CN113468461A - Oil-immersed transformer fault diagnosis method based on support vector machine and genetic algorithm - Google Patents
Oil-immersed transformer fault diagnosis method based on support vector machine and genetic algorithm Download PDFInfo
- Publication number
- CN113468461A CN113468461A CN202010236460.XA CN202010236460A CN113468461A CN 113468461 A CN113468461 A CN 113468461A CN 202010236460 A CN202010236460 A CN 202010236460A CN 113468461 A CN113468461 A CN 113468461A
- Authority
- CN
- China
- Prior art keywords
- svm
- fault
- transformer
- oil
- support vector
- 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.)
- Withdrawn
Links
- 238000012706 support-vector machine Methods 0.000 title claims abstract description 73
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000003745 diagnosis Methods 0.000 title claims abstract description 45
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 34
- 230000002068 genetic effect Effects 0.000 title claims abstract description 26
- 239000007789 gas Substances 0.000 claims abstract description 40
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 claims abstract description 29
- 238000012549 training Methods 0.000 claims abstract description 17
- HSFWRNGVRCDJHI-UHFFFAOYSA-N alpha-acetylene Natural products C#C HSFWRNGVRCDJHI-UHFFFAOYSA-N 0.000 claims abstract description 15
- 238000002790 cross-validation Methods 0.000 claims abstract description 13
- 125000002534 ethynyl group Chemical group [H]C#C* 0.000 claims abstract description 13
- 238000012360 testing method Methods 0.000 claims abstract description 13
- 239000005977 Ethylene Substances 0.000 claims abstract description 11
- UFHFLCQGNIYNRP-UHFFFAOYSA-N Hydrogen Chemical compound [H][H] UFHFLCQGNIYNRP-UHFFFAOYSA-N 0.000 claims abstract description 11
- 229910052739 hydrogen Inorganic materials 0.000 claims abstract description 11
- 239000001257 hydrogen Substances 0.000 claims abstract description 11
- VGGSQFUCUMXWEO-UHFFFAOYSA-N Ethene Chemical compound C=C VGGSQFUCUMXWEO-UHFFFAOYSA-N 0.000 claims abstract description 10
- OTMSDBZUPAUEDD-UHFFFAOYSA-N Ethane Chemical compound CC OTMSDBZUPAUEDD-UHFFFAOYSA-N 0.000 claims abstract description 9
- 238000004868 gas analysis Methods 0.000 claims abstract description 4
- 108090000623 proteins and genes Proteins 0.000 claims description 12
- 238000005457 optimization Methods 0.000 claims description 11
- 230000009977 dual effect Effects 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 238000013021 overheating Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- 238000000354 decomposition reaction Methods 0.000 claims description 4
- 238000012216 screening Methods 0.000 claims description 4
- 239000004215 Carbon black (E152) Substances 0.000 claims description 3
- AZFKQCNGMSSWDS-UHFFFAOYSA-N MCPA-thioethyl Chemical compound CCSC(=O)COC1=CC=C(Cl)C=C1C AZFKQCNGMSSWDS-UHFFFAOYSA-N 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 3
- 238000010891 electric arc Methods 0.000 claims description 3
- 229930195733 hydrocarbon Natural products 0.000 claims description 3
- 150000002430 hydrocarbons Chemical class 0.000 claims description 3
- 239000011810 insulating material Substances 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 7
- 230000006870 function Effects 0.000 description 10
- CURLTUGMZLYLDI-UHFFFAOYSA-N Carbon dioxide Chemical compound O=C=O CURLTUGMZLYLDI-UHFFFAOYSA-N 0.000 description 4
- 238000013473 artificial intelligence Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 4
- 230000035772 mutation Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 229910002092 carbon dioxide Inorganic materials 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- UGFAIRIUMAVXCW-UHFFFAOYSA-N Carbon monoxide Chemical compound [O+]#[C-] UGFAIRIUMAVXCW-UHFFFAOYSA-N 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000001569 carbon dioxide Substances 0.000 description 1
- 229910002091 carbon monoxide Inorganic materials 0.000 description 1
- 229920002678 cellulose Polymers 0.000 description 1
- 239000001913 cellulose Substances 0.000 description 1
- 238000004587 chromatography analysis Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000002405 diagnostic procedure Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000009413 insulation Methods 0.000 description 1
- 230000010076 replication Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Physiology (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Algebra (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Mathematical Optimization (AREA)
- Genetics & Genomics (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Housings And Mounting Of Transformers (AREA)
Abstract
An oil-immersed transformer fault diagnosis method based on a support vector machine and a genetic algorithm analyzes five common gases in transformer oil by utilizing dissolved GAs analysis based on SVM and GA, wherein the five common gases are methane, ethane, ethylene, acetylene and hydrogen; s2, establishing a fitness function based on five-fold cross validation by taking the ratio of the five dissolved gases as input, and taking the accuracy as a judgment standard of the function; s3, finding out a more optimal SVM parameter C/sigma by using cross validation and GA; and S4, training the model according to the parameters C/sigma of the optimal SVM, and verifying the model performance by taking the fault case as a test set. The oil-immersed transformer fault diagnosis method based on the support vector machine and the genetic algorithm has high diagnosis efficiency and accuracy, can effectively and accurately identify the transformer fault, and has an excellent use effect.
Description
Technical Field
The invention relates to the technical field of transformer fault diagnosis, in particular to an oil-immersed transformer fault diagnosis method based on a support vector machine and a genetic algorithm.
Background
In a power system, a power transformer can work safely, stably and reliably, and has a very important significance for the whole power system, namely one of very important devices, once operation faults occur, the faults can cause immeasurable loss to the society, and therefore, the research on diagnosis methods of different faults of the transformer is of a vital significance. Currently, there are many methods for diagnosing faults of power transformers, such as insulation performance detection, chromatography, partial discharge detection, and the like. DGA, which is based on a method for analyzing dissolved gases in oil, has the advantages of on-line observation and simple operation, and is one of the most common and effective methods for determining the fault of the power transformer at present. The method can accurately find the hidden fault of the transformer in the working process in advance, so that a countermeasure can be taken in advance, and the method plays a key role in prolonging the service life of the transformer and ensuring the safety and the stability of a power system.
In the soluble gas in the transformer oil, the composition, the proportion, the gas generation speed and the like are closely related to the fault types, and the characteristic gas content of the gas is more capable of reflecting the fault types to a certain extent, and the gas comprises hydrogen H2, methane CH4, acetylene C2H2, ethylene C2H4 and ethane C2H6, carbon monoxide CO and carbon dioxide CO2 generated by the decomposition of the transformer oil and cellulose paper. In recent years, in addition to analysis of dissolved gases in oil, various diagnostic methods have been developed, such as a key gas method, CUSC method, Doernerburg ratio method, Rogers ratio method, modified Rogers ratio method, and the like. However, most of the obstacles are determined based on field experience, and compared with the conventional transformer fault diagnosis method, most of the methods have disadvantages, such as incomplete coding, overlarge coding boundary and the like, and determination errors often occur in practical application.
With the continuous development of theoretical techniques such as Artificial Intelligence (AI) and Machine Learning (ML), a complex nonlinear relationship between the content of dissolved gas in oil and transformer faults can be constructed through the AI, and the DGA has the advantages of continuous learning and updating compared with the conventional DGA. Clustering-based methods (CBT), Fuzzy Logic Inference Systems (FLIS), expert systems (expert systems), artificial neural networks (artificial neural networks), Support Vector Machines (SVMs) and other AI techniques are widely used for transformer fault diagnosis, which improves the efficiency of fault diagnosis and obtains better feedback [4 ]. However, all AI methods have certain disadvantages, for example, although CBT can divide a fault sample into a plurality of subclass sets of different classes, CBT cannot find out which type the fault belongs to; the core of FLIS is related to experience, and the reasoning process and the result are connected with the experience too much; the network structure and weights of the ANN are difficult to determine and are easily trapped in "local minima" and "overfitting".
Disclosure of Invention
Objects of the invention
In order to solve the technical problems in the background art, the invention provides the oil-immersed transformer fault diagnosis method based on the support vector machine and the genetic algorithm, which has higher diagnosis efficiency and accuracy, can effectively and accurately identify the transformer fault and has excellent use effect.
(II) technical scheme
The invention provides a fault diagnosis method of an oil-immersed transformer based on a support vector machine and a genetic algorithm, which comprises the following steps:
s1, analyzing five common gases in the transformer oil by utilizing dissolved GAs analysis based on SVM and GA, wherein the five common gases are methane, ethane, ethylene, acetylene and hydrogen;
s2, establishing a fitness function based on five-fold cross validation by taking the ratio of the five dissolved gases as input, and taking the accuracy as a judgment standard of the function;
s3, finding out a more optimal SVM parameter C/sigma by using cross validation and GA;
and S4, training the model according to the parameters C/sigma of the optimal SVM, and verifying the model performance by taking the fault case as a test set.
Preferably, the SVM obtains an optimal result of the feature space classification hyperplane by using the structure risk minimization theory, finds an optimal hyperplane by the SVM, enlarges the areas on both sides of the hyperplane as much as possible, ensures the classification accuracy, and simultaneously adopts a convex optimization measure to convert the solution with the largest solution distance into the solution of a convex quadratic programming problem by using a dual method, wherein the specific operation of the SVM is as follows:
a transformer fault sample is defined as a positive sample, and a normal sample is defined as a negative sample;
the fault sample is distinguished from the normal sample by constructing a classification hyperplane, and the parameter optimization of the SVM is carried out by utilizing an intelligent algorithm in the construction of the classification hyperplane so as to solve the problem of hyperplane deviation;
for non-linear sample data, positive sample data and negative sample data are mapped to a high-dimensional feature space to be linear separable or approximately linear separable, and a relaxation variable xi i is added to adapt to a classification error.
Preferably, in a classifier constructed by SVM, the independent classification hyperplane is generally understood as: f (x) wTx+b＝0(1)；
To further simplify the optimization problem, the relaxation variable ξ and the regularization term are added to make it transform
The following steps are changed:
wherein, the punishment parameter C and the optimal interface are obtained by solving the dual problem of the equation (2); aiming at the situation of solving the nonlinear problem, mapping nonlinear data in an initial low-dimensional space in the SVM into a high-dimensional space by using a kernel function, thereby obtaining a better result; the selection of the appropriate penalty factor C and the kernel parameter σ from the SVM parameters will directly affect the accuracy of the diagnosis.
Preferably, the specific operation of GA is as follows:
generating a certain number of initial populations in a search space, further initializing relevant parameters and calculating individual fitness, and selecting the optimal individual through multiple times of selection, crossing, variation and screening.
Preferably, the initial population of GA is searched for possible answers to the questions, and each specific individual in the population has a specific gene code, so that the association between the expression of the population and the gene is first realized, and the expression of the individual is mapped to the gene, i.e. the gene coding work of the population is performed.
Preferably, the gas contained in the transformer oil can effectively indicate the decomposition condition around the fault point, and the content of the characteristic gas in the transformer oil is different due to the difference of fault types, energy and insulating materials; during various transformer faults, the degree of unsaturation and fault energy of the hydrocarbon gas it generates also varies, as follows:
characteristic gases generated in the transformer overheating fault are methane and ethylene, and characteristic gases generally generated in the discharge fault are acetylene and hydrogen; the ratio between methane and hydrogen can be used for determining whether the transformer fails to be overheated or discharged, and the content of methane can rise along with the temperature rise when the transformer fails, so that the failures with different overheating types can be distinguished by ethylene and ethane; the failure acetylene gas of the transformer is related to the intensity of discharge, and the acetylene is not generally generated when the discharge is weak.
Preferably, in S3, the GA is used to optimize the penalty factor C and the kernel function σ of the SVM, and a fault diagnosis model based on the GA-SVM power transformer is established, which includes the following specific steps:
initialization: setting relevant parameters of SVM, carrying out binary coding operation on them and setting relevant parameters of GA to screen out optimal (C, sigma);
establishing a fault processing sample set: aiming at a fault sample set, in order to improve the fault judgment accuracy, zero-one normalization processing is carried out on data, errors caused by quantity values are reduced, training sets and test set division are carried out on the data, and the normalization operation is as follows:
wherein, the sum is the characteristic value before and after conversion; and the sum is respectively the maximum eigenvalue and the minimum eigenvalue in the fault sample set;
model training: substituting the established fault training set into a GA-SVM model for training;
and (3) detecting the model: and substituting the established fault test set into the GA-SVM model for testing.
Preferably, the SVM of the transformer fault model is trained and tested by utilizing a libsvm extension packet of Matlab, and Matlab2016a is adopted as simulation software.
Preferably, the transformer faults are mainly divided into thermal faults and electrical faults, and specifically can be divided into low-temperature overheat faults, medium-temperature overheat faults, high-temperature overheat faults, low-energy discharge and overheat faults, arc discharge and overheat faults and partial discharge faults.
The technical scheme of the invention has the following beneficial technical effects:
the SVM (support vector machine) method is based on the structural risk minimization criterion, the defect that the traditional neural network diagnosis method is based on the empirical risk minimization principle is overcome, the diagnosis effect is good even if the sample capacity is small in the actual transformer fault diagnosis, and the optimal diagnosis accuracy can be effectively improved and obtained by using the GA (genetic algorithm) to optimize the parameters of the support vector machine, so that the GA-SVM-based transformer fault diagnosis model is proved to have high accuracy and applicability and has excellent use effect;
the method has higher diagnosis efficiency and accuracy, and can effectively and accurately identify the transformer fault.
Drawings
Fig. 1 is a flowchart of a fault diagnosis method for an oil-immersed transformer based on a support vector machine and a genetic algorithm according to the present invention.
Fig. 2 is a flowchart of a Genetic Algorithm (GA) in the oil-immersed transformer fault diagnosis method based on a support vector machine and the genetic algorithm.
Fig. 3 is a flowchart of an algorithm for optimizing SVM parameters by using GA in the oil-immersed transformer fault diagnosis method based on a support vector machine and a genetic algorithm according to the present invention.
Detailed Description
In order to make 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 in conjunction with the following detailed description. It should be understood that the description is intended to be exemplary only, and is not intended to limit the scope of the present invention. Moreover, in the following description, descriptions of well-known structures and techniques are omitted so as to not unnecessarily obscure the concepts of the present invention.
As shown in fig. 1 to 3, the method for diagnosing the fault of the oil-immersed transformer based on the support vector machine and the genetic algorithm, provided by the invention, comprises the following steps:
s1, analyzing five common gases in the transformer oil by utilizing dissolved GAs analysis based on SVM (support vector machine) and GA (genetic algorithm), wherein the five common gases are methane, ethane, ethylene, acetylene and hydrogen;
s2, establishing a fitness function based on five-fold Cross Validation (CV) by taking the ratio of the five dissolved gases as input, and taking the accuracy of the fitness function as a judgment standard of the fitness function;
s3, finding out a more optimal SVM parameter C/sigma by using cross validation and GA;
and S4, training the model according to the parameters C/sigma of the optimal SVM, and verifying the model performance by taking the fault case as a test set.
In an alternative embodiment, the SVM obtains the optimal result of the feature space classification hyperplane using the structure risk minimization theory, finds an optimal hyperplane, enlarges the areas on both sides of the hyperplane as much as possible, ensures the classification accuracy, and simultaneously adopts a convex optimization measure to transform the solution with the largest solution distance into the solution of a convex quadratic programming problem by a dual method, wherein the specific operation of the SVM is as follows:
a transformer fault sample is defined as a positive sample, and a normal sample is defined as a negative sample;
the fault sample is distinguished from the normal sample by constructing a classification hyperplane, and the parameter optimization of the SVM is carried out by utilizing an intelligent algorithm in the construction of the classification hyperplane so as to solve the problem of hyperplane deviation;
for non-linear sample data, positive sample data and negative sample data are mapped to a high-dimensional feature space to be linear separable or approximately linear separable, and a relaxation variable xi i is added to adapt to a classification error.
In an alternative embodiment, in a classifier constructed by an SVM, the independent classification hyperplane is generally understood to be: f (x) wTx+b＝0(1)；
To further simplify the optimization problem, the relaxation variable ξ and the regularization term are added to make it transform
The following steps are changed:
wherein, the punishment parameter C and the optimal interface are obtained by solving the dual problem of the equation (2); aiming at the situation of solving the nonlinear problem, mapping nonlinear data in an initial low-dimensional space in the SVM into a high-dimensional space by using a kernel function, thereby obtaining a better result; the selection of the appropriate penalty factor C and the kernel parameter σ from the SVM parameters will directly affect the accuracy of the diagnosis.
In an alternative embodiment, the GA is derived from replication, crossover and mutation phenomena in biogenetics, which is evolved from an initial population, and a group of better and more environment-suitable individuals is selected through operations such as selection, crossover and mutation, so that the population can be evolved in a better area step by step in a search space, and through cyclic evolution of such a generation, the optimal individual is obtained finally, i.e. the optimal solution of the problem is obtained, and the specific operations of the GA are as follows:
generating a certain number of initial populations in a search space, further initializing relevant parameters and calculating individual fitness, and selecting the optimal individual through multiple times of selection, crossing, variation and screening.
In an alternative embodiment, the initial population of GA is searched from possible answers to the questions, and each specific individual in the population has a specific gene code, so that the connection between the expression of the population and the gene is firstly realized, and the expression of the individual is mapped to the gene, i.e. the gene coding work of the population is carried out; after an initial population is randomly generated in a search space, through a plurality of screening measures of selection, intersection and variation, the fitness of the individual is calculated every time, a more optimal new solution set population is generated through gradual evolution, and finally the optimal individual in the final population is the approximate optimal solution.
In an alternative embodiment, the gas contained in the transformer oil can effectively indicate the decomposition condition around the fault point, and the different fault types, the different energy and the different insulating materials can cause different content of the characteristic gas in the transformer oil; during various transformer faults, the degree of unsaturation and fault energy of the hydrocarbon gas it generates also varies, as follows:
characteristic gases generated in the transformer overheating fault are methane and ethylene, and characteristic gases generally generated in the discharge fault are acetylene and hydrogen; the ratio between methane and hydrogen can be used for determining whether the transformer fails to be overheated or discharged, and the content of methane can rise along with the temperature rise when the transformer fails, so that the failures with different overheating types can be distinguished by ethylene and ethane; the fault acetylene gas of the transformer is related to the discharge intensity, and the acetylene is not generated generally when the discharge intensity is weak;
the transformer faults are mainly divided into thermal faults and electrical faults, and specifically can be divided into low-temperature overheat faults, medium-temperature overheat faults, high-temperature overheat faults, low-energy discharge and overheat faults, arc discharge and overheat faults and partial discharge faults.
In an alternative embodiment, in the SVM, the difference between the penalty factor C and the kernel parameter σ directly affects the diagnostic accuracy of the model, and different C and σ directly cause the classifier to have different diagnostic performance; the conventional algorithm such as grid division for finding the optimal parameters needs a large amount of calculation, the calculation complexity is high, and a large amount of time is consumed, so that an advanced algorithm is required to optimize the parameters of the SVM to obtain the optimal diagnostic performance;
in S3, a penalty factor C and a kernel function sigma of the SVM are optimized by using GA, a fault diagnosis model based on the GA-SVM power transformer is established, the global optimal solution in the solving space can be rapidly and accurately solved by using the characteristics of GA, such as practicability, high efficiency, strong robustness and the like, a trap trapped in a local optimal solution which suddenly drops is not caused, SVM parameters are rapidly and accurately determined, and therefore the fault diagnosis model of the GA-SVM transformer is finally determined, and the specific implementation steps are as follows:
initialization: setting relevant parameters of SVM, carrying out binary coding operation on them and setting relevant parameters of GA to screen out optimal (C, sigma);
establishing a fault processing sample set: aiming at a fault sample set, in order to improve the fault judgment accuracy, zero-one normalization processing is carried out on data, errors caused by quantity values are reduced, training sets and test set division are carried out on the data, and the normalization operation is as follows:
wherein, the sum is the characteristic value before and after conversion; and the sum is respectively the maximum eigenvalue and the minimum eigenvalue in the fault sample set;
model training: substituting the established fault training set into a GA-SVM model for training;
and (3) detecting the model: and substituting the established fault test set into the GA-SVM model for testing.
In an alternative embodiment, the transformer fault model SVM is trained and tested by using a libsvm extension packet of Matlab, and in the Cross Validation (CV) sense, the accuracy of a training set is used as a fitness function in GA, and CV is based on 5-fold cross validation; the flow chart of the overall algorithm model for performing optimization on SVM parameters by using GA (genetic algorithm) is shown in FIG. 2, wherein the experimental error is reduced by executing the experimental model algorithm twenty times, the optimal value is taken as output; the experimental genetic algorithm parameters are set as follows: the maximum population number is set to be 20, the maximum evolution algebra is set to be 200, the cross probability is 0.7, the mutation probability is 0.05, the optimization interval of the penalty factor C is [0-1000], the optimization interval of the kernel parameter sigma is [0-100], and Matlab2016a is adopted by simulation software.
In the invention, an SVM (support vector machine) method is based on a structural risk minimization criterion, and overcomes the defect that the traditional neural network diagnosis method is based on an empirical risk minimization principle, the diagnosis effect is good even if the sample capacity is small in the actual transformer fault diagnosis, and the optimal diagnosis accuracy can be effectively improved and obtained by using GA (genetic algorithm) to optimize the parameters of the support vector machine, which proves that the GA-SVM-based transformer fault diagnosis model has higher accuracy and is applicable, and the use effect is excellent; the method has higher diagnosis efficiency, and can effectively and accurately identify the transformer fault.
It is to be understood that the above-described embodiments of the present invention are merely illustrative of or explaining the principles of the invention and are not to be construed as limiting the invention. Therefore, any modification, equivalent replacement, improvement and the like made without departing from the spirit and scope of the present invention should be included in the protection scope of the present invention. Further, it is intended that the appended claims cover all such variations and modifications as fall within the scope and boundaries of the appended claims or the equivalents of such scope and boundaries.
Claims (9)
1. The oil-immersed transformer fault diagnosis method based on the support vector machine and the genetic algorithm is characterized by comprising the following steps:
s1, analyzing five common gases in the transformer oil by utilizing dissolved GAs analysis based on SVM and GA, wherein the five common gases are methane, ethane, ethylene, acetylene and hydrogen;
s2, establishing a fitness function based on five-fold cross validation by taking the ratio of the five dissolved gases as input, and taking the accuracy as a judgment standard of the function;
s3, finding out a more optimal SVM parameter C/sigma by using cross validation and GA;
and S4, training the model according to the parameters C/sigma of the optimal SVM, and verifying the model performance by taking the fault case as a test set.
2. The oil-immersed transformer fault diagnosis method based on the support vector machine and the genetic algorithm according to claim 1, characterized in that the SVM uses a structure risk minimization theory to obtain an optimal result of a characteristic space classification hyperplane, the SVM finds an optimal hyperplane, enlarges the areas on both sides of the hyperplane as much as possible, ensures the classification accuracy thereof, and simultaneously adopts a convex optimization measure to convert the maximum solution distance into a problem of solving a convex quadratic programming by a dual method, and the specific operation of the SVM is as follows:
a transformer fault sample is defined as a positive sample, and a normal sample is defined as a negative sample;
the method comprises the steps of (1) distinguishing a fault sample from a normal sample by constructing a classification hyperplane H, wherein w, phi (x) + b is 0, and in the construction of the classification hyperplane, optimizing parameters of an SVM (support vector machine) by using an intelligent algorithm to solve the problem of hyperplane deviation;
for non-linear sample data, positive sample data and negative sample data are mapped to a high-dimensional feature space to be linear separable or approximately linear separable, and a relaxation variable xi i is added to adapt to a classification error.
3. Oil-filled transformer fault diagnosis method based on support vector machine and genetic algorithm according to claim 2, characterized in that in a classifier constructed by SVM, the independent classification hyperplane is generally understood as:
f(x)＝wTx+b＝0 (1)；
to further simplify the optimization problem, we turn it into:
wherein, the punishment parameter C and the optimal interface are obtained by solving the dual problem of the equation (2); aiming at the situation of solving the nonlinear problem, mapping nonlinear data in an initial low-dimensional space in the SVM into a high-dimensional space by using a kernel function, thereby obtaining a better result; the selection of the appropriate penalty factor C and the kernel parameter σ from the SVM parameters will directly affect the accuracy of the diagnosis.
4. The oil-immersed transformer fault diagnosis method based on the support vector machine and the genetic algorithm according to claim 1, characterized in that the specific operation of the GA is as follows:
generating a certain number of initial populations in a search space, further initializing relevant parameters and calculating individual fitness, and selecting the optimal individual through multiple times of selection, crossing, variation and screening.
5. The oil-filled transformer fault diagnosis method based on the support vector machine and the genetic algorithm is characterized in that an initial population of GAs is searched from possible answers of a question, each specific individual in the population has a specific gene code, and therefore, firstly, the connection between the expression of the population and the gene is realized, and the expression of the individual is mapped to the gene, namely, the gene coding work of the population is carried out.
6. The oil-immersed transformer fault diagnosis method based on the support vector machine and the genetic algorithm is characterized in that gas contained in transformer oil can effectively explain decomposition conditions around fault points, and the content of characteristic gas in the transformer oil is different due to the difference of fault types, energy and insulating materials; during various transformer faults, the degree of unsaturation and fault energy of the hydrocarbon gas it generates also varies, as follows:
characteristic gases generated in the transformer overheating fault are methane and ethylene, and characteristic gases generally generated in the discharge fault are acetylene and hydrogen; the ratio between methane and hydrogen can be used for determining whether the transformer fails to be overheated or discharged, and the content of methane can rise along with the temperature rise when the transformer fails, so that the failures with different overheating types can be distinguished by ethylene and ethane; the failure acetylene gas of the transformer is related to the intensity of discharge, and the acetylene is not generally generated when the discharge is weak.
7. The oil-immersed transformer fault diagnosis method based on the support vector machine and the genetic algorithm according to claim 1, characterized in that in S3, a penalty factor C and a kernel function sigma of an SVM are optimized by using GA to establish a fault diagnosis model based on a GA-SVM power transformer, and the specific implementation steps are as follows:
initialization: setting relevant parameters of SVM, carrying out binary coding operation on them and setting relevant parameters of GA to screen out optimal (C, sigma);
establishing a fault processing sample set: aiming at a fault sample set, in order to improve the fault judgment accuracy, zero-one normalization processing is carried out on data, errors caused by quantity values are reduced, training sets and test set division are carried out on the data, and the normalization operation is as follows:
wherein xi and xsi are characteristic values before and after conversion respectively; and xi.max and xi.min are respectively the maximum characteristic value and the minimum characteristic value in the fault sample set;
model training: substituting the established fault training set into a GA-SVM model for training;
and (3) detecting the model: and substituting the established fault test set into the GA-SVM model for testing.
8. The oil-immersed transformer fault diagnosis method based on the support vector machine and the genetic algorithm is characterized in that an SVM of a transformer fault model is trained and tested by using a libsvm extension packet of Matlab, and Matlab2016a is adopted as simulation software.
9. The oil-immersed transformer fault diagnosis method based on the support vector machine and the genetic algorithm according to claim 6, wherein the transformer faults are mainly classified into thermal faults and electrical faults, and specifically can be classified into low-temperature overheat faults, medium-temperature overheat faults, high-temperature overheat faults, low-energy discharge and overheat faults, arc discharge and overheat faults and partial discharge faults.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010236460.XA CN113468461A (en) | 2020-03-30 | 2020-03-30 | Oil-immersed transformer fault diagnosis method based on support vector machine and genetic algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010236460.XA CN113468461A (en) | 2020-03-30 | 2020-03-30 | Oil-immersed transformer fault diagnosis method based on support vector machine and genetic algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113468461A true CN113468461A (en) | 2021-10-01 |
Family
ID=77864864
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010236460.XA Withdrawn CN113468461A (en) | 2020-03-30 | 2020-03-30 | Oil-immersed transformer fault diagnosis method based on support vector machine and genetic algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113468461A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113899997A (en) * | 2021-11-10 | 2022-01-07 | 哈尔滨理工大学 | Transformer insulation state diagnosis method based on improved support vector machine |
| CN114034344A (en) * | 2021-11-12 | 2022-02-11 | 广东电网有限责任公司江门供电局 | Transformer model measurement analysis method |
| CN114896883A (en) * | 2022-05-13 | 2022-08-12 | 西安工程大学 | Transformer fault diagnosis method based on MEA-SVM classifier |
| CN117335571A (en) * | 2023-10-09 | 2024-01-02 | 国网河南省电力公司濮阳供电公司 | Intelligent fault early warning management system and method for power distribution network |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107169514A (en) * | 2017-05-05 | 2017-09-15 | 清华大学 | The method for building up of diagnosing fault of power transformer model |
| CN109270390A (en) * | 2018-09-14 | 2019-01-25 | 广西电网有限责任公司电力科学研究院 | Diagnosis Method of Transformer Faults based on Gaussian transformation Yu global optimizing SVM |
| CN109902339A (en) * | 2019-01-18 | 2019-06-18 | 昆明理工大学 | A kind of Fault Diagnosis of Roller Bearings based on IAGA-SVM |
| CN109919178A (en) * | 2019-01-23 | 2019-06-21 | 广西大学 | Failure prediction method based on characteristic quantity preferably with Wavelet Kernel Function LSSVM |
-
2020
- 2020-03-30 CN CN202010236460.XA patent/CN113468461A/en not_active Withdrawn
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107169514A (en) * | 2017-05-05 | 2017-09-15 | 清华大学 | The method for building up of diagnosing fault of power transformer model |
| CN109270390A (en) * | 2018-09-14 | 2019-01-25 | 广西电网有限责任公司电力科学研究院 | Diagnosis Method of Transformer Faults based on Gaussian transformation Yu global optimizing SVM |
| CN109902339A (en) * | 2019-01-18 | 2019-06-18 | 昆明理工大学 | A kind of Fault Diagnosis of Roller Bearings based on IAGA-SVM |
| CN109919178A (en) * | 2019-01-23 | 2019-06-21 | 广西大学 | Failure prediction method based on characteristic quantity preferably with Wavelet Kernel Function LSSVM |
Non-Patent Citations (1)
| Title |
|---|
| 邬蓉蓉 等: "综合遗传优化支持向量机的变压器故障诊断模型" * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113899997A (en) * | 2021-11-10 | 2022-01-07 | 哈尔滨理工大学 | Transformer insulation state diagnosis method based on improved support vector machine |
| CN114034344A (en) * | 2021-11-12 | 2022-02-11 | 广东电网有限责任公司江门供电局 | Transformer model measurement analysis method |
| CN114896883A (en) * | 2022-05-13 | 2022-08-12 | 西安工程大学 | Transformer fault diagnosis method based on MEA-SVM classifier |
| CN117335571A (en) * | 2023-10-09 | 2024-01-02 | 国网河南省电力公司濮阳供电公司 | Intelligent fault early warning management system and method for power distribution network |
| CN117335571B (en) * | 2023-10-09 | 2024-05-03 | 国网河南省电力公司濮阳供电公司 | Intelligent fault early warning management system and method for power distribution network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113468461A (en) | Oil-immersed transformer fault diagnosis method based on support vector machine and genetic algorithm | |
| Malik et al. | Application of gene expression programming (GEP) in power transformers fault diagnosis using DGA | |
| Ma et al. | Power transformer fault diagnosis under measurement originated uncertainties | |
| de Andrade Lopes et al. | Incipient fault diagnosis in power transformers by data-driven models with over-sampled dataset | |
| CN114372693B (en) | Transformer fault diagnosis method based on cloud model and improved DS evidence theory | |
| CN110569888A (en) | transformer fault diagnosis method and device based on directed acyclic graph support vector machine | |
| CN116562114A (en) | Power transformer fault diagnosis method based on graph convolution neural network | |
| Nanfak et al. | Hybrid DGA method for power transformer faults diagnosis based on evolutionary k-means clustering and dissolved gas subsets analysis | |
| Guo et al. | Fault diagnosis of power transformers based on comprehensive machine learning of dissolved gas analysis | |
| CN117113166A (en) | Industrial boiler fault detection method based on improved integrated learning | |
| Jian et al. | Transformer fault diagnosis model with unbalanced samples based on SMOTE algorithm and focal loss | |
| Zhang et al. | Improved GWO-MCSVM algorithm based on nonlinear convergence factor and tent chaotic mapping and its application in transformer condition assessment | |
| Cheemala et al. | Transformer incipient fault diagnosis using machine learning classifiers | |
| Wang et al. | A Power Transformer Fault Diagnosis Method Based on Random Vector Functional‐Link Neural Network | |
| CN112085064A (en) | Transformer fault diagnosis method based on multi-classification probability output of support vector machine | |
| Nanfak et al. | Hybrid Method for Power Transformers Faults Diagnosis Based on Ensemble Bagged Tree Classification and Training Subsets Using Rogers and Gouda Ratios. | |
| Qaedi et al. | Improvement in power transformer intelligent dissolved gas analysis method | |
| CN116522138A (en) | Multi-element time sequence anomaly detection method and system based on weak supervision integration | |
| Zhang et al. | A new method for transformer fault diagnosis by using improved clustering method | |
| CN116611003A (en) | Transformer fault diagnosis method, device and medium | |
| Singh et al. | Dissolved gas analysis of power transformer using K-means and support vector machine | |
| CN113721002A (en) | Transformer fault diagnosis method based on hard voting ensemble learning | |
| Yue et al. | Similarity Makes Difference: SSHTN for Generalized Zero-Shot Industrial Fault Diagnosis by Leveraging Auxiliary Set | |
| CN113283479A (en) | Feature extraction and diagnosis method suitable for power transformer faults | |
| Sun et al. | Fault diagnosis of power transformer based on DGA and information fusion |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20211001 |
|
| WW01 | Invention patent application withdrawn after publication |