CN-121980184-A - Transformer direct current magnetic bias degree assessment method based on multi-feature quantity fusion

CN121980184ACN 121980184 ACN121980184 ACN 121980184ACN-121980184-A

Abstract

The invention discloses a transformer direct current magnetic bias degree assessment method based on multi-feature quantity fusion. Firstly, synchronously collecting a transformer grounded neutral point current signal, a box surface vibration signal and a surrounding noise signal. Next, characteristic quantities representing the dc bias state, including an average value of the current signal, a low-frequency energy ratio, kurtosis of the vibration/noise signal, and a harmonic energy ratio, are extracted from the time domain and the frequency domain. And then, combining subjective weight determined by an analytic hierarchy process with objective weight determined by a coefficient of variation process, calculating fuzzy density of each characteristic quantity by weighted average, and solving lambda value and fuzzy measure. And then, establishing a membership function and a fuzzy evaluation matrix of the characteristic quantity to the DC magnetic bias degree of the transformer. And finally, fusing the multi-characteristic-quantity information by adopting a fuzzy integral algorithm, and evaluating the DC magnetic bias degree grade of the transformer according to the principle of maximum membership.

Inventors

LI XIAOPENG
ZHOU WENYUE
ZHANG HUAJIE
LUO YIPING
HAN RUI
WANG HAO
ZHANG CHUN
LUO RONGSEN

Assignees

国网四川省电力公司电力科学研究院

Dates

Publication Date: 20260505
Application Date: 20260123

Claims (10)

1. A transformer direct current magnetic bias degree evaluation method based on multi-feature quantity fusion is characterized by comprising the following steps: Acquiring a multi-source signal when the transformer operates; Extracting a plurality of characteristic quantities used for representing the DC magnetic bias degree from the multi-source signal; Based on the feature quantities, fusing subjective weights determined by an analytic hierarchy process and objective weights determined by a coefficient of variation process to obtain comprehensive weights of the feature quantities as fuzzy densities; According to the fuzzy density, determining interaction parameters for representing interaction relations among the features, and calculating fuzzy measures of any non-empty subsets in the feature quantities based on the interaction parameters, wherein the fuzzy measures are used for representing the importance of the feature subsets and the interaction relations among the feature subsets; establishing membership functions of each characteristic quantity for a plurality of preset direct current magnetic bias degree grades so as to form a fuzzy membership evaluation matrix; aiming at each bias magnetic degree grade, based on the membership vector corresponding to the grade in the fuzzy membership evaluation matrix and the fuzzy measure, performing fusion calculation to obtain comprehensive evaluation values of all grades; And determining the final DC magnetic bias degree of the transformer according to the comprehensive evaluation values of the various grades.
2. The method of claim 1, wherein the step of acquiring the multi-source signal of the transformer during operation comprises: Acquiring a grounded neutral point current signal through a current sensor arranged at a grounded neutral point of the transformer; Acquiring a tank vibration signal through a vibration sensor arranged on the surface of a transformer tank; Acquiring an environmental noise signal through a noise sensor arranged at a preset position around the transformer; and carrying out synchronous time synchronization processing on the grounded neutral point current signal, the box body vibration signal and the environmental noise signal.
3. The method of claim 1, wherein the multi-source signal comprises a grounded neutral current signal, a tank vibration signal, and an ambient noise signal, and wherein the step of extracting a plurality of features from the multi-source signal that characterize the degree of DC bias comprises: preprocessing the grounded neutral point current signal, the box vibration signal and the environmental noise signal respectively; Extracting time domain statistical characteristics and frequency domain energy distribution characteristics based on power spectrum density from the preprocessed grounded neutral point current signal; And respectively extracting time domain kurtosis characteristics and harmonic energy ratio characteristics based on power spectrum density from the preprocessed box vibration signals and the preprocessed environment noise signals, wherein the extracted various characteristic quantities jointly form a plurality of characteristic quantities for representing the DC magnetic bias degree.
4. The method according to claim 3, wherein the step of fusing subjective weights determined by a hierarchical analysis method with objective weights determined by a coefficient of variation method based on the plurality of feature quantities to obtain an integrated weight of each feature quantity as a blur density comprises: constructing a judgment matrix based on the relative importance of each characteristic quantity to the DC magnetic bias degree, and obtaining the subjective weight through solving the maximum characteristic value of the judgment matrix and the corresponding characteristic vector thereof and carrying out normalization processing; calculating variation coefficients of all the characteristic quantities, and carrying out normalization processing on the variation coefficients of all the characteristic quantities to obtain the objective weight; and fusing the subjective weight and the objective weight in a weighted average mode to obtain comprehensive weight of each characteristic quantity, and determining the comprehensive weight as fuzzy density of each characteristic quantity, wherein the weighted average coefficient is a preset preference factor and is used for adjusting preference degree of the subjective weight or the objective weight.
5. The method of claim 4, wherein the step of determining interaction parameters for characterizing the interaction relationship between features based on the blur density and calculating a blur measure for any non-empty subset of the plurality of feature quantities based on the interaction parameters comprises: Solving an interaction parameter according to the fuzzy density of each feature quantity, wherein the value of the interaction parameter is obtained by solving a polynomial equation taking each fuzzy density as a variable, wherein the polynomial equation is configured to be a value of the interaction parameter plus one and is equal to a continuous product corresponding to each feature quantity, and the continuous product is a continuous product of the interaction parameter multiplied by the product of the fuzzy density; For any one non-empty subset of the feature quantities, the fuzzy measure value is calculated by firstly calculating a continuous product of an interaction parameter and a fuzzy density product corresponding to each feature quantity in the subset, and then dividing the continuous product by one by the interaction parameter to obtain a quotient, namely the fuzzy measure value of the feature subset.
6. The method of claim 4, wherein the step of establishing membership functions of each feature quantity for a plurality of preset dc bias level classes to form a fuzzy membership evaluation matrix comprises: Calculating the deviation degree of the current running state according to the actual measured value and a preset reference value of each characteristic quantity; Defining a trapezoid membership function taking the deviation degree as an independent variable for each characteristic quantity aiming at each preset direct current magnetic bias degree grade, wherein the shape of the trapezoid membership function is determined by a threshold parameter preset for the direct current magnetic bias degree grade, and the preset direct current magnetic bias degree grades comprise four grades which are respectively in a normal state, slightly magnetic bias, moderately magnetic bias and severely magnetic bias; calculating the membership degree of the deviation degree of each characteristic quantity to each direct-current magnetic bias degree grade based on the trapezoidal membership function; And constructing a fuzzy membership evaluation matrix according to all the membership obtained through calculation, wherein the row of the fuzzy membership evaluation matrix corresponds to the characteristic quantity, the column corresponds to the bias degree grade, and the matrix element is the membership of the grade corresponding to the characteristic quantity.
7. The method of claim 6, wherein the step of performing a fusion calculation for each bias level based on the membership vector corresponding to the level in the fuzzy membership evaluation matrix and the fuzzy measure to obtain the comprehensive evaluation value of each level comprises: Extracting membership vectors corresponding to the levels in the fuzzy membership evaluation matrix aiming at each bias degree level, and sequencing the membership vectors according to the order of membership values from large to small; performing Choquet integral operation based on the sorted membership vector and the fuzzy measure to obtain a comprehensive evaluation value of the bias magnetic degree grade; the step of determining the final DC magnetic bias degree of the transformer according to the comprehensive evaluation values of all the levels comprises the following steps: And comparing the comprehensive evaluation values of all the bias magnetic degree grades, and determining the bias magnetic degree grade with the maximum comprehensive evaluation value as the final direct current bias magnetic degree of the transformer.
8. The utility model provides a transformer direct current magnetic bias degree evaluation device based on many feature quantity fuses which characterized in that includes: the acquisition module is used for acquiring multi-source signals when the transformer runs; An extraction module for extracting a plurality of characteristic quantities for representing the DC magnetic bias degree from the multi-source signal; The fuzzy density fusion module is used for fusing subjective weights determined by a analytic hierarchy process and objective weights determined by a coefficient of variation process based on the plurality of feature quantities to obtain comprehensive weights of the feature quantities as fuzzy densities; The fuzzy measure determining module is used for determining interaction parameters for representing interaction relations among the features according to the fuzzy density, and calculating fuzzy measures of any non-empty subsets in the feature quantities based on the interaction parameters, wherein the fuzzy measures are used for representing the importance of the feature subsets and the interaction relations among the feature subsets; The establishing module is used for establishing membership functions of each characteristic quantity to a plurality of preset direct current magnetic bias degree grades so as to form a fuzzy membership evaluation matrix; The comprehensive evaluation value calculation module is used for carrying out fusion calculation on the basis of the membership vector corresponding to the level in the fuzzy membership evaluation matrix and the fuzzy measure according to each bias magnetic degree level to obtain comprehensive evaluation values of all levels; And the determining module is used for determining the final direct current magnetic bias degree of the transformer according to the comprehensive evaluation value of each grade.
9. An electronic device, comprising: A memory, and one or more processors communicatively coupled to the memory; stored in the memory are instructions executable by the one or more processors to cause the one or more processors to implement the method of any one of claims 1 to 7.
10. A computer readable storage medium, characterized in that the computer program is stored in the readable storage medium, which computer program, when being executed by a processor, implements the method of any of claims 1 to 7.

Description

Transformer direct current magnetic bias degree assessment method based on multi-feature quantity fusion Technical Field The invention relates to the technical field of transformer DC magnetic bias degree evaluation, in particular to a transformer DC magnetic bias degree evaluation method based on multi-feature quantity fusion. Background The grid structure of the power grid in China is complex, the electrical connection is tight, the invasion points of direct current magnetic bias current of the power grid are numerous, and the phenomenon that the direct current bias current such as the ground current and the geomagnetic induction current of the high-voltage direct current power transmission system invade the power grid is more frequent is caused. The bias current invading the power grid accelerates the corrosion of metal structures such as a grounding grid, buried oil/gas pipelines and the like, and induces the main transformer DC bias of the urban power grid to cause serious consequences such as aggravation of non-stable vibration of the transformer, relay protection failure and the like, thereby seriously endangering the safety of urban infrastructure. The method is used for accurately evaluating the DC magnetic bias degree of the transformer, and is a basis for acquiring the health state of the transformer and developing corresponding magnetic bias inhibition measures. However, the traditional DC magnetic bias degree evaluation method based on the single characteristic quantity cannot effectively cope with complex operation conditions of the transformer, is easily influenced by load fluctuation, environmental conditions and the like, has the defects of low reliability and the like, and cannot be used for evaluating the DC magnetic bias degree of the transformer once the sensor fails. The transformer direct current magnetic bias degree assessment method based on multi-feature quantity fusion has the advantages of strong anti-interference performance, high assessment precision and the like by comprehensively considering multi-dimensional data information and cross-verifying multi-dimensional features, but also faces the problems that the feature selection is difficult to accurately represent the direct current magnetic bias degree, the multi-feature quantity is difficult to fusion and assess, and the like. Therefore, it is necessary to study the evaluation method of the dc bias degree of the transformer based on the multi-feature fusion to guide the development of the dc bias response and suppression of the transformer. In summary, in the related art, when the dc bias degree of the transformer is evaluated, the evaluation is performed depending on a single feature quantity, which is easily interfered by the working condition, and the interaction relationship between multiple feature quantities cannot be effectively represented by adopting a simple weighted fusion method, so that the technical problem of insufficient reliability and accuracy of the evaluation result is caused. Disclosure of Invention The technical problem to be solved by the invention is that in the related technology, when the DC magnetic bias degree of the transformer is evaluated, the evaluation is easy to be interfered by working conditions depending on a single characteristic quantity, and the interaction relation among multiple characteristic quantities cannot be effectively represented by adopting a simple weighted fusion method, so that the technical problem of insufficient reliability and accuracy of an evaluation result is caused. The transformer direct current magnetic bias degree evaluation method based on multi-feature quantity fusion solves the technical problems of insufficient reliability and accuracy of evaluation results. The invention is realized by the following technical scheme: In a first aspect, the invention provides a transformer direct current magnetic bias degree evaluation method based on multi-feature fusion, which comprises the following steps: Acquiring a multi-source signal when the transformer operates; Extracting a plurality of characteristic quantities used for representing the DC magnetic bias degree from the multi-source signal; Based on the feature quantities, fusing subjective weights determined by an analytic hierarchy process and objective weights determined by a coefficient of variation process to obtain comprehensive weights of the feature quantities as fuzzy densities; According to the fuzzy density, determining interaction parameters for representing interaction relations among the features, and calculating fuzzy measures of any non-empty subsets in the feature quantities based on the interaction parameters, wherein the fuzzy measures are used for representing the importance of the feature subsets and the interaction relations among the feature subsets; establishing membership functions of each characteristic quantity for a plurality of preset direct current magnetic bias degree grades so as to