CN108491503B - Method and system for determining fault type of transformer based on data analysis - Google Patents

Method and system for determining fault type of transformer based on data analysis Download PDF

Info

Publication number
CN108491503B
CN108491503B CN201810233161.3A CN201810233161A CN108491503B CN 108491503 B CN108491503 B CN 108491503B CN 201810233161 A CN201810233161 A CN 201810233161A CN 108491503 B CN108491503 B CN 108491503B
Authority
CN
China
Prior art keywords
fault
data
sample
standard data
determining
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201810233161.3A
Other languages
Chinese (zh)
Other versions
CN108491503A (en
Inventor
林春耀
李德波
周丹
马志钦
杨贤
赵永发
赵东生
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Electric Power Research Institute of Guangdong Power Grid Co Ltd
Original Assignee
Electric Power Research Institute of Guangdong Power Grid Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Electric Power Research Institute of Guangdong Power Grid Co Ltd filed Critical Electric Power Research Institute of Guangdong Power Grid Co Ltd
Priority to CN201810233161.3A priority Critical patent/CN108491503B/en
Publication of CN108491503A publication Critical patent/CN108491503A/en
Application granted granted Critical
Publication of CN108491503B publication Critical patent/CN108491503B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/20—Administration of product repair or maintenance

Landscapes

  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Engineering & Computer Science (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Other Investigation Or Analysis Of Materials By Electrical Means (AREA)

Abstract

The invention discloses a method and a system for determining the fault type of a transformer based on data analysis, wherein the method comprises the following steps: acquiring historical fault data associated with a plurality of transformers, and performing data format conversion to generate a plurality of standard data samples; determining a central sample in the plurality of standard data samples, and determining the membership degree of each characteristic gas in each standard data sample; determining the similarity between any two standard data samples in the plurality of data samples according to the membership degree of each characteristic gas in each standard data sample to generate a similarity matrix; classifying the plurality of standard data samples according to the similarity matrix to generate a fault diagnosis structure comprising a plurality of classification nodes; acquiring characteristic gas data of each transformer in a plurality of transformers to be monitored from the monitoring node; and matching the characteristic gas data with each classification node in the plurality of classification nodes, and determining the fault type of the transformer according to the matching result.

Description

Method and system for determining fault type of transformer based on data analysis
Technical Field
The present invention relates to the field of transformer fault diagnosis technologies, and in particular, to a method and a system for determining a fault type of a transformer based on data analysis.
Background
Because the fault classification of the transformer has ambiguity, faults in a ratio interval with an ambiguous three-ratio coding boundary are prone to misjudgment, such as misjudgment of low-energy discharge and high-energy discharge, and when various faults occur, the corresponding ratios cannot be found. And the fault diagnosis method based on the neural network, for example: BPNN and RBFNN, due to the lack of empirical risk minimization principle, require a large number of training sample supports to obtain good diagnostic results. When the actual fault is diagnosed, even if a very detailed quantitative representation is adopted, the traditional fault diagnosis method based on the dissolved gas in the oil is difficult to accurately diagnose the fault condition of the equipment, and the diagnosed conclusion is single.
Due to the fact that a large-scale transformer fault diagnosis lacks of a definite causal model and needs a large amount of experience, an actual fault case can provide more information than a group of classification rules, and the case is easy to acquire compared with the rule, and therefore, the method for diagnosing the transformer fault based on case reasoning is very valuable. Therefore, a method for diagnosing transformer faults based on case data analysis is needed.
Disclosure of Invention
The invention provides a method and a system for determining the fault type of a transformer based on data analysis, which aim to solve the problem of accurately and effectively diagnosing the fault of the transformer.
In order to solve the above problem, according to an aspect of the present invention, there is provided a method of determining a fault type of a transformer based on data analysis, the method including:
acquiring historical fault data associated with a plurality of transformers, and performing data format conversion on each fault sample data in the historical fault data to generate a plurality of standard data samples, wherein each fault sample data comprises: fault type and corresponding content of various characteristic gases;
determining a central sample in the plurality of standard data samples and determining a degree of membership of each characteristic gas in each standard data sample based on the central sample;
determining the similarity between any two standard data samples in the plurality of data samples according to the membership degree of each characteristic gas in each standard data sample to generate a similarity matrix;
classifying the plurality of standard data samples according to the similarity matrix to generate a fault diagnosis structure comprising a plurality of classification nodes, wherein each classification node corresponds to a single fault type;
acquiring characteristic gas data of each transformer in a plurality of transformers to be monitored from the monitoring node;
and matching the characteristic gas data with each classification node in the plurality of classification nodes, and determining the fault type of the transformer according to the matching result.
Preferably, the characteristic gas is dissolved gas in transformer oil, and comprises: h2、CH4、C2H2、C2H4And C2H6。
Preferably, the data format converting each fault sample data in the historical fault data to generate a plurality of standard data samples includes:
PC2H2＝CC2H2/Num，
PC2H4＝CC2H4/Num，
PC2H6＝CC2H6/Num，
PCH4＝CCH4/Num，
PH2＝CH2/(Num+CH2)，
Num＝CC2H2+CC2H4+CC2H6+CCH4，
wherein, CC2H2Sample data C for each fault2H2The hydrocarbon number of (a); cC2H4Sample data C for each fault2H4The hydrocarbon number of (a); cC2H6Sample data C for each fault2H6The hydrocarbon number of (a); cCH4For CH in each failure sample data4The hydrocarbon number of (a); NUM is the total hydrocarbon value of each fault sample data; pC2H2、PC2H4、PC2H6、PCH4And PH2Respectively converted standardC in data sample2H2、C2H4、C2H6、CH4And H2The content of (a).
Preferably, the determining a center sample in the plurality of standard data samples and determining a degree of membership of each characteristic gas in each standard data sample based on the center sample comprises:
Figure GDA0001692221870000031
Figure GDA0001692221870000032
Figure GDA0001692221870000033
wherein the content of the first and second substances,
Figure GDA0001692221870000034
the average component of the jth characteristic gas in the central sample; n is the number of standard data samples; x is the number ofijThe content of the jth characteristic gas in the ith standard data sample is shown; mu.sijAnd the membership degree of the jth characteristic gas in the ith standard data sample.
Preferably, the determining the similarity between any two standard data samples in the plurality of data samples according to the membership of each characteristic gas in each standard data sample to generate a similarity matrix includes:
Figure GDA0001692221870000035
Figure GDA0001692221870000036
rij＝1-cNij 1≤c≤1.5，
wherein, the ith fault class has ni groups of samples, rijRepresenting the similarity between sample i and sample j; c is a constant, and ω k is the weight of the kth characteristic gas, which is proportional to the averaging error of the kth characteristic gas in the class of samples.
Preferably, the classifying the plurality of standard data samples according to the similarity matrix to generate a fault diagnosis structure including a plurality of classification nodes includes:
and generating classification nodes, wherein each classification node corresponds to a fault class, in the process, firstly verifying whether a first fault data sample i is classified into a previous class, if not, using the first fault data sample i as first sample data of a new fault class, then determining whether a jth row and an ith column are equal to 1, if so, classifying the fault classes into the class, and finally, compiling the generated fault classes into an array to become a sub-fault class array of the current fault class, and generating a fault diagnosis structure comprising a plurality of classification nodes.
Preferably, wherein the method further comprises:
further refining the classification result by using the idea of the ISODATA dynamic clustering method, which comprises the following steps: calculating the similarity between each group of standard data samples and the standard samples of the determined sub-fault classes, selecting the class with the minimum similarity as a new sub-fault class, and calculating new fuzzy statistics according to the following formula:
Figure GDA0001692221870000041
wherein tmp _ fuzzy is a fuzzy statistic;
when the difference between the fuzzy statistic and the former fuzzy statistic is less than a certain constant, ending; otherwise, performing refinement calculation again, and storing the refined result as the optimal classification in the subclass array.
Preferably, the matching the characteristic gas data with each classification node in the plurality of classification nodes and determining the fault type of the transformer according to the matching result includes:
and searching a best matching node in the next layer of nodes of the root node from the root node, and then iterating by taking the best matching node as the root node until the iteration is finished when the next layer of nodes are not available in the matching node, wherein the last node is the fault type of the transformer.
Preferably, the matching the characteristic gas data with each classification node in the plurality of classification nodes and determining the fault type of the transformer according to the matching result includes:
and directly searching all leaf nodes, taking all leaf nodes as the same layer, and selecting the most matched node as the fault type of the transformer.
According to another aspect of the present invention, there is provided a system for determining a fault type of a transformer based on data analysis, the system comprising:
the standard data sample determining unit is used for acquiring historical fault data associated with a plurality of transformers and performing data format conversion on each fault sample data in the historical fault data to generate a plurality of standard data samples, wherein each fault sample data comprises: fault type and corresponding content of various characteristic gases;
a membership degree determination unit for determining a center sample among the plurality of standard data samples and determining a membership degree of each characteristic gas in each standard data sample based on the center sample;
the similarity matrix determining unit is used for determining the similarity between any two standard data samples in the plurality of data samples according to the membership of each characteristic gas in each standard data sample so as to generate a similarity matrix;
a fault diagnosis structure determination unit, configured to classify the multiple standard data samples according to the similarity matrix to generate a fault diagnosis structure including multiple classification nodes, where each classification node corresponds to a single fault type;
the characteristic gas data acquisition unit is used for acquiring characteristic gas data in each transformer of the plurality of transformers to be monitored from the monitoring node;
and the fault type determining unit is used for matching the characteristic gas data with each classification node in the plurality of classification nodes and determining the fault type of the transformer according to the matching result.
Preferably, the characteristic gas is dissolved gas in transformer oil, and comprises: h2、CH4、C2H2、C2H4And C2H6。
Preferably, the standard data sample determining unit performs data format conversion on each fault sample data in the historical fault data to generate a plurality of standard data samples, and includes:
PC2H2＝CC2H2/Num，
PC2H4＝CC2H4/Num，
PC2H6＝CC2H6/Num，
PCH4＝CCH4/Num，
PH2＝CH2/(Num+CH2)，
Num＝CC2H2+CC2H4+CC2H6+CCH4，
wherein, CC2H2Sample data C for each fault2H2The hydrocarbon number of (a); cC2H4Sample data C for each fault2H4The hydrocarbon number of (a); cC2H6Sample data C for each fault2H6The hydrocarbon number of (a); cCH4For CH in each failure sample data4The hydrocarbon number of (a); NUM is the total hydrocarbon value of each fault sample data; pC2H2、PC2H4、PC2H6、PCH4And PH2Respectively C in the converted standard data samples2H2、C2H4、C2H6、CH4And H2The content of (a).
Preferably, the membership determining unit determines a center sample among the plurality of standard data samples, and determines the membership of each characteristic gas in each standard data sample based on the center sample, including:
Figure GDA0001692221870000061
Figure GDA0001692221870000062
Figure GDA0001692221870000063
wherein the content of the first and second substances,
Figure GDA0001692221870000064
the average component of the jth characteristic gas in the central sample; n is the number of standard data samples; x is the number ofijThe content of the jth characteristic gas in the ith standard data sample is shown; mu.sijAnd the membership degree of the jth characteristic gas in the ith standard data sample.
Preferably, the similarity matrix determining unit determines the similarity between any two standard data samples in the plurality of data samples according to the membership of each characteristic gas in each standard data sample to generate the similarity matrix, including:
Figure GDA0001692221870000065
Figure GDA0001692221870000066
rij＝1-cNij 1≤c≤1.5，
wherein, the ith fault class has ni groups of samples, rijRepresenting the similarity between sample i and sample j; c is a constant, ω k is the weight of the kth characteristic gas, which is the average of the kth characteristic gas in the class of samplesThe error is proportional.
Preferably, the fault diagnosis structure determining unit classifies the plurality of standard data samples according to the similarity matrix to generate a fault diagnosis structure including a plurality of classification nodes, and includes:
and generating classification nodes, wherein each classification node corresponds to a fault class, in the process, firstly verifying whether a first fault data sample i is classified into a previous class, if not, using the first fault data sample i as first sample data of a new fault class, then determining whether a jth row and an ith column are equal to 1, if so, classifying the fault classes into the class, and finally, compiling the generated fault classes into an array to become a sub-fault class array of the current fault class, and generating a fault diagnosis structure comprising a plurality of classification nodes.
Preferably, the failure diagnosis structure determination unit further includes:
further refining the classification result by using the idea of the ISODATA dynamic clustering method, which comprises the following steps: calculating the similarity between each group of standard data samples and the standard samples of the determined sub-fault classes, selecting the class with the minimum similarity as a new sub-fault class, and calculating new fuzzy statistics according to the following formula:
Figure GDA0001692221870000071
wherein tmp _ fuzzy is a fuzzy statistic;
when the difference between the fuzzy statistic and the former fuzzy statistic is less than a certain constant, ending; otherwise, performing refinement calculation again, and storing the refined result as the optimal classification in the subclass array.
Preferably, the fault type determining unit matches the characteristic gas data with each of the plurality of classification nodes, and determines the fault type of the transformer according to a matching result, including:
and searching a best matching node in the next layer of nodes of the root node from the root node, and then iterating by taking the best matching node as the root node until the iteration is finished when the next layer of nodes are not available in the matching node, wherein the last node is the fault type of the transformer.
Preferably, the fault type determining unit matches the characteristic gas data with each of the plurality of classification nodes, and determines the fault type of the transformer according to a matching result, including:
and directly searching all leaf nodes, taking all leaf nodes as the same layer, and selecting the most matched node as the fault type of the transformer.
The invention provides a method and a system for determining the fault type of a transformer based on data analysis. The historical fault data associated with the plurality of transformers is obtained from existing case history, repair records or other data, the fault type of the transformers can be rapidly and accurately diagnosed by analyzing the historical fault data, the determined diagnosis result can be reused based on data analysis without reckoning again, and the analysis efficiency of new problems is greatly improved.
Drawings
A more complete understanding of exemplary embodiments of the present invention may be had by reference to the following drawings in which:
FIG. 1 is a flow diagram of a method 100 for determining a fault type of a transformer based on data analysis according to an embodiment of the present invention; and
fig. 2 is a schematic diagram of a system 200 for determining a fault type of a transformer based on data analysis according to an embodiment of the present invention.
Detailed Description
The exemplary embodiments of the present invention will now be described with reference to the accompanying drawings, however, the present invention may be embodied in many different forms and is not limited to the embodiments described herein, which are provided for complete and complete disclosure of the present invention and to fully convey the scope of the present invention to those skilled in the art. The terminology used in the exemplary embodiments illustrated in the accompanying drawings is not intended to be limiting of the invention. In the drawings, the same units/elements are denoted by the same reference numerals.
Unless otherwise defined, terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Further, it will be understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and will not be interpreted in an idealized or overly formal sense.
Fig. 1 is a flow chart of a method 100 of determining a fault type of a transformer based on data analysis according to an embodiment of the present invention. As shown in fig. 1, the method for determining the fault type of the transformer based on the data analysis according to the embodiment of the present invention performs inference based on an actual transformer fault case, and determines the fault type of the transformer mainly based on the analysis and diagnosis method of the gas analysis and the electrical test data because the gas analysis and the electrical test data can reflect the insulation condition and the fault symptom of the transformer most. The historical fault data associated with the plurality of transformers is obtained from existing case history, repair records or other data, the fault type of the transformers can be rapidly and accurately diagnosed by analyzing the historical fault data, the determined diagnosis result can be reused based on data analysis without reckoning again, and the analysis efficiency of new problems is greatly improved. The method 100 for determining the fault type of the transformer based on the data analysis starts at step 101, acquires historical fault data associated with a plurality of transformers at step 101, and performs data format conversion on each fault sample data in the historical fault data to generate a plurality of standard data samples, wherein each fault sample data comprises: the type of fault and the corresponding content of the characteristic gases.
Preferably, wherein the characteristic gasThe body is dissolved gas in transformer oil, including: h2、CH4、C2H2、C2H4And C2H6。
Preferably, the data format converting each fault sample data in the historical fault data to generate a plurality of standard data samples includes:
PC2H2＝CC2H2/Num，
PC2H4＝CC2H4/Num，
PC2H6＝CC2H6/Num，
PCH4＝CCH4/Num，
PH2＝CH2/(Num+CH2)，
Num＝CC2H2+CC2H4+CC2H6+CCH4，
wherein, CC2H2Sample data C for each fault2H2The hydrocarbon number of (a); cC2H4Sample data C for each fault2H4The hydrocarbon number of (a); cC2H6Sample data C for each fault2H6The hydrocarbon number of (a); cCH4For CH in each failure sample data4The hydrocarbon number of (a); NUM is the total hydrocarbon value of each fault sample data; pC2H2、PC2H4、PC2H6、PCH4And PH2Respectively C in the converted standard data samples2H2、C2H4、C2H6、CH4And H2The content of (a).
Preferably, a central sample is determined in the plurality of standard data samples at step 102, and a degree of membership for each characteristic gas in each standard data sample is determined based on the central sample.
Preferably, the determining a center sample in the plurality of standard data samples and determining a degree of membership of each characteristic gas in each standard data sample based on the center sample comprises:
Figure GDA0001692221870000101
Figure GDA0001692221870000102
Figure GDA0001692221870000103
wherein the content of the first and second substances,
Figure GDA0001692221870000104
the average component of the jth characteristic gas in the central sample; n is the number of standard data samples; x is the number ofijThe content of the jth characteristic gas in the ith standard data sample is shown; mu.sijAnd the membership degree of the jth characteristic gas in the ith standard data sample.
In the pattern recognition problem of fault diagnosis, data is processed by classifying the data with small difference into one class, and classifying the data with large difference into different classes, so that samples in the same class are close to the center sample, and samples at a certain distance are further far away from the center, thereby forming a Martian effect. Based on this consideration, and the requirement of normalization in the fuzzy processing, the embodiment of the present invention proposes a "double cosine" type membership function to solve the membership degree of the original data. The method comprises the following specific steps:
step 1: calculating the total hydrocarbon number Num ═ CC2H2+CC2H4+CC2H6+CCH4。
Step 2: calculating C in converted standard data sample2H2、C2H4、C2H6、CH4And H2The content of (a). Wherein, the calculation formula is:
PC2H2＝CC2H2/Num，
PC2H4＝CC2H4/Num，
PC2H6＝CC2H6/Num，
PCH4＝CCH4/Num，
PH2＝CH2/(Num+CH2)，
in the following calculation, the characteristic gas C2H2、C2H4、C2H6、CH4And H2The content of (c) is substituted by the converted ratio.
Step 3: setting a fault class center sample, setting the number of the fault class center samples as N, and calculating the average component of the jth characteristic gas in the center sample according to the following formula:
Figure GDA0001692221870000111
step 4: the fourth step: determining a degree of membership for each characteristic gas in each standard data sample based on the central sample using the formula:
Figure GDA0001692221870000112
Figure GDA0001692221870000113
after the transformation, the membership degree of the central sample is 0.5, the data in the vicinity of the central sample is transformed to be further reduced in difference, and the samples farther away from the center. The reason why the membership function is again gentle at both ends is to take into account the normalization requirement, and the data at both ends will be separated when the center of the subclass is formed.
Preferably, in step 103, the similarity between any two standard data samples in the plurality of data samples is determined according to the membership of each characteristic gas in each standard data sample, so as to generate a similarity matrix.
Preferably, the determining the similarity between any two standard data samples in the plurality of data samples according to the membership of each characteristic gas in each standard data sample to generate a similarity matrix includes:
Figure GDA0001692221870000121
Figure GDA0001692221870000122
rij＝1-cNij 1≤c≤1.5，
wherein, the ith fault class has ni groups of samples, rijRepresenting the similarity between sample i and sample j; c is a constant, and ω k is the weight of the kth characteristic gas, which is proportional to the averaging error of the kth characteristic gas in the class of samples.
Preferably, the plurality of standard data samples are classified according to the similarity matrix to generate a fault diagnosis structure comprising a plurality of classification nodes, wherein each classification node corresponds to a single fault type, in step 104.
Preferably, the classifying the plurality of standard data samples according to the similarity matrix to generate a fault diagnosis structure including a plurality of classification nodes includes:
and generating classification nodes, wherein each classification node corresponds to a fault class, in the process, firstly verifying whether a first fault data sample i is classified into a previous class, if not, using the first fault data sample i as first sample data of a new fault class, then determining whether a jth row and an ith column are equal to 1, if so, classifying the fault classes into the class, and finally, compiling the generated fault classes into an array to become a sub-fault class array of the current fault class, and generating a fault diagnosis structure comprising a plurality of classification nodes.
Preferably, wherein the method further comprises:
further refining the classification result by using the idea of the ISODATA dynamic clustering method, which comprises the following steps: calculating the similarity between each group of standard data samples and the standard samples of the determined sub-fault classes, selecting the class with the minimum similarity as a new sub-fault class, and calculating new fuzzy statistics according to the following formula:
Figure GDA0001692221870000131
wherein tmp _ fuzzy is a fuzzy statistic;
when the difference between the fuzzy statistic and the former fuzzy statistic is less than a certain constant, ending; otherwise, performing refinement calculation again, and storing the refined result as the optimal classification in the subclass array.
In the implementation mode of the invention, aiming at the chromatographic analysis data of the dissolved gas in the transformer oil, the provided clustering algorithm combines the technology of a fuzzy clustering algorithm and a dynamic clustering algorithm.
For the fuzzy similarity relation matrix R, i.e. the similarity matrix, the element rij in the fuzzy similarity relation matrix R represents the similarity degree of the sample i and the sample j in a similarity concept. Therefore, selecting a good similarity definition is significant for the similarity relation matrix R. The invention selects weighted Euclidean distance closeness as similarity measurement between samples.
Figure GDA0001692221870000132
rij＝1-cNij 1≤c≤1.5，
Where c is a constant and ω k is the weight of the kth component, which is proportional to the averaging error of the kth component in the class of samples. Assuming there are ni groups of samples in the ith fault class, ω ik can be calculated by the following equation:
Figure GDA0001692221870000133
the step of solving the transfer closure t (R) of the fuzzy similarity relation matrix R is as follows:
1) setting the number of fault samples to be classified by N as N, and making k equal to log2N+1
2)
Figure GDA0001692221870000134
3) Let k-1 if k >0 let rij-r' ij (i-1-N, j-1-N) l go to 2)
4) Let t (R) ═ R', end.
And dynamic classification, wherein the step of obtaining the optimal classification structure comprises the following steps:
1) let the cutoff λ be 0.99 and the blur statistic m _ fuzzy be 0 as an initial variable.
2) Solving a lambda intercept matrix (t (R)) lambda of an equivalence closure
if(r’ij≥λ)
then tij＝1
else tij＝0)
3) Classification nodes are generated, each classification node corresponding to a fault class. In this process, it is first verified whether the first element i has been classified into the previous class, and if not, it is taken as the first sample data of the new failed class, and then it is checked whether the ith column in the next row j is equal to 1, and if so, it is classified into this class. Finally, the generated fault classes are programmed into an array to become a sub-fault class array of the current fault class.
4) Calculating a temporary fuzzy statistic tmp _ fuzzy, and if N samples are already classified into w classes, and the number of samples in each subclass is ni (i is 1-w), then:
Figure GDA0001692221870000141
5) let λ -0.01, judge whether λ is greater than 0.5, if yes, go to 2); otherwise, the next step is carried out.
6) And (4) classifying and refining levels, and refining the previous classification results by using the idea of an ISODATA dynamic clustering method for reference. The refinement process is to find the weighted Euclidean distance between each group of samples and the previously found standard samples of each sub-fault class, and the class with the minimum distance becomes the new homing class of the sample. Thus, a group of new sub-fault classes is obtained, new fuzzy statistic is obtained according to the formula, when the difference between the new fuzzy statistic and the previous fuzzy statistic is smaller than a certain constant, the refinement process is finished, otherwise, a new round of refinement is started. And storing the refined result as the optimal classification in the subclass array until the fuzzy clustering process is completely finished.
The fault diagnosis method of the invention is characterized in that the fault redundancy classification structure is adopted. The structure fully confirms the existence of ambiguity in transformer insulation fault diagnosis, and when fault diagnosis is carried out, after the strength degree of a certain discharge or heating fault expression is found according to the content of the characteristic gas, the fault cause of the phenomenon can be confirmed only by taking the past fault samples (the sample group is reduced to a smaller range) as reference, and then carrying out targeted inspection work.
Preferably, characteristic gas data is acquired from the monitoring node in step 105 for each of the plurality of transformers to be monitored.
Preferably, the characteristic gas data is matched with each classification node in the plurality of classification nodes in step 106, and the fault type of the transformer is determined according to the matching result.
Preferably, the matching the characteristic gas data with each classification node in the plurality of classification nodes and determining the fault type of the transformer according to the matching result includes:
and searching a best matching node in the next layer of nodes of the root node from the root node, and then iterating by taking the best matching node as the root node until the iteration is finished when the next layer of nodes are not available in the matching node, wherein the last node is the fault type of the transformer.
Preferably, the matching the characteristic gas data with each classification node in the plurality of classification nodes and determining the fault type of the transformer according to the matching result includes:
and directly searching all leaf nodes, taking all leaf nodes as the same layer, and selecting the most matched node as the fault type of the transformer.
In the embodiment of the present invention, the fuzzy fault diagnosis is a process of finding the most similar class in the previous classification result, and the process is completed in two ways: one is that from the root node, the best matching node is searched in the next layer of nodes of the root node as the first step solution, then the best matching node is used as the root node for the next round of search, and the iteration is continuously carried out until the last matching node has no next layer of nodes, and the iteration is finished, and the last node is the diagnosis result; the other is to search all leaf nodes directly, and all leaf nodes are used as the same layer, and the most matched node is the diagnosis result.
For example, the total hydrocarbon content of the C phase of the main transformer of the 500kV Pingguo transformer substation #2 exceeds the standard, and the C phase is analyzed to be a low-temperature overheating fault of the transformer. Tables 4-4 show the oil chromatographic trace.
TABLE 4-4 main transformer low-temperature overheat fault oil chromatogram data
Figure GDA0001692221870000151
The case is analyzed by using the fault diagnosis method, and the analysis result is consistent with the actual situation.
A first group:
closeness degree 1: the value: 193.77 diagnostic results: low temperature superheating
Closeness 2: the value: 203.22 diagnostic results: superheating
Closeness 3: the value: 222.17 diagnostic results: low voltage lead wire welding out burning loss
The final diagnosis result is: low temperature superheating
Second group:
closeness degree 1: the value: 137.72 diagnostic results: low temperature superheating
Closeness 2: the value: 153.172 diagnostic results: superheating
Closeness 3: the value: 176.1 diagnosis result: continuous spark discharge of suspended potential or oil breakdown between solid materials
The final diagnosis result is: low temperature superheating
Fig. 2 is a schematic diagram of a system 200 for determining a fault type of a transformer based on data analysis according to an embodiment of the present invention. As shown in fig. 2, a system 200 for determining a fault type of a transformer based on data analysis according to an embodiment of the present invention includes: the system comprises a standard data sample determining unit 201, a membership degree determining unit 202, a similarity matrix determining unit 203, a fault diagnosis structure determining unit 204, a characteristic gas data acquiring unit 205 and a fault type determining unit 206. Preferably, at the standard data sample determining unit 201, historical fault data associated with a plurality of transformers is acquired, and data format conversion is performed on each fault sample data in the historical fault data to generate a plurality of standard data samples, where each fault sample data includes: the type of fault and the corresponding content of the characteristic gases.
Preferably, the characteristic gas is dissolved gas in transformer oil, and comprises: h2、CH4、C2H2、C2H4And C2H6。
Preferably, the standard data sample determining unit performs data format conversion on each fault sample data in the historical fault data to generate a plurality of standard data samples, and includes:
PC2H2＝CC2H2/Num，
PC2H4＝CC2H4/Num，
PC2H6＝CC2H6/Num，
PCH4＝CCH4/Num，
PH2＝CH2/(Num+CH2)，
Num＝CC2H2+CC2H4+CC2H6+CCH4，
wherein, CC2H2Sample data C for each fault2H2The hydrocarbon number of (a); cC2H4Sample data C for each fault2H4The hydrocarbon number of (a); cC2H6Sample data C for each fault2H6The hydrocarbon number of (a); cCH4For CH in each failure sample data4The hydrocarbon number of (a);NUM is the total hydrocarbon value of each fault sample data; pC2H2、PC2H4、PC2H6、PCH4And PH2Respectively C in the converted standard data samples2H2、C2H4、C2H6、CH4And H2The content of (a).
Preferably, in the membership degree determining unit 202, a center sample is determined among the plurality of standard data samples, and the membership degree of each characteristic gas in each standard data sample is determined based on the center sample.
Preferably, the membership determining unit determines a center sample among the plurality of standard data samples, and determines the membership of each characteristic gas in each standard data sample based on the center sample, including:
Figure GDA0001692221870000171
Figure GDA0001692221870000172
Figure GDA0001692221870000173
wherein the content of the first and second substances,
Figure GDA0001692221870000174
the average component of the jth characteristic gas in the central sample; n is the number of standard data samples; x is the number ofijThe content of the jth characteristic gas in the ith standard data sample is shown; mu.sijAnd the membership degree of the jth characteristic gas in the ith standard data sample.
Preferably, in the similarity matrix determining unit 203, the similarity between any two standard data samples in the plurality of data samples is determined according to the membership of each characteristic gas in each standard data sample, so as to generate a similarity matrix.
Preferably, the similarity matrix determining unit determines the similarity between any two standard data samples in the plurality of data samples according to the membership of each characteristic gas in each standard data sample to generate the similarity matrix, including:
Figure GDA0001692221870000181
Figure GDA0001692221870000182
rij＝1-cNij 1≤c≤1.5，
wherein, the ith fault class has ni groups of samples, rijRepresenting the similarity between sample i and sample j; c is a constant, and ω k is the weight of the kth characteristic gas, which is proportional to the averaging error of the kth characteristic gas in the class of samples.
Preferably, at the fault diagnosis structure determining unit 204, the plurality of standard data samples are classified according to the similarity matrix to generate a fault diagnosis structure including a plurality of classification nodes, wherein each classification node corresponds to a single fault type.
Preferably, the fault diagnosis structure determining unit classifies the plurality of standard data samples according to the similarity matrix to generate a fault diagnosis structure including a plurality of classification nodes, and includes: and generating classification nodes, wherein each classification node corresponds to a fault class, in the process, firstly verifying whether a first fault data sample i is classified into a previous class, if not, using the first fault data sample i as first sample data of a new fault class, then determining whether a jth row and an ith column are equal to 1, if so, classifying the fault classes into the class, and finally, compiling the generated fault classes into an array to become a sub-fault class array of the current fault class, and generating a fault diagnosis structure comprising a plurality of classification nodes.
Preferably, the failure diagnosis structure determination unit further includes: further refining the classification result by using the idea of the ISODATA dynamic clustering method, which comprises the following steps: calculating the similarity between each group of standard data samples and the standard samples of the determined sub-fault classes, selecting the class with the minimum similarity as a new sub-fault class, and calculating new fuzzy statistics according to the following formula:
Figure GDA0001692221870000191
wherein tmp _ fuzzy is a fuzzy statistic;
when the difference between the fuzzy statistic and the former fuzzy statistic is less than a certain constant, ending; otherwise, performing refinement calculation again, and storing the refined result as the optimal classification in the subclass array.
Preferably, in the characteristic gas data obtaining unit 205, the characteristic gas data in each of the plurality of transformers to be monitored is obtained from the monitoring node.
Preferably, in the fault type determining unit 206, the characteristic gas data is matched with each of the plurality of classification nodes, and the fault type of the transformer is determined according to a matching result.
Preferably, the fault type determining unit matches the characteristic gas data with each of the plurality of classification nodes, and determines the fault type of the transformer according to a matching result, including: and searching a best matching node in the next layer of nodes of the root node from the root node, and then iterating by taking the best matching node as the root node until the iteration is finished when the next layer of nodes are not available in the matching node, wherein the last node is the fault type of the transformer.
Preferably, the fault type determining unit matches the characteristic gas data with each of the plurality of classification nodes, and determines the fault type of the transformer according to a matching result, including: and directly searching all leaf nodes, taking all leaf nodes as the same layer, and selecting the most matched node as the fault type of the transformer.
The system 200 for determining the fault type of the transformer based on the data analysis according to the embodiment of the present invention corresponds to the method 100 for determining the fault type of the transformer based on the data analysis according to another embodiment of the present invention, and is not described herein again.
The invention has been described with reference to a few embodiments. However, other embodiments of the invention than the one disclosed above are equally possible within the scope of the invention, as would be apparent to a person skilled in the art from the appended patent claims.
Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the technical field, unless explicitly defined otherwise herein. All references to "a/an/the [ device, component, etc ]" are to be interpreted openly as referring to at least one instance of said device, component, etc., unless explicitly stated otherwise. The steps of any method disclosed herein do not have to be performed in the exact order disclosed, unless explicitly stated.

Claims (16)

1. A method of determining a fault type of a transformer based on data analysis, the method comprising:
acquiring historical fault data associated with a plurality of transformers, and performing data format conversion on each fault sample data in the historical fault data to generate a plurality of standard data samples, wherein each fault sample data comprises: fault type and corresponding content of various characteristic gases;
determining a central sample in the plurality of standard data samples and determining a degree of membership of each characteristic gas in each standard data sample based on the central sample;
determining the similarity between any two standard data samples in the plurality of data samples according to the membership degree of each characteristic gas in each standard data sample to generate a similarity matrix;
classifying the plurality of standard data samples according to the similarity matrix to generate a fault diagnosis structure comprising a plurality of classification nodes, wherein each classification node corresponds to a single fault type;
acquiring characteristic gas data of each transformer in a plurality of transformers to be monitored from the monitoring node;
matching the characteristic gas data with each classification node in the plurality of classification nodes, and determining the fault type of the transformer according to the matching result;
the determining a center sample in the plurality of standard data samples and determining a degree of membership of each characteristic gas in each standard data sample based on the center sample comprises:
Figure FDA0002827686930000011
Figure FDA0002827686930000012
Figure FDA0002827686930000013
wherein the content of the first and second substances,
Figure FDA0002827686930000021
the average component of the jth characteristic gas in the central sample; n is the number of standard data samples; x is the number ofijThe content of the jth characteristic gas in the ith standard data sample is shown; mu.sijAnd the membership degree of the jth characteristic gas in the ith standard data sample.
2. The method of claim 1, wherein the characteristic gas is a dissolved gas in transformer oil, comprising: h2、CH4、C2H2、C2H4And C2H6。
3. The method of claim 2, wherein the converting the data format of each fault sample data in the historical fault data to generate a plurality of standard data samples comprises:
PC2H2＝CC2H2/Num，
PC2H4＝CC2H4/Num，
PC2H6＝CC2H6/Num，
PCH4＝CCH4/Num，
PH2＝CH2/(Num+CH2)，
Num＝CC2H2+CC2H4+CC2H6+CCH4，
wherein, CC2H2Sample data C for each fault2H2The hydrocarbon number of (a); cC2H4Sample data C for each fault2H4The hydrocarbon number of (a); cC2H6Sample data C for each fault2H6The hydrocarbon number of (a); cCH4For CH in each failure sample data4The hydrocarbon number of (a); NUM is the total hydrocarbon value of each fault sample data; pC2H2、PC2H4、PC2H6、PCH4And PH2Respectively C in the converted standard data samples2H2、C2H4、C2H6、CH4And H2The content of (a).
4. The method of claim 1, wherein determining the similarity between any two standard data samples in the plurality of data samples according to the membership of each signature gas in each standard data sample to generate a similarity matrix comprises:
Figure FDA0002827686930000022
Figure FDA0002827686930000023
rij＝1-cNij 1≤c≤1.5，
wherein, the ith fault class has ni groups of samples, rijRepresenting the similarity between sample i and sample j; c is a constant, and ω k is the weight of the kth characteristic gas, which is proportional to the averaging error of the kth characteristic gas in the class of samples.
5. The method of claim 4, wherein the classifying the plurality of standard data samples according to the similarity matrix to generate a fault diagnosis structure comprising a plurality of classification nodes comprises:
and generating classification nodes, wherein each classification node corresponds to a fault class, in the process, firstly verifying whether a first fault data sample i is classified into a previous class, if not, using the first fault data sample i as first sample data of a new fault class, then determining whether a jth row and an ith column are equal to 1, if so, classifying the fault classes into the class, and finally, compiling the generated fault classes into an array to become a sub-fault class array of the current fault class, and generating a fault diagnosis structure comprising a plurality of classification nodes.
6. The method of claim 5, further comprising:
further refining the classification result by using the idea of the ISODATA dynamic clustering method, which comprises the following steps: calculating the similarity between each group of standard data samples and the standard samples of the determined sub-fault classes, selecting the class with the minimum similarity as a new sub-fault class, and calculating new fuzzy statistics according to the following formula:
Figure FDA0002827686930000031
wherein tmp _ fuzzy is a fuzzy statistic;
when the difference between the fuzzy statistic and the former fuzzy statistic is less than a certain constant, ending; otherwise, performing refinement calculation again, and storing the refined result as the optimal classification in the subclass array.
7. The method of claim 1, wherein matching the signature gas data with each of the plurality of classification nodes and determining a fault type of the transformer based on the matching comprises:
and searching a best matching node in the next layer of nodes of the root node from the root node, and then iterating by taking the best matching node as the root node until the iteration is finished when the next layer of nodes are not available in the matching node, wherein the last node is the fault type of the transformer.
8. The method of claim 1, wherein matching the signature gas data with each of the plurality of classification nodes and determining a fault type of the transformer based on the matching comprises:
and directly searching all leaf nodes, taking all leaf nodes as the same layer, and selecting the most matched node as the fault type of the transformer.
9. A system for determining a fault type of a transformer based on data analysis, the system comprising:
the standard data sample determining unit is used for acquiring historical fault data associated with a plurality of transformers and performing data format conversion on each fault sample data in the historical fault data to generate a plurality of standard data samples, wherein each fault sample data comprises: fault type and corresponding content of various characteristic gases;
a membership degree determination unit for determining a center sample among the plurality of standard data samples and determining a membership degree of each characteristic gas in each standard data sample based on the center sample;
the similarity matrix determining unit is used for determining the similarity between any two standard data samples in the plurality of data samples according to the membership of each characteristic gas in each standard data sample so as to generate a similarity matrix;
a fault diagnosis structure determination unit, configured to classify the multiple standard data samples according to the similarity matrix to generate a fault diagnosis structure including multiple classification nodes, where each classification node corresponds to a single fault type;
the characteristic gas data acquisition unit is used for acquiring characteristic gas data in each transformer of the plurality of transformers to be monitored from the monitoring node;
the fault type determining unit is used for matching the characteristic gas data with each classification node in the plurality of classification nodes and determining the fault type of the transformer according to the matching result;
the membership degree determination unit determines a center sample among the plurality of standard data samples and determines a membership degree of each characteristic gas in each standard data sample based on the center sample, including:
Figure FDA0002827686930000051
Figure FDA0002827686930000052
Figure FDA0002827686930000053
wherein the content of the first and second substances,
Figure FDA0002827686930000054
the average component of the jth characteristic gas in the central sample; n is the number of standard data samples; x is the number ofijThe content of the jth characteristic gas in the ith standard data sample is shown; mu.sijAnd the membership degree of the jth characteristic gas in the ith standard data sample.
10. The system of claim 9, wherein the characteristic gas is a dissolved gas in transformer oil, comprising: h2、CH4、C2H2、C2H4And C2H6。
11. The system according to claim 10, wherein the standard data sample determination unit performs data format conversion on each fault sample data in the historical fault data to generate a plurality of standard data samples, including:
PC2H2＝CC2H2/Num，
PC2H4＝CC2H4/Num，
PC2H6＝CC2H6/Num，
PCH4＝CCH4/Num，
PH2＝CH2/(Num+CH2)，
Num＝CC2H2+CC2H4+CC2H6+CCH4，
wherein, CC2H2Sample data C for each fault2H2The hydrocarbon number of (a); cC2H4Sample data C for each fault2H4The hydrocarbon number of (a); cC2H6Sample data C for each fault2H6The hydrocarbon number of (a); cCH4For CH in each failure sample data4The hydrocarbon number of (a); NUM is the total hydrocarbon value of each fault sample data; pC2H2、PC2H4、PC2H6、PCH4And PH2Respectively C in the converted standard data samples2H2、C2H4、C2H6、CH4And H2The content of (a).
12. The system of claim 9, wherein the similarity matrix determining unit determines the similarity between any two standard data samples in the plurality of data samples according to the membership of each characteristic gas in each standard data sample to generate the similarity matrix, and comprises:
Figure FDA0002827686930000061
Figure FDA0002827686930000062
rij＝1-cNij 1≤c≤1.5，
wherein, the ith fault class has ni groups of samples, rijRepresenting the similarity between sample i and sample j; c is a constant, and ω k is the weight of the kth characteristic gas, which is proportional to the averaging error of the kth characteristic gas in the class of samples.
13. The system according to claim 12, wherein the failure diagnosis structure determining unit classifies the plurality of standard data samples according to the similarity matrix to generate a failure diagnosis structure including a plurality of classification nodes, including:
and generating classification nodes, wherein each classification node corresponds to a fault class, in the process, firstly verifying whether a first fault data sample i is classified into a previous class, if not, using the first fault data sample i as first sample data of a new fault class, then determining whether a jth row and an ith column are equal to 1, if so, classifying the fault classes into the class, and finally, compiling the generated fault classes into an array to become a sub-fault class array of the current fault class, and generating a fault diagnosis structure comprising a plurality of classification nodes.
14. The system of claim 13, wherein the fault diagnosis structure determination unit further comprises:
further refining the classification result by using the idea of the ISODATA dynamic clustering method, which comprises the following steps: calculating the similarity between each group of standard data samples and the standard samples of the determined sub-fault classes, selecting the class with the minimum similarity as a new sub-fault class, and calculating new fuzzy statistics according to the following formula:
Figure FDA0002827686930000071
wherein tmp _ fuzzy is a fuzzy statistic;
when the difference between the fuzzy statistic and the former fuzzy statistic is less than a certain constant, ending; otherwise, performing refinement calculation again, and storing the refined result as the optimal classification in the subclass array.
15. The system of claim 9, wherein the fault type determination unit matches the characteristic gas data with each of the plurality of classification nodes and determines the fault type of the transformer according to the matching result, and comprises:
and searching a best matching node in the next layer of nodes of the root node from the root node, and then iterating by taking the best matching node as the root node until the iteration is finished when the next layer of nodes are not available in the matching node, wherein the last node is the fault type of the transformer.
16. The system of claim 9, wherein the fault type determination unit matches the characteristic gas data with each of the plurality of classification nodes and determines the fault type of the transformer according to the matching result, and comprises:
and directly searching all leaf nodes, taking all leaf nodes as the same layer, and selecting the most matched node as the fault type of the transformer.
CN201810233161.3A 2018-03-21 2018-03-21 Method and system for determining fault type of transformer based on data analysis Active CN108491503B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810233161.3A CN108491503B (en) 2018-03-21 2018-03-21 Method and system for determining fault type of transformer based on data analysis

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810233161.3A CN108491503B (en) 2018-03-21 2018-03-21 Method and system for determining fault type of transformer based on data analysis

Publications (2)

Publication Number Publication Date
CN108491503A CN108491503A (en) 2018-09-04
CN108491503B true CN108491503B (en) 2021-03-12

Family

ID=63318777

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810233161.3A Active CN108491503B (en) 2018-03-21 2018-03-21 Method and system for determining fault type of transformer based on data analysis

Country Status (1)

Country Link
CN (1) CN108491503B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109374324A (en) * 2018-09-29 2019-02-22 国网山西省电力公司阳泉供电公司 Transformer mechanical failure diagnostic method based on vibration signal characteristics matrix similarity
CN113379210A (en) * 2021-05-31 2021-09-10 三一重型装备有限公司 Motor fault detection method and device, heading machine and readable storage medium
CN114545294B (en) * 2022-01-14 2023-06-16 国电南瑞科技股份有限公司 Transformer fault diagnosis method, system, storage medium and computing device

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110182522A1 (en) * 2010-01-25 2011-07-28 King Jen Chang Method for multi-layer classifier

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104101794A (en) * 2014-02-19 2014-10-15 江苏倍尔科技发展有限公司 Integrated control system for power transformer
CN104156568A (en) * 2014-07-22 2014-11-19 国家电网公司 Transformer fault diagnosis method on basis of weighted gray correlation and fuzzy clustering
CN106896219B (en) * 2017-03-28 2019-01-29 浙江大学 The identification of transformer sub-health state and average remaining lifetime estimation method based on Gases Dissolved in Transformer Oil data
CN107063349A (en) * 2017-04-17 2017-08-18 云南电网有限责任公司电力科学研究院 A kind of method and device of Fault Diagnosis Method of Power Transformer
CN107656154B (en) * 2017-09-18 2019-11-26 杭州安脉盛智能技术有限公司 Based on the Diagnosis Method of Transformer Faults for improving Fuzzy C-Means Cluster Algorithm

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110182522A1 (en) * 2010-01-25 2011-07-28 King Jen Chang Method for multi-layer classifier

Also Published As

Publication number Publication date
CN108491503A (en) 2018-09-04

Similar Documents

Publication Publication Date Title
CN112699913B (en) Method and device for diagnosing abnormal relationship of household transformer in transformer area
CN109685138B (en) XLPE power cable partial discharge type identification method
CN108491503B (en) Method and system for determining fault type of transformer based on data analysis
Taha et al. Optimal ratio limits of rogers' four-ratios and IEC 60599 code methods using particle swarm optimization fuzzy-logic approach
Sun et al. Fault diagnosis of power transformers using computational intelligence: A review
CN105930901B (en) A kind of Diagnosis Method of Transformer Faults based on RBPNN
CN110929847A (en) Converter transformer fault diagnosis method based on deep convolutional neural network
Malik et al. Fuzzy reinforcement learning based intelligent classifier for power transformer faults
CN107037306B (en) Transformer fault dynamic early-warning method based on Hidden Markov Model
CN104535865A (en) Comprehensive diagnosing method for operation troubles of power transformer based on multiple parameters
EP2410312A1 (en) A method for computer-assisted analyzing of a technical system
Siddique et al. Artificial neural networks based incipient fault diagnosis for power transformers
CN108021942A (en) A kind of power transformer incipient fault diagnostic method
CN113987294A (en) CVT (continuously variable transmission) online fault diagnosis method based on genetic optimization GRU (generalized regression Unit) neural network
CN109840548A (en) One kind being based on BP neural network Diagnosis Method of Transformer Faults
CN115881238A (en) Model training method, transformer fault diagnosis method and related device
CN111999591B (en) Method for identifying abnormal state of primary equipment of power distribution network
CN117113166A (en) Industrial boiler fault detection method based on improved integrated learning
Luo et al. Prediction for dissolved gas in power transformer oil based on TCN and GCN
CN105137238A (en) Fault diagnosis system for gas insulation combination electric appliance
CN105741184A (en) Transformer state evaluation method and apparatus
CN112085064A (en) Transformer fault diagnosis method based on multi-classification probability output of support vector machine
Qaedi et al. Improvement in power transformer intelligent dissolved gas analysis method
Dehghani et al. Distribution feeder classification based on self organized maps (case study: Lorestan province, Iran)
Pacori et al. Identification of internal failure in power transformers using fuzzy logic through the dissolved gas analysis in mineral insulating oil

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
GR01 Patent grant
GR01 Patent grant