CN109657720A - A kind of inline diagnosis method of power transformer shorted-turn fault - Google Patents
A kind of inline diagnosis method of power transformer shorted-turn fault Download PDFInfo
- Publication number
- CN109657720A CN109657720A CN201811560473.1A CN201811560473A CN109657720A CN 109657720 A CN109657720 A CN 109657720A CN 201811560473 A CN201811560473 A CN 201811560473A CN 109657720 A CN109657720 A CN 109657720A
- Authority
- CN
- China
- Prior art keywords
- transformer
- measured
- short circuit
- monitoring
- monitoring data
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Testing Of Short-Circuits, Discontinuities, Leakage, Or Incorrect Line Connections (AREA)
- Housings And Mounting Of Transformers (AREA)
Abstract
The invention discloses a kind of inline diagnosis methods of power transformer shorted-turn fault, it include: the monitoring data before and after short circuit that (1) collects each monitoring index record of case transformer, it examines whether each monitoring index occurs conspicuousness variation before and after short circuit, filters out the monitoring index for thering is conspicuousness to change before and after short circuit；(2) monitoring data for the monitoring index record for having conspicuousness to change before and after short circuit are input in Random Forest model, training short trouble analysis model on line；(3) in the Random Forest model after the monitoring data to be measured input optimization of transformer to be measured, will export to the inline diagnosis of monitoring data to be measured as a result, judge transformer to be measured whether failure.The above method only need to carry out remote analysis to the monitoring data to be measured of transformer to be measured, and field test is not needed for offline inspection method, saves testing expenses, improves diagnosis efficiency.
Description
Technical field
The present invention relates to fault diagnosis fields, and in particular to a kind of inline diagnosis side of power transformer shorted-turn fault
Method.
Background technique
Power transformer is one of important equipment of network system, carries the work of power grid hinge.Prevent electric power in time
The potential faults of transformer have a very important significance the safe and stable operation for ensureing power grid.According to statistics, about 25% change
Depressor turn-to-turn accident is caused after being deformed by transformer winding after short-circuit impact, the severity and short circuit current of failure
Size it is related.
The classification of short circuit current bad condition can be divided into three classes: it is short that I grades of bad conditions refer to that short circuit current can be born in maximum
Between the 50%~70% of road electric current, and number has reached 6 times or more；II grades of bad conditions refer to short circuit current in maximum
It can bear between the 70%~100% of short circuit current；III level bad condition refers to that short circuit current can bear short circuit current in maximum
100% or more.
Currently, the diagnosis of Domestic Transformers shorted-turn fault relies primarily on observation main transformer protection action situation, oil sample point
Analyse situation and short-circuit current value.If main transformer, which does not trip, oil sample is without exception and short-circuit current value is less than 50% can bear maximum short circuit
Current value, then it is assumed that obvious fault is not present in main transformer, reinforces subsequent oil chromatography tracking and monitoring.If main transformer does not trip, oil sample is abnormal
Or short-circuit current value can bear maximum short circuit current value greater than 50%, then need to arrange to have a power failure as early as possible and test, including frequency response
Test, short-circuit impedance test, capacitance test etc., comprehensive all test results carry out specific analysis, judge whether winding is rushed
Hit have occurred it is severely deformed, to determine that can transformer continue to run.If main transformer non-charge protection act is needed to the transformer
Arrange to test Mobile state anti-short circuit capability analysis of going forward side by side, comprehensive all test results carry out specific analyses to decide whether carry out office
It puts test or hangs cover, strip inspection.
The above short trouble diagnostic method finally requires to carry out power failure test or strip inspection to confirm transformer winding
It is deformed after whether being hit, belongs to offline diagnosis, there are complicated for operation, test difficulty, dependence artificial experiences etc. to lack
Point.
The diagnostic method of do not need the have a power failure test opposite with offline diagnosis is known as inline diagnosis method, is such as believed with ultrasonic wave
Number or other broadband signals variation come detect whether deformation, measured using radar map around packet size and observation whether deform with
And deformation is detected whether using the variation of magnetic flux distribution etc..
Above method is easy to operate, and can directly observe inside transformer structure, makes very accurate judging result,
But the disadvantage is that new technology and the application cost of new equipment are high, these methods are also failed in investment practical application and a wide range of popularization.
The patent specification of 108197639 A of Publication No. CN discloses a kind of transformer fault based on random forest
Diagnostic method, comprising: dissolved gas in oil concentration data and corresponding fault category are as training in acquisition transformer
Sample；Establish failure decision tree according to the generation step of decision tree according to the training sample: according to failure decision tree establish with
Machine forest model；Acquire unknown failure classification failure gas concentration data, be input to the Random Forest model, with it is described with
Machine forest model obtains fault category.It is established using dissolved gas in oil concentration data in transformer as training sample
Random Forest model has accurate diagnosis to entire transformer fault, and stability is high, can be applied to transformer diagnosis technology
Field.
The patent specification of 107025514 A of Publication No. CN discloses a kind of dynamic evaluation transformer equipment state
Evaluation method, comprising the following steps: count the historical data of each quantity of state of transformer equipment；Based on going through for each quantity of state
History data calculate the different degree of each quantity of state using random forests algorithm；By the different degree of each quantity of state according to modifying factor
The quantized value of element is modified, then by revised quantity of state different degree discretization, is converted into preset parameter, and reflection is every
The different degree grade of a quantity of state；Status assessment is carried out to transformer equipment according to the different degree grade of each quantity of state, is determined
The real-time status of equipment；Using the different degree grade of each quantity of state in method for evaluating state as historical data, and repeat with
Upper step, realize evaluation method it is white I optimize and dynamic evaluation.
Summary of the invention
For shortcoming existing for this field, the present invention provides a kind of the online of power transformer shorted-turn fault
Diagnostic method is distinguished testing data by random forests algorithm for the transformer that deformation occurred, has deformation operation data
Operation data is deformed with the history of the transformer and operates normally data and is matched, and is realized long-distance intelligent diagnosis, is not only saved examination
Expense is tested, the multi-source information of online monitoring data is also combined, is more fully diagnosed.
A kind of inline diagnosis method of power transformer shorted-turn fault, comprising:
(1) monitoring data before and after short circuit for collecting each monitoring index record of case transformer, examine each monitoring to refer to
It is marked on whether short circuit front and back occurs conspicuousness variation, filters out the monitoring index for thering is conspicuousness to change before and after short circuit；
(2) using the monitoring data for the monitoring index record for thering is before and after short circuit conspicuousness to change as training set, be input to
In machine forest model, training short trouble analysis model on line；
(3) it by the Random Forest model after the monitoring data to be measured input optimization of transformer to be measured, exports to prison to be measured
The inline diagnosis of measured data as a result, judge transformer to be measured whether failure.
In step (1), it is preferable that each monitoring index is voltage, electric current, oil temperature and power.
Preferably, whether each monitoring index of inspection occurs the specific steps packet of conspicuousness variation before and after short circuit
It includes:
A. test of normality is carried out to the distribution of the online monitoring data of monitoring index record；
If b. meeting normal distribution, the arithmetic mean of instantaneous value of the monitoring data after examining short circuit preceding and short circuit using ANOVA is
It is no that there were significant differences；If being unsatisfactory for normal distribution, monitoring index after examining short circuit preceding and short circuit using Kruskal-Wallis
Median, whether there were significant differences for very poor and distribution.
Test of normality, ANOVA inspection and the Kruskal-Wallis inspection can be carried out by SPSS software.
When the significance value of monitoring index is less than 0.05, it is believed that there are significant differences before and after short circuit for the monitoring index.
In step (2), the random forest is a kind of important integrated learning approach based on Bagging, can be located
Discrete data is managed, can also handle continuous data, and also have good applicability to high dimensional data.
Preferably, the specific steps of the training short trouble analysis model on line include:
A. it is concentrated use in Bootstraping method from training and puts back to sampling at random and select m sample, progress n times are adopted altogether
Sample generates n sub- training sets；
B. for n sub- training sets, n decision tree is respectively trained；
C. it is divided according to information gain than maximum feature when every decision tree division, until all instructions of the node
Practice sample and belongs to same class；
D. more decision trees of generation are formed into random forest.
Preferably, adjustment can be optimized to Random Forest model described in step (2), promoted diagnostic accuracy, can be adopted
Specific method has:
1, phase difference is constructed as new feature according to the three-phase imbalance principle of transformer.
Building phase difference is as the foundation of new feature can be by A, B, C three-phase windings to the same voltage class of transformer
Amplitude-frequency response characteristic is compared, and calculates the unbalance factor of three-phase, to judge whether transformer winding deforms.Principle is, A,
B, the voltage and current of C three-phase under normal circumstances amplitude be it is equal, if a certain phase deforms, three-phase would be at injustice
Weighing apparatus state.
2, decision tree quantity gradient is set, classification performance comparison is carried out, weighs diagnostic accuracy and efficiency of algorithm.
Classification performance is improved with the increase of decision tree subtree quantity, but increase rate can be gradually reduced.Therefore, it is necessary to
Decision tree quantity is reasonably selected, classification performance and the comprehensive preferably result of operation duration are obtained.It is excellent for higher efficiency of algorithm
Selection of land, operation duration are 5~10 minutes.
Output described in step (3) to the inline diagnosis of monitoring data to be measured as a result, judge transformer to be measured whether failure
Specific method can be each decision tree of output to the classification results of the monitoring data to be measured, if classification results are failure
It is normal decision tree quantity that decision tree quantity, which is less than classification results, judges that transformer to be measured is normal；If classification results are event
It is normal decision tree quantity that the decision tree quantity of barrier, which is greater than classification results, judges transformer to be measured for failure.
Compared with prior art, the present invention major advantage includes:
(1) need to only remote analysis be carried out to the monitoring data to be measured of transformer to be measured, for offline inspection method not
Field test is needed, testing expenses are saved, improves diagnosis efficiency.
(2) multi-source informations of the online monitoring datas such as transformer voltage, electric current, power and oil temperature is utilized in fusion, can be into
Row more comprehensively diagnoses.
Detailed description of the invention
Fig. 1 is the flow chart of the inline diagnosis method of the power transformer shorted-turn fault of embodiment 1.
Specific embodiment
With reference to the accompanying drawing and specific embodiment, the present invention is further explained.It should be understood that these embodiments are merely to illustrate
The present invention rather than limit the scope of the invention.In the following examples, the experimental methods for specific conditions are not specified, usually according to
Normal condition, or according to the normal condition proposed by manufacturer.
Embodiment 1
As shown in Figure 1, the process of the inline diagnosis method of power transformer shorted-turn fault includes:
S01, collects the monitoring data before and after short circuit of each monitoring index record of case transformer, and monitoring index includes
Voltage, four class of electric current, oil temperature and power.
S02, examines whether above-mentioned monitoring index occurs conspicuousness variation before and after short circuit.If above-mentioned monitoring data meet just
State distribution, whether there were significant differences for the arithmetic mean of instantaneous value of the monitoring data after examining short circuit preceding and short circuit using ANOVA；If discontented
Sufficient normal distribution, examine short circuit preceding using Kruskal-Wallis is with the median of monitoring index, very poor and distribution after short circuit
It is no that there were significant differences.
S03 filters out the monitoring index for having conspicuousness to change before and after short circuit, is input in Random Forest model, and training is short
Road on-line fault diagnosis model；
S04 optimizes adjustment to Random Forest model, constructs phase difference according to the three-phase imbalance principle of transformer and makees
For new feature, it is 30,100 and 200 that decision tree quantity, which is respectively set, carries out classification performance comparison, and preferentially parameter is to weigh diagnosis
Precision and efficiency of algorithm；
The monitoring data to be measured of transformer to be measured are input in the Random Forest model after optimization, export each by S05
Decision tree is to the classification results of the monitoring data to be measured, if the decision tree quantity that classification results are failure is positive less than classification results
Normal decision tree quantity then judges that the transformer to be measured is normal；If the decision tree quantity that classification results are failure is greater than classification
As a result it is normal decision tree quantity, then judges the transformer to be measured for failure.
(1) the transformer conduct to have deformed after short circuit through offline inspection display winding occurs using Zhejiang power grid six
Case transformer.Collect voltage, electric current, oil temperature and the power of the short circuit front and back monitored under above-mentioned six Three-Phase Transformer three windings
Data, according to being divided into normal group and failure group before short circuit and after short circuit.
Whether the arithmetic mean of instantaneous value for carrying out the monitoring data of the normal group of variance analysis judgement and failure group first is variant.Side
Difference analysis needs to meet three assumed conditions: normality, neat variance and independence.Test of normality the results show that each index P
Therefore value is 0.001 hereinafter, refuse normal distribution null hypothesis, it is believed that all indexs all significantly disobey normal distribution, therefore
It cannot be examined using independence T and common ANOVA, consideration are used without the non-parametric test side for meeting normal assumption
Method Kruskal-Wallis is examined.
Kruskal-Wallis inspection is otherwise known as rank sum test, can be used to differentiate calculate two independent samples from
Whether two overall distribution positions have difference.Using Kruskal-Wallis examine failure group and normal group data median,
Very poor and distribution whether there is difference, and the results are shown in Table 1.Generally, it is considered that recognizing when the conspicuousness of statistic is less than 0.05
There is significant difference between the two groups for the statistic.Therefore, in addition to the side 110kV A phase current magnitude, the side 220kV A phase current width
Value and the side 220kV have the median of work value other than failure group and normal group difference be not significant without work value, the side 220kV, failure group
The median for remaining monitoring index normally organized, very poor and distribution have differences, and illustrate electric current, voltage, power and oil temperature
Four class monitoring indexes are for judging whether failure can provide effective information to transformer.
1 nonparametric Kruskal-Wallis of table is examined
Therefore selected voltage, electric current, power and four class monitoring index of oil temperature are regard as feature, is input to random forest
Model is trained, using ten folding cross validations, to judge the precision of diagnostic model.Mainly use three precision indexs: quasi-
True rate (Precision), recall rate (Recall) and F1 score.What accuracy rate was measured is the erroneous judgement situation to failure, accuracy
Higher, False Rate is smaller.What recall rate was measured is to the situation of failing to judge of failure, and recall rate is higher, and misdetection rate is smaller.Ordinary circumstance
Under, recall rate can be reduced while improving accuracy rate, F1 score is the index for integrating the two.The calculating side of three kinds of indexs
Method is as follows:
As shown in table 2, comprehensive observing accuracy rate, recall rate and F1 score, Random Forest model is in addition on transformer 2
Other than diagnosis effect is slightly lower, for the F1 score on other five transformers all 80% or more, diagnosis effect is preferable.
Diagnostic result of 2 Random Forest model of table on 6 transformers
| Transformer 1 | Transformer 2 | Transformer 3 | Transformer 4 | Transformer 5 | Transformer 6 | |
| Accuracy rate | 0.9861 | 0.8792 | 0.9887 | 0.9667 | 0.9836 | 0.8595 |
| Recall rate | 0.985 | 0.6947 | 0.9741 | 0.8325 | 0.9681 | 0.7804 |
| F1 score | 0.9856 | 0.7761 | 0.9813 | 0.8946 | 0.9758 | 0.818 |
In order to further increase diagnostic accuracy, consider to increase phase difference as new feature, foundation is can be by transformer
A, B, C three-phase windings amplitude-frequency response characteristic of same voltage class are compared, and the unbalance factor of three-phase are calculated, to judge transformation
Whether device winding deforms.Principle is, the voltage and current of A, B, C three-phase under normal circumstances amplitude be it is equal, if a certain
It mutually deforms, three-phase would be at non-equilibrium state.
Therefore, when feature merges, consider to calculate the AB phase of every side and the electric current and voltage difference of BC phase, make
For new feature.For the raw information held up to, voltage originally, current data still retain, and have only newly increased 12
Characteristic value.
After increasing phase difference index, classification performance of the Random Forest model on each transformer is calculated again.Such as 3 institute of table
Show, after increasing phase difference, Random Forest model improves 13.71%, F1 score to the recall rate of transformer 2 and improves
6.4%.Meanwhile the Random Forest model for increasing phase difference improves 5% left side to the recall rate and F1 score of transformer 6
It is right.
Table 3 increases the diagnostic result comparison of the Random Forest model before and after phase difference index
In Random Forest model, more decision tree subtree can allow model to have better performance, but subtree is excessive, can allow
Model running obtains very slow.In the present embodiment, the record number of case transformer is all between 20,000~100,000, decision tree quantity
When being set as 500, runing time has just been more than 20 minutes.
After comprehensively considering, it is 30,100 and 200 that decision tree quantity, which is respectively set, carries out classification performance comparison, random forest
The results are shown in Table 4 for model running, and classification performance is improved with the increase of decision tree subtree quantity, but increase rate gradually subtracts
It is small.Therefore, when trade-off decision tree quantity is 200, a classification performance can be obtained and operation duration is comprehensive preferably as a result, operation
Shi Changwei 5~10 minutes.
The operation result of the Random Forest model of the different decision tree quantity of table 4
After feature merging and arameter optimization, the inline diagnosis side of the power transformer shorted-turn fault of the present embodiment
Method has a more considerable diagnostic result on six transformers, to after short circuit winding deformation fault identification it is accurate
Rate, recall rate and F1 score have all reached 85% or more.
In addition, it should also be understood that, those skilled in the art can be to this hair after having read foregoing description content of the invention
Bright to make various changes or modifications, these equivalent forms also fall within the scope of the appended claims of the present application.
Claims (8)
1. a kind of inline diagnosis method of power transformer shorted-turn fault, comprising:
(1) monitoring data before and after short circuit for collecting each monitoring index record of case transformer, examine each monitoring index to exist
Whether short-circuit front and back occurs conspicuousness variation, filters out the monitoring index for having conspicuousness to change before and after short circuit；
(2) it using the monitoring data for the monitoring index record for thering is conspicuousness to change before and after short circuit as training set, is input to random gloomy
In woods model, training short trouble analysis model on line；
(3) it by the Random Forest model after the monitoring data to be measured input optimization of transformer to be measured, exports to monitoring number to be measured
According to inline diagnosis as a result, judge transformer to be measured whether failure.
2. the inline diagnosis method of power transformer shorted-turn fault according to claim 1, which is characterized in that described
Each monitoring index be voltage, electric current, oil temperature and power.
3. the inline diagnosis method of power transformer shorted-turn fault according to claim 1, which is characterized in that described
Each monitoring index of inspection the specific steps of conspicuousness variation whether occur before and after short circuit include:
A. test of normality is carried out to the distribution of the online monitoring data of monitoring index record；
If b. meeting normal distribution, whether the arithmetic mean of instantaneous value of the monitoring data after examining short circuit preceding and short circuit using ANOVA has
Significant difference；If being unsatisfactory for normal distribution, the middle position of monitoring index after examining short circuit preceding and short circuit using Kruskal-Wallis
Whether there were significant differences for several, very poor and distribution.
4. the inline diagnosis method of power transformer shorted-turn fault according to claim 1, which is characterized in that described
The specific steps of training short trouble analysis model on line include:
A. it is concentrated use in Bootstraping method from training and puts back to sampling at random and select m sample, altogether progress n times sampling,
Generate n sub- training sets；
B. for n sub- training sets, n decision tree is respectively trained；
C. it is divided according to information gain than maximum feature when every decision tree division, until all trained samples of the node
Example belongs to same class；
D. more decision trees of generation are formed into random forest.
5. the inline diagnosis method of power transformer shorted-turn fault according to claim 1, which is characterized in that step
Suddenly Random Forest model described in (2) optimizes adjustment, promotes diagnostic accuracy.
6. the inline diagnosis method of power transformer shorted-turn fault according to claim 5, which is characterized in that described
Random Forest model optimize adjustment specific method be according to the three-phase imbalance principle of transformer construct phase difference make
For new feature.
7. the inline diagnosis method of power transformer shorted-turn fault according to claim 5, which is characterized in that described
Random Forest model optimize the specific method of adjustment as setting decision tree quantity gradient, carry out classification performance comparison, power
The diagnostic accuracy that weighs and efficiency of algorithm.
8. the inline diagnosis method of power transformer shorted-turn fault according to claim 1, which is characterized in that described
Output to the inline diagnosis of monitoring data to be measured, as a result, judging transformer to be measured, whether the specific method of failure is each to export
Decision tree is to the classification results of the monitoring data to be measured, if the decision tree quantity that classification results are failure is less than classification results
Normal decision tree quantity, judges that transformer to be measured is normal；If the decision tree quantity that classification results are failure is greater than classification knot
Fruit is normal decision tree quantity, judges transformer to be measured for failure.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811560473.1A CN109657720B (en) | 2018-12-20 | 2018-12-20 | On-line diagnosis method for turn-to-turn short circuit fault of power transformer |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811560473.1A CN109657720B (en) | 2018-12-20 | 2018-12-20 | On-line diagnosis method for turn-to-turn short circuit fault of power transformer |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109657720A true CN109657720A (en) | 2019-04-19 |
| CN109657720B CN109657720B (en) | 2021-05-11 |
Family
ID=66115132
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811560473.1A Active CN109657720B (en) | 2018-12-20 | 2018-12-20 | On-line diagnosis method for turn-to-turn short circuit fault of power transformer |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109657720B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113872200A (en) * | 2021-10-13 | 2021-12-31 | 国网江苏省电力有限公司电力科学研究院 | Medium-voltage distribution network power failure event diagnosis and identification method, system and storage medium |
| CN114019298A (en) * | 2021-09-28 | 2022-02-08 | 中电华创(苏州)电力技术研究有限公司 | PCC-SVM-based generator rotor turn-to-turn short circuit online monitoring method |
| US11474163B2 (en) * | 2019-08-12 | 2022-10-18 | Wuhan University | Power transformer winding fault positioning method based on deep convolutional neural network integrated with visual identification |
| CN116799741A (en) * | 2023-05-24 | 2023-09-22 | 华斗数字科技(上海)有限公司 | Precision equipment short-circuit monitoring protection method and system based on slope detection |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1441257A (en) * | 2003-03-27 | 2003-09-10 | 河海大学 | In-situ fault diagnosing technology for turn-to-turn short-circuit of transformer windings based on change in loss |
| US20160343133A1 (en) * | 2014-07-04 | 2016-11-24 | Arc Devices Limited | Non-Touch Optical Detection of Vital Signs From Variation Amplification Subsequent to Multiple Frequency Filters |
| CN106372747A (en) * | 2016-08-27 | 2017-02-01 | 天津大学 | Random forest-based zone area reasonable line loss rate estimation method |
| CN107516108A (en) * | 2017-08-15 | 2017-12-26 | 国网四川省电力公司电力科学研究院 | Grader creation method and partial discharge of transformer method of fault pattern recognition |
| CN107782442A (en) * | 2017-10-24 | 2018-03-09 | 华北电力大学（保定） | Transformer multiple features parameter selection method based on big data and random forest |
| CN108062571A (en) * | 2017-12-27 | 2018-05-22 | 福州大学 | Diagnosing failure of photovoltaic array method based on differential evolution random forest grader |
| CN108197639A (en) * | 2017-12-14 | 2018-06-22 | 佛山科学技术学院 | A kind of Diagnosis Method of Transformer Faults based on random forest |
| CN108398612A (en) * | 2018-01-12 | 2018-08-14 | 广州市扬新技术研究有限责任公司 | A kind of urban track traffic DC power-supply system short trouble localization method |
| CN108594067A (en) * | 2018-04-02 | 2018-09-28 | 湖南大学 | A kind of Multi-port direct-current distribution network line short fault distance measuring method |
| CN108693478A (en) * | 2018-04-17 | 2018-10-23 | 北京理工大学 | A kind of method for detecting leakage of lithium-ion-power cell |
| CN108872778A (en) * | 2018-07-03 | 2018-11-23 | 安徽理工大学 | A kind of transformer winding fault diagnostic system based on correlation dimension and random forest |
| CN108959732A (en) * | 2018-06-15 | 2018-12-07 | 西安科技大学 | A kind of transmission line malfunction kind identification method based on convolutional neural networks |
-
2018
- 2018-12-20 CN CN201811560473.1A patent/CN109657720B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1441257A (en) * | 2003-03-27 | 2003-09-10 | 河海大学 | In-situ fault diagnosing technology for turn-to-turn short-circuit of transformer windings based on change in loss |
| US20160343133A1 (en) * | 2014-07-04 | 2016-11-24 | Arc Devices Limited | Non-Touch Optical Detection of Vital Signs From Variation Amplification Subsequent to Multiple Frequency Filters |
| CN106372747A (en) * | 2016-08-27 | 2017-02-01 | 天津大学 | Random forest-based zone area reasonable line loss rate estimation method |
| CN107516108A (en) * | 2017-08-15 | 2017-12-26 | 国网四川省电力公司电力科学研究院 | Grader creation method and partial discharge of transformer method of fault pattern recognition |
| CN107782442A (en) * | 2017-10-24 | 2018-03-09 | 华北电力大学（保定） | Transformer multiple features parameter selection method based on big data and random forest |
| CN108197639A (en) * | 2017-12-14 | 2018-06-22 | 佛山科学技术学院 | A kind of Diagnosis Method of Transformer Faults based on random forest |
| CN108062571A (en) * | 2017-12-27 | 2018-05-22 | 福州大学 | Diagnosing failure of photovoltaic array method based on differential evolution random forest grader |
| CN108398612A (en) * | 2018-01-12 | 2018-08-14 | 广州市扬新技术研究有限责任公司 | A kind of urban track traffic DC power-supply system short trouble localization method |
| CN108594067A (en) * | 2018-04-02 | 2018-09-28 | 湖南大学 | A kind of Multi-port direct-current distribution network line short fault distance measuring method |
| CN108693478A (en) * | 2018-04-17 | 2018-10-23 | 北京理工大学 | A kind of method for detecting leakage of lithium-ion-power cell |
| CN108959732A (en) * | 2018-06-15 | 2018-12-07 | 西安科技大学 | A kind of transmission line malfunction kind identification method based on convolutional neural networks |
| CN108872778A (en) * | 2018-07-03 | 2018-11-23 | 安徽理工大学 | A kind of transformer winding fault diagnostic system based on correlation dimension and random forest |
Non-Patent Citations (2)
| Title |
|---|
| TIAGO DOS SANTOS.ET.: "Stator winding short-circuit fault diagnosis in induction motors using random forest", 《2017 IEEE INTERNATIONAL ELECTRIC MACHINES AND DRIVES CONFERENCE》 * |
| 单林森: "配电变压器故障分析及随机森林诊断实现", 《科技风》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11474163B2 (en) * | 2019-08-12 | 2022-10-18 | Wuhan University | Power transformer winding fault positioning method based on deep convolutional neural network integrated with visual identification |
| CN114019298A (en) * | 2021-09-28 | 2022-02-08 | 中电华创(苏州)电力技术研究有限公司 | PCC-SVM-based generator rotor turn-to-turn short circuit online monitoring method |
| CN114019298B (en) * | 2021-09-28 | 2023-12-05 | 中电华创(苏州)电力技术研究有限公司 | On-line monitoring method for turn-to-turn short circuit of generator rotor based on PCC-SVM |
| CN113872200A (en) * | 2021-10-13 | 2021-12-31 | 国网江苏省电力有限公司电力科学研究院 | Medium-voltage distribution network power failure event diagnosis and identification method, system and storage medium |
| CN116799741A (en) * | 2023-05-24 | 2023-09-22 | 华斗数字科技(上海)有限公司 | Precision equipment short-circuit monitoring protection method and system based on slope detection |
| CN116799741B (en) * | 2023-05-24 | 2024-03-26 | 华斗数字科技(上海)有限公司 | Precision equipment short-circuit monitoring protection method and system based on slope detection |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109657720B (en) | 2021-05-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109657720A (en) | A kind of inline diagnosis method of power transformer shorted-turn fault | |
| CN109670242B (en) | Transformer winding deformation unsupervised online monitoring method based on elliptical envelope curve | |
| CN104156568A (en) | Transformer fault diagnosis method on basis of weighted gray correlation and fuzzy clustering | |
| CN106777984A (en) | A kind of method that photovoltaic array Working state analysis and fault diagnosis are realized based on density clustering algorithm | |
| CN113124929A (en) | Transformer substation multi-parameter signal acquisition comprehensive analysis system and method | |
| CN103884943B (en) | A kind of comprehensive analysis and diagnosis method of deformation of transformer winding | |
| CN104215323B (en) | Method for determining sensitivity of each sensor in mechanical equipment vibrating sensor network | |
| CN109359271A (en) | A kind of deformation of transformer winding degree online test method that logic-based returns | |
| CN108802535B (en) | Screening method, main interference source identification method and device, server and storage medium | |
| CN107844067B (en) | A kind of gate of hydropower station on-line condition monitoring control method and monitoring system | |
| CN110647924B (en) | GIS equipment state evaluation method based on support vector description and K-nearest neighbor algorithm | |
| CN103439631A (en) | Method and system for detecting corrosion state of grounding grid | |
| US20130151461A1 (en) | Identification of power system events using fuzzy logic | |
| CN116184265A (en) | Lightning arrester leakage current detection method and system based on multi-classification SVM | |
| CN113447766A (en) | Method, device, equipment and storage medium for detecting high-resistance ground fault | |
| CN105911407A (en) | Transformer state fuzzy set pair assessment method based on matter-element augmentation extensive correlation | |
| CN110632455A (en) | Fault detection and positioning method based on distribution network synchronous measurement big data | |
| CN104655967A (en) | Extraction method for vibration signal characteristic quantity of winding of distribution transformer | |
| CN112801135B (en) | Fault line selection method and device for power plant service power system based on characteristic quantity correlation | |
| CN113611568A (en) | Vacuum circuit breaker based on genetic convolution depth network | |
| CN117520951A (en) | Transformer health assessment method and system based on multiple characteristic quantities | |
| CN115343579B (en) | Power grid fault analysis method and device and electronic equipment | |
| CN116990633A (en) | Fault studying and judging method based on multiple characteristic quantities | |
| CN107832264A (en) | The overall evaluation system and evaluation method of a kind of design of electrical motor | |
| CN114091593A (en) | Network-level arc fault diagnosis method based on multi-scale feature 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |