CN105468911A - Transformer state evaluation method based on analytic hierarchy process and rough set - Google Patents

Abstract

The invention discloses a transformer state evaluation method based on an analytic hierarchy process and a rough set. The method relates to the technical field of power equipment state evaluation and solves the technical problem of transformer state evaluation. The method comprises: dividing a transformer into two structure levels of components and state parameters, carrying out relative cracking processing on each state parameter to obtain the state grade value of each state parameter; calculating the subject weights and object weights of all the components and state parameters of the transformer by using the analytic hierarchy process and the rough set theory; carrying out combinatorial optimization to the subject weights and object weights of the components and state parameters, wherein the proportions of the components and state parameters in influencing the operation state of the transformer are reflected by optimized weights; and calculating the state scoring value of the transformer according to the optimized weights and the sate grade values, and evaluating the state of the transformer according to the state scoring value. According to the method provided by the invention, the operation state of the transformer is effectively evaluated.

Description

Based on the Transformer condition evaluation of analytical hierarchy process and rough set

Technical field

The present invention relates to evaluating status of electric power technology, particularly relate to a kind of technology of the Transformer condition evaluation based on analytical hierarchy process and rough set.

Background technology

Transformer is responsible for the vital role of electric energy conversion and transmission, is one of most important equipment in electric system.Therefore, important guarantee to power system security, reliability service when transformer normal runs, assessing the health status of transformer is the important evidence of transformer state maintenance, can effectively prevent and reduce the generation of transformer fault and accident.In recent years, along with the development of transformer on-line condition monitoring technology and the further investigation of state estimation theory, transformer health state evaluation obtained the great attention of State Grid Corporation of China.

Method for evaluating state at present for transformer all also exists a lot of defect, effectively cannot assess running state of transformer.Such as: the shortcoming adopting analytical hierarchy process to evaluate transformer state is the subjective experience too relying on expert, and when quantity of state is numerous, is difficult to correct location relative importance therebetween in the mode compared between two, and computation process is also too loaded down with trivial details.Although adopt Gray Correlation to evaluate transformer state can process because of inconsistent the caused weight allocation problem of expert's subjective experience, the resolution ratio of its method determines it is an insoluble problem.Blur method is adopted to evaluate transformer state when factor of evaluation is more, its adaptive faculty is poor, although artificial neural network is when complex data, there is certain adaptive faculty, but need a large amount of data to train its network, often cause net training time long, the situation that speed of convergence does not even restrain slowly occurs.Adopt Fault Tree evaluation transformer state then to depend critically upon authenticity and the completeness of data, the degree by data influence is large, excavates potential information ability weak.The input data of rough set are necessary for discrete data, and the description for problem is too dull, well can not reflect the essence of problem.

Summary of the invention

For the defect existed in above-mentioned prior art, technical matters to be solved by this invention is to provide a kind of Transformer condition evaluation based on analytical hierarchy process and rough set that can carry out Efficient Evaluation to running state of transformer.

In order to solve the problems of the technologies described above, a kind of Transformer condition evaluation based on analytical hierarchy process and rough set provided by the present invention, it is characterized in that, concrete steps are as follows:

1) oil chromatography test, electrical test, Optimum Experiment are carried out to transformer;

2) transformer is divided into parts and quantity of state two layer of structure, each quantity of state is belonged to all parts respectively, and each quantity of state of transformer is carried out relative cracking process, try to achieve the state grade score value of each quantity of state;

3) analytical hierarchy process is utilized to solve all parts of transformer and the subjective weight of each quantity of state, and the state grade score value of each quantity of state according to transformer, utilize rough set theory to solve all parts of transformer and the objective weight of each quantity of state;

4) using all parts of transformer and each quantity of state as optimization element, optimize the subjective weight of element to each, objective weight carries out Combinatorial Optimization, concrete Combinatorial Optimization formula is:

w _i＝μw _ai+(1-μ)w _σi

Wherein, w _ibe i-th optimization weight optimizing element, w _aibe i-th subjective weight optimizing element, w _{σ i}be i-th objective weight optimizing element, μ is coefficient ratio that is main, objective weight, has μ=0.382;

5) for each quantity of state of transformer, ask for the product that it optimizes weight and state grade score value, obtain the score value of this quantity of state, and will the score value of score value sum as these parts of each quantity of state of same parts be belonged to;

6) for each parts of transformer, ask for the product that it optimizes weight and score value, obtain this unit status score value, and using the condition grading value of the state score sum of all parts as transformer;

If the condition grading value of transformer is greater than 80 and is less than or equal to 100, then show that transformer is in normal condition;

If the condition grading value of transformer is greater than 60 and is less than or equal to 80, then show that transformer is on the alert;

If the condition grading value of transformer is greater than 20 and is less than or equal to 60, then show that transformer is in abnormality;

If the condition grading value of transformer is more than or equal to 0 and is less than or equal to 20, then show that transformer is in severe conditions.

Transformer condition evaluation based on analytical hierarchy process and rough set provided by the invention, by the classification to the numerous quantity of state of transformer, for transformer state single evaluation the method deficiency existed and the complicated feature of quantity of state affecting transformer performance, carry out dividing with the hierarchical structure of parts, quantity of state to transformer; Use for reference expertise, try to achieve the subjective weight of each parts, quantity of state with analytical hierarchy process; In conjunction with transformer testing data, Analysis of Rough Set Theory is used to go out the objective weight of each parts, quantity of state; Then master, objective weight being optimized combination, reflecting that transformer all parts, quantity of state are affecting proportion shared in running state of transformer, effectively can assess running state of transformer with optimal combination weight.

Accompanying drawing explanation

Fig. 1 is the estimation flow figure of the Transformer condition evaluation based on analytical hierarchy process and rough set of the embodiment of the present invention.

Embodiment

Illustrate below in conjunction with accompanying drawing and embodiments of the invention are described in further detail; but the present embodiment is not limited to the present invention; every employing analog structure of the present invention and similar change thereof, all should list protection scope of the present invention in, the pause mark in the present invention all represent and relation.

As shown in Figure 1, a kind of Transformer condition evaluation based on analytical hierarchy process and rough set that the embodiment of the present invention provides, it is characterized in that, concrete steps are as follows:

1) oil chromatography test, electrical test, Optimum Experiment are carried out to transformer;

2) transformer is divided into parts and quantity of state two layer of structure, each quantity of state is belonged to all parts respectively, and each quantity of state of transformer is carried out relative cracking process, try to achieve the state grade score value of each quantity of state, the method asking for the state grade score value of each quantity of state is prior art;

If the numerical value of quantity of state is larger, show that transformer state is poorer, then the computing formula of the state grade score value of this quantity of state is:

$m_t=\left\{\begin{array}{cc}1& M_t\&GreaterEqual;M_e\\ \frac{M_t-M_s}{M_e-M_s}& M_s\&le;M_t\&le;M_e\\ 0& M_t<M_s\end{array}\right.$

If the numerical value of quantity of state is less, show that transformer state is poorer, then the computing formula of the state grade score value of this quantity of state is:

$m_t=\left\{\begin{array}{cc}1& M_t<M_e\\ \frac{M_s-M_t}{M_s-M_e}& M_e\&le;M_t\&le;M_s\\ 0& M_t>M_s\end{array}\right.$

Wherein, m _tfor the state grade score value of quantity of state, M _tfor the measured value of quantity of state, M _sfor the initial value of quantity of state, M _efor the warning value of quantity of state, M _sand M _evalue be the ratings preset;

3) analytical hierarchy process is utilized to solve all parts of transformer and the subjective weight of each quantity of state, and the state grade score value of each quantity of state according to transformer, utilize rough set theory to solve all parts of transformer and the objective weight of each quantity of state;

4) using all parts of transformer and each quantity of state as optimization element, the subjective weight of element is optimized to each, objective weight carries out Combinatorial Optimization, reflect that transformer all parts, each quantity of state are affecting proportion shared in running state of transformer to optimize weight, concrete Combinatorial Optimization formula is:

w _i＝μw _ai+(1-μ)w _σi

Wherein, w _ibe i-th optimization weight optimizing element, w _aibe i-th subjective weight optimizing element, w _{σ i}be i-th objective weight optimizing element, μ is coefficient ratio that is main, objective weight, has μ=0.382;

5) for each quantity of state of transformer, ask for the product that it optimizes weight and state grade score value, obtain the score value of this quantity of state, and will the score value of score value sum as these parts of each quantity of state of same parts be belonged to;

6) for each parts of transformer, ask for the product that it optimizes weight and score value, obtain this unit status score value, and using the condition grading value of the state score sum of all parts as transformer;

If the condition grading value of transformer is greater than 80 and is less than or equal to 100, then show that transformer is in normal condition;

If the condition grading value of transformer is greater than 60 and is less than or equal to 80, then show that transformer is on the alert;

If the condition grading value of transformer is greater than 20 and is less than or equal to 60, then show that transformer is in abnormality;

If the condition grading value of transformer is more than or equal to 0 and is less than or equal to 20, then show that transformer is in severe conditions.

In the embodiment of the present invention, analytical hierarchy process is utilized to solve the method for weight for prior art, power transformer inner structure is extremely complicated, and it is of a great variety to affect the factor that transformer normally runs, to carry out comprehensively the running status of transformer, objective appraisal, select the state index that normally runs of reflection transformer to be primary factor to Transformer State Assessment accurately and efficiently, in current transformer performance test, main with oil dissolved gas, electrical test, insulating oil is tested, operating maintenance, transformer accessory operating condition is as the Main Basis of assessment transformer state, common transformer fault has short trouble, body fault, fault unshakable in one's determination, shunting switch fault, sleeve pipe fault, annex and other faults, when these faults produce, have numerous relevant transformer state index can change simultaneously, the kind that the present embodiment breaks down in conjunction with expertise and transformer, transformer is divided into body, sleeve pipe, shunting switch, cooling system, insulating oil, fuel tank, non-ionizing energy loss system 7 parts,

The correlation behavior amount of body comprises: furfural content, winding temperature, winding insulation resistance in total hydrocarbon and hydrogen, winding resistance, core inductance resistance, winding frequency spectrum, acetylene, winding dielectric loss, winding Leakage Current, the paper insulation degree of polymerization, oil;

The correlation behavior amount of sleeve pipe comprises: sleeve pipe dielectric loss, sleeve pipe oil level, capacitance of bushing amount, sleeve pipe anti-soil level, sleeve pipe insulation resistance, oil level, connector temperature, sealing;

The correlation behavior amount of shunting switch comprises: load ratio bridging switch, tap change operation;

The correlation behavior amount of cooling system comprises: Cooler Power switching, the filth of oil pump temperature, cooling system ruuning situation, fan ruuning situation, refrigeratory;

The correlation behavior amount of insulating oil comprises: insulating oil dielectric loss, oil breakdown voltage, oily micro-water, air content, oily short form test;

The correlation behavior amount of fuel tank comprises: strong oil cooler sealing, top-oil temperature, fuel tank temperature;

The correlation behavior amount of non-ionizing energy loss system comprises: pressure relief valve, grave gas signal, thermometer letter signal, light Gas Signal;

The weight of each quantity of state of transformer determines it is the key of accurate evaluation transformer state, analytical hierarchy process resolves into each compositing factor challenge, again these factors are formed recursive hierarchy structure by dominance relation grouping, use analytical hierarchy process effectively can carry out Classification Management to each quantity of state of transformer;

Analytical hierarchy process resolves into each compositing factor challenge, and again these factors are formed recursive hierarchy structure by dominance relation grouping, the step of the subjective weight of each parts of application level analytic approach determination transformer and each quantity of state is as follows:

3.1.1) using each for transformer parts as calculating element;

3.1.2) element is calculated on the impact of transformer operating condition according to each, element is calculated to each and carries out Mode of Level Simple Sequence by significance level, and the significance level of element is calculated according to each, element is calculated to each and adopts the mode compared between two to construct a weight judgment matrix A;

3.1.3) check the consistance of weight judgment matrix A, inspection formula is:

$CI=\frac{{\mathrm{\&lambda;}}_{max}-n}{n-1}$

AW＝λ _maxW

$CR=\frac{CI}{RI}$

Wherein, CR is the consistency ration of matrix A, λ _maxfor the eigenvalue of maximum of matrix A, W is λ _maxcorresponding proper vector, n is the exponent number of matrix A, and RI is the Aver-age Random Consistency Index of matrix A;

If CR<0.1, then trip current A is by consistency check, then go to step 3.1.4), otherwise then return step 3.1.2) re-construct this matrix;

3.1.4) the eigenvalue of maximum λ of compute matrix A _maxand characteristic of correspondence vector W, and tried to achieve characteristic vector W is made normalized, obtain the subjective weight that each calculates element;

3.1.5) using each for transformer quantity of state as calculating element, and repeat step 3.1.2) to step 3.1.4).

In the embodiment of the present invention, utilize rough set theory to solve the method for weight for prior art, utilize rough set theory to solve the step of all parts of transformer and the objective weight of each quantity of state as follows:

3.2.1) using each for transformer parts as calculating element;

3.2.1) by each calculating element composition conditional attribute collection C, transformer state evaluation is as decision kind set D, and the typical fault set choosing transformer is combined into domain U, attribute codomain integrates as V={1, and 2,3, in 4}, V, each element respectively is normal grade, warning level, abnormal level, serious level;

3.2.2) domain U is divided into different equivalence classes;

3.2.3) the positive territory POS in domain U is solved _p(Q);

Wherein, P and Q is the relation of equivalence in U, and the positive territory of Q and P is designated as POS _p(Q);

3.2.4) the different dependency degree γ calculated between element is solved _p(Q);

Wherein, dependency degree γ _p(Q)=| POS _p(Q) |/| U|, and 0≤γ _p(Q)≤1;

3.2.5) the importance degree σ calculating element is solved _cD(a) and attribute weight w _{σ i}(objective weight);

Wherein, importance degree σ _pQ(a)=γ _p(Q)-γ _{p-P '}(Q), γ _{p-P '}(Q) for deducting the dependency degree of P ' attribute in P;

Wherein, attribute weight $w_{\mathrm{\&sigma;}i}={\mathrm{\&sigma;}}_{PQ}\left(P_i^{\&prime;}\right)/\underset{i=1}{\overset{m}{\&Sigma;}}{\mathrm{\&sigma;}}_{PQ}\left(P_i^{\&prime;}\right),$ P ' is the attribute in P;

3.2.6) using each for transformer quantity of state as calculating element, and repeat step 3.2.2) to step 3.2.5).

Claims (1)

1., based on a Transformer condition evaluation for analytical hierarchy process and rough set, it is characterized in that, concrete steps are as follows:

1) oil chromatography test, electrical test, Optimum Experiment are carried out to transformer;

2) transformer is divided into parts and quantity of state two layer of structure, each quantity of state is belonged to all parts respectively, and each quantity of state of transformer is carried out relative cracking process, try to achieve the state grade score value of each quantity of state;

3) analytical hierarchy process is utilized to solve all parts of transformer and the subjective weight of each quantity of state, and the state grade score value of each quantity of state according to transformer, utilize rough set theory to solve all parts of transformer and the objective weight of each quantity of state;

4) using all parts of transformer and each quantity of state as optimization element, optimize the subjective weight of element to each, objective weight carries out Combinatorial Optimization, concrete Combinatorial Optimization formula is:

w _i＝μw _ai+(1-μ)w _si

Wherein, w _ibe i-th optimization weight optimizing element, w _aibe i-th subjective weight optimizing element, w _sibe i-th objective weight optimizing element, μ is coefficient ratio that is main, objective weight, has μ=0.382;

5) for each quantity of state of transformer, ask for the product that it optimizes weight and state grade score value, obtain the score value of this quantity of state, and will the score value of score value sum as these parts of each quantity of state of same parts be belonged to;

6) for each parts of transformer, ask for the product that it optimizes weight and score value, obtain this unit status score value, and using the condition grading value of the state score sum of all parts as transformer;

If the condition grading value of transformer is greater than 80 and is less than or equal to 100, then show that transformer is in normal condition;

If the condition grading value of transformer is greater than 60 and is less than or equal to 80, then show that transformer is on the alert;

If the condition grading value of transformer is greater than 20 and is less than or equal to 60, then show that transformer is in abnormality;

If the condition grading value of transformer is more than or equal to 0 and is less than or equal to 20, then show that transformer is in severe conditions.