair based on condition of component and state evaluation, and have started to implement.But above-mentioned fire protection technology is still Based on expertise, only relying on linear model and carry out simple evaluation, scoring item weight is fixed, and indices is with static evaluation Being main, lack and dynamically pass judgment on, the accuracy of evaluation haves much room for improvement.For above-mentioned deficiency, existing document has carried out Associated.

Wu Jiang, Cai Zexiang, Hu Chunchao, Cao Jiandong have delivered a kind of relay protection based on Fuzzy normal distribution membership function Unit state evaluation methodology (protecting electrical power system and control, 2012,05:48-52), the method is by membership function and fuzzy evaluation In theoretical introducing relay protection state evaluation, there is certain guidance meaning, but still continue to use the method that weight expert directly determines, objective Property not enough.

Peng Zhifeng, Yang Mingyu delivered a kind of relay protection method for evaluating state based on variable-weight theory (China Power, 2014,04:92-96), the method establishes the basic ideas of variable weight in relay protection state evaluation, but the change introduced is weighed Function is discontinuous, if the suitable satisfied power condition value that becomes must be discussed.

Song outstanding personality, Chen Yuming have delivered a kind of based on the relay protection state fuzzy synthetic appraisement method (electricity becoming weight coefficient Force system protection and control, 2016,03:46-50), the method uses merely Information Entropy to establish normal weight coefficient, and transition relies on history Monitoring Data, it is impossible to effectively utilizing expert has had experience, the change power balance function that it introduces simultaneously is " class " variable weight function, Applicable elements is limited.

Shi Yuankang, Jiang Zhenchao, peace cun has so been delivered and has been ground about relay protection device of intelligent substation state estimation is practical Studying carefully (protecting electrical power system and control, 2016,10:119-125), this researchs and proposes membership function fuzzy discrimination matrix dynamic corrections Method, improves the conventional fuzzy assessment method shortcoming for rapid degradation index reaction sensitivity deficiency, has certain guidance to anticipate Justice, but calculate the most loaded down with trivial details.

Summary of the invention

The technical problem to be solved is the defect overcoming above-mentioned prior art to exist, it is provided that a kind of consideration is subordinate to Spend variable relay protection and become power fuzzy synthesis method for evaluating state, to react the virtual condition of relay protection device exactly, Improving the accuracy for relay protection state evaluation, carrying out repair based on condition of component for relay protection provides decision-making foundation.

To this end, the present invention adopts the following technical scheme that: consider the relay protection change power fuzzy synthesis shape that degree of membership is variable State evaluation methodology, its step is as follows:

1) state quantity data X is obtained；Set up Comment gathers V, and determine span；Choose evaluation index, set up quantity of state Factor of evaluation collection U；

2) state quantity data standardization, utilizes Information Entropy to calculate the normal weight coefficient of Information Entropy；FAHP method is utilized to calculate FAHP method averagely often weight coefficient；Comprehensive Chang Quan is constituted by FAHP method averagely often weight coefficient and the combination of Information Entropy normal weight coefficient Weight coefficient；

3) set up impairment grade model, substitute into state quantity data X and calculate impairment grade；Choose variable weight function, substitute into impairment grade meter Calculate and become weight coefficient, become power effect according to reality and decide whether to adjust variable weight function parameter；If do not adjusted, directly with comprehensive Chang Quan Weight coefficient synthesizes final comprehensive weight coefficient；As adjusted, the parameter after adjustment returns again to recalculate to variable weight function；

4) choose variable membership function, parameter system is set；Calculate fuzzy matrix, such as the ginseng of membership function to be adjusted Number, then be back in variable membership function recalculate, otherwise judge whether dynamically to adjust fuzzy matrix numerical value, as reached to adjust Whole fuzzy matrix trigger condition is then back to recalculate in fuzzy matrix, otherwise enters next step；

5) the final comprehensive weight coefficient described in and fuzzy matrix carry out fuzzy linear transformation and obtain evaluation result.

The present invention is improved based on fuzzy synthesis method for evaluating state existing: the most perfect relay protection shape State assessment indicator system, adds new evaluation index and impairment grade computation model；Propose, normal distribution variable based on coefficient And the variable membership function model that Comment gathers is 4th district, and consider the dynamic corrections problem of degree of membership fuzzy matrix, it is proposed that The fuzzy matrix dynamic correcting method that effect becomes apparent from；The amplitude of constructing is had any different and becomes weight coefficient adjustable punishment excitation Type variable weight function.

Further, step 2) in, it is contemplated that different experts there will be one when protective relaying device is carried out state evaluation Determine difference, the FAHP method weight to different expertsCarry out arithmetic mean, finally give FAHP method averagely often weight coefficient w^**。

Further, step 2) in, the computing formula of comprehensive often weight coefficient is

$w_i^{\′}=\frac{w_i^{**}{w}_i^{*\mathrm{\α}}}{\Σw_i^{**}{w}_i^{*\mathrm{\α}}}$

In formula,The i-th averagely often weight coefficient determined for FAHP method,The normal weight coefficient determined for Information Entropy, w′_iFor comprehensive often weight coefficient, α is for becoming weight coefficient.

Further, step 3) in, for temperature factor of evaluation, use half normal distyribution function degradation model；For it His factor of evaluation, uses trapezoidal point of function degradation model.

Further, step 3) in, variable weight function formula is as follows:

$S_i\left(x\right)=\left\{\begin{array}{cc}ln{(0.7/x)}^{\mathrm{\&beta;}}+1.0& 0.01\&le;x\&le;0.7\\ 1& 0.7<x\&le;0.9\\ ln{(0.2/(1.1-x\left)\right)}^{\mathrm{\&beta;}}+1.0& 0.9<x\&le;1\end{array}\right.$

In formula, β is for becoming weight coefficient, and its function curve is constituted by 3 sections, i.e. weight punishment, weight are not punished and weight excitation； When metrics evaluation value is in numerical range 0.01≤x≤0.7, the heaviest punishment dynamics is imposed for variable weight function, as 0.7 ＜ During x≤0.9, neither punish and do not encourage；When 0.9 ＜ x≤1, use the dynamical encourage of weak intensity.

Further, step 4) in, if the method adjusting dynamically adjustment fuzzy matrix numerical value is as follows: set jth evaluation The quantity of state collection value impairment grade of factor t-Δ t and t is respectively c (t-Δ t), c (t)；Another jth factor of evaluation t Degree of membership result of calculation corresponding to moment quantity of state collection value is r_ij、r_i+1,j, corresponding comment factor is v_i、v_i+1If, full Foot: (t-Δ t)-c (t) >=0.2, to t degree of membership r for c_ij、r_i+1,jIt is modified:

$r_{ij}^{\&prime;}=\frac{r_{ij}}{a},$

r′_i+1,j=min (a × r_i+1,j, 1),

r′_ij、r′_i+1,jFor revised degree of membership, a is regulation coefficient, and a ＞ 1.

The multiple method advantage of set of the present invention, both in view of objective Chang Quan and the combination of subjective Chang Quan, introduces again change Power theory and fuzzy evaluation theory so that the weight coefficient of factor of evaluation can adjust in real time with the change of quantity of state.Become in power It is further contemplated that multiple method, it is proposed that increase and become weight coefficient and two kinds of methods of degree of membership dynamic corrections, and have following improvement and innovation: (1) the most perfect relay protection state evaluation index system, introduces the indexs such as relay protection unbalance voltage, the temperature difference, Propose the impairment grade computation model of the index such as temperature and unbalance voltage；(2) Information Entropy and Fuzzy AHP are used (FAHP) calculation is often weighed in combination, and introduces multidigit expert score in FAHP；(3) introduce variable-weight theory, construct width Degree is had any different and becomes weight coefficient adjustable punishment stimulable type variable weight function；(4) introduce fuzzy evaluation theory, and propose one The membership function fuzzy discrimination matrix dynamic correcting method that effect becomes apparent from, and compare and simple become power and simple degree of membership is repaiied Positive actual effect.

The present invention can relatively accurately react the virtual condition of relay protection device, improves for relay protection state evaluation Accuracy, carry out repair based on condition of component for relay protection and provide decision-making foundation.

Accompanying drawing explanation

Fig. 1 is the flow chart of the present invention.

Detailed description of the invention

Below in conjunction with specification drawings and specific embodiments, the invention will be further described.

The variable relay protection of consideration degree of membership as shown in Figure 1 becomes power fuzzy synthesis method for evaluating state, and its step is such as Under:

Step 1) obtain state quantity data X；Set up Comment gathers V,

V=[v_j]={ v₁,v₂,v₃,…,v_m}

Comment factor v in formula_j, choose " normally ", " attention ", "abnormal" and " severely subnormal " 4 comment variablees, and really Determine span.

Choose evaluation index, set up quantity of state factor of evaluation collection U；

U=[u_j]={ u₁,u₂,u₃,…,u_m}

Factor of evaluation u in formula_jComprise temperature, the temperature difference, insulate, sample, passage, difference stream and unbalance voltage.

Step 2) state quantity data standardization, utilize Information Entropy to calculate the normal weight coefficient of Information Entropy；Utilize FAHP method Calculate FAHP method averagely often weight coefficient；Constituted comprehensive by FAHP method averagely often weight coefficient and the combination of Information Entropy normal weight coefficient Often weight coefficient.

A) the concrete calculating process of the normal weight coefficient of Information Entropy is as follows:

1) data normalization processes, and closes if U is a group data set, comprises n group sampled data, and often group is containing m factor, u_ijFor The numerical value of the jth factor in i-th group of sampled data,

NoteFor the data after standardization,

For forward index,

$u_{ij}^{*}=\frac{u_{ij}-min\left\{u_j\right\}}{max\left\{u_j\right\}-\min \left\{u_j\right\}}---\left(10\right)$

For negative sense index,

$u_{ij}^{*}=\frac{max\left\{u_j\right\}-u_{ij}}{\max \left\{u_j\right\}-min\left\{u_j\right\}}---\left(11\right)$

2) factor weight of jth in i-th group of quantity of state is calculated

$p_{ij}=\frac{u_{ij}^{*}}{\underset{i=1}{\overset{n}{\&Sigma;}}{u}_{ij}^{*}},j=1,2,...,m---\left(12\right)$

3) entropy of jth item factor is calculated

$e_j=-k\underset{i=1}{\overset{n}{\&Sigma;}}{p}_{ij}\&times;\ln p_{ij}---\left(13\right)$

In formula, k=1/ln m,

4) each factor weight coefficient is calculated

${w_j}^{*}=\frac{1-e_j}{m-\underset{j=1}{\overset{m}{\&Sigma;}}{e}_j},j=1,2,\mathrm{...},m---\left(14\right)$

B) calculating of FAHP method averagely often weight coefficient

Fuzzy AHP (FAHP) is a kind of improved H (AHP), introduces Fuzzy Consistent and judges Matrix, compares original analytic hierarchy process (AHP), and the method is easier to whether test and judge matrix meets coherence request.

(1) Judgement Matrix with Fuzzy Consistency is set up

If d₁…d_mFor factor of evaluation, a_ijFor factor of evaluation d_iRelative to factor of evaluation d_jSignificance level fiducial value.

(2) adjusting Judgement Matrix with Fuzzy Consistency makes it meet concordance

(3) weight coefficient is calculated

$w_i^1=\frac{1}{m}-\frac{1}{2b}+\frac{1}{mb}\underset{j=1}{\overset{m}{\&Sigma;}}{a}_{ij},i=1,2,...,m---\left(17\right)$

In formula,For FAHP method weight coefficient, b is for adjusting parameter, b >=(m-1)/2.

Different is there will be when protective relaying device being carried out state evaluation in view of different experts, special to difference herein The FAHP method weight of familyCarry out arithmetic mean, finally give FAHP method averagely often weight coefficient w^**。

C) computing formula of comprehensive often weight coefficient is

$w_i^{\&prime;}=\frac{w_i^{**}{w}_i^{*\mathrm{\&alpha;}}}{\&Sigma;w_i^{**}{w}_i^{*\mathrm{\&alpha;}}}---\left(18\right)$

In formula,The i-th averagely often weight coefficient determined for FAHP method,The normal weight coefficient determined for Information Entropy, w′_iFor comprehensive often weight coefficient, α is for becoming weight coefficient, the present embodiment α=1.

Step 3) set up impairment grade model, substitute into state quantity data X and calculate impairment grade；Choose variable weight function, substitute into deterioration Degree calculates and becomes weight coefficient, becomes power effect according to reality and decides whether to adjust variable weight function parameter；If do not adjusted, directly with comprehensive Often weight coefficient synthesizes final comprehensive weight coefficient；As adjusted, the parameter after adjustment returns again to again count to variable weight function Calculate.

A) for temperature factor of evaluation, half normal distyribution function degradation model is used, i.e.

$c\left(\mathrm{\&mu;}\right)=\left\{\begin{array}{ccc}1& \mathrm{\&mu;}\&le;{\mathrm{\&mu;}}_0& \\ e^{-\frac{{(\mathrm{\&mu;}-{\mathrm{\&mu;}}_0)}^2}{2{\mathrm{\&sigma;}}^2}};& \mathrm{\&sigma;}=\frac{{\mathrm{\&mu;}}_{max}-{\mathrm{\&mu;}}_0}{3},& \mathrm{\&mu;}>{\mathrm{\&mu;}}_0\end{array}\right.---\left(7\right)$

For other factors of evaluation, use trapezoidal point of function degradation model, i.e.

$c\left(\mathrm{\&mu;}\right)=\left\{\begin{array}{cc}1& \mathrm{\&mu;}\&le;{\mathrm{\&mu;}}_0\\ 1-0.99\&times;\frac{(\mathrm{\&mu;}-{\mathrm{\&mu;}}_0)}{{\mathrm{\&mu;}}_0}& \mathrm{\&mu;}>{\mathrm{\&mu;}}_0\end{array}\right.---\left(8\right)$

In two formula above, c is impairment grade, and μ is state magnitude, μ₀For each index border normal range, μ_maxFor respectively Index " severely subnormal " border, μ₁、μ₂It is respectively " attention " and "abnormal" border.Value model in view of variable weight function independent variable Enclosing and ensure enough deterioration precision, the span of above-mentioned impairment grade c is c ∈ [0.01,1].For forward index, such as Device insulation index, inverted rear unified for negative sense index.Degradation curve and the span of different model device there may be Difference, can be adjusted by actual parameter in application.Table 1 gives Certain Evaluation Guideline and span thereof.

Table 1 evaluation index and span thereof

B) variable weight function and change weight coefficient determine

Become and weigh formula:

$w_i\left(x\right)=\frac{w_i^0{S}_i\left(x\right)}{\&Sigma;w_i^0{S}_i\left(x\right)}---\left(19\right)$

S_iX () is variable weight function,For becoming the front weight of power, w_iX () is weight after change power.

The present invention constructs local punishment stimulable type variable weight function formula

$S_i\left(x\right)=\left\{\begin{array}{cc}ln{(0.7/x)}^{\mathrm{\&beta;}}+1.0& 0.01\&le;x\&le;0.7\\ 1& 0.7<x\&le;0.9\\ \ln {(0.2/(1.1-x\left)\right)}^{\mathrm{\&beta;}}+1.0& 0.9<x\&le;1\end{array}\right.---\left(20\right)$

In formula, β is for becoming weight coefficient, and its function curve is constituted by 3 sections, i.e. weight punishment, weight are not punished and weight excitation； When metrics evaluation value is in numerical range 0.01≤x≤0.7, the heaviest punishment dynamics is imposed for variable weight function, as 0.7 ＜ During x≤0.9, neither punish and do not encourage；When 0.9 ＜ x≤1, use the dynamical encourage of weak intensity.

Step 4) choose variable membership function, parameter system is set；Calculate fuzzy matrix, such as membership function to be adjusted Parameter, then be back in variable membership function recalculate, otherwise determine whether to dynamically adjust fuzzy matrix numerical value, as Reach to adjust fuzzy matrix trigger condition to be then back to fuzzy matrix recalculates, otherwise enter next step.

A) variable membership function

The present invention selects modified model normal state trapezoidal membership function.Corresponding comment factor v₁、v₂、v₃、v₄, i.e. " normally ", " note ", the membership function expression formula of "abnormal" and " severely subnormal " 4 comment linguistic variables as follows.

$r_{i1}\left(x\right)=\left\{\begin{array}{ccc}1& & \\ e^{-\frac{{(x-{\mathrm{\&mu;}}_1)}^2}{2{\mathrm{\&sigma;}}_1^2}},& {\mathrm{\&sigma;}}_1=\frac{{\mathrm{\&mu;}}_1-{\mathrm{\&mu;}}_0}{a_1},& {\mathrm{\&mu;}}_0<x\&le;{\mathrm{\&mu;}}_1\\ 0& & x>{\mathrm{\&mu;}}_1\end{array}\right.---\left(21\right)$

$r_{i2}\left(x\right)=\left\{\begin{array}{ccc}{e}^{-\frac{{(x-{\mathrm{\&mu;}}_1)}^2}{2{\mathrm{\&sigma;}}_2^2}},& {\mathrm{\&sigma;}}_2=\frac{{\mathrm{\&mu;}}_1-{\mathrm{\&mu;}}_0}{a_2},& {\mathrm{\&mu;}}_0\&le;x\&le;{\mathrm{\&mu;}}_1\\ e^{-\frac{{(x-{\mathrm{\&mu;}}_1)}^2}{2{\mathrm{\&sigma;}}_3^2}},& {\mathrm{\&sigma;}}_3=\frac{{\mathrm{\&mu;}}_2-{\mathrm{\&mu;}}_1}{a_3},& {\mathrm{\&mu;}}_1\&le;x\&le;{\mathrm{\&mu;}}_2\\ 0& x<{\mathrm{\&mu;}}_0,& x>{\mathrm{\&mu;}}_2\end{array}\right.---\left(22\right)$

$r_{i3}\left(x\right)=\left\{\begin{array}{ccc}{e}^{-\frac{{(x-{\mathrm{\&mu;}}_2)}^2}{2{\mathrm{\&sigma;}}_4^2}},& {\mathrm{\&sigma;}}_4=\frac{{\mathrm{\&mu;}}_2-{\mathrm{\&mu;}}_1}{a_4},& {\mathrm{\&mu;}}_1\&le;x\&le;{\mathrm{\&mu;}}_2\\ e^{-\frac{{(x-{\mathrm{\&mu;}}_2)}^2}{2{\mathrm{\&sigma;}}_5^2}},& {\mathrm{\&sigma;}}_5=\frac{{\mathrm{\&mu;}}_{\max }-{\mathrm{\&mu;}}_2}{a_5},& {\mathrm{\&mu;}}_2\&le;x\&le;{\mathrm{\&mu;}}_{\max }\\ 0& x<{\mathrm{\&mu;}}_1,& x>{\mathrm{\&mu;}}_{\max }\end{array}\right.---\left(23\right)$

$r_{i4}\left(x\right)=\left\{\begin{array}{ccc}{e}^{-\frac{{(x-{\mathrm{\&mu;}}_{\max })}^2}{2{\mathrm{\&sigma;}}_6^2}},& {\mathrm{\&sigma;}}_6=\frac{{\mathrm{\&mu;}}_{\max }-{\mathrm{\&mu;}}_2}{a_6},& {\mathrm{\&mu;}}_2\&le;x\&le;{\mathrm{\&mu;}}_{\max }\\ 1& & x>{\mathrm{\&mu;}}_{\max }\\ 0& & x<{\mathrm{\&mu;}}_2\end{array}\right.---\left(24\right)$

A in formula₁-a₆For membership function adjustable parameter, can adjust according to reality, the normal function of the least correspondence of its value is more Precipitous, it will to shield the information that more degree of membership is low.

B) degree of membership dynamic corrections

If the quantity of state collection value impairment grade of jth factor of evaluation t-Δ t and t is respectively c (t-Δ t), c (t)；The degree of membership result of calculation that separately jth factor of evaluation t quantity of state collection value is corresponding is r_ij、r_i+1,j, corresponding comment Factor is v_i、v_i+1If met: (t-Δ t)-c (t) >=0.2, to t degree of membership r for c_ij、r_i+1,jIt is modified:

$r_{ij}^{\&prime;}=\frac{r_{ij}}{a},$

r′_i+1,j=min (a × r_i+1,j, 1),

r′_ij、r′_i+1,jFor revised degree of membership, a is regulation coefficient, and a ＞ 1.

(1) if r_i+1,j=0, then it is meant that comment factor v of correspondence_iNot changing, degree of membership is still in " normally " State, now need not revise；

(2) if r_i,j=0, then it is meant that comment factor v of correspondence_i+1Come into " severely subnormal " state, r_i+1,j=1, Reach maximum, therefore without correction.

Step 5) described in final comprehensive weight coefficient and fuzzy matrix carry out fuzzy linear transformation and obtain evaluation result.

Application examples

As a example by certain protective relaying device, the variable relay protection change power fuzzy synthesis state that is subordinate to of the present invention is used to comment Valency method carries out state evaluation to it.

Table 2 is 10 groups of quantity of state collection values in a certain monitoring cycle.In addition: respectively obtain two in t-Δ t and t Group quantity of state collection value U₁、U₂,

U₂={ 31 1.1 0.073 0.029 0.000 8.0 2.5}, U₁={ 31 1.1 0.073 0.029 0.000 8.0 1.1} utilize table 2 data and formula (10)-formula (14) to calculate entropy weight:

w^*={ 0.132 0.127 0.187 0.136 0.159 0.151 0.108}

Utilize weight questionnaire table (herein omiting) of 5 experts, and formula (15)-formula (17) calculating FAHP weight be:

w^**={ 0.081 0.098 0.152 0.126 0.164 0.195 0.198}

Utilize formula (18) to calculate often power combining weights to be:

W '={ 0.073 0.085 0.196 0.118 0.179 0.202 0.146}

Utilize U₁、U₂Data set, and formula (7)-formula (8) calculates t and t-Δ t impairment grade:

C (t)={ 0.9982 0.9993 1 0.8349 1 0.934 0.2575}

C (t-Δ t)={ 0.9982 0.9993 1 0.8349 1 0.934 0.9670}

Understand impairment grade to meet:

c₇(t-Δt)-c₇(t) ＞ 0.2

Utilize formula (20) calculate variable weight function:

S (x)={ 1.6753 1.6861 1.6931 1.0000 1.6931 1.1863 2.0001}

Utilize formula (19) to calculate and become weight after power:

W={0.079 0.093 0.214 0.076 0.196 0.154 0.188}

Table 2 data and formula (21)-formula (24) is utilized to calculate fuzzy matrix R:

$R=\left\{\begin{array}{ccccccc}0.9704& 0.9925& 1.0000& 1.0000& 1.0000& 0.3394& 0.0000\\ 0.0879& 0.0666& 0.0000& 0.0000& 0.0000& 0.6187& 0.0000\\ 0.0000& 0.0000& 0.0000& 0.0000& 0.0000& 0.0000& 0.4722\\ 0.0000& 0.0000& 0.0000& 0.0000& 0.0000& 0.0000& 0.4722\end{array}\right\}$

Utilize formula

Carry out fuzzy synthesis state evaluation, obtain evaluation result B:

B={0.711 0.109 0.090 0.090}

Not considering that change is weighed, normally becomes power, increase change weight coefficient, considered degree of membership dynamic corrections and based on assessment guidelines The evaluation result that the 5 kinds of evaluation methodologys that grade obtain lists table 3 in.

Table 3 evaluation result contrasts

According to degree of membership analytical table 3 result it is known that become power method (normal value: 0.711) and use Chang Quanfa (normal Value: 0.723) evaluation result that obtains is consistent with the evaluation result (normal value: 0.883) scored by certain valuation of enterprise directive/guide , it is normal that evaluation result is illustrated as device, but compares assessment guidelines scoring method, becomes power method and Chang Quanfa can be the most timely accurate Really reflect the ANOMALOUS VARIATIONS of device, although suggestion device is generally in normal condition, but occurred that some factor of evaluation is good for The situation that health situation declines.Change power method (exceptional value: 0.090) is sensitiveer than Chang Quanfa (exceptional value: 0.070), particularly increases After big change weight coefficient (exceptional value: 0.117), effect becomes apparent from.And use proposed degree of membership dynamic correcting method (exceptional value: 0.228), effect is the most obvious, and exceptional value is brought up to 0.228 by 0.090, this height that be enough to cause evaluation personnel Note.

Above example is used for illustrating the present invention, but is not limited to the scope of the present invention, without departing substantially from present invention spirit On the basis of modifications or improvements, belong to the scope of protection of present invention.If not specializing, used in embodiment The conventional means that is well known to those skilled in the art of technological means.

Claims (6)

1. considering the relay protection change power fuzzy synthesis method for evaluating state that degree of membership is variable, its step is as follows:

1) state quantity data X is obtained；Set up Comment gathers V, and determine span；Choose evaluation index, set up quantity of state evaluation Set of factors U；

2) state quantity data standardization, utilizes Information Entropy to calculate the normal weight coefficient of Information Entropy；FAHP method is utilized to calculate FAHP Method averagely often weight coefficient；Comprehensive Chang Quanchong system is constituted by FAHP method averagely often weight coefficient and the combination of Information Entropy normal weight coefficient Number；

3) set up impairment grade model, substitute into state quantity data X and calculate impairment grade；Choose variable weight function, substitute into impairment grade and calculate change Weight coefficient, becomes power effect according to reality and decides whether to adjust variable weight function parameter；If do not adjusted, directly with comprehensive Chang Quanchong system Number synthesizes final comprehensive weight coefficient；As adjusted, the parameter after adjustment returns again to recalculate to variable weight function；

4) choose variable membership function, parameter system is set；Substitute into state quantity data X and calculate fuzzy matrix, such as person in servitude to be adjusted The parameter of genus degree function, then be back in variable membership function recalculate, otherwise determine whether to dynamically adjust fuzzy square Battle array numerical value, is then back to recalculate in fuzzy matrix as reached adjustment fuzzy matrix trigger condition, otherwise enters next step；

5) the final comprehensive weight coefficient described in and fuzzy matrix carry out fuzzy linear transformation and obtain evaluation result.

The relay protection change power fuzzy synthesis method for evaluating state that consideration degree of membership the most according to claim 1 is variable, its It is characterised by,

Step 2) in, it is contemplated that different experts there will be different, to difference when protective relaying device is carried out state evaluation The FAHP method weight of expertCarry out arithmetic mean, finally give FAHP method averagely often weight coefficient w^**。

The relay protection change power fuzzy synthesis method for evaluating state that consideration degree of membership the most according to claim 2 is variable, its It is characterised by,

Step 2) in, the computing formula of comprehensive often weight coefficient is

$w_i^{\&prime;}=\frac{w_i^{**}{w}_i^{*\mathrm{\&alpha;}}}{{\mathrm{\&Sigma;w}}_i^{**}{w}_i^{*\mathrm{\&alpha;}}}$

In formula,The i-th averagely often weight coefficient determined for FAHP method,The normal weight coefficient determined for Information Entropy, w '_iFor Comprehensive often weight coefficient, α is for becoming weight coefficient.

The relay protection change power fuzzy synthesis method for evaluating state that consideration degree of membership the most according to claim 1 is variable, its It is characterised by,

Step 3) in, for temperature factor of evaluation, use half normal distyribution function degradation model；For other factors of evaluation, adopt With trapezoidal point of function degradation model.

The relay protection change power fuzzy synthesis method for evaluating state that consideration degree of membership the most according to claim 1 is variable, its It is characterised by,

Step 3) in, variable weight function formula is as follows:

$S_i\left(x\right)=\left\{\begin{array}{cc}ln{(0.7/x)}^{\mathrm{\&beta;}}+1.0& 0.01\&le;x\&le;0.7\\ 1& 0.7<x\&le;0.9\\ \ln {(0.2/(1.1-x\left)\right)}^{\mathrm{\&beta;}}+1.0& 0.9<x\&le;1\end{array}\right.$

In formula, β is for becoming weight coefficient, and its function curve is constituted by 3 sections, i.e. weight punishment, weight are not punished and weight excitation；Work as finger Mark evaluation of estimate when being in numerical range 0.01≤x≤0.7, imposes the heaviest punishment dynamics for variable weight function, when 0.7 ＜ x≤ When 0.9, neither punish and do not encourage；When 0.9 ＜ x≤1, use the dynamical encourage of weak intensity.

The relay protection change power fuzzy synthesis method for evaluating state that consideration degree of membership the most according to claim 1 is variable, its It is characterised by,

Step 4) in, if the method adjusting dynamically adjustment fuzzy matrix numerical value is as follows: set jth factor of evaluation t-Δ t It is respectively c (t-Δ t), c (t) with the quantity of state collection value impairment grade of t；Another jth factor of evaluation t quantity of state gathers The degree of membership result of calculation of value correspondence is r_ij、r_i+1,j, corresponding comment factor is v_i、v_i+1If met: c (t-Δ t)-c (t) >=0.2, to t degree of membership r_ij、r_i+1,jIt is modified:

$r_{ij}^{\&prime;}=\frac{r_{ij}}{a},$

r′_i+1,j=min (a × r_i+1,j, 1),

r′_ij、r_i+1,j' for revised degree of membership, a is regulation coefficient, and a ＞ 1.