CN105911407B - Associated transformer state fuzzy set can be opened up to appraisal procedure based on matter-element augmentation - Google Patents
Associated transformer state fuzzy set can be opened up to appraisal procedure based on matter-element augmentation Download PDFInfo
- Publication number
- CN105911407B CN105911407B CN201610211816.8A CN201610211816A CN105911407B CN 105911407 B CN105911407 B CN 105911407B CN 201610211816 A CN201610211816 A CN 201610211816A CN 105911407 B CN105911407 B CN 105911407B
- Authority
- CN
- China
- Prior art keywords
- fault
- transformer
- symptom
- state
- sample
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/50—Testing of electric apparatus, lines, cables or components for short-circuits, continuity, leakage current or incorrect line connections
- G01R31/62—Testing of transformers
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/12—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing
- G01R31/1227—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing of components, parts or materials
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
本发明公开了一种基于物元增广可拓关联的变压器状态模糊集对评估方法，包括：将变压器故障样本集按照既定故障类型分类，通过随机加权法进行分类故障样本集等量增广；基于增广故障样本集构造故障类型‑故障征兆关联辨识矩阵，得到各故障类型的简化关联频繁故障征兆集以及优化的故障征兆常权系数；结合物元可拓理论，构造变压器故障征兆集合数据字典经典域和节域，进而得到各故障征兆的物元关联函数，求取各故障类型对应的变权系数；通过梯岭形模糊隶属度函数计算变压器状态评估所需的同异反评价矩阵，结合优化变权系数得到各故障类型和变压器整体运行状态的集对分析多元联系度，实现对变压器运行状态的分析评估。
The invention discloses a transformer state fuzzy set pair evaluation method based on matter-element augmentation and extension association, which includes: classifying transformer fault sample sets according to predetermined fault types, and performing equal augmentation of classified fault sample sets by a random weighting method; Based on the augmented fault sample set, the fault type-fault symptom association identification matrix is constructed, and the simplified associated frequent fault symptom sets and optimized fault symptom constant weight coefficients of each fault type are obtained; combined with the matter-element extension theory, a transformer fault symptom collection data dictionary is constructed Classical domain and node domain, and then obtain the matter-element correlation function of each fault symptom, and obtain the variable weight coefficient corresponding to each fault type; calculate the similarity, difference and reverse evaluation matrix required for transformer state evaluation through the ladder-shaped fuzzy membership function, combined with The variable weight coefficient is optimized to obtain the set-pair analysis multivariate connection degree of each fault type and the overall operation state of the transformer, so as to realize the analysis and evaluation of the operation state of the transformer.
Description
Technical Field
The invention relates to the technical field of transformer state evaluation, in particular to a transformer state fuzzy set pair evaluation method based on matter element augmentation and extension association.
Background
The working state of a power transformer, especially an oil-immersed power transformer, is an important reference for evaluating the safe reliability of the operation of the power transformer. A power transformer failure may result in a serious power system outage. Therefore, the detection of potential initial faults and the evaluation of working conditions of the power transformer are particularly important, which is an important technical means for preventing and avoiding serious faults of the power transformer at present.
Among the existing transformer fault detection and condition assessment methods, the dissolved gas in oil analysis (DGA) method is commonly employed. The DGA method has the working mechanism that the expression form of the power transformer fault mainly comprises overheating, arc discharge, partial discharge and the like, the specific value information of the characteristic dissolved gas concentration is obtained by analyzing the dissolved gas in the insulating oil of the power transformer, and the phenomenon can be effectively detected and judged. By means of the above analysis of the dissolved gas in the oil, one or more possible fault states of the power transformer can be identified and necessary preventive measures can be taken.
The main disadvantage of the above method is that the establishment of fault diagnosis and condition evaluation criteria requires the accumulation of empirical knowledge. In some cases, the measured dissolved gas concentration or ratio information may not be sufficiently complete and comprehensive to match existing decision criteria, such that some faults that may occur inside the power transformer may not be effectively identified. On the other hand, in engineering practice, different DGA decision methods based on the same set of dissolved gas record samples may lead to different diagnostic and assessment conclusions. Therefore, it is often necessary to comprehensively evaluate the current operating status of the transformer by combining various DGA methods and other related information for diagnosing the transformer, such as the operation history of the transformer, the recent service records, and the on-load tap-changer status. Obviously, the effectiveness and robustness of the above evaluation method cannot be effectively guaranteed.
Generally, when a power transformer fails, not only the state quantity of DGA is changed, but also the state quantities of furfural content, partial discharge data, other transformer insulation parameters and the like in oil are changed, so in order to solve the problems of robustness, adaptability and the like existing in a DGA method, comprehensive analysis needs to be performed on the state quantity change conditions of the transformer to judge the operation state and potential faults of the transformer, and accordingly, an extension analysis method based on state quantity extension evaluation, an association rule algorithm based on comprehensive state quantity correlation analysis, a pair analysis method comprehensively considering system certainty and uncertainty factors and the like are proposed at present to improve the effectiveness and robustness of transformer fault diagnosis and state evaluation.
However, when the weight calculation is performed on the analysis state evaluation method, an expert comprehensive evaluation method and an analytic hierarchy process with strong subjective dependence are generally adopted, so that the support of subjective experience is required, and the robustness is poor; in the existing intelligent algorithm fusing association rules, only simplified support degree and confidence degree criteria are adopted in the process of association rule mining and weight calculation, particularly in weight coefficient calculation based on the confidence degree, although the subjectivity of an expert comprehensive score and an analytic hierarchy process is improved, the inherent correlation among fault symptoms and the complexity of rules are not effectively considered, the influence of sample balance on the transformer state feature association extraction correctness is not considered, the influence of a negative boundary is blurred in the same difference inverse evaluation degradation analysis, and a triangle membership function is rough in the process of simulating the ambiguity among states and the like; and the extension analysis based on the object element association introduces the object element association idea in the degradation analysis, effectively considers the difference of the degradation degree change of each state parameter of the transformer in the normal and fault value intervals, but the evaluation process still needs the accumulation and support of an expert system and experience knowledge, and the state evaluation robustness needs to be improved.
Disclosure of Invention
The invention aims to solve the problems, and provides a transformer state fuzzy set pair evaluation method based on matter element expansion and extension association, so as to improve the influence of unbalanced sample set on the transformer state feature association extraction correctness, effectively consider the inherent correlation among fault symptoms and the complexity of rules, globally optimize the fitness constant of the fault symptom set on the basis, reduce the influence of subjective dependence of the constant calculation process on transformer fault diagnosis and state evaluation validity, reasonably quantize the difference of degradation degree change of each state parameter of a transformer in a normal and fault value interval, and finally improve the robustness and validity of the transformer fault diagnosis and state evaluation process.
In order to achieve the purpose, the invention adopts the following specific scheme:
the transformer state fuzzy set pair evaluation method based on the matter element augmentation and extension association comprises the following steps:
step 1: dividing possible fault types of the power transformer, and then classifying the collected transformer fault sample set according to the divided fault types; equivalently amplifying the classified fault sample set to form an amplified fault sample set;
step 2: selecting a fault symptom set of the transformer, constructing a fault type-fault symptom correlation identification matrix based on the augmented fault sample set, and calculating an optimized fault symptom constant weight coefficient;
and step 3: constructing a variable classical domain and a variable quality node domain of transformer fault symptom set data so as to obtain an object element correlation function of each fault symptom, and calculating an optimal variable weight coefficient corresponding to each fault type by combining the optimized fault symptom constant weight coefficient;
and 4, step 4: normalizing the object element correlation function of each fault symptom, and defining a mapping table of the object element correlation function of the fault symptom and the state level of the fault parameter of the transformer;
and 5: and calculating an identity-difference inverse evaluation matrix required by the transformer state evaluation, obtaining a set pair analysis multivariate degree of each fault type and the whole operation state of the power transformer according to the optimal variable weight coefficient, and analyzing and evaluating the operation state of the transformer by combining an object element correlation function of fault symptoms and a transformer fault parameter state grade mapping table.
Further, in step 1, the method for equivalently expanding the classification fault sample set includes:
classifying the collected transformer fault sample set according to the divided power transformer fault types to obtain a classified sample set;
and (4) amplifying the classified sample set through a k-time set obtained by random bootstrap to obtain an equivalent amplification fault sample set.
Further, in the step 2,
dividing each fault type augmentation sample in the equivalent augmentation fault sample set by k-fold according to the verification set sample and the training set sample;
establishing a fault type-fault symptom association identification matrix by using a threshold Apriori algorithm;
initializing a fault sign constant weight fitness function weight based on a fault type-fault sign correlation identification matrix, and performing global PSO optimization on the fault sign constant weight fitness function weight by taking the minimum error rate of K-fold verification set average fault diagnosis as an optimization target; and obtaining the optimized fault symptom constant weight coefficient.
Further, the method for establishing the fault type-fault symptom association identification matrix specifically comprises the following steps:
for each fault type FTiThe training set samples of (a) are,by including fault symptoms FSjThe sample proportion of (1) is a criterion to form a 1-candidate set association identification mapping vector;
by each fault type FTjAnd constructing a fault type-fault symptom correlation identification matrix by the corresponding 1-candidate set correlation identification mapping vector.
Further, the average fault diagnosis misjudgment rate of the k-fold verification set specifically is as follows:
wherein,number of misjudged samples for fault diagnosis in the ith verification set, NiThe number of samples in the ith verification set is the number of samples in the ith verification set; k is the number of cross-validation fold repetitions.
Further, in step 3, the relationship function of the object element of each fault sign is as follows:
fault symptom state parameter X in variable classical domain XlAbove with respect to x0The effective moment of and the fault symptom state parameter X are related to the domain Xl∪XzThe ratio of the positions of (a); wherein the variable classical domain Xl＝[a,b]Region of metamorphic order Xz＝[b,c]And has a>b>c。
Further, in step 3, the optimal variable weight coefficient corresponding to each fault type is:
wherein,as a fault type FTjThe constant weight coefficient of (a) is,β is the variable weight coefficient of the weight,for the fault sign FSiWeighted sum of fault types FTjThe value of (a).
Further, in step 4, the evaluation matrix of the same type and the different type required for the state evaluation of the transformer specifically includes:
wherein,a constant weight coefficient for fault signs;for fault sign FSiFor fault type FTjThe same or different evaluation vectors of (1); 1, 2.
Further, in step 4, the set pair analysis multivariate connection degree of each fault type and the overall operation state of the power transformer is specifically
Wherein,the optimal variable weight coefficient corresponding to the fault type,as a fault type FTjThe degree of association of (c).
Further, the calculation of the average fault diagnosis misjudgment rate of the K-fold verification sets requires a fault diagnosis process for each sample of each verification set crossed by the K-fold, and each fault diagnosis process is performed under the current constant weight fitness function weight; and the fault diagnosis process self-calls the process of the transformer state fuzzy set pair evaluation method based on the matter element augmentation and extension association.
The invention has the beneficial effects that:
compared with the prior art, in the transformer state fuzzy set pair evaluation method based on the matter element augmentation and extension association provided by the invention, the influence of unbalance of the sample set on the transformer state feature correlation extraction correctness is improved through equivalent augmentation of the classification fault sample set, by constructing a weighted fitness function that effectively accounts for the inherent correlation between fault symptoms and the complexity of the rules, on the basis, the fitness constant of the fault symptom set is globally optimized through a threshold Apriori-PSO-CV algorithm, the influence of subjective dependence of the constant calculation process on transformer fault diagnosis and state evaluation effectiveness is reduced, and reasonably quantifying the difference of the degradation degree change of each state parameter of the transformer in the normal and fault value intervals by combining a matter element extension analysis theory, and finally improving the robustness and effectiveness of the transformer fault diagnosis and state evaluation process.
Drawings
FIG. 1 is a flow chart of an implementation of a transformer state fuzzy set pair evaluation method based on matter element augmentation and extension association;
FIG. 2 is a flow chart of a threshold Apriori-PSO-CV algorithm implementation;
FIG. 3 is a flow chart of the implementation of the fault diagnosis verification of the threshold Apriori-PSO-CV optimization process.
The specific implementation mode is as follows:
the invention is described in detail below with reference to the accompanying drawings:
the invention provides a transformer state fuzzy set pair evaluation method based on matter element augmentation and extension association. An implementation flow of the transformer state fuzzy set pair evaluation method based on the matter element augmentation and extension association is shown in fig. 1. The method comprises the following steps of firstly, dividing possible fault types of the power transformer according to oil immersed transformer (reactor) state evaluation guide rules and actual operation experience, then classifying collected transformer fault sample sets according to set fault types, and performing classification fault sample set equivalent augmentation through a random weighting method; then, constructing a fault type-fault symptom correlation identification matrix based on the augmented fault sample set, and obtaining a simplified correlation frequent fault symptom set and an optimized fault symptom constant weight coefficient of each fault type through a threshold value Apriori-PSO-CV algorithm;
then, constructing a classical domain and a node domain of a transformer fault symptom set data dictionary by combining a matter element extension theory, further obtaining a matter element correlation function of each fault symptom, and solving a variable weight coefficient corresponding to each fault type; and finally, calculating an identity-difference evaluation matrix required by the transformer state evaluation through a gradient ridge fuzzy membership function, and combining the optimized variable weight coefficient to obtain a set pair analysis multivariate degree of each fault type and the whole transformer operation state so as to realize the analysis evaluation of the transformer operation state.
The specific implementation method comprises the following steps:
step 1: dividing possible fault types of the power transformer according to the oil immersed transformer (reactor) state evaluation guide rule and actual operation experience, classifying the collected transformer fault sample set according to the set fault type, and performing classification fault sample set equivalent augmentation through a random Bootstrap method (Bootstrap);
step 2: selecting a fault symptom set of the transformer from representative state quantities with complete measurable parameters, constructing a fault type-fault symptom association identification matrix based on an augmented fault sample set, and obtaining a simplified association frequent fault symptom set and an optimized fault symptom constant weight coefficient of each fault type through a threshold Apriori-PSO-CV algorithm;
and step 3: constructing a classical domain and a node domain of a transformer fault symptom set data dictionary by combining a matter element extension theory, further obtaining a matter element correlation function of each fault symptom, and calculating an optimal variable weight coefficient corresponding to each fault type by combining an optimized fault symptom constant weight coefficient matrix;
and 4, step 4: and calculating an identity-difference evaluation matrix required by transformer state evaluation through a gradient-ridge fuzzy membership function, obtaining a set pair analysis multivariate degree of each fault type and the whole operation state of the power transformer by combining the optimal variable weight coefficient, and analyzing and evaluating the operation state of the transformer.
In step 1 of the embodiment of the present invention, the divided power transformer fault types and fault symptom types are shown in tables 1 and 2, respectively, according to "oil-immersed transformer (reactor) state evaluation guide rules" and actual operation experience.
TABLE 1 Fault types of Power transformers
TABLE 2 types of fault symptoms for power transformers
| Sign of failureIdentification | Type of fault symptom | Fault symptom identification | Type of fault symptom |
| FS1 | Dielectric loss of insulating oil | FS13 | Relative rate of CO production |
| FS2 | Water content in oil | FS14 | CO2Relative gas production rate |
| FS3 | Breakdown voltage of oil | FS15 | Initial value difference of winding short circuit impedance |
| FS4 | Insulation resistance absorption ratio | FS16 | Dielectric loss of winding insulation |
| FS5 | Index of polarization | FS17 | Initial value difference of winding capacitance |
| FS6 | Volume resistivity | FS18 | C2H2Content (wt.) |
| FS7 | H2Content (wt.) | FS19 | Amount of partial discharge |
| FS8 | Iron core grounding current | FS20 | Gas content in oil |
| FS9 | Iron core insulation resistance | FS21 | CH4Content (wt.) |
| FS10 | C2H6Content (wt.) | FS22 | Neutral point oil flow electrostatic current |
| FS11 | C2H4Content (wt.) | FS23 | Content of Furfural |
| FS12 | Mutual difference of direct-current resistance of winding | FS24 | Degree of polymerization of paper board |
In step 1 of the embodiment of the present invention, a process of performing equivalent augmentation on a classification fault sample set by using a random bootstrapping method (Bootstrap) includes:
classifying the collected transformer fault sample set according to the fault types of the power transformers in the table 1 to obtain a classified sample set I:
I＝{Di},(i＝1,2,...,9) (1)
wherein D isiFor the ith sample set of fault types (transaction database) of the corresponding list of fault types,
and n is the number of samples in the ith fault type sample set. DiIs itself a collection of samples that implies an unknown distribution, and samples are collectedAre independent of each other. Randomly bootstrapping the resulting k-th set
An efficient estimation of the unknown distribution can be achieved and can therefore be used for the set of samples DiThe amplification is carried out, wherein,for the m-th time from the sample set XiSampled at medium probabilityWith a decimation probability of n-1. The resulting final augmented sample set D for the corresponding fault type ii*Is composed of
When random bootstrap is performed, the number of samples of each fault type sample set is ensured to be approximately equal, so that the influence of unbalanced sample set on the transformer state feature correlation extraction correctness can be eliminated on the premise of ensuring that the probability distribution characteristic is unchanged, and the obtained equivalent augmentation sample set I*Comprises the following steps:
I*＝{Di*}，(i＝1，2，...，9) (5)
in step 2 of the embodiment of the present invention, an implementation flow of the embodiment of the threshold Apriori-PSO-CV algorithm is shown in fig. 2. The technical scheme is that firstly, an equivalent augmentation sample set I is collected*Each fault type augmentation sample in the system is randomly divided into K subsets which are independent mutually, the number of the samples of each subset is equal, one subset is taken from each equally divided fault type augmentation sample and is recombined into a verification set, and the rest K-1 subsets of each equally divided fault type augmentation sample are combined into a training set; then establishing a fault type-fault symptom association identification mapping set by using a threshold Apriori algorithm; then initializing the function weight of the normal weight fitness degree of the fault symptom based on the fault type-fault symptom correlation identification matrix, and carrying out global PSO optimization on the normal weight fitness degree by taking the minimum error rate of the K-fold verification set average fault diagnosis as an optimization target; and finally, checking whether the PSO process ending condition is met, namely if the optimization reaches the maximum evolution algebra or the average fault diagnosis misjudgment rate of the K-fold verification set is less than a given value, ending the optimization process, and outputting the current fault symptom constant weight fitness function weight optimization result for calculating the fault symptom constant weight.
Wherein, the process of establishing the fault type-fault symptom correlation identification matrix by the K-1 training set comprises the following steps:
for each fault type FTiK-1 training set of (1), consisting of (1) samples containing fault symptoms FSj(j ═ 1,2 …, m) sample ratio as criterion1-the candidate set association identifies the mapping vector as,
for different fault types, note each fault type FTjThe corresponding 1-candidate set associated identification mapping vector has a length ofWherein the fault symptom ratioTraining a fault type FT in a set for K-1jSample failure sign FSiNumber of times and failure type FTjBy a ratio of the number of samples of (c), a fault symptom correlation threshold epsilonFS-FT＝0.7。
Each fault type FT described by equation (6)jThe corresponding 1-candidate set association identification mapping vector constructs a fault type-fault symptom association identification mapping set which is:
counting the occurrence times of the K-1 training set according to the fault types and the fault symptoms respectively, and recording the occurrence times of each fault type as(j ═ 1, 2.. multidot.n.) the number of occurrences of each symptom of failure is(i 1,2, …, m), and
the expression of the constant weight fitness function is
Wherein the weight vector in the formula (9) satisfies
w1+w2+w3＝1 (10)
In the formula (9), the reaction mixture is,as a fault type FTjThe sample contains fault symptoms FSiFault type FT ofjSample proportion;for fault sign FSiReflecting fault type FT in samplejFailure sign FS ofiSample proportion; by fault type FTjThe inverse of the length of the associated identification mapping vectorCharacterizing the fault type FTjIs associated with the fault symptom of (i.e. with fault type FT)jThe more fault symptoms of the associated mapping, the corresponding fault symptom versus fault type FTjThe smaller the baseline contribution of (c).
Therefore, the constant weight is calculated by the constant weight fitness function expressed by the equation (9), and the weight of each fault symptom in the associated mapping fault type can be more reasonably expressed by sufficiently considering the influences of the fault symptom proportion weight, the fault type proportion weight and the fault type-fault symptom associated mapping complexity. Combining a Particle Swarm Optimization (PSO) algorithm and a Cross Validation (CV) algorithm pair, taking the fault diagnosis misjudgment rate of the test set as an optimization target, and setting a weight vector w of the constant weight fitness function as [ w ═ w1,w2,w3]And optimizing.
In the threshold value Apriori-PSO-CV algorithm, the mean fault diagnosis misjudgment rate of a K-fold verification set is defined as
Wherein N isiFor the number of samples in the ith validation set,and misjudging the number of samples for the fault diagnosis in the verification set.
The calculation of the average fault diagnosis misjudgment rate of the K-fold verification sets in the formula (11) requires a fault diagnosis process for each sample of each verification set crossed by the K-fold, and each fault diagnosis process is performed under the current constant weight fitness function weight. The implementation flow of each fault diagnosis process embodiment is shown in fig. 3, where step 3 and step 4 are the same as the implementation process of the transformer state fuzzy set pair evaluation method based on the matter element augmentation and extension association, that is, the transformer state fuzzy set pair evaluation method based on the matter element augmentation and extension association includes a self-adaptive process in each fault diagnosis false rate calculation in the normal weight fitness optimization process.
Through the constant weight fitness optimizing process, the final fault type-fault sign constant weight matrix which takes the fault signs as the row signs and the fault types as the column signs is obtained
In step 3 described in the embodiment of the present invention, the basic idea of the matter element extension process is: generally, the operation state parameters of the power transformer only progress towards unilateral degradation, so the operation state values can be quantized into two basic intervals, one is a variable classical domain of normal operation, and the other is a variable quality domain with prominent state degradation. In the variable classical domain, the operation state parameters of the transformer become variable and deteriorate along with the increase of the operation time, during which the transformer can still normally operate and the fault symptoms can not appear; as the operating state variable continues to deteriorate, when the operating state variable of the transformer enters the quality-change range, it means that the transformer will exhibit certain fault symptoms. According to the regulations, the variable classical and variable qualitative domains of the various power transformers for measurable signs of failure are shown in table 3.
Generally, in the process of deterioration of the operating state parameters of the transformer, when a fault symptom occurs, the deterioration of the fault state of the transformer is aggravated, and the phenomenon is reflected on a time scale that the transformer rapidly enters an abnormal or even serious fault state and is reflected in a quantitative value mapping, that is, the numerical degradation slope between a variable classical domain and a fault node domain of the operating state parameters is different, which is not considered in the calculation of the fault symptom deterioration of the existing transformer.
TABLE 3 variable classical domain and variable quality node domain of power transformer fault symptom
Based on the above analysis, in step 3 in the embodiment of the present invention, the calculation process of the physical element extension degradation is as follows:
for negative degradation fault symptom state parameter X, the order variable classical domain Xl＝[a,b]Region of metamorphic order Xz＝[b,c]And has a>b>c, then the parameter X is related to the domain Xl∪XzIs defined as the position of (a) in which,
wherein d (X, X)l) And d (X, X)z) Respectively representParameter X and variable classical field XlAnd the metamorphic nodal region XzA moment of
Optimal point x for negative degradation0A, then the parameter X is in the quantization domain XlAbove with respect to x0Is defined as an effective moment
d(x,Xl,x0)＝a-x (15)
With the combined equations (13) - (15), the negative degradation fault symptom state parameter X can be obtained in the region Xl∪XzAbove with respect to x0The deterioration function of the matter element extension correlation is
Similarly, for the positive degradation fault symptom state parameter x, a < b < c, and equations (13) to (15) are changed to
In order to facilitate the fuzzy membership calculation related to the matter element extension related degradation function (16), the normalization processing is carried out on the formula (16), and a fault sign FS can be obtainediFor fault type FTjThe physical element extension normalization degradation degree is
The defined degradation function values and transformer fault parameter state level mappings are shown in table 4 based on the normalization element extension correlation degradation function (16) and the relation between the fault symptom state parameter attention value and the warning value specified in the relevant regulations, wherein s0＝0，s1＝5/18，s2＝7/18，s3＝4/9，s4＝5/9，s5＝11/18，s6＝13/18，s7＝7/9，s8＝8/9，s9＝1。
Table 4 object element extension association degradation function value and transformer fault parameter state grade mapping table
| Operating state | Is normal | Attention is paid to | Light and slight | Abnormality (S) | Severe severity of disease |
| Interval symbol association identifier | s7-s8-s9 | s5-s6-s7-s8 | s3-s4-s5-s6 | s1-s2-s3-s4 | s0-s1-s2 |
| Matter element extension associated degradation function interval | [1,0.5] | [0.5,0] | [0,-0.5] | [-0.5,-1] | [-1,-2] |
| Normalized interval/multivariate coefficient | [1,5/6] | [5/6,2/3] | [2/3,1/2] | [1/2,1/3] | [1/3,0] |
Wherein, a certain fault symptom of the transformer is in a normal or attention running state, which indicates that the fault symptom state parameter of the transformer is in a variable classical domain, a certain fault symptom is in a slight, abnormal or serious fault running state, which indicates that the fault symptom state parameter of the transformer is in a variable quality node domain, the state interval division of the object element extension association mapping is different from the traditional equal interval division mode as shown in the table 4, and the division of the original image state parameters of the object element extension association mapping is also different from the traditional equal interval parameter gradual change mode, it is based on the physical fact that, during the actual transformer state change, the quantitative evolution between the states is not smoothly gradual, namely, the quantitative change process and the qualitative change process of the state parameter are different, and the phenomenon cannot be reflected in the traditional uniform interval smooth quantization normalization process.
Combining a fault type-fault symptom constant weight matrix (12) and a normalization matter element extension correlation function(16) Let us order Andrespectively represent the fault symptoms FSiThe trial value and the optimum value of (c),anda variable classical domain and a variable qualitative domain representing a fault symptom, respectively, thereby addressing a fault symptom FSiWeighted sum available fault type FTjHas a score value of
Considering the influence of the fault type decision state variable score on the weight configuration, the constant weight coefficient is corrected through variable weight calculation,
wherein,as a fault type FTjThe constant weight coefficient reflects the characterization accuracy of the fault type on the running state of the transformer, and if each fault type can accurately characterize the running state of one aspect of the transformer, then the constant weight coefficient hasβ is a variable weight coefficient, and β is 0.2 to take into account that the safety of the whole transformer is affected by the serious deviation of the transformer state index parameter.
In step 4 of the embodiment of the present invention, in order to balance the simulation performance of the state level ambiguity of the transformer and ensure the validity of the state level information of the transformer as much as possible, a semi-staircase-semi-ridge type fuzzy membership function is used to quantify the matter element extension associated degradation degree of each fault symptom of the transformer,
the expression of the fuzzy membership identity different-inverse evaluation state corresponding to the normalized matter element extension correlation function (19) is shown as a formula (23).
In step 4 of the embodiment of the present invention, based on the above analysis, a failure symptom FS is obtainediFor fault type FTjThe same or different inverse evaluation vector of
If the type of failure FTjThe number of associated fault symptoms is NjIn combination with the respective fault symptoms FSiFor fault type FTjConstant weight coefficient of (1), available fault type FTjThe same or different inverse evaluation vector of
In step 4 described in the embodiment of the present invention, further, the fault type FTjHas a degree of association of
Wherein the same or different inverse coefficient vector VSRE is
VSRE＝[VSRE1,VSRE2,…,VSRE5]T(27)
In step 4 of the embodiment of the present invention, due to the nonlinear characteristic of the object element extension correlation degradation function, the equal-difference coefficient vector of the multivariate connectivity cannot be solved by using the equipartition method, the multivariate connectivity is divided according to the operation state normalization interval of table 4, the interval median is used as the equal-difference coefficient vector element, and there areSince the identity and opposite coefficient usually takes 1 and-1, the scale expansion of the identity and opposite coefficient vector can be obtained
In step 4 of the embodiment of the present invention, further, the fault type FT is usedjThe relation of the overall operation state of the transformer can be obtained
In summary, the equations (26) and (28) represent the fault types FT of the power transformer, respectivelyjAnd comparing the quantitative evaluation indexes corresponding to the overall operation state with the operation state normalized quantitative interval shown in the table 4 to obtain the operation state of the current operation power transformer and the possible fault type.
Although the embodiments of the present invention have been described with reference to the accompanying drawings, it is not intended to limit the scope of the present invention, and it should be understood by those skilled in the art that various modifications and variations can be made without inventive efforts by those skilled in the art based on the technical solution of the present invention.
Claims (10)
1. The transformer state fuzzy set pair evaluation method based on the matter element augmentation and extension association is characterized by comprising the following steps:
step 1: dividing possible fault types of the power transformer, and then classifying the collected transformer fault sample set according to the divided fault types; equivalently augmenting the classified fault sample set to form an equivalently augmented fault sample set;
step 2: selecting a fault symptom set of the transformer, constructing a fault type-fault symptom correlation identification matrix based on the equivalent augmented fault sample set, and calculating an optimized fault symptom constant weight coefficient;
and step 3: constructing a variable classical domain and a variable quality node domain of transformer fault symptom set data so as to obtain an object element correlation function of each fault symptom, and calculating an optimal variable weight coefficient corresponding to each fault type by combining the optimized fault symptom constant weight coefficient;
and 4, step 4: normalizing the object element correlation function of each fault symptom, and defining a mapping table of the object element correlation function of the fault symptom and the state level of the fault parameter of the transformer;
and 5: and calculating an identity-difference inverse evaluation matrix required by the transformer state evaluation, obtaining a set pair analysis multivariate degree of each fault type and the whole operation state of the power transformer according to the optimal variable weight coefficient, and analyzing and evaluating the operation state of the transformer by combining an object element correlation function of fault symptoms and a transformer fault parameter state grade mapping table.
2. The transformer state fuzzy set pair evaluation method based on the matter element augmentation extensive association as claimed in claim 1, wherein in the step 1, the method for equivalently augmenting the classification fault sample set comprises:
classifying the collected transformer fault sample set according to the divided power transformer fault types to obtain a classified sample set;
and (4) amplifying the classified sample set through a k-time set obtained by random bootstrap to obtain an equivalent amplification fault sample set.
3. The method as claimed in claim 1, wherein in step 2,
dividing each fault type augmentation sample in the equivalent augmentation fault sample set by k-fold according to the verification set sample and the training set sample;
establishing a fault type-fault symptom association identification matrix by using a threshold Apriori algorithm;
initializing a fault sign constant weight fitness function weight based on a fault type-fault sign correlation identification matrix, and performing global PSO optimization on the fault sign constant weight fitness function weight by taking the minimum error rate of K-fold verification set average fault diagnosis as an optimization target; and obtaining the optimized fault symptom constant weight coefficient.
4. The transformer state fuzzy set pair evaluation method based on the matter element augmented and extensive correlation as claimed in claim 3, wherein the method for establishing the fault type-fault symptom correlation identification matrix specifically comprises:
for each fault type FTjThe training set samples of (2) are selected from the group consisting of FSjThe sample proportion of (1) is a criterion to form a 1-candidate set association identification mapping vector;
by each fault type FTjAnd constructing a fault type-fault symptom correlation identification matrix by the corresponding 1-candidate set correlation identification mapping vector.
5. The method for evaluating the fuzzy set pair of transformer states based on the element-augmented extensive correlation as claimed in claim 3, wherein the K-fold validation set average fault diagnosis misjudgment rate is specifically:
wherein,number of misjudged samples for fault diagnosis in the ith verification set, NiThe number of samples in the ith verification set is the number of samples in the ith verification set; k is the number of cross-validation fold repetitions.
6. The transformer state fuzzy set pair evaluation method based on matter element augmentation extensive association as claimed in claim 1, wherein in said step 3, the matter element association function of each fault symptom is:
fault symptom state parameter X in variable classical domain XlAbove with respect to x0Effective moment ofBarrier symptom state parameter X with respect to domain Xl∪XzThe ratio of the positions of (a); wherein x is0For the optimum point of negative degradation, the variable classical domain Xl＝[a,b]Region of metamorphic order Xz＝[b,c]And has a>b>c。
7. The transformer state fuzzy set pair evaluation method based on the matter element augmentation and extension association as claimed in claim 1, wherein in the step 3, the optimal variable weight coefficient corresponding to each fault type is:
wherein,as a fault type FTjThe constant weight coefficient of (a) is,β is the variable weight coefficient of the weight,for the fault sign FSiWeighted sum of fault types FTjThe value of (a).
8. The method according to claim 1, wherein in step 5, the evaluation matrix of identity and difference of the transformer state required for evaluation is specifically:
wherein,a constant weight coefficient for fault signs;for fault sign FSiFor fault type FTjThe same or different evaluation vectors of (1); 1, 2.
9. The method as claimed in claim 1, wherein in step 5, the set of fault types and overall operating states of the power transformer is analyzed for multivariate relations
Wherein,the optimal variable weight coefficient corresponding to the fault type,as a fault type FTjThe degree of association of (c).
10. The method for evaluating the fuzzy set pair of transformer states based on the element-augmented and extensive correlation as claimed in claim 3, wherein the calculation of the mean fault diagnosis misjudgment rate of the K-fold validation sets requires performing a fault diagnosis process on each sample crossing each validation set by K-fold, and each fault diagnosis process is performed under the current weight fitness function weight; and the fault diagnosis process self-calls the process of the transformer state fuzzy set pair evaluation method based on the matter element augmentation and extension association.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610211816.8A CN105911407B (en) | 2016-04-07 | 2016-04-07 | Associated transformer state fuzzy set can be opened up to appraisal procedure based on matter-element augmentation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610211816.8A CN105911407B (en) | 2016-04-07 | 2016-04-07 | Associated transformer state fuzzy set can be opened up to appraisal procedure based on matter-element augmentation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105911407A CN105911407A (en) | 2016-08-31 |
| CN105911407B true CN105911407B (en) | 2018-08-17 |
Family
ID=56744652
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610211816.8A Active CN105911407B (en) | 2016-04-07 | 2016-04-07 | Associated transformer state fuzzy set can be opened up to appraisal procedure based on matter-element augmentation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105911407B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107958292B (en) * | 2017-10-19 | 2022-02-25 | 山东科技大学 | Fault Diagnosis Method Based on Cost-Sensitive Learning for Transformer Fuzzy Careful Reasoning |
| CN107808044A (en) * | 2017-10-20 | 2018-03-16 | 广东电网有限责任公司河源供电局 | A kind of transformer insulating paper deterioration appraisal procedure for considering running temperature and influenceing |
| CN110110816A (en) * | 2019-06-05 | 2019-08-09 | 国网上海市电力公司 | A kind of Power Transformer Faults deduction method |
| CN110333414A (en) * | 2019-08-02 | 2019-10-15 | 华北电力大学（保定） | The multi-level state evaluating method of power transformer |
| CN112231624B (en) * | 2020-09-16 | 2024-03-26 | 中电电气(江苏)变压器制造有限公司 | Real-time evaluation system for short-circuit resistance of multi-transformer winding based on Internet of things |
| CN115407053B (en) * | 2022-08-19 | 2024-03-15 | 山东科技大学 | Indication optimization method, computer equipment and readable storage medium |
| CN118780662A (en) * | 2024-06-17 | 2024-10-15 | 云南滇能(集团)控股公司 | A method and system for safety management of hydropower station equipment operation and maintenance |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6901389B1 (en) * | 2000-02-25 | 2005-05-31 | Information Decision Technologies, Llc | Method to augment imagery to display 3-D probabilistic object locations |
| CN101251564A (en) * | 2008-04-08 | 2008-08-27 | 昆明理工大学 | A Fault Diagnosis Method of Power Transformer Using Extenics and Rough Set Theory |
| CN102495318A (en) * | 2011-12-19 | 2012-06-13 | 华北电力大学（保定） | Fault diagnosis method of capacitive equipment |
| CN102522705A (en) * | 2011-11-22 | 2012-06-27 | 广东电网公司电力科学研究院 | Transformer maintenance method and device |
| CN103576050A (en) * | 2013-11-21 | 2014-02-12 | 贵阳供电局 | Operating state assessment method of capacitor voltage transformer |
| CN103678765A (en) * | 2013-10-31 | 2014-03-26 | 上海交通大学 | Transformer operating state comprehensive evaluation method based on on-line monitoring |
| CN103871000A (en) * | 2014-03-13 | 2014-06-18 | 石家庄铁道大学 | Probability whole life cycle cost-based power transformer performance comprehensive evaluation method |
| CN103954884A (en) * | 2014-04-21 | 2014-07-30 | 昆明理工大学 | Distribution network fault line selection method based on extensional fusion of pattern spectrum analysis and morphological peak and valley detection |
| CN104101794A (en) * | 2014-02-19 | 2014-10-15 | 江苏倍尔科技发展有限公司 | Integrated control system for power transformer |
| CN105160484A (en) * | 2015-09-11 | 2015-12-16 | 国家电网公司 | Electric power energy saving integrated assessment method of charge source combined peak regulating plan |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7120119B2 (en) * | 2000-06-08 | 2006-10-10 | International Business Machines Corporation | Management of protocol information in PNNI hierarchical networks |
-
2016
- 2016-04-07 CN CN201610211816.8A patent/CN105911407B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6901389B1 (en) * | 2000-02-25 | 2005-05-31 | Information Decision Technologies, Llc | Method to augment imagery to display 3-D probabilistic object locations |
| CN101251564A (en) * | 2008-04-08 | 2008-08-27 | 昆明理工大学 | A Fault Diagnosis Method of Power Transformer Using Extenics and Rough Set Theory |
| CN102522705A (en) * | 2011-11-22 | 2012-06-27 | 广东电网公司电力科学研究院 | Transformer maintenance method and device |
| CN102495318A (en) * | 2011-12-19 | 2012-06-13 | 华北电力大学（保定） | Fault diagnosis method of capacitive equipment |
| CN103678765A (en) * | 2013-10-31 | 2014-03-26 | 上海交通大学 | Transformer operating state comprehensive evaluation method based on on-line monitoring |
| CN103576050A (en) * | 2013-11-21 | 2014-02-12 | 贵阳供电局 | Operating state assessment method of capacitor voltage transformer |
| CN104101794A (en) * | 2014-02-19 | 2014-10-15 | 江苏倍尔科技发展有限公司 | Integrated control system for power transformer |
| CN103871000A (en) * | 2014-03-13 | 2014-06-18 | 石家庄铁道大学 | Probability whole life cycle cost-based power transformer performance comprehensive evaluation method |
| CN103954884A (en) * | 2014-04-21 | 2014-07-30 | 昆明理工大学 | Distribution network fault line selection method based on extensional fusion of pattern spectrum analysis and morphological peak and valley detection |
| CN105160484A (en) * | 2015-09-11 | 2015-12-16 | 国家电网公司 | Electric power energy saving integrated assessment method of charge source combined peak regulating plan |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105911407A (en) | 2016-08-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105911407B (en) | Associated transformer state fuzzy set can be opened up to appraisal procedure based on matter-element augmentation | |
| Tamma et al. | High voltage power transformer condition assessment considering the health index value and its decreasing rate | |
| Taha et al. | Conditional probability‐based interpretation of dissolved gas analysis for transformer incipient faults | |
| CN106251059B (en) | Cable state evaluation method based on probabilistic neural network algorithm | |
| CN107063349A (en) | A kind of method and device of Fault Diagnosis Method of Power Transformer | |
| Tran et al. | Comparative investigations on reference models for fault detection and diagnosis in centrifugal chiller systems | |
| Nosratian Ahour et al. | Determination and localisation of turn‐to‐turn fault in transformer winding using frequency response analysis | |
| CN106529124A (en) | A transformer insulation state assessment method based on principal component analysis and support vector machines | |
| Yan et al. | Transformer fault diagnosis based on BP‐Adaboost and PNN series connection | |
| CN103389430A (en) | Oil-immersed type transformer fault detection method based on Bayesian discrimination theory | |
| JP2011185880A (en) | Reliability evaluation device, and program and method of the same | |
| Kari et al. | Power transformer fault diagnosis using FCM and improved PCA | |
| Islam et al. | Application of Parzen Window estimation for incipient fault diagnosis in power transformers | |
| CN115563494B (en) | A Data-Driven Fault Detection Method for Multimodal Flow Industrial Processes Based on Transfer Learning | |
| CN110110816A (en) | A kind of Power Transformer Faults deduction method | |
| CN112085064B (en) | Transformer fault diagnosis method based on multi-classification probability output of support vector machine | |
| CN106569069A (en) | Power transformer fault diagnosis method | |
| CN108802576B (en) | Subjective and objective integration assessment method for running state of oil-immersed capacitor bushing | |
| CN114236332A (en) | A method and system for judging the insulation state of a power cable | |
| Behkam et al. | Condition monitoring of distribution transformers using frequency response traces and artificial neural network to detect the extent of windings axial displacements | |
| Tian et al. | Subsystem measurement-based condition assessment for power transformers via joint inference of data and knowledge | |
| CN116701932A (en) | A Transformer Fault Diagnosis Method Based on NGO-GBDT | |
| CN109557434B (en) | Partial discharge signal identification method under strong background noise based on composite dictionary sparse representation classification | |
| Jahan et al. | Detecting the location and severity of transformer winding deformation by a novel adaptive particle swarm optimization algorithm | |
| Carrijo et al. | Bushing anomaly detection: an initial assessment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |