CN116990633A - Fault studying and judging method based on multiple characteristic quantities - Google Patents
Fault studying and judging method based on multiple characteristic quantities Download PDFInfo
- Publication number
- CN116990633A CN116990633A CN202310757582.7A CN202310757582A CN116990633A CN 116990633 A CN116990633 A CN 116990633A CN 202310757582 A CN202310757582 A CN 202310757582A CN 116990633 A CN116990633 A CN 116990633A
- Authority
- CN
- China
- Prior art keywords
- fault
- real
- sample
- feature
- historical
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 59
- 238000012545 processing Methods 0.000 claims abstract description 45
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 19
- 238000012544 monitoring process Methods 0.000 claims abstract description 5
- 239000011159 matrix material Substances 0.000 claims description 39
- 230000008569 process Effects 0.000 claims description 16
- 238000003745 diagnosis Methods 0.000 claims description 14
- 239000013598 vector Substances 0.000 claims description 13
- 238000004458 analytical method Methods 0.000 claims description 11
- 230000009466 transformation Effects 0.000 claims description 10
- 238000010606 normalization Methods 0.000 claims description 7
- 238000012850 discrimination method Methods 0.000 claims description 6
- 238000011160 research Methods 0.000 abstract description 15
- 238000010801 machine learning Methods 0.000 description 8
- 238000004088 simulation Methods 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 238000011156 evaluation Methods 0.000 description 6
- 230000001186 cumulative effect Effects 0.000 description 5
- 238000002474 experimental method Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000000513 principal component analysis Methods 0.000 description 4
- 238000012795 verification Methods 0.000 description 4
- 230000003247 decreasing effect Effects 0.000 description 3
- 230000007935 neutral effect Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000001052 transient effect Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000006399 behavior Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000007417 hierarchical cluster analysis Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000002955 isolation Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000011418 maintenance treatment Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 238000011282 treatment Methods 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
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
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/081—Locating faults in cables, transmission lines, or networks according to type of conductors
- G01R31/086—Locating faults in cables, transmission lines, or networks according to type of conductors in power transmission or distribution networks, i.e. with interconnected conductors
-
- 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
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/088—Aspects of digital computing
-
- 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
- G01R31/52—Testing for short-circuits, leakage current or ground faults
-
- 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
- G01R31/58—Testing of lines, cables or conductors
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Theoretical Computer Science (AREA)
- Locating Faults (AREA)
Abstract
The invention discloses a fault studying and judging method based on multiple characteristic quantities, which belongs to the technical field of power distribution networks and comprises the following steps: s1, acquiring historical fault characteristic quantities, constructing a historical characteristic sample and performing standardization processing; s2, classifying the historical feature samples based on a fuzzy membership matching algorithm to obtain the types of the historical feature samples, and obtaining corresponding membership matching centers; s3, monitoring the zero sequence voltage value of the power distribution network in real time, and if the zero sequence voltage value exceeds a preset threshold value, starting a non-setting protection scheme; s4, collecting real-time fault characteristic quantity and constructing a real-time characteristic sample; s5, carrying out standardized processing on the real-time characteristic sample, and calculating the distance from the real-time characteristic sample to the membership matching center; s6, determining the type of the real-time characteristic sample data, classifying the real-time characteristic sample into the historical data of the corresponding type, and executing S3. According to the scheme, the types of the characteristic samples are determined through the distances from the characteristic samples to the membership matching center, so that the fault research and judgment of the power distribution network is realized, and the accuracy of the fault research and judgment is improved.
Description
Technical Field
The invention belongs to the technical field of power distribution networks, and particularly relates to a fault studying and judging method based on multiple characteristic quantities.
Background
In a power system, ground faults occur on overhead lines of a power distribution network, wherein the single-phase short circuit ground faults account for more than 80% of all fault types, the misjudgment rate of arc light and high-resistance ground faults by the existing fault criteria is still higher, the system runs with the ground faults for a long time, and the running safety of the power distribution network is seriously threatened; therefore, after the single-phase earth fault occurs, how to quickly and accurately identify fault electric information is a prerequisite for fault isolation, elimination and load transfer under the condition of changing the operation working condition of the power distribution network, and has great significance for realizing self-healing control of the power grid.
Chinese patent, publication No.: CN111965475a, publication date: 11 months and 20 days in 2020, a comprehensive fault research and judgment method of a power distribution network based on zero-sequence current distribution characteristics is disclosed, and the fault characteristics of a low-current grounding system are researched by aiming at the analysis of the zero-sequence current distribution characteristics; constructing fault section identification criteria according to the zero sequence current distribution characteristic difference of the sound circuit and the fault circuit; identifying a power distribution network fault section based on a hierarchical clustering algorithm; the fault studying and judging method is based on zero sequence fundamental frequency components, has low sampling requirements and strong noise resistance, and improves the adaptability of the section identification method; the multi-period fault data are used, the difference of zero sequence current amplitudes at two ends of the section is amplified, the anti-interference capability is high, and the reliability of the section identification method is improved; performing hierarchical clustering analysis on the fault data, defining a distance ratio, screening fault lines, and improving the accuracy of the section identification method, wherein the method is simple and visual; this solution has the following problems: the zero sequence current distribution characteristic is limited by the influence of factors such as load change of a circuit, a power grid topological structure, system parameters and the like, so that the accuracy of the fault research and judgment of the power distribution network is low, a hierarchical clustering algorithm is used for identifying the fault section of the power distribution network, the influence of parameter selection and data preprocessing can be caused, the inaccuracy of a clustering result is caused, and the accuracy of the fault section identification is low.
Disclosure of Invention
The invention aims to solve the problem of low accuracy of a power distribution network fault judging method in the prior art, provides a fault judging method based on multiple feature quantities, realizes classification of historical feature samples through a fuzzy membership matching algorithm, obtains types of the historical feature samples and corresponding membership matching centers thereof, calculates distances between real-time feature samples and non-fault membership matching centers and fault membership matching centers respectively through a distance judging method, determines types of the feature samples through the size relation between the real-time feature samples and the distances between the non-fault membership matching centers and the fault membership matching centers, realizes fault judging of the power distribution network, and confirms a protection state through comparison of a real-time monitoring zero sequence voltage value and a preset threshold value, so that the accuracy of fault judging can be remarkably improved.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
the fault studying and judging method based on the multiple characteristic quantities is characterized by comprising the following steps:
s1, acquiring historical fault characteristic quantities of a protected circuit under different operation conditions, constructing a historical characteristic sample based on the historical fault characteristic quantities, and carrying out standardized processing on the historical characteristic sample;
s2, classifying the standardized historical feature samples based on a fuzzy membership matching algorithm to obtain the types of the historical feature samples and membership matching centers corresponding to the types of the historical feature samples;
s3, collecting real-time fault characteristic quantities from protected feeder lines, and constructing a real-time characteristic sample based on the real-time fault characteristic quantities;
s4, carrying out standardization processing on the real-time characteristic sample, and obtaining the distance d between the real-time characteristic sample and the non-fault membership matching center and the fault membership matching center respectively through a distance discrimination method 1g 、d 2g ;
S5, based on d 1g 、d 2g Judging the type of the real-time characteristic sample data, and classifying the current real-time characteristic sample into the historical data of the corresponding type;
s6, monitoring a zero sequence voltage value of the power distribution network in real time, and if the zero sequence voltage of the power distribution network is smaller than a preset threshold value, executing S3; and if the zero sequence voltage of the power distribution network is larger than a preset threshold value, starting a non-setting protection scheme.
In the scheme, the historical feature samples are constructed through the historical fault feature quantity and subjected to standardized processing, classification of the historical feature samples is achieved through the fuzzy membership matching algorithm, the real-time fault feature quantity is collected to construct the real-time feature samples, the distance from the real-time feature samples to the membership matching center is calculated through the distance discrimination method, the types of the real-time feature samples are judged through the obtained distance result, the protection scheme is determined based on the real-time zero sequence voltage value of the power distribution network, the influence of faults on the system can be reduced, and the accuracy of fault research and judgment is improved.
Preferably, the step S1 includes the following steps:
s11, collecting original data, establishing an original data matrix, and performing isotactical transformation on the original data matrix;
s12, carrying out standardization processing on the original data matrix subjected to the isotacticity transformation to obtain a standardized matrix;
s13, establishing a correlation coefficient matrix based on the standardized matrix, and calculating an accumulated variance contribution rate based on the correlation coefficient matrix; and S14, determining data of a preferable historical characteristic sample based on the accumulated variance contribution rate to obtain the preferable historical characteristic sample.
In the scheme, noise and interference in the original data can be eliminated by carrying out homodromous transformation on the original data, so that the subsequent processing is more accurate and reliable; the value ranges of different characteristic quantities can be unified to the same scale through standardization processing, so that dimensional differences among the different characteristic quantities are eliminated, and the data are comparable; establishing a correlation coefficient matrix based on the standardized matrix, and measuring the correlation between different characteristic quantities; the contribution degree of each feature quantity to the total variance can be evaluated by calculating the cumulative variance contribution rate based on the correlation coefficient matrix, so that feature selection and dimension reduction can be realized, the data dimension is reduced, the subsequent processing process is simplified, and the calculation efficiency is improved.
Preferably, the step S2 includes the following steps:
s21, setting a history characteristic sample type and a corresponding membership matching center based on the running condition of the power distribution network;
s22, calculating the membership degree between each history feature sample and the membership matching center based on a fuzzy membership matching algorithm;
s23, determining the type of the historical feature sample based on the membership degree of each historical feature sample, and recording the corresponding membership matching center.
In the scheme, the matching degree of the historical characteristic sample and each type can be evaluated by calculating the membership degree; the type of the sample is determined based on the membership degree of the historical feature sample, and the corresponding membership matching center is recorded, so that important reference information can be provided for subsequent fault research and judgment, classification and marking of the historical feature sample can be realized, a fault judgment model can be conveniently established, fault identification can be conveniently carried out, and the accuracy and reliability of fault research and judgment are improved.
Preferably, the fuzzy membership matching algorithm is implemented by a balanced iterative equation, the balanced iterative equation including:
wherein p is i As a membership center, u ik Representing membership matching samples x k Membership degree belonging to the ith membership matching type, m being a weighted index;
and i is more than or equal to 1 and less than or equal to c, k is more than or equal to 1 and less than or equal to n, wherein c is the number of target categories of membership matching analysis, and I are matrix norms for representing the space distance between a sample to be tested and a membership matching center.
In the scheme, a fuzzy membership matching algorithm realized by a balance iteration equation is adopted to process the data of a plurality of characteristic quantities, so that the accuracy of fault research and judgment can be improved.
Preferably, the history feature sample type includes a fault class and a non-fault class, and the membership matching center includes a fault class membership matching center and a non-fault class membership matching center.
Preferably, the step S4 includes the following steps:
s41, carrying out standardized processing on the real-time characteristic samples to ensure that the data range of each characteristic is consistent;
s42, representing the real-time feature sample as a feature vector, and respectively representing a non-fault class membership matching center and a fault class membership matching center as two feature vectors;
s43, calculating the distance between the real-time feature sample and the non-fault membership matching center and the distance d between the real-time feature sample and the fault membership matching center through a Euclidean distance formula 1g 、d 2g 。
In the scheme, the feature vector can be more accurate and reliable in calculating the distance and the similarity through the standardization processing; by representing the samples and the matching centers as feature vectors, the processing and calculation of the problem can be simplified, and the efficiency of the algorithm can be improved; the distance between the real-time characteristic sample and the class membership matching center is calculated through the Euclidean distance formula, so that the degree of difference between the real-time characteristic sample and the two membership matching centers can be quantified.
Preferably, the euclidean distance formula is:wherein d ig Representing real-time feature samples x gj To membership match centre p gj Is a distance of (3).
In the scheme, the Euclidean distance formula is used for calculating the distance from the real-time characteristic sample to the membership matching center, so that the calculation is simple and visual.
Preferably, the step S5 includes the following steps:
s51, if d 1g <d 2g The real-time characteristic sample data is of a non-fault type, the protected feeder line does not have faults, and the real-time characteristic sample is classified into the non-fault type historical data;
s52, if d 1g >d 2g And if the real-time characteristic sample data is in a fault class, the protected feeder line fails, and the real-time characteristic sample data is classified into the fault class historical data.
In the scheme, by judging d 1g And d 2g The size relation of the system is used for judging the type of the real-time characteristic sample, so that the accuracy can be improved, false alarm is reduced, fault positioning and timely processing can be realized, reasonable classification judgment is carried out in the fault judging process, and the reliability of the system and the efficiency of fault processing are improved.
Preferably, the step S6 includes the following steps:
s61, collecting voltage data of each node of the power distribution network in real time, wherein the voltage data comprise real-time data of fault occurrence time;
s62, calculating a voltage average value as zero sequence voltage based on the voltage data;
s63, judging the magnitude relation between the zero sequence voltage and a preset threshold value, and executing S64 if the zero sequence voltage is larger than the preset threshold value; if the zero sequence voltage is smaller than a preset threshold value, executing S3;
s64, fault location is carried out through a fault location method based on the voltage data;
s65, sending an alarm and processing faults, and executing S3.
In the scheme, the zero sequence voltage is calculated through the real-time voltage data, whether fault positioning and fault handling are carried out or not is determined through judging the magnitude relation between the zero sequence voltage and the preset threshold value, accurate fault information can be provided, measures can be timely taken for maintenance treatment, and the operation efficiency and reliability of the power distribution network are improved.
Preferably, the normalization process includes the steps of:
for each feature quantity, calculating the average value of all samples, and subtracting the feature value of each sample from the average value so that the average value of the feature quantity is zero;
the standard deviation of each feature quantity is calculated, and the feature value of each sample is divided by the standard deviation thereof so that the variance of the feature quantity is one.
In the scheme, the average value of the characteristic quantity is subtracted from the characteristic value of each sample, so that the average value of the characteristic quantity is zero, the offset influence of the characteristic quantity can be eliminated, and the relative difference of the characteristic values is highlighted; by dividing the characteristic value of each sample by the standard deviation of the characteristic value, so that the variance of the characteristic value is one, different characteristic values can have similar importance, and the influence of a large numerical range of some characteristic values on an analysis result is avoided.
The invention has the beneficial effects that: the method comprises the steps of constructing a historical feature sample through historical fault feature quantity, carrying out standardized processing, realizing classification of the historical feature sample by using a fuzzy membership matching algorithm, collecting real-time fault feature quantity to construct a real-time feature sample, calculating the distance from the real-time feature sample to a membership matching center through a distance discrimination method, judging the type of the real-time feature sample through an obtained distance result, determining a protection scheme based on a real-time zero sequence voltage value of a power distribution network, reducing the influence of faults on a system, and improving the accuracy of fault research and judgment; the type of the real-time characteristic sample is judged by judging the size relation from the real-time characteristic sample to the non-fault class membership matching center and the fault class membership matching center, so that the accuracy can be improved, false alarm is reduced, fault positioning and timely processing can be realized, reasonable classification judgment is carried out in the process of power distribution network fault research and judgment, and the reliability of a system and the efficiency of fault processing are improved.
The foregoing summary is merely an overview of the present invention, and is intended to be implemented in accordance with the teachings of the present invention in order that the same may be more fully understood, and in order that the same or additional objects, features and advantages of the present invention may be more fully understood.
Drawings
FIG. 1 is a flow chart of a fault diagnosis method based on multiple feature quantities according to the present invention;
FIG. 2 is a schematic diagram of a set of historical feature samples according to the present invention;
FIG. 3 is a graph showing a two-dimensional fault signature spatial distribution of a sample set to be tested according to the present invention;
FIG. 4 is a three-dimensional fault signature spatial distribution diagram of a sample set to be tested according to the present invention;
FIG. 5 is a simulation model diagram of a fault diagnosis method based on multiple feature quantities according to the present invention;
FIG. 6 is a principal component contribution graph of each fault feature of the two-dimensional visualization of the present invention;
FIG. 7 is a diagram of a distribution of a sample set to be tested in a high-dimensional fault signature space in accordance with the present invention;
FIG. 8 is a graph showing the results of verification experiments of fault signature values below 800 Ω in various grounding modes of the present invention;
FIG. 9 shows the results of the fault feature and arc ground fault verification experiments under various grounding modes of the invention, wherein the fault feature is more than 800 omega.
Detailed Description
Other features, objects and advantages of the present invention will become more apparent upon reading of the detailed description of non-limiting embodiments made with reference to the following drawings. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to designate like parts throughout the figures.
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings and examples, it being understood that the detailed description herein is merely a preferred embodiment of the present invention, which is intended to illustrate the present invention, and not to limit the scope of the invention, as all other embodiments obtained by those skilled in the art without making any inventive effort fall within the scope of the present invention.
Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although a flowchart depicts operations (or steps) as a sequential process, many of the operations (or steps) can be performed in parallel, concurrently, or at the same time. Furthermore, the order of the operations may be rearranged. The process may be terminated when its operations are completed, but may have additional steps not included in the figures; the processes may correspond to methods, functions, procedures, subroutines, and the like.
Examples:
as shown in fig. 1, a fault diagnosis method based on multiple feature quantities includes the following steps:
s1, collecting historical fault characteristic quantities of a protected circuit under different operation conditions, constructing a historical characteristic sample based on the historical fault characteristic quantities, and carrying out standardized processing on the historical characteristic sample.
Specifically, S1 includes the steps of:
s11, collecting original data, establishing an original data matrix, and performing homodromous transformation on the original data matrix;
s12, carrying out standardization processing on the original data matrix subjected to the isotacticity transformation to obtain a standardized matrix;
s13, establishing a correlation coefficient matrix based on the standardized matrix, and calculating an accumulated variance contribution rate based on the correlation coefficient matrix;
s14, determining data of a preferred historical characteristic sample based on the accumulated variance contribution rate to obtain the preferred historical characteristic sample.
In the embodiment, noise and interference in the original data can be eliminated by performing homodromous transformation on the original data, so that the subsequent processing is more accurate and reliable; the value ranges of different characteristic quantities can be unified to the same scale through standardization processing, so that dimensional differences among the different characteristic quantities are eliminated, and the data are comparable; establishing a correlation coefficient matrix based on the standardized matrix, and measuring the correlation between different characteristic quantities; the contribution degree of each feature quantity to the total variance can be evaluated by calculating the cumulative variance contribution rate based on the correlation coefficient matrix, so that feature selection and dimension reduction can be realized, the data dimension is reduced, the subsequent processing process is simplified, and the calculation efficiency is improved.
Specifically, n samples are provided, each sample has p original fault feature values, and system evaluation and data acquisition are performed on the original data to obtain an original data matrix x= (X) ij ) n×p =[X 1 ,X 2 ,…,X p ]Wherein x is ij Is the value of the ith scheme with respect to the jth index.
Specifically, principal component analysis is performed on the original fault feature quantities, wherein the principal component analysis is actually that p original variables are converted into linear combination problems for discussing p variables, so as to obtain a plurality of fault feature principal components y 1 ,y 2 ,…,y m (m is less than or equal to p), the main information quantity in the original fault characteristic quantity is reserved as far as possible for each fault characteristic main component, and the fault characteristic sample set is fully reflected but the fault characteristic main components are mutually independent; the principal component analysis is basically to perform orthogonal transformation on X to find the linear combination Y of the original variables:
the formula needs to meet the following conditions: the sum of squares of the coefficients of each principal component is 1, namely:
the principal components are not related to each other, and the eigenvalue λi of the ith principal component yi is the variance of the principal component, namely:
the variance of the main components is gradually decreased, and the respective importance degrees are also gradually decreased, namely:
var(Y 1 )≥var(Y 2 )≥…≥var(Y p )
the above conditions can judge the importance proportion of each main component in the actual problem, the first main component is the largest, and then the first main component is reduced in turn; to simplify the problem for further analysis, several principal components with the greatest variance are typically chosen, the remainder being ignored.
Specifically, the elements in the original data matrix X are subjected to isotactic transformation, so that the variation trend of each element is synchronized as much as possible, and the vector in the original data matrix X is subjected to standardization processing for eliminating the influence of dimension and normalizing the measurement value range of the evaluation indexWherein->Obtaining a standardized matrix Z:
establishing a correlation coefficient matrix R= (R) on the basis of a standardized matrix ij ) p×p The method is used for representing the degree of interrelationship among elements in a matrix, and the expression is as follows:
cov(z i ,z j ) As index z i And z j Is a covariance of (2); characteristic equation R-lambda I of decorrelation coefficient matrix R p The value of =0, p eigenvalues λ 1 ≥λ 2 ≥…≥λ p 0 or more, wherein lambda i =var(Y i ) Then, the variance contribution ratio of the principal component can be found as:
variance contribution ratio w i The percentage of the original variable information contained in the ith principal component is reflected, so that the first principal component variance contribution rate is maximum, and the second principal component is gradually decreased; the cumulative variance contribution is expressed as follows:
in order to achieve the purpose of reducing dimension and simplifying analysis, if the accumulated variance contribution rate rho of the first m main components is more than or equal to 75%, the first m main components which can be reserved are used as new variables containing most historical sample data, wherein the contribution degree of each original evaluation index is different, the first m main components are analyzed, the correlation coefficient between the first m main components and the original evaluation index is represented by a feature vector, the larger the value of the correlation coefficient is represented by the feature vector, the stronger the correlation is represented, the larger the contribution degree of the evaluation index to the main components is, the description data information of the evaluation index can be better is indicated, the optimal purpose is achieved, and the fault research judgment optimal feature sample library is obtained.
Specifically, as shown in fig. 2, a schematic diagram is formed by a history feature sample set: defining a plurality of fault characteristic information in each case as a history characteristic sample, and setting s samples constructed by fault characteristic quantities under the kth operating condition as: x's' k =(x' k1 ,...,x' kj ,...,x' ks ) T Wherein: x's' k1 ,x' kj ,x' ks The specific values of the s fault feature quantities extracted under the kth operating condition are respectively.
S2, classifying the standardized historical feature samples based on a fuzzy membership matching algorithm to obtain the types of the historical feature samples and the membership matching centers corresponding to the types of the historical feature samples.
Specifically, S2 includes the steps of:
s21, setting a history characteristic sample type and a corresponding membership matching center based on the running condition of the power distribution network;
s22, calculating the membership degree between each history feature sample and the membership matching center based on a fuzzy membership matching algorithm;
s23, determining the type of the historical feature sample based on the membership degree of each historical feature sample, and recording the corresponding membership matching center.
In this embodiment, the matching degree of the history feature sample and each type can be evaluated by calculating the membership degree; the type of the sample is determined based on the membership degree of the historical feature sample, and the corresponding membership matching center is recorded, so that important reference information can be provided for subsequent fault research and judgment, classification and marking of the historical feature sample can be realized, a fault judgment model can be conveniently established, fault identification can be conveniently carried out, and the accuracy and reliability of fault research and judgment are improved.
Specifically, when fault feature similarity fuzzy membership matching is performed, any sample x to be tested k Each can be represented as a determined coordinate point in a multi-dimensional coordinate space; establishing a high-dimensional fault feature space according to the number of fault feature quantities contained in a sample to be tested, wherein the closer the space positions occupied by different sample points are, the higher the similarity degree of the essential characteristics is; the s fault feature quantities in the sample to be tested have a one-to-one correspondence with the s coordinate axes in the feature space.
Specifically, the establishment of the high-dimensional fault feature space realizes the imaging of the similarity degree between the samples to be tested, as shown in fig. 3, the distribution situation of the samples to be tested in the s=2 feature space is shown, namely, the two-dimensional fault feature space distribution of the sample set to be tested is shown; as shown in fig. 4, the distribution of the samples to be tested in the s=3 feature space, that is, the three-dimensional fault feature space distribution of the sample set to be tested is shown.
Specifically, similarity membership matching analysis is performed on the optimized feature quantity sample library, and each sample x to be detected is realized by optimizing an objective function and a balance iteration equation 1 ,…,x n Is used to divide the samples into a fault class and a non-fault class: specifically, the objective function is:wherein p is i For clerical effectBelongs to the center; u (u) ik ∈[0,1]Representing membership matching samples x k Membership degree belonging to the ith membership matching type, satisfy +.>m is a weighted index, taking m=2.
Specifically, the equilibrium iterative equation includes:
equilibrium iteration equation 1:
equilibrium iteration equation 2:
wherein c is the number of target categories subject to matching analysis; and the I & ltI & gt is a matrix norm for representing the space distance between the sample to be tested and the membership matching center.
Specifically, the equilibrium iterative equation can be described as: firstly, randomly selecting two sets of membership matching centers p i Will p i Substituting the obtained membership degree u into the equilibrium iterative equation 2 ik The calculated membership degree u ik Substituted equilibrium iterative equation 1 is applied to membership center p i Make corrections whenWhen the iteration process is terminated, epsilon is the iteration stop threshold, and epsilon=1.0e-6 is preferable.
In this embodiment, a fuzzy membership matching algorithm implemented by a balanced iterative equation is used to process data of a plurality of feature quantities, so that accuracy of fault determination can be improved.
Specifically, the history feature sample types comprise a fault class and a non-fault class, and the membership matching center comprises a fault class membership matching center and a non-fault class membership matching center.
S3, collecting real-time fault characteristic quantities from the protected feeder line, and constructing a real-time characteristic sample based on the real-time fault characteristic quantities.
S4, carrying out standardization processing on the real-time characteristic sample, and obtaining the distance d between the real-time characteristic sample and the non-fault membership matching center and the fault membership matching center respectively through a distance discrimination method 1g 、d 2g 。
Specifically, S4 includes the steps of:
s41, carrying out standardized processing on the real-time characteristic samples to ensure that the data range of each characteristic is consistent;
s42, representing the real-time feature sample as a feature vector, and respectively representing a non-fault class membership matching center and a fault class membership matching center as two feature vectors;
s43, calculating the distance between the real-time feature sample and the non-fault membership matching center and the distance d between the real-time feature sample and the fault membership matching center through a Euclidean distance formula 1g 、d 2g 。
In the embodiment, the feature vector can be more accurate and reliable in calculating the distance and the similarity through the standardization processing; by representing the samples and the matching centers as feature vectors, the processing and calculation of the problem can be simplified, and the efficiency of the algorithm can be improved; the distance between the real-time characteristic sample and the membership matching center is calculated through the Euclidean distance formula, so that the degree of difference between the real-time characteristic sample and the two membership matching centers can be quantified.
Specifically, the Euclidean distance formula is:wherein d ig Representing real-time feature samples x gj To membership match centre p gj Is a distance of (3).
In this embodiment, the Euclidean distance formula is used to calculate the distance from the real-time feature sample to the membership matching center, which is simple and intuitive.
S5, based on d 1g 、d 2g And (3) judging the type of the real-time characteristic sample data, and classifying the current real-time characteristic sample into the historical data of the corresponding type.
Specifically, S5 includes the steps of:
s51, if d 1g <d 2g The real-time characteristic sample data is of a non-fault type, the protected feeder line does not have faults, and the real-time characteristic sample is classified into the non-fault type historical data;
s52, if d 1g >d 2g And if the real-time characteristic sample data is in a fault class, the protected feeder line fails, and the real-time characteristic sample data is classified into the fault class historical data.
In the present embodiment, by judging d 1g And d 2g The size relation of the system is used for judging the type of the real-time characteristic sample, so that the accuracy can be improved, false alarm is reduced, fault positioning and timely processing can be realized, reasonable classification judgment is carried out in the fault judging process, and the reliability of the system and the efficiency of fault processing are improved.
S6, monitoring a zero-sequence voltage value of the power distribution network in real time, and starting a non-setting protection scheme if the zero-sequence voltage of the power distribution network is greater than a preset threshold value; and if the zero sequence voltage of the power distribution network is smaller than a preset threshold value, executing S3.
Specifically, in order to further improve the performance of the fault line selection scheme, a zero sequence voltage value in the power distribution network is selected as a triggering condition of the protection scheme, and U is taken out 0set And the E is a system phase voltage value, and when the power distribution zero sequence voltage value meets the triggering condition of the protection scheme, the non-setting protection scheme is started to carry out protection judgment.
Preferably, S6 comprises the steps of:
s61, collecting voltage data of each node of the power distribution network in real time, wherein the voltage data comprise real-time data of fault occurrence time;
s62, calculating a voltage average value based on the voltage data to serve as zero sequence voltage;
s63, judging the magnitude relation between the zero sequence voltage and a preset threshold value, and executing S64 if the zero sequence voltage is greater than the preset threshold value; if the zero sequence voltage is smaller than a preset threshold value, executing S3;
s64, performing fault location through a fault location method based on the voltage data;
s65, sending an alarm and processing faults, and executing S3.
In the embodiment, the zero sequence voltage is calculated through the real-time voltage data, and whether fault positioning and fault handling are performed or not is determined by judging the magnitude relation between the zero sequence voltage and the preset threshold value, so that accurate fault information can be provided, measures can be taken in time to perform maintenance processing, and the operation efficiency and reliability of the power distribution network are improved.
Specifically, in this embodiment, an impedance distribution method is combined with a machine learning method to perform fault location of the power distribution network, where the principle of the impedance distribution method is as follows: locating the fault based on the different impedances before and after the fault; the principle of the machine learning method is based on the modern computer technology and the data processing technology, and diagnosis and positioning are carried out on faults through training by establishing an expert system or an artificial intelligent model; the impedance learning method for fault location has the following advantages: the positioning accuracy is high, and the sensitivity to external interference is low; the machine learning method for fault location has the following advantages: and a large amount of complex data can be better processed, and automatic real-time positioning is realized.
Specifically, the fault location using the impedance distribution method in combination with the machine learning method includes the steps of:
feature extraction: extracting characteristic data of the real-time voltage data;
and (3) data marking: marking the historical fault feature samples, namely marking the fault positions marked by each historical fault feature sample; establishing a machine learning model: taking the extracted characteristic data as input, taking the fault position of the historical fault characteristic as output, performing model training, and establishing a machine learning model;
fault location: calculating impedance values of all nodes of the power distribution network based on real-time voltage data, obtaining preliminary fault position estimation based on impedance distribution, and then taking the estimation result as input, and carrying out further fault positioning prediction through a machine learning model;
fault location assessment: and comparing the actual fault position with the fault position predicted by the machine learning model, evaluating the accuracy and performance of the model, and optimizing and adjusting the model according to the requirement.
Specifically, since various fault characteristic amounts differ in dimension, it is necessary to perform normalization processing for each fault sample, the normalization processing including the steps of:
for each feature quantity, calculating the average value of all samples, and subtracting the feature value of each sample from the average value so that the average value of the feature quantity is zero;
the standard deviation of each feature quantity is calculated, and the feature value of each sample is divided by the standard deviation thereof so that the variance of the feature quantity is one.
Specifically, the normalization processing calculation mode is as follows:
in which x is kj Sample data after standardized processing;the sample mean value of the j-th fault characteristic quantity; s (x' j ) Sample standard deviation of the j-th fault feature quantity; after normalization, the historical feature sample set is denoted as X n×s ={x 1 ,...,x k ,...,x n }。
In this embodiment, by subtracting the average value of the feature values from the feature value of each sample so that the average value of the feature values is zero, the offset effect of the feature values can be eliminated, and the relative difference of the feature values is highlighted; by dividing the characteristic value of each sample by the standard deviation of the characteristic value, so that the variance of the characteristic value is one, different characteristic values can have similar importance, and the influence of a large numerical range of some characteristic values on an analysis result is avoided.
Specifically, in the embodiment, simulation analysis is performed on a typical 35kV power distribution network by adopting EMTP electromagnetic transient simulation software, and a simulation model diagram is shown in FIG. 5; the simulation model of the embodiment is realized by selecting 4 feeder lines, 3 overhead lines and 1 cable line connected with a bus, wherein a measuring point is arranged at the inlet of the feeder line 4, fault characteristic quantities are extracted, and specific parameters of the lines are shown in table 1:
TABLE 1 line parameters
By varying the fault resistance R f Fault current phase angle θ and fault location X f Extracting a historical characteristic sample of the feeder line 4; simulation takes a lot of data to examine the behavior of the protection device installed at the outlet of line 4 when single-phase earth faults occurred in line 3 (external fault) and line 4 (internal fault), respectively, and the historical feature samples are listed in tables 2 and 3:
TABLE 2 historical steady state feature sample set for neutral ungrounded systems
TABLE 3 transient characteristics sample set of neutral ungrounded system history
*X f Indicating the distance between the fault point and the bus
In the feeder line grounding protection, the historical data collected by the protection device is classified by taking multi-source fault information as an index, and the sizes of the dimensions, units and numerical values of various fault characteristic quantities are different, so that in order to obtain a better membership matching result, normalization pretreatment is required to be carried out on the original data, and the obtained membership matching sample data after treatment is shown in a table 4;
TABLE 4 normalized pre-processed membership match sample data
Carrying out principal component analysis operation on the normalized pretreatment data of the sample set to obtain characteristic values, contribution rates and accumulated contribution rates of the principal components; as shown in the table, the cumulative contribution rate of the first main component and the second main component is up to 86%, so that the requirement of most of original data information is met, and only the first two main components can be taken for analysis;
TABLE 5 principal component eigenvalue distribution of fault eigenvalues
Fault signature principal component | Eigenvalue lambda i | Contribution rate omega i | Cumulative contribution rate ρ i |
y 1 | 5.5162 | 0.5516 | 0.5516 |
y 2 | 3.0848 | 0.3085 | 0.8601 |
y 3 | 0.6867 | 0.0687 | 0.9288 |
y 4 | 0.3459 | 0.0346 | 0.9634 |
y 5 | 0.2214 | 0.0221 | 0.9855 |
y 6 | 0.1056 | 0.0106 | 0.9961 |
y 7 | 0.0279 | 0.0028 | 0.9989 |
y 8 | 0.0068 | 0.0007 | 0.9996 |
y 9 | 0.0027 | 0.0003 | 0.9999 |
y 10 | 0.0006 | 0.0001 | ≈1 |
As shown in FIG. 6, the principal component contribution graph of each fault feature is visualized in two dimensions, the most optimal fault feature with larger contribution degree is selected, and a new fault test sample set is formed by x k1 ,x k3 ,x k5 ,x k6 And x k7 。
The normalized sample data in the table 4 is calculated according to the objective function, the membership matrix is repeatedly compared with a preset threshold value, and the membership degree data of each sample obtained finally are listed in the table 6;
TABLE 6 sample membership
According to the membership value in the table, it can be found that the history sample X is subjected to normalized pretreatment 1 ~X 8 Is divided into non-fault classes, X 9 ~X 16 The simulation system is divided into fault classes, and is consistent with simulation conditions; therefore, the multi-source fault characteristic quantity samples can be effectively and accurately classified by calculating the membership degree of each sample class;
TABLE 7 membership center coordinates of historical sample data
Calculating to obtain a non-fault class membership center p 1 With fault class membership centre p 2 As shown in table 7; as shown in FIG. 7, the location of the membership matching center in the fault signature space effectively characterizes the space of the fault class history signature sample and the non-fault class history signature sampleDistribution conditions;
TABLE 8 real-time characterization sample of neutral ungrounded system portion
TABLE 9 simulation results
As shown in Table 9, european space distance is adopted as a fault protection criterion, and a fault sample can still be accurately identified under the condition that the acquisition process of a real-time characteristic sample is influenced by interference factors, so that fault judgment is realized.
As shown in fig. 8, which shows the verification experiment results of the fault feature quantity below 800 Ω in multiple modes, and as shown in fig. 9, which shows the verification experiment results of the fault feature quantity above 800 Ω and arc ground fault in multiple ground modes, it can be seen from fig. 8 and 9 that the fault diagnosis method provided by the embodiment combines the advantages of multiple fault feature quantities in different fault scenes, and has higher accuracy and adaptability compared with the fault diagnosis method based on single fault feature quantity; the number of the preferable fault feature quantities is reduced to slightly reduce the fault judging accuracy, and the number of the non-preferable fault feature quantities is increased to hardly affect the fault judging accuracy, so that the effect of the multi-fault feature quantity preferable in screening the applicable fault feature quantities for different fault scenes is reflected.
The beneficial effects of this embodiment are: the method comprises the steps of constructing a historical feature sample through historical fault feature quantity, carrying out standardized processing, realizing classification of the historical feature sample by using a fuzzy membership matching algorithm, collecting real-time fault feature quantity to construct a real-time feature sample, calculating the distance from the real-time feature sample to a membership matching center through a distance discrimination method, judging the type of the real-time feature sample through an obtained distance result, determining a protection scheme based on a real-time zero sequence voltage value of a power distribution network, reducing the influence of faults on a system, and improving the accuracy of fault research and judgment; the type of the real-time characteristic sample is judged by judging the size relation from the real-time characteristic sample to the non-fault class membership matching center and the fault class membership matching center, so that the accuracy can be improved, false alarm is reduced, fault positioning and timely processing can be realized, reasonable classification judgment is carried out in the process of power distribution network fault research and judgment, and the reliability of a system and the efficiency of fault processing are improved.
The above embodiments are preferred embodiments of the fault diagnosis method based on multiple feature values, and are not intended to limit the scope of the present invention, which includes but is not limited to the embodiments, and equivalent changes of shape and structure according to the present invention are all within the scope of the present invention.
Claims (10)
1. The fault studying and judging method based on the multiple characteristic quantities is characterized by comprising the following steps:
s1, acquiring historical fault characteristic quantities of a protected circuit under different operation conditions, constructing a historical characteristic sample based on the historical fault characteristic quantities, and carrying out standardized processing on the historical characteristic sample;
s2, classifying the standardized historical feature samples based on a fuzzy membership matching algorithm to obtain the types of the historical feature samples and membership matching centers corresponding to the types of the historical feature samples;
s3, collecting real-time fault characteristic quantities from protected feeder lines, and constructing a real-time characteristic sample based on the real-time fault characteristic quantities;
s4, carrying out standardization processing on the real-time characteristic sample, and obtaining the distance d between the real-time characteristic sample and the non-fault membership matching center and the fault membership matching center respectively through a distance discrimination method 1g 、d 2g ;
S5, based on d 1g 、d 2g Judging the type of the real-time characteristic sample data according to the comparison result of the database, and classifying the current real-time characteristic sample into the historical data of the corresponding typeIn (a) and (b);
s6, monitoring a zero sequence voltage value of the power distribution network in real time, and if the zero sequence voltage of the power distribution network is smaller than a preset threshold value, executing S3; and if the zero sequence voltage of the power distribution network is larger than a preset threshold value, starting a non-setting protection scheme.
2. The method for fault diagnosis and judgment based on multiple feature values according to claim 1, wherein S1 comprises the steps of:
s11, collecting original data, establishing an original data matrix, and performing isotactical transformation on the original data matrix;
s12, carrying out standardization processing on the original data matrix subjected to the isotacticity transformation to obtain a standardized matrix;
s13, establishing a correlation coefficient matrix based on the standardized matrix, and calculating an accumulated variance contribution rate based on the correlation coefficient matrix;
and S14, determining data of a preferable historical characteristic sample based on the accumulated variance contribution rate to obtain the preferable historical characteristic sample.
3. The method for fault diagnosis and judgment based on multiple feature values according to claim 1, wherein S2 comprises the steps of:
s21, setting a history characteristic sample type and a corresponding membership matching center based on the running condition of the power distribution network;
s22, calculating the membership degree between each history feature sample and the membership matching center based on a fuzzy membership matching algorithm;
s23, determining the type of the historical feature sample based on the membership degree of each historical feature sample, and recording the corresponding membership matching center.
4. A multi-feature-based fault finding method as claimed in claim 1 or 3, wherein the fuzzy membership matching algorithm is implemented by a balanced iterative equation comprising:
wherein p is i As a membership center, u ik Representing membership matching samples x k Membership degree belonging to the ith membership matching type, m being a weighted index; />Wherein c is the number of target categories of the membership matching analysis, and I I.I. I are matrix norms for representing the space distance between the sample to be tested and the membership matching center.
5. A method of fault diagnosis based on multiple feature values according to claim 1 or 3, wherein the historical feature sample types include a fault class and a non-fault class, and the membership matching center includes a fault class membership matching center and a non-fault class membership matching center.
6. The method for fault diagnosis and judgment based on multiple feature values according to claim 1, wherein S4 comprises the steps of:
s41, carrying out standardized processing on the real-time characteristic samples to ensure that the data range of each characteristic is consistent;
s42, representing the real-time feature sample as a feature vector, and respectively representing a non-fault class membership matching center and a fault class membership matching center as two feature vectors;
s43, calculating the distance d between the real-time characteristic sample and the non-fault class membership matching center through a Euclidean distance formula 1g And distance d between real-time characteristic sample and fault class membership matching center 2g 。
7. The method for fault diagnosis and judgment based on multiple feature values according to claim 6, wherein the euclidean distance formula is:wherein d ig Representing real-time feature samples x gj To non-fault class membership matching centre p gj Is a distance of (3).
8. The method for fault diagnosis and judgment based on multiple feature values according to claim 1, wherein S5 comprises the steps of:
s51, if d 1g <d 2g The real-time characteristic sample data is of a non-fault type, the protected feeder line does not have faults, and the real-time characteristic sample is classified into the non-fault type historical data;
s62, if d 1g >d 2g And if the real-time characteristic sample data is in a fault class, the protected feeder line fails, and the real-time characteristic sample data is classified into the fault class historical data.
9. The method for fault diagnosis and judgment based on multiple feature values according to claim 1, wherein S6 comprises the steps of:
s61, collecting voltage data of each node of the power distribution network in real time, wherein the voltage data comprise real-time data of fault occurrence time;
s62, calculating a voltage average value as zero sequence voltage based on the voltage data;
s63, judging the magnitude relation between the zero sequence voltage and a preset threshold value, and executing S64 if the zero sequence voltage is larger than the preset threshold value; if the zero sequence voltage is smaller than a preset threshold value, executing S3;
s64, fault location is carried out through a fault location method based on the voltage data;
s65, sending an alarm and processing faults, and executing S3.
10. The multi-feature-based fault finding method as claimed in claim 1 or 2 or 6, wherein the normalization process includes the steps of:
for each feature quantity, calculating an average value of all samples, and subtracting the feature value of each sample from the average value so that the average value of the feature quantity is 0;
the standard deviation of each feature quantity is calculated, and the feature value of each sample is divided by the standard deviation thereof so that the variance of the feature quantity is 1.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310757582.7A CN116990633A (en) | 2023-06-25 | 2023-06-25 | Fault studying and judging method based on multiple characteristic quantities |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310757582.7A CN116990633A (en) | 2023-06-25 | 2023-06-25 | Fault studying and judging method based on multiple characteristic quantities |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116990633A true CN116990633A (en) | 2023-11-03 |
Family
ID=88530997
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310757582.7A Pending CN116990633A (en) | 2023-06-25 | 2023-06-25 | Fault studying and judging method based on multiple characteristic quantities |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116990633A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117791597A (en) * | 2024-02-23 | 2024-03-29 | 广东电网有限责任公司广州供电局 | Power distribution network fault self-healing method and system based on machine learning |
-
2023
- 2023-06-25 CN CN202310757582.7A patent/CN116990633A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117791597A (en) * | 2024-02-23 | 2024-03-29 | 广东电网有限责任公司广州供电局 | Power distribution network fault self-healing method and system based on machine learning |
CN117791597B (en) * | 2024-02-23 | 2024-05-28 | 广东电网有限责任公司广州供电局 | Power distribution network fault self-healing method and system based on machine learning |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112699913A (en) | Transformer area household variable relation abnormity diagnosis method and device | |
CN111768082A (en) | Power equipment state evaluation method based on big data analysis | |
CN110609200B (en) | Power distribution network earth fault protection method based on fuzzy metric fusion criterion | |
CN105759201B (en) | High-voltage circuitbreaker self-diagnosing method based on exceptional sample identification | |
CN111596167A (en) | Fault section positioning method and device based on fuzzy c-means clustering algorithm | |
CN109829497B (en) | Supervised learning-based station area user identification and discrimination method | |
CN109389325B (en) | Method for evaluating state of electronic transformer of transformer substation based on wavelet neural network | |
CN109409425B (en) | Fault type identification method based on neighbor component analysis | |
CN112200694B (en) | Dominant instability mode identification model construction and application method based on graph neural network | |
CN111382809B (en) | Isolating switch mechanical fault diagnosis method based on motor output power | |
CN116990633A (en) | Fault studying and judging method based on multiple characteristic quantities | |
CN112964959A (en) | Power distribution network earth fault line selection method based on correlation coefficient measurement criterion | |
CN109710661A (en) | Based on Global Genetic Simulated Annealing Algorithm to the method for high-pressure heater state analysis | |
CN110569876A (en) | Non-invasive load identification method and device and computing equipment | |
CN110927527A (en) | Transient fault sensing method based on transient energy double channels and information fusion | |
CN111582406A (en) | Power equipment state monitoring data clustering method and system | |
CN114062850B (en) | Double-threshold power grid early fault detection method | |
CN111273125A (en) | RST-CNN-based power cable channel fault diagnosis method | |
CN112684295A (en) | Power distribution network fault line selection method and system under high permeability based on similarity separation degree | |
CN112801135B (en) | Fault line selection method and device for power plant service power system based on characteristic quantity correlation | |
CN111856209A (en) | Power transmission line fault classification method and device | |
CN113125903A (en) | Line loss anomaly detection method, device, equipment and computer-readable storage medium | |
CN117150244A (en) | Intelligent power distribution cabinet state monitoring method and system based on electrical parameter analysis | |
CN109389313B (en) | Fault classification diagnosis method based on weighted neighbor decision | |
CN116644358A (en) | Power system transient stability evaluation method based on Bayesian convolutional neural network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |