CN106841905A - A kind of recognition methods of transformer short circuit fault and device - Google Patents
A kind of recognition methods of transformer short circuit fault and device Download PDFInfo
- Publication number
- CN106841905A CN106841905A CN201710242712.8A CN201710242712A CN106841905A CN 106841905 A CN106841905 A CN 106841905A CN 201710242712 A CN201710242712 A CN 201710242712A CN 106841905 A CN106841905 A CN 106841905A
- Authority
- CN
- China
- Prior art keywords
- transformer
- characteristic value
- statistical characteristic
- identified
- electric discharge
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/50—Testing of electric apparatus, lines, cables or components for short-circuits, continuity, leakage current or incorrect line connections
- G01R31/62—Testing of transformers
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/12—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing
- G01R31/1227—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing of components, parts or materials
- G01R31/1263—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing of components, parts or materials of solid or fluid materials, e.g. insulation films, bulk material; of semiconductors or LV electronic components or parts; of cable, line or wire insulation
- G01R31/1272—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing of components, parts or materials of solid or fluid materials, e.g. insulation films, bulk material; of semiconductors or LV electronic components or parts; of cable, line or wire insulation of cable, line or wire insulation, e.g. using partial discharge measurements
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Testing Relating To Insulation (AREA)
Abstract
Recognition methods and device this application discloses a kind of transformer short circuit fault, are related to technical field of power systems, are invented to solve the problems, such as short trouble rate of correct diagnosis low.Main method includes：The shelf depreciation collection of illustrative plates of transformer is obtained, the transformer is the transformer that there is short trouble；According to preset Fourier algorithm, the multiple dimensioned signal of the shelf depreciation collection of illustrative plates is extracted；The statistical characteristic value of the multiple dimensioned signal is calculated, the statistical characteristic value refers to the statistical information of the multiple dimensioned signal；The statistical characteristic value is integrated, knowledge base is built；Using the statistical characteristic value in the knowledge base as training sample and test sample, RBF nerve network is built and trained；Statistical characteristic value to be identified is input to the RBF neural, the short trouble type of the transformer to be identified corresponding to the statistical characteristic value to be identified is recognized.Present invention is mainly applied to recognize transformer short circuit fault.
Description
Technical field
The application is related to technical field of power systems, more particularly to a kind of transformer short circuit fault recognition methods and dress
Put.
Background technology
The running status of transformer is the key factor for influenceing electric energy transmission reliability, and because some idols in practical operation
Right or non-accidental the reason for, transformer occurred various failures, and the failure that wherein transformer is easiest to occur is short trouble.
Short trouble causes mainly due to insulation ag(e)ing or insulation reduction, wherein, partial discharge phenomenon is transformer insulated water
The increased important symbol of possibility of the low failure that is short-circuited of pancake, therefore the result that shelf depreciation is tested imports analogue system
In Classification and Identification is subject to electric discharge type, it becomes possible to accurately and rapidly judge the potential short trouble of inside transformer, so as to and
When the failure of transformer is investigated, it is ensured that transformer can continue reliably to run, and the normal fortune of whole power network is ensured with this
OK.
Fig. 1 is the transformer schematic diagram of simulated interior short circuit in winding state.Transformer includes：Shell 1, lead sleeve pipe 2, height
Low pressure winding 3, iron core 4, spiral range finding conducting rod 5, winding inter-turn tap 6 and winding interlayer tap 7.Spiral range finding is led
The insulating sleeve of electric pole 5 is fixed on shell 1, the certain spiral range finding conducting rod 5 of conductive lengths is changed, by polytetrafluoroethylene (PTFE)
Insulating bar finds range in the precession insulating sleeve of conducting rod 5 spiral, is allowed to be flushed with zero graduation.Select the short-circuit condition for needing, meter
The spiral range finding rotary distance of conducting rod 5 is calculated, by conducting rod and turn-to-turn (interlayer) winding tap short circuit, short-circuit condition is simulated.Need
When changing the number of turn (number of plies) of short-circuited winding, spiral range finding conducting rod 5 is screwed out, change the conducting rod of appropriate length, again
In fixed and precession transformer.Partial discharge of transformer test refers in preset time, preset voltage difference to be applied to tested
On examination transformer so that the insulation vulnerable area of tested transformer occurs electric discharge phenomena.The most direct phenomenon of shelf depreciation is
Electric charge is moved between causing electrode, and is caused on tested transformer outer electrode by dielectric with a number of electric charge
Voltage change, the discharge pulse of short time can produce the electromagnetic signal of high frequency to external radiation.
In the prior art, neural network algorithm is applied to transformer fault diagnosis aspect, system is tested by shelf depreciation
Unite and the discharge pulse collection of illustrative plates of different transformers is collected, power spectrumanalysis is carried out to discharge pulse, obtained from power spectrum
Characteristic quantity in extract training sample and test sample, build BP (BackproPagation, backpropagation) neutral net, instruction
Practice BP neural network, last test BP neural network.Can pass through discharge pulse atlas diagnostic by the BP neural network tested
Transformer fault.From the above-mentioned parameter setting that can be seen that in the prior art during fault diagnosis training BP neural network,
Influence the correct recognition rata of fault diagnosis.
The content of the invention
Recognition methods and device this application provides a kind of transformer short circuit fault are correct to solve short trouble diagnosis
The low problem of rate.
In a first aspect, this application provides a kind of recognition methods of transformer short circuit fault, the method includes：Obtain transformation
The shelf depreciation collection of illustrative plates of device, the transformer is the transformer that there is short trouble；According to preset Fourier algorithm, extract described
The multiple dimensioned signal of shelf depreciation collection of illustrative plates；The statistical characteristic value of the multiple dimensioned signal is calculated, the statistical characteristic value refers to institute
State the statistical information of multiple dimensioned signal；The statistical characteristic value is integrated, knowledge base is built；By in the knowledge base
The statistical characteristic value builds and trains RBF nerve network as training sample and test sample；Will be to be identified
Statistical characteristic value is input to the RBF neural, recognizes the transformer to be identified corresponding to the statistical characteristic value to be identified
Short trouble type.Using this implementation, it is possible to increase transformer short circuit fault diagnosis efficiency, and rapidly find to become
The short trouble that depressor is present, can be used for transformer self-inspection, so as to improve the quality of transformer.
With reference in a first aspect, in first aspect in the first possible implementation, the electric discharge class that the transformer is present
Type includes turn-to-turn short circuit electric discharge type, layer short circuit electric discharge type and turn-to-turn and interlayer short circuit dischange type simultaneously；The acquisition
The shelf depreciation collection of illustrative plates of transformer, including：Record the electric discharge type of the transformer；Obtain by the described of discharge system test
The shelf depreciation collection of illustrative plates of transformer；With the electric discharge type, the shelf depreciation collection of illustrative plates is marked.Using this implementation, can
Improve the accuracy of transformer short circuit fault diagnosis.
It is described to integrate the statistical nature with reference in a first aspect, in second possible implementation of first aspect
Amount, before building knowledge base, methods described also includes：Obtain the electric discharge type of the shelf depreciation collection of illustrative plates；With institute
Electric discharge type is stated, the statistical characteristic value is marked.
With reference in a first aspect, in the third possible implementation of first aspect, RBF nerve network
The activation primitive of selection isWherein,
Andη in formulaaIt is a learning rates, αaIt is the factor of momentum of a,AndIn formula
ηbIt is the learning rate of b, αbIt is the factor of momentum of b.
It is described by statistical nature to be identified with reference in a first aspect, in the 4th kind of possible implementation of first aspect
Before amount is input to the RBF neural, methods described also includes：Obtain the shelf depreciation figure to be identified of transformer to be identified
Spectrum；According to the preset Fourier algorithm, the to be identified multiple dimensioned signal of the shelf depreciation collection of illustrative plates to be identified is extracted；Calculate institute
State the statistical characteristic value to be identified of multiple dimensioned signal to be identified.
With reference in a first aspect, in the 5th kind of possible implementation of first aspect, the identification change to be identified
The short trouble type of depressor, including：Obtain the electric discharge class of the statistical characteristic value described to be identified of the RBF neural output
Type；According to fault dictionary, the corresponding short trouble type of the electric discharge type is searched.
Second aspect, present invention also provides a kind of identifying device of transformer short circuit fault, described device includes being used for
Perform the unit of method and step in the various implementations of first aspect.Specifically include：First acquisition unit, for obtaining transformer
Shelf depreciation collection of illustrative plates, the transformer is the transformer that there is short trouble；Extraction unit, for being calculated according to preset Fourier
Method, extracts the multiple dimensioned signal of the shelf depreciation collection of illustrative plates；Computing unit, the statistical nature for calculating the multiple dimensioned signal
Amount, the statistical characteristic value refers to the statistical information of the multiple dimensioned signal；Integral unit, for integrating the statistical nature
Amount, builds knowledge base；Construction unit, for using the statistical characteristic value in the knowledge base as training sample
And test sample, build and train RBF nerve network；Recognition unit, for statistical characteristic value to be identified is defeated
Enter the short trouble class that the transformer to be identified corresponding to the statistical characteristic value to be identified is recognized to the RBF neural
Type.
The third aspect, present invention also provides a kind of terminal, including：Processor and memory；The processor can be held
The program stored in the row memory or instruction, so as to realize with transformer short-circuit described in the various implementations of first aspect
The recognition methods of failure.
Fourth aspect, present invention also provides a kind of storage medium, the computer-readable storage medium can have program stored therein, the journey
Sequence can be realized including part or all of in each embodiment of recognition methods of the transformer short circuit fault of the application offer when performing
Step.
A kind of recognition methods of transformer short circuit fault that the application is provided and device, are put by the part for obtaining transformer
Electrograph is composed, and then according to preset Fourier algorithm, extracts the multiple dimensioned signal of shelf depreciation collection of illustrative plates, then calculate multiple dimensioned signal
Statistical characteristic value, then integrate statistical characteristic value, builds knowledge base, then using the statistical characteristic value in knowledge base as instruction
Practice sample and test sample, build and train RBF nerve network, be finally input to statistical characteristic value to be identified
RBF neural, recognizes the short trouble type of the how corresponding transformer to be identified of statistical characteristic value to be identified.With prior art
Compare, the application can use RBF neural algorithm, quickly and accurately recognize transformer short circuit fault.The application can
Transformer fault diagnosis ground efficiency is improved, and does not need the time-consuming laborious transformer of disassembling of professional to carry out tracing trouble type,
A kind of self-inspection means of transformer are provided indirectly, so as to provide the quality of transformer, there is provided the stability of distribution network operation.
Brief description of the drawings
In order to illustrate more clearly of the technical scheme of the application, letter will be made to the accompanying drawing to be used needed for embodiment below
Singly introduce, it should be apparent that, for those of ordinary skills, without having to pay creative labor,
Other accompanying drawings can also be obtained according to these accompanying drawings.
A kind of transformer schematic diagram of simulated interior short circuit in winding state that Fig. 1 is provided for the application；
A kind of flow chart of one embodiment of the recognition methods of transformer short circuit fault that Fig. 2 is provided for the application；
The flow chart of one embodiment of the recognition methods of another transformer short circuit fault that Fig. 3 is provided for the application；
A kind of structural representation of one embodiment of the identifying device of transformer short circuit fault that Fig. 4 is provided for the application
Figure；
The structural representation of one embodiment of the identifying device of another transformer short circuit fault that Fig. 5 is provided for the application
Figure.
Specific embodiment
It is a kind of structural representation of the transformer schematic diagram of simulated interior short circuit in winding state referring to Fig. 1.In transformer
In be easiest to occur failure be short trouble.The method for diagnosing faults of transformer, including short-circuit test, direct current resistance m easurem ent,
There are dissolved gas constituent analysis, shelf depreciation diagnosis, the analysis of the dielectric degree of polymerization, frequency response etc..Wherein shelf depreciation
The test method of diagnosis can be divided into Electric Method and the major class of non-electrical method two.Have in Electric Method pulse current method, dielectric loss method and
Electromagnetic radiation method.There are sonic method, photometry, calorimetry and physical-chemical process in non-electrical method.The sensitivity of Electric Method is compared with non-electrical
Gas method is high, so, it is general to use Electric Method more.In Electric Method using it is most be pulse current method.In non-electrical method, often
Using being used for sonic method, especially sonic method playing a game, portion's discharge source is positioned more.
Referring to Fig. 2, a kind of flow chart of the recognition methods one embodiment for the transformer short circuit fault provided for the application,
The method comprises the following steps：
Step 201, obtains the shelf depreciation collection of illustrative plates of transformer.
Shelf depreciation collection of illustrative plates, refers to the collection of illustrative plates of transformer generation when discharge test is carried out, including phase collection of illustrative plates, classification chart
Spectrum etc..The application provides the recognition methods of transformer short circuit fault, refers to which kind of short trouble identification transformer has,
So the corresponding transformer of shelf depreciation collection of illustrative plates is the transformer that there is short trouble.
In order to the identification of transformer short circuit fault is more accurate, it is necessary to obtain the office of the transformer that there is typical short-circuit failure
Discharge collection of illustrative plates in portion.Wherein typical short-circuit failure includes turn-to-turn short circuit electric discharge type, layer short circuit electric discharge type and turn-to-turn and interlayer
While short circuit dischange type.Every kind of typical short-circuit failure all has the model machine of multiple identical materials, volume and structure, and model machine is by people
Work is made.Ensure that the corresponding shelf depreciation collection of illustrative plates of short trouble has certain representativeness, energy by the model machine being manually made
Ensure that the shelf depreciation collection of illustrative plates for obtaining is the transformer with single short trouble so that the shelf depreciation collection of illustrative plates of acquisition can have
There is the characteristic of the short trouble.Oilpaper electric discharge to model machine carries out multi collect, to exclude due to some burst reasons
The shelf depreciation collection of illustrative plates that causes is widely varied, so that recognition result of the influence to transformer short circuit fault.
Step 202, according to preset Fourier algorithm, extracts the multiple dimensioned signal of shelf depreciation collection of illustrative plates.
Fourier algorithm, is a kind of method of signal Analysis, is capable of the composition of signal Analysis, it is also possible to which these compositions synthesize
Signal.Many waveforms can be used as the composition of signal, such as sine wave, square wave, sawtooth waveforms etc., Fourier transformation sine wave conduct
The composition of signal.Fourier algorithm includes：Fourier algorithm, fast Fourier algorithm etc. basic Fourier algorithm is spread out in short-term
Raw algorithm.Fourier algorithm is the most frequently used Time-Frequency Analysis Method in short-term, and it identifies certain by the segment signal in time window
The signal characteristic at one moment.During Short Time Fourier Transform, the length of window determines the temporal resolution and frequency of spectrogram
Resolution ratio, window is long, and the signal of interception is more long, and signal is more long, and frequency resolution is higher after Fourier transformation, temporal resolution
It is poorer；Conversely, window is long shorter, the signal of interception is shorter, and frequency resolution is poorer, and temporal resolution is better, that is, in short-term
In Fourier transformation, can not be got both between temporal resolution and frequency, it should accepted or rejected according to real needs.Preferably by
Fourier algorithm, is conducive in temporal resolution demand and frequency needs in short-term, is effectively adjusted according to demand.
Multiple dimensioned signal refers to certain signal through multiscale analysis, the signal of the multiple single yardstick of acquisition.According to different
Scale calibration, extracts required signal, and be referred to as multiple dimensioned signal from shelf depreciation collection of illustrative plates.According to preset Fourier
Algorithm, extracts the multiple dimensioned signal of shelf depreciation collection of illustrative plates.When multiple dimensioned signal is extracted, wavelet algorithm can also be used.At this
Optimization algorithm is Fourier algorithm in application.
Step 203, calculates the statistical characteristic value of multiple dimensioned signal.
Statistical characteristic value refers to the statistical information of multiple dimensioned signal.Statistical characteristic value can include discharge capacity and discharge time
Phase distribution, electric discharge amplitude and energy distribution statistical information.According to the characteristic of shelf depreciation collection of illustrative plates, due to different spies
Levy the wave band for belonging to different, it is considered to the correlation between different-waveband, calculate the statistical characteristic value of multiple dimensioned signal.
Step 204, integrates statistical characteristic value, builds knowledge base.
Calculate the transformer of different short troubles, identical short trouble transformer and the different tests of same transformer
The statistical characteristic value of number of times, all of statistical characteristic value is integrated, and builds knowledge base.In knowledge base, comprising
The statistical characteristic value of the multiple dimensioned signal of the shelf depreciation collection of illustrative plates of different short troubles, that is, all short trouble model machines
Fault signature data.
Step 205, using the statistical characteristic value in knowledge base as training sample and test sample, builds and trains footpath
To basic function RBF neural.
Short trouble model machine has different typical short-circuit failures, but trains RBF (Radical Basis
Function, RBF) neutral net is in order to recognize different short trouble types, so needing different short circuits
Failure, is respectively trained.The statistical characteristic value of same fault type, every group of statistical characteristic value correspondence one are chosen from knowledge base
The shelf depreciation collection of illustrative plates of individual short trouble transformer.By statistical characteristic value according to 3:1 ratio is respectively as training sample and survey
Sample sheet.Training sample learns for RBF neural, and test sample is verified for RBF neural.The RBF nerves of selection
Network can make weights converge to optimal value rapidly.Illustrate in specific training process, definition grading function is less than
The end condition of 0.0001 RBF neural algorithm the most, the input layer number of RBF neural is set to 20, output layer
Neuron number positioning 3, three kinds of short troubles of correspondence, selection desired output is respectively turn-to-turn short circuit electric discharge (1,0,0), layer short circuit
Electric discharge (0,1,0) and turn-to-turn and interlayer are while short circuit dischange (0,0,1).
By the RBF neural by training and test, as the identification neutral net of transformer short circuit fault.
Step 206, RBF neural is input to by statistical characteristic value to be identified, recognizes that statistical characteristic value institute to be identified is right
The short trouble type of the transformer to be identified answered.
Statistical characteristic value to be identified, refers to the statistical characteristic value of the transformer of short trouble to be identified.Implement in the present invention
Computational methods in example to the statistical characteristic value of the transformer of short trouble to be identified are not limited.By statistical characteristic value to be identified
It is input in the RBF neural for completing training, recognizes the short circuit of the transformer to be identified corresponding to statistical characteristic value to be identified
Fault type.The transformer of transformer to be identified i.e. short trouble to be identified.
From above-described embodiment as can be seen that the application provide a kind of transformer short circuit fault recognition methods, by obtaining
The shelf depreciation collection of illustrative plates of transformer is taken, then according to preset Fourier algorithm, the multiple dimensioned signal of shelf depreciation collection of illustrative plates is extracted, then
The statistical characteristic value of multiple dimensioned signal is calculated, then integrates statistical characteristic value, build knowledge base, then by knowledge base
Statistical characteristic value builds and trains RBF nerve network as training sample and test sample, finally will be to be identified
Statistical characteristic value is input to RBF neural, recognizes the short circuit event of the how corresponding transformer to be identified of statistical characteristic value to be identified
Barrier type.Compared with prior art, the application can use RBF neural algorithm, quickly and accurately recognize that transformer is short
Road failure.The application can improve transformer fault diagnosis ground efficiency, and not need that professional is time-consuming laborious to disassemble transformation
Device carrys out tracing trouble type, and a kind of self-inspection means of transformer are provided indirectly, so as to provide the quality of transformer, there is provided distribution
The stability of network operation.
Referring to Fig. 3, the flow of the recognition methods one embodiment for another transformer short circuit fault provided for the application
Figure, the method comprises the following steps：
Step 301, obtains the shelf depreciation collection of illustrative plates of transformer.
Transformer is the transformer that there is short trouble.The electric discharge type that transformer is present includes turn-to-turn short circuit electric discharge class
Type, layer short circuit electric discharge type and turn-to-turn and interlayer are while short circuit dischange type；Obtain the shelf depreciation collection of illustrative plates of transformer, bag
Include：Record the electric discharge type of transformer；Obtain the shelf depreciation collection of illustrative plates of the transformer tested by discharge system；With class of discharging
Type, marks shelf depreciation collection of illustrative plates.The corresponding relation of record electric discharge type and shelf depreciation collection of illustrative plates, is subsequent authentication RBF nerve nets
The accuracy of network provides foundation.
Step 302, according to preset Fourier algorithm, extracts the multiple dimensioned signal of shelf depreciation collection of illustrative plates.
This step is identical with the method described in the step 202 shown in Fig. 2, repeats no more here.
Step 303, calculates the statistical characteristic value of multiple dimensioned signal.
Statistical characteristic value refers to the statistical information of multiple dimensioned signal.This step and the method described in the step 203 shown in Fig. 2
It is identical, repeat no more here.
Step 304, obtains the electric discharge type of shelf depreciation collection of illustrative plates.
Obtain the label information of shelf depreciation collection of illustrative plates, the as corresponding electric discharge type of shelf depreciation collection of illustrative plates.
Step 305, with electric discharge type, tokens statisticses characteristic quantity.
The statistical characteristic value that step 303 is calculated, is marked with electric discharge type, and wherein electric discharge type discharges including turn-to-turn short circuit
Type, layer short circuit electric discharge type and turn-to-turn and interlayer are while short circuit dischange type.
Step 306, integrates statistical characteristic value, builds knowledge base.
With electric discharge type, statistical characteristic value is integrated in classification, builds knowledge base.
Step 307, using the statistical characteristic value in knowledge base as training sample and test sample, builds and trains footpath
To basic function RBF neural.
Short trouble model machine has different typical short-circuit failures, but trains RBF (Radical Basis
Function, RBF) neutral net is in order to recognize different short trouble types, so needing different short circuits
Failure, trains successively.The statistical characteristic value of same fault type, every group of statistical characteristic value correspondence one are chosen from knowledge base
The shelf depreciation collection of illustrative plates of individual short trouble transformer.By statistical characteristic value according to 3:1 ratio is used as training sample and test specimens
This.Training sample learns for RBF neural, and test sample is verified for RBF neural.The RBF neural of selection
Weights can be made to converge to optimal value rapidly.Illustrate in specific training process, definition grading function is less than 0.0001 most
It is the end condition of RBF neural algorithm, the input layer number of RBF neural is set to 20, output layer neuron number
Positioning 3, correspondence three kinds of short troubles, selection desired output be respectively turn-to-turn short circuit electric discharge (1,0,0), layer short circuit electric discharge (0,
And turn-to-turn and interlayer short circuit dischange (0,0,1) simultaneously 1,0).
Radial basis function neural network is a kind of efficient feed forward type neutral net, and there are other feedforward networks not have for it
The optimal approximation capability and global optimum's characteristic having, and simple structure, training speed are fast.RBF nerve network
The activation primitive of selection isWherein,
Andη in formulaaIt is a learning rates, αaIt is the factor of momentum of a,AndFormula
Middle ηbIt is the learning rate of b, αbIt is the factor of momentum of b.
Step 308, obtains the shelf depreciation collection of illustrative plates to be identified of transformer to be identified.
Transformer to be identified, can be the transformer for having broken down, or the transformer not dispatched from the factory, in this hair
To transformer to be identified whether using not limiting in bright embodiment.Transformer to be identified is not the transformation for working in a word
Device, identification method is directed to the current transformer being stopped.
Step 309, according to preset Fourier algorithm, extracts the to be identified multiple dimensioned signal of shelf depreciation collection of illustrative plates to be identified.
Because Fourier algorithm is the most frequently used video analysis method in short-term, be conducive in temporal resolution demand and frequency
In resolution requirements, effectively adjusted as needed, the preset Fourier algorithm for preferably using is Fourier in short-term
Algorithm.Extract the to be identified multiple dimensioned signal of shelf depreciation collection of illustrative plates to be identified.Multiple dimensioned signal refers to that certain signal passed through yardstick
Analysis, the signal of the multiple single yardstick of acquisition.
Step 310, calculates the statistical characteristic value to be identified of multiple dimensioned signal to be identified.
Statistical characteristic value to be identified refers to the statistical information of multiple dimensioned signal to be identified.According to the spy of shelf depreciation collection of illustrative plates
Property, because different features belongs to different wave bands, it is considered to the correlation between different-waveband, calculate multiple dimensioned signal to be identified
Statistical characteristic value to be identified.Statistical characteristic value to be identified can include phase distribution, the electric discharge width of discharge capacity and discharge time
The statistical information of the distribution of value and energy.
Step 311, RBF neural is input to by statistical characteristic value to be identified, recognizes that statistical characteristic value institute to be identified is right
The short trouble type of the transformer to be identified answered.
Statistical characteristic value to be identified is input in RBF neural, RBF neural recognizes statistical nature to be identified
Amount, specifically includes：Obtain the electric discharge type of the statistical characteristic value to be identified of RBF neural output；According to fault dictionary, search
The corresponding short trouble type of electric discharge type.
Fault dictionary is worked out after being collected by substantial amounts of experiment and experience by prior art, manually experience generation,
It is the existing data basis for judging short trouble type.
From above-described embodiment as can be seen that the application provide a kind of transformer short circuit fault recognition methods, by obtaining
The shelf depreciation collection of illustrative plates of transformer is taken, then according to preset Fourier algorithm, the multiple dimensioned signal of shelf depreciation collection of illustrative plates is extracted, then
The statistical characteristic value of multiple dimensioned signal is calculated, then integrates statistical characteristic value, build knowledge base, then by knowledge base
Statistical characteristic value builds and trains RBF nerve network as training sample and test sample, finally will be to be identified
Statistical characteristic value is input to RBF neural, recognizes the short circuit event of the how corresponding transformer to be identified of statistical characteristic value to be identified
Barrier type.Compared with prior art, the application can use RBF neural algorithm, quickly and accurately recognize that transformer is short
Road failure.The application can improve transformer fault diagnosis ground efficiency, and not need that professional is time-consuming laborious to disassemble transformation
Device carrys out tracing trouble type, and a kind of self-inspection means of transformer are provided indirectly, so as to provide the quality of transformer, there is provided distribution
The stability of network operation.
It is a kind of structural representation of identifying device one embodiment of transformer short circuit fault of the application referring to Fig. 4, uses
In the recognition methods for performing the transformer short circuit fault corresponding to Fig. 2 and Fig. 3.
As shown in figure 4, the device includes：First acquisition unit 41, extraction unit 42, computing unit 43, integral unit 44,
Construction unit 45 and recognition unit 46.Wherein,
First acquisition unit 41, the shelf depreciation collection of illustrative plates for obtaining transformer, transformer is the change that there is short trouble
Depressor；
Extraction unit 42, for according to preset Fourier algorithm, extracting the multiple dimensioned signal of shelf depreciation collection of illustrative plates；
Computing unit 43, the statistical characteristic value for calculating multiple dimensioned signal, statistical characteristic value refers to multiple dimensioned signal
Statistical information；
Integral unit 44, for integrating statistical characteristic value, builds knowledge base；
Construction unit 45, for using the statistical characteristic value in knowledge base as training sample and test sample, building
And train RBF nerve network；
Recognition unit 46, for statistical characteristic value to be identified to be input into RBF neural, recognizes statistical nature to be identified
The short trouble type of the corresponding transformer to be identified of amount.
Further, transformer exist electric discharge type include turn-to-turn short circuit electric discharge type, layer short circuit electric discharge type and
Turn-to-turn and interlayer are while short circuit dischange type；
First acquisition unit 41, including：
Logging modle 411, the electric discharge type for recording transformer；
Acquisition module 412, the shelf depreciation collection of illustrative plates for obtaining the transformer tested by discharge system；
Mark module 413, for electric discharge type, marking shelf depreciation collection of illustrative plates.
Further, as shown in figure 5, the device also includes：
Second acquisition unit 47, for integrating statistical characteristic value, before building knowledge base, obtains shelf depreciation collection of illustrative plates
Electric discharge type；
Indexing unit 48, for electric discharge type, tokens statisticses characteristic quantity.
Further, the activation primitive of RBF nerve network selection is
Wherein,
And
η in formulaaIt is a learning rates, αaIt is the factor of momentum of a,
And
η in formulabIt is the learning rate of b, αbIt is the factor of momentum of b.
Further, the device also includes：
First acquisition unit 41, for before statistical characteristic value to be identified is input into RBF neural, obtaining to be identified
The shelf depreciation collection of illustrative plates to be identified of transformer；
Extraction unit 42, for according to preset Fourier algorithm, extracting to be identified many chis of shelf depreciation collection of illustrative plates to be identified
Degree signal；
Computing unit 43, the statistical characteristic value to be identified for calculating multiple dimensioned signal to be identified.
Further, as shown in figure 5, recognition unit 46, including：
Acquisition module 461, the electric discharge type of the statistical characteristic value to be identified for obtaining RBF neural output；
Searching modul 462, for according to fault dictionary, searching the corresponding short trouble type of electric discharge type.
From above-described embodiment as can be seen that the application provide a kind of transformer short circuit fault identifying device, by obtaining
The shelf depreciation collection of illustrative plates of transformer is taken, then according to preset Fourier algorithm, the multiple dimensioned signal of shelf depreciation collection of illustrative plates is extracted, then
The statistical characteristic value of multiple dimensioned signal is calculated, then integrates statistical characteristic value, build knowledge base, then by knowledge base
Statistical characteristic value builds and trains RBF nerve network as training sample and test sample, finally will be to be identified
Statistical characteristic value is input to RBF neural, recognizes the short circuit event of the how corresponding transformer to be identified of statistical characteristic value to be identified
Barrier type.Compared with prior art, the application can use RBF neural algorithm, quickly and accurately recognize that transformer is short
Road failure.The application can improve transformer fault diagnosis ground efficiency, and not need that professional is time-consuming laborious to disassemble transformation
Device carrys out tracing trouble type, and a kind of self-inspection means of transformer are provided indirectly, so as to provide the quality of transformer, there is provided distribution
The stability of network operation.
In implementing, the present invention also provides a kind of computer-readable storage medium, wherein, the computer-readable storage medium can be stored
There is program, the program may include the part or all of step in each embodiment of the method for calling of the present invention offer when performing.Institute
The storage medium stated can be magnetic disc, CD, read-only memory (English：Read-only memory, referred to as：ROM) or with
Machine storage memory (English：Random access memory, referred to as：RAM) etc..
It is required that those skilled in the art can be understood that the technology in the embodiment of the present invention can add by software
The mode of general hardware platform realize.Based on such understanding, the technical scheme in the embodiment of the present invention substantially or
Say that the part contributed to prior art can be embodied in the form of software product, the computer software product can be deposited
Storage in storage medium, such as ROM/RAM, magnetic disc, CD, including some instructions are used to so that computer equipment (can be with
It is personal computer, server, or network equipment etc.) perform some part institutes of each embodiment of the invention or embodiment
The method stated.
In this specification between each embodiment identical similar part mutually referring to.Especially for ... implement
For example, because it is substantially similar to embodiment of the method, so description is fairly simple, related part is referring in embodiment of the method
Explanation.
Invention described above implementation method is not intended to limit the scope of the present invention..
Claims (10)
1. a kind of recognition methods of transformer short circuit fault, it is characterised in that methods described includes：
The shelf depreciation collection of illustrative plates of transformer is obtained, the transformer is the transformer that there is short trouble；
According to preset Fourier algorithm, the multiple dimensioned signal of the shelf depreciation collection of illustrative plates is extracted；
The statistical characteristic value of the multiple dimensioned signal is calculated, the statistical characteristic value refers to the statistics letter of the multiple dimensioned signal
Breath；
The statistical characteristic value is integrated, knowledge base is built；
Using the statistical characteristic value in the knowledge base as training sample and test sample, radial direction base is built and trained
Function RBF neural；
Statistical characteristic value to be identified is input to the RBF neural, is recognized corresponding to the statistical characteristic value to be identified
The short trouble type of transformer to be identified.
2. method according to claim 1, it is characterised in that the electric discharge type that the transformer is present includes turn-to-turn short circuit
Electric discharge type, layer short circuit electric discharge type and turn-to-turn and interlayer are while short circuit dischange type；
The shelf depreciation collection of illustrative plates for obtaining transformer, including：
Record the electric discharge type of the transformer；
Obtain the shelf depreciation collection of illustrative plates of the transformer tested by discharge system；
With the electric discharge type, the shelf depreciation collection of illustrative plates is marked.
3. method according to claim 2, it is characterised in that the integration statistical characteristic value, builds fault signature
Before storehouse, methods described also includes：
Obtain the electric discharge type of the shelf depreciation collection of illustrative plates；
With the electric discharge type, the statistical characteristic value is marked.
4. method according to claim 1, it is characterised in that the activation that the RBF nerve network is chosen
Function isWherein,
And
η in formulaaIt is a learning rates, αaIt is the factor of momentum of a,
And
η in formulabIt is the learning rate of b, αbIt is the factor of momentum of b.
5. method according to claim 1, it is characterised in that described that statistical characteristic value to be identified is input to the RBF
Before neutral net, methods described also includes：
Obtain the shelf depreciation collection of illustrative plates to be identified of transformer to be identified；
According to the preset Fourier algorithm, the to be identified multiple dimensioned signal of the shelf depreciation collection of illustrative plates to be identified is extracted；
Calculate the statistical characteristic value to be identified of the multiple dimensioned signal to be identified.
6. method according to claim 1, it is characterised in that the short trouble class of the identification transformer to be identified
Type, including：
Obtain the electric discharge type of the statistical characteristic value described to be identified of the RBF neural output；
According to fault dictionary, the corresponding short trouble type of the electric discharge type is searched.
7. a kind of identifying device of transformer short circuit fault, it is characterised in that described device includes：
First acquisition unit, the shelf depreciation collection of illustrative plates for obtaining transformer, the transformer is the transformation that there is short trouble
Device；
Extraction unit, for according to preset Fourier algorithm, extracting the multiple dimensioned signal of the shelf depreciation collection of illustrative plates；
Computing unit, the statistical characteristic value for calculating the multiple dimensioned signal, the statistical characteristic value refers to described multiple dimensioned
The statistical information of signal；
Integral unit, for integrating the statistical characteristic value, builds knowledge base；
Construction unit, for using the statistical characteristic value in the knowledge base as training sample and test sample, structure
Build and train RBF nerve network；
Recognition unit, for statistical characteristic value to be identified to be input into the RBF neural, recognizes that the statistics to be identified is special
The short trouble type of the transformer to be identified corresponding to the amount of levying.
8. device according to claim 7, it is characterised in that the electric discharge type that the transformer is present includes turn-to-turn short circuit
Electric discharge type, layer short circuit electric discharge type and turn-to-turn and interlayer are while short circuit dischange type；
The first acquisition unit, including：
Logging modle, the electric discharge type for recording the transformer；
Acquisition module, the shelf depreciation collection of illustrative plates for obtaining the transformer tested by discharge system；
Mark module, for the electric discharge type, marking the shelf depreciation collection of illustrative plates.
9. device according to claim 8, it is characterised in that described device also includes：
Second acquisition unit, the statistical characteristic value is integrated for described, before building knowledge base, is obtained the part and is put
The electric discharge type of electrograph spectrum；
Indexing unit, for the electric discharge type, marking the statistical characteristic value.
10. device according to claim 7, it is characterised in that the recognition unit, including：
Acquisition module, the electric discharge type of the statistical characteristic value described to be identified for obtaining the RBF neural output；
Searching modul, for according to fault dictionary, searching the corresponding short trouble type of the electric discharge type.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710242712.8A CN106841905A (en) | 2017-04-14 | 2017-04-14 | A kind of recognition methods of transformer short circuit fault and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710242712.8A CN106841905A (en) | 2017-04-14 | 2017-04-14 | A kind of recognition methods of transformer short circuit fault and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106841905A true CN106841905A (en) | 2017-06-13 |
Family
ID=59147631
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710242712.8A Pending CN106841905A (en) | 2017-04-14 | 2017-04-14 | A kind of recognition methods of transformer short circuit fault and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106841905A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108520339A (en) * | 2018-03-23 | 2018-09-11 | 佛山科学技术学院 | A kind of electric power networks fault diagnosis system and method |
| CN109191448A (en) * | 2018-09-03 | 2019-01-11 | 重庆大学 | The transformer fault identification technology with Digital Image Processing is drawn based on three-dimensional coordinate |
| CN110850228A (en) * | 2019-11-04 | 2020-02-28 | 国网江苏省电力有限公司泰州供电分公司 | Low-voltage transformer area fault positioning method based on equivalent impedance characteristics of phase change switch |
| CN111272222A (en) * | 2020-02-28 | 2020-06-12 | 西南交通大学 | Transformer fault diagnosis method based on characteristic quantity set |
| CN111289837A (en) * | 2020-02-06 | 2020-06-16 | 云南电网有限责任公司电力科学研究院 | Method and system for evaluating latent fault of power distribution network equipment |
| CN111308268A (en) * | 2020-02-06 | 2020-06-19 | 云南电网有限责任公司电力科学研究院 | Method and system for evaluating latent fault of power distribution network equipment |
| CN112595998A (en) * | 2020-12-01 | 2021-04-02 | 清华大学 | Frequency response testing method based on transformer broadband model and application |
| CN115542101A (en) * | 2022-11-30 | 2022-12-30 | 杭州兆华电子股份有限公司 | Voiceprint preprocessing method of transformer voiceprint detection system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101505058A (en) * | 2009-03-19 | 2009-08-12 | 东北大学 | Dual recursive neutral element sensitivity load network modeling apparatus and method based double network |
| KR101025058B1 (en) * | 2009-06-18 | 2011-03-25 | 대림대학 산학협력단 | A method for tracking signal detection of low voltage lines using fuzzy RBF neural network algorithm |
| CN102221651A (en) * | 2011-03-11 | 2011-10-19 | 太原理工大学 | Fault on-line diagnosis and early warning method of flameproof dry-type transformer for mine |
| CN103440497A (en) * | 2013-08-13 | 2013-12-11 | 上海交通大学 | GIS insulation defect partial discharge atlas pattern recognition method |
| CN105425076A (en) * | 2015-12-11 | 2016-03-23 | 厦门理工学院 | Method of carrying out transformer fault identification based on BP neural network algorithm |
| CN105606966A (en) * | 2015-12-21 | 2016-05-25 | 安徽理工大学 | Partial discharge pattern recognition method based on mixed neural network algorithm |
-
2017
- 2017-04-14 CN CN201710242712.8A patent/CN106841905A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101505058A (en) * | 2009-03-19 | 2009-08-12 | 东北大学 | Dual recursive neutral element sensitivity load network modeling apparatus and method based double network |
| KR101025058B1 (en) * | 2009-06-18 | 2011-03-25 | 대림대학 산학협력단 | A method for tracking signal detection of low voltage lines using fuzzy RBF neural network algorithm |
| CN102221651A (en) * | 2011-03-11 | 2011-10-19 | 太原理工大学 | Fault on-line diagnosis and early warning method of flameproof dry-type transformer for mine |
| CN103440497A (en) * | 2013-08-13 | 2013-12-11 | 上海交通大学 | GIS insulation defect partial discharge atlas pattern recognition method |
| CN105425076A (en) * | 2015-12-11 | 2016-03-23 | 厦门理工学院 | Method of carrying out transformer fault identification based on BP neural network algorithm |
| CN105606966A (en) * | 2015-12-21 | 2016-05-25 | 安徽理工大学 | Partial discharge pattern recognition method based on mixed neural network algorithm |
Non-Patent Citations (5)
| Title |
|---|
| 化希耀: "基于RBF神经网络与粗糙集的挖掘电力变压器故障诊断研究", 《基于RBF神经网络与粗糙集的挖掘电力变压器故障诊断研究》 * |
| 禹建丽: "基于径向基神经网络的变压器故障诊断", 《科技通报》 * |
| 邵安海: "基于局部放电监测的变压器故障诊断***研究", 《四川电力技术》 * |
| 郭佰胜等: "激活函数可调的RBF神经网络模型", 《微计算机信息》 * |
| 龚瑞昆等: "基于RBF神经网络结合LIBS的变压器故障诊断", 《华北理工大学学报》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108520339A (en) * | 2018-03-23 | 2018-09-11 | 佛山科学技术学院 | A kind of electric power networks fault diagnosis system and method |
| CN109191448A (en) * | 2018-09-03 | 2019-01-11 | 重庆大学 | The transformer fault identification technology with Digital Image Processing is drawn based on three-dimensional coordinate |
| CN109191448B (en) * | 2018-09-03 | 2021-11-02 | 重庆大学 | Transformer fault identification method based on three-dimensional coordinate drawing and digital image processing |
| CN110850228A (en) * | 2019-11-04 | 2020-02-28 | 国网江苏省电力有限公司泰州供电分公司 | Low-voltage transformer area fault positioning method based on equivalent impedance characteristics of phase change switch |
| CN111289837A (en) * | 2020-02-06 | 2020-06-16 | 云南电网有限责任公司电力科学研究院 | Method and system for evaluating latent fault of power distribution network equipment |
| CN111308268A (en) * | 2020-02-06 | 2020-06-19 | 云南电网有限责任公司电力科学研究院 | Method and system for evaluating latent fault of power distribution network equipment |
| CN111289837B (en) * | 2020-02-06 | 2022-04-08 | 云南电网有限责任公司电力科学研究院 | Method and system for evaluating latent fault of power distribution network equipment |
| CN111308268B (en) * | 2020-02-06 | 2022-04-12 | 云南电网有限责任公司电力科学研究院 | Method and system for evaluating latent fault of power distribution network equipment |
| CN111272222A (en) * | 2020-02-28 | 2020-06-12 | 西南交通大学 | Transformer fault diagnosis method based on characteristic quantity set |
| CN112595998A (en) * | 2020-12-01 | 2021-04-02 | 清华大学 | Frequency response testing method based on transformer broadband model and application |
| CN115542101A (en) * | 2022-11-30 | 2022-12-30 | 杭州兆华电子股份有限公司 | Voiceprint preprocessing method of transformer voiceprint detection system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106841905A (en) | A kind of recognition methods of transformer short circuit fault and device | |
| CN109685138B (en) | XLPE power cable partial discharge type identification method | |
| CN110082640B (en) | Distribution network single-phase earth fault identification method based on long-time memory network | |
| CN105425118B (en) | A kind of Multi-information acquisition fault arc detection method and device | |
| CN110514924A (en) | The Winding in Power Transformer Fault Locating Method of depth convolutional neural networks fusion visual recognition | |
| CN109116203A (en) | Power equipment partial discharges fault diagnostic method based on convolutional neural networks | |
| CN109932627A (en) | A kind of localization method and system of GIS pressure test shelf depreciation | |
| CN109871746A (en) | A kind of mode identification method based on the diagnosis of GIS insulation defect | |
| CN110108992B (en) | Cable partial discharge fault identification method and system based on improved random forest algorithm | |
| CN107589341B (en) | Single-phase grounding online fault positioning method based on distribution automation main station | |
| CN107340456A (en) | Power distribution network operating mode intelligent identification Method based on multiple features analysis | |
| CN114414942A (en) | Power transmission line fault identification classifier, identification method and system based on transient waveform image identification | |
| WO2014101579A1 (en) | Method for identifying a transformer local discharge mode based on a singular value decomposition algorithm | |
| CN109100627A (en) | A kind of power equipment partial discharges fault diagnostic method based on end-to-end mode | |
| CN103091612A (en) | Separation and recognition algorithm for transformer oiled paper insulation multiple partial discharging source signals | |
| CN107907799A (en) | The recognition methods of shelf depreciation defect type based on convolutional neural networks and system | |
| CN105372528B (en) | A kind of state maintenance method of Power Transformer Internal Faults and New Transformer | |
| CN108761287A (en) | Transformer partial discharge type identification method | |
| CN110247420B (en) | Intelligent fault identification method for HVDC transmission line | |
| CN110376497A (en) | Low-voltage distribution system series fault arc identification method based on full-phase deep learning | |
| CN107942210A (en) | The classification of transformer pulse electric current Partial Discharge and denoising method and system | |
| De et al. | Recognition of impulse fault patterns in transformers using Kohonen's self-organizing feature map | |
| CN110261746B (en) | Cable defect detection method based on periodic attenuation characteristics of oscillating wave voltage | |
| CN105334436A (en) | Cross-linked cable partial discharge mode identification method based on SOM-BP combined neural network | |
| CN109142963B (en) | Interturn short circuit fault positioning system and method |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170613 |