CN113655348B - Power equipment partial discharge fault diagnosis method, system terminal and readable storage medium based on deep twin network - Google Patents
Power equipment partial discharge fault diagnosis method, system terminal and readable storage medium based on deep twin network Download PDFInfo
- Publication number
- CN113655348B CN113655348B CN202110854734.6A CN202110854734A CN113655348B CN 113655348 B CN113655348 B CN 113655348B CN 202110854734 A CN202110854734 A CN 202110854734A CN 113655348 B CN113655348 B CN 113655348B
- Authority
- CN
- China
- Prior art keywords
- sample
- test sample
- partial discharge
- support
- depth
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000003745 diagnosis Methods 0.000 title claims abstract description 37
- 238000012360 testing method Methods 0.000 claims abstract description 89
- 238000001228 spectrum Methods 0.000 claims abstract description 32
- 238000013507 mapping Methods 0.000 claims abstract description 31
- 102100025142 Beta-microseminoprotein Human genes 0.000 claims description 33
- 101000576812 Homo sapiens Beta-microseminoprotein Proteins 0.000 claims description 33
- 238000001208 nuclear magnetic resonance pulse sequence Methods 0.000 claims description 19
- 230000008569 process Effects 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 5
- 239000002923 metal particle Substances 0.000 claims description 5
- 101100042474 Homo sapiens SFTPD gene Proteins 0.000 claims description 4
- 102100027845 Pulmonary surfactant-associated protein D Human genes 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 238000011176 pooling Methods 0.000 claims description 3
- 238000005339 levitation Methods 0.000 claims description 2
- 238000001514 detection method Methods 0.000 abstract description 8
- 208000028659 discharge Diseases 0.000 description 95
- 238000004891 communication Methods 0.000 description 6
- 238000013135 deep learning Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 238000012549 training Methods 0.000 description 5
- 238000009413 insulation Methods 0.000 description 4
- 230000007547 defect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000003491 array Methods 0.000 description 2
- 230000000295 complement effect Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 239000000725 suspension Substances 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000003213 activating effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000007667 floating Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011524 similarity measure Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/12—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
Abstract
The invention discloses a method, a system, a terminal and a readable storage medium for diagnosing partial discharge faults of power equipment based on a deep twin network, wherein the method comprises the following steps: step 1: acquiring a characteristic spectrum of the power equipment to be tested and taking the characteristic spectrum as a test sample, and acquiring the characteristic spectrum of the power equipment under various partial discharge faults and without partial discharge faults and taking the characteristic spectrum as a support set sample; step 2: inputting the characteristic maps of the test sample and the support set sample into two depth twin network models correspondingly to obtain respective corresponding depth characteristics; step 3: calculating a feature mapping distance based on the depth features of the test sample and the depth features corresponding to each type of support sample in the support set sample; step 4: and determining a partial discharge fault diagnosis result of the test sample according to the feature mapping distance between the test sample and each type of support sample. The method provided by the invention meets the high-precision detection of partial discharge faults under the condition of few fault samples.
Description
Technical Field
The invention belongs to the technical field of power equipment partial discharge fault diagnosis, and particularly relates to a power equipment partial discharge fault diagnosis method, a system terminal and a readable storage medium based on a deep twin network.
Background
During long-term operation of electrical equipment (e.g., gas-insulated switchgear, transformers), various insulation defects may exist, resulting in partial discharge. If the partial discharge fault in the power equipment cannot be detected in time, the partial discharge can be changed into discharge breakdown or spark discharge, so that the power equipment is damaged, and huge economic loss is caused. At present, the diagnosis of the partial discharge fault of the power equipment mainly depends on manual analysis and discrimination of the acquired characteristic spectrum. The method has the advantages of high labor cost, long detection period and high false detection rate. Therefore, an intelligent and reliable partial discharge fault diagnosis method is provided, and further active maintenance measures are taken, so that the method has important practical significance for guaranteeing the safe operation of the power equipment.
In recent years, with the rapid development of deep learning and computer vision technology, a series of image recognition models have emerged. The models can construct fitting relations between input and output through a large amount of training data, so that the purpose of distinguishing images of different categories is achieved. In light of the above, some researchers begin to apply the image recognition model to the diagnosis of the partial discharge fault of the power equipment, and a general means is to use a large number of data samples as model input, take the diagnosis result of the partial discharge fault as model output, and build a diagnosis model of the partial discharge fault based on deep learning through training. However, the quality of current model performance depends on whether there are a large number of data samples for model training. In the practical application process, the fault samples of partial discharge are quite rare, and it is difficult to acquire enough data samples to train an image recognition model. Therefore, under the condition of few fault samples, the detection precision of the existing image recognition model is often unsatisfactory, and the partial discharge fault of the power equipment cannot be well diagnosed.
Disclosure of Invention
The invention aims to provide a power equipment partial discharge fault diagnosis method, a system terminal and a readable storage medium based on a deep twin network, aiming at the problem that the existing partial discharge diagnosis model is not high in inspection precision under the condition of few fault samples. The method is characterized in that the map information of various partial discharge faults is used as a reference, the map features are extracted by using a deep learning network, and similar calculation is carried out on the map features of the power equipment to be detected and the map features corresponding to the various partial discharge faults based on the map features, so that the partial discharge fault detection result of the power equipment to be detected is obtained. The method provided by the invention meets the high-precision detection of partial discharge faults under the condition of few fault samples.
In one aspect, the invention provides a method for diagnosing partial discharge faults of power equipment based on a deep twin network, which comprises the following steps:
step 1: acquiring a characteristic spectrum of the power equipment to be tested and taking the characteristic spectrum as a test sample, and acquiring the characteristic spectrum of the power equipment under various partial discharge faults and without partial discharge faults and taking the characteristic spectrum as a support set sample;
wherein the characteristic map comprises partial discharge characteristic information;
step 2: inputting the characteristic maps of the test sample and the support set sample into two depth twin network models correspondingly to obtain respective corresponding depth characteristics;
step 3: calculating a feature mapping distance based on the depth features of the test sample and the depth features corresponding to each type of support sample in the support set sample;
step 4: determining a partial discharge fault diagnosis result of a test sample according to a feature mapping distance corresponding to each type of support sample;
the larger the feature mapping distance is, the closer the test sample is to the corresponding support sample.
Optionally, the characteristic spectrum is one or a combination of a pulse sequence spectrum PRPS and a phase spectrum PRPD.
Optionally, if the feature profile includes a pulse sequence profile PRPS and a phase profile PRPD, the depth features of the test sample and the support sample in step 2 are respectively expressed as:
wherein O is t In order to test the corresponding depth features of the sample,for testing sample pulsesSequence pattern->Corresponding output result,/->For testing sample phase pattern->Outputting a corresponding result; o (O) s (i) For supporting the corresponding depth features of the sample +.>Pulse sequence profile for supporting samples +.>Corresponding output result,/->Phase profile for supporting sample->Corresponding output results, i corresponds to the type of support sample;
the feature mapping distances of the test sample and the support sample in the step 3 include similar distances on the PRPD map and similar distances on the PRPS map, expressed as:
wherein E is PRPD (i) Representing the similar distance on the PRPD pattern between the test sample and the support sample of type i, E PSPD (i) Representing the similar distance on the PRPS profile between the test sample and the class i support sample.
Optionally, in step 4, the partial discharge fault diagnosis result of the test sample is determined according to the feature mapping distance corresponding to each type of support sample, and the specific process is as follows:
firstly, calculating the sum of similar distances between a test sample and each type of support sample on a PRPD map and a PRPS map;
E(i)＝E PRPD (i)+E PRPS (i)
e (i) is the sum of the similar distances of the test sample and the class i support sample on the PRPD and PRPS profiles;
then, the maximum value E in the sum of the similar distances is selected max The type of the selected support sample corresponds to the partial discharge fault diagnosis result of the test sample;
E max ＝Softmax{E(i)}
wherein Softmax is a function of the maximum value.
Optionally, the similar distance is a euclidean distance.
Optionally, the support set sample comprises: metal particle discharge, levitation discharge, corona discharge, and insulating gap discharge, and normal partial discharge-free type 5 samples.
Optionally, the depth twin network model is a VGG-16 model, which is composed of a multi-layer convolution layer, a Sigmoid activation function, a Mean-Pooling layer, and a fully connected layer.
In a second aspect, the present invention provides a system based on the above-mentioned method for diagnosing a partial discharge fault of an electrical device, including:
the sample acquisition module is used for acquiring a characteristic spectrum of the power equipment to be tested and taking the characteristic spectrum as a test sample, and acquiring the characteristic spectrum of the power equipment under various partial discharge faults and without the partial discharge faults and taking the characteristic spectrum as a support set sample;
wherein the characteristic map comprises partial discharge characteristic information;
the depth feature acquisition module is used for inputting the feature maps of the test sample and the support set sample into two depth twin network models correspondingly to obtain corresponding depth features;
the distance calculation module is used for calculating a feature mapping distance based on the depth features of the test sample and the depth features corresponding to each type of support sample in the support set samples;
the diagnosis module is used for determining a partial discharge fault diagnosis result of the test sample according to the feature mapping distance corresponding to the test sample and each type of support sample;
the larger the feature mapping distance is, the closer the test sample is to the corresponding support sample.
In a third aspect, the present invention provides a terminal comprising one or more processors and a memory storing one or more programs, the processors invoking the programs in the memory to implement:
a method for diagnosing partial discharge faults of power equipment based on a deep twin network.
In a fourth aspect, the present invention provides a readable storage medium storing a computer program, the computer program being invoked by a processor to implement:
a method for diagnosing partial discharge faults of power equipment based on a deep twin network.
Advantageous effects
The method for diagnosing the partial discharge fault of the power equipment based on the deep twin network provided by the invention aims at inputting the detection sample and the support set sample into a pre-trained deep learning model for similarity measurement under the condition of few fault samples, evaluates the depth characteristic difference between the two input samples, realizes high-precision diagnosis of the type of the partial discharge fault, and has great practical significance for diagnosing the partial discharge fault of the power equipment. The technical obstacle that a large number of samples are required in the aspect of the existing image recognition model is solved.
Drawings
Fig. 1 is a flowchart of a method for diagnosing a partial discharge fault of a power device based on a deep twin network according to an embodiment of the present invention.
FIG. 2 is a schematic diagram of a similarity measure for a test sample and a support set sample using a deep twin network in accordance with an embodiment of the present invention.
Fig. 3 is a schematic structural diagram of a twin network according to an embodiment of the present invention.
Fig. 4 is a PRPS and PRPD profile of a support set sample and one test sample.
Detailed Description
The invention provides a deep twin network-based power equipment partial discharge fault diagnosis method which is different from the traditional partial discharge fault classification model based on machine learning or deep learning, and does not need a large number of samples for model training. The method of the invention uses the spectrum information of various partial discharge faults as reference, utilizes a deep learning network to extract the spectrum characteristics, and then carries out similar calculation on the spectrum characteristics of the power equipment to be detected and the spectrum characteristics corresponding to various partial discharge faults based on the spectrum characteristics to obtain the detection result of the partial discharge faults of the power equipment to be detected. The process does not need a large number of samples, and solves the technical obstacle that the existing image recognition model has a large number of sample requirements.
The technical solutions in the embodiments of the present invention will be fully and clearly described below with reference to the accompanying drawings in the embodiments of the present invention.
Example 1:
the method for diagnosing the partial discharge fault of the power equipment based on the deep twin network comprises the following steps:
s1: and acquiring a characteristic map of the electric equipment to be tested and taking the characteristic map as a test sample. In this embodiment, two types of characteristic patterns, namely a phase pattern (Phase resolved partial discharge, PRPD) and a pulse sequence pattern (Phase resolved pulse sequence, PRPS), are selected, so that the electrical equipment to be tested is analyzed by an ultrahigh frequency method to obtain the phase pattern PRPD and the pulse sequence pattern PRPS.
The phase map PRPD can record the relationship among the phase, the discharge signal amplitude and the discharge frequency in a plurality of periods, and can be used as the classification basis of different discharge types. The map is generated by using a wavelet transform or a hilbert-yellow transform. The pulse sequence pattern PRPS is a 3-dimensional distribution of discharge amplitude with respect to phase and power frequency period. The PRPS graph can extract partial discharge information such as phase distribution, and the PRPS shows similar distribution under the same defect, and shows obvious difference under different defects. Therefore, the PRPS map can also be used as the classification basis of different discharge types. Because the two kinds of maps are used for analyzing the same power equipment, the partial discharge faults of the power equipment can be reflected from different angles, and the two kinds of maps are used together, so that the complementary information between the maps can be fully mined, and the fault diagnosis precision of the partial discharge is improved.
In other possible embodiments, the phase profile PRPD or the pulse sequence profile PRPS may be selected alone as a feature profile, or may be selected to be combined with other profiles, which is not specifically limited in the present invention, and it should be understood that the selected feature profile includes a partial discharge fault feature, and may be used to distinguish the type of partial discharge fault.
S2: and acquiring characteristic maps of the power equipment under various partial discharge faults and under no partial discharge faults, and taking the characteristic maps as a support set sample.
In this embodiment, pulse sequence patterns and phase patterns of 5 kinds of images in total of metal particle discharge, suspension discharge, corona discharge, insulation gap discharge and normal partial discharge-free samples are obtained in advance. I.e. metal particle discharge X 1 Suspension discharge X 2 Corona discharge X 3 And an insulation gap discharge X 4 Totally 4 types of fault discharge samples and normal partial discharge-free sample X 5 Together 5 types of samples. 1 picture is selected from each type of sample, and a total of 5 pictures are taken as support set samples. A support set sample and one input test sample are illustrated in fig. 4. Fig. 4 (a) - (e) are respectively a normal no-discharge sample image, a floating discharge sample image, a metal particle discharge sample image, an insulation gap discharge sample image, and a corona discharge sample image in the support set sample. Fig. 4 (f) shows an input test sample image. In the figure, the left subgraph is the PRPS map, and the right subgraph is the PRPD map.
Wherein,and->Represented as a phase profile PRPD and a pulse sequence profile PRPS, respectively, of the test sample. />And->Respectively denoted as phase profile PRPD and pulse sequence profile PRPS of the support set samples, where i=1, 2,3,4,5, respectively denoted as 5 categories of support set. The test sample comprises a phase pattern->And pulse sequence profile->Namely:
the support set sample also contains a phase mapAnd pulse sequence profile->Namely:
s3: and respectively inputting the characteristic maps of the test sample and the support set sample into two depth twin network models correspondingly to obtain respective corresponding depth characteristics.
Fig. 2 shows a schematic diagram of similarity measurement of a test sample and a support set sample using a deep twin network in this embodiment. Wherein, a twin network obtains depth characteristic information of the characterization test sample and the support set sample on the PRPD map, and the depth characteristic information is used for calculating the similarity. The other twin network obtains depth characteristic information characterizing the test sample and the support set sample on the PRPS map for calculating the similarity.
As shown in fig. 3, the deep twin network is composed of two VGG-16 models with identical structure and shared parameters. The VGG-16 model consists of a multi-layer convolution layer, a Sigmoid activation function, a Mean-Pooling layer and a full connection layer. And taking the test sample image and the support set sample as two inputs of the depth twin network, and respectively extracting depth characteristic information of the test picture and the support set picture. Carrying out abstract processing on the test picture and the support set picture through a VGG-16 model respectively to obtain respective depth characteristic information:
O t ＝G(X t ) (3)
wherein G (X) t ) Representing that the abstract processing is carried out by VGG-16 model to input test picture X t Step (1), O t Is X t And outputting the depth characteristic information.
O s (i)＝G(X s (i)) (4)
Similarly, G (X) s (i) Representing the abstract processing of the VGG-16 model to input the support set picture X s (i) Step (1), O s (i) Is equal to and X s (i) And outputting the depth characteristic information.
Depth feature O of test sample output t Comprises two parts:
in the method, in the process of the invention,is pulse sequence pattern->Output result of->Is a phase pattern->Output results of (2). Depth feature O of support set sample output s (i) Comprises two parts:
in the method, in the process of the invention,is pulse sequence pattern->Output result of->Is a phase pattern->Output results of (2). Wherein i=1, 2,3,4,5, respectively correspond to 5 categories in the support set. Because the VGG-16 model is a network model with strong generalization capability, depth features among images can be well dug by adopting the model, and the difference among different types of images can be evaluated.
Wherein it should be appreciated that in other possible embodiments, where the number of feature maps is different, the corresponding depth features will vary adaptively.
S4: and calculating a feature mapping distance based on the depth features of the test sample and the depth features corresponding to each type of support sample in the support set samples.
In this embodiment, the feature map distance is selected to be represented by the euclidean distance. The feature mapping distance of the test sample and the support set sample output on the PRPD map is calculated by adopting Euclidean distance, and the feature mapping distance is as follows:
wherein E is PRPD (i) The test samples are shown as being at similar distances on the PRPD pattern from the 5 types of samples in the support set, respectively. Similarly, the Euclidean distance is adopted to calculate the feature mapping distance of the test sample and the support set sample output on the PRPS map,
wherein E is PSPD (i) The test samples are shown as being at similar distances on the PRPS profile from the 5 types of samples in the support set, respectively. The network adopts an activating function as a Sigmoid function, so that the numerical value of an output result is 0,1]Between them.
As can be seen from the above, in this embodiment, the feature mapping distance between the test sample and any type of support sample includes the similar distance on the PRPD map and the similar distance on the PRPS map.
S5: determining a partial discharge fault diagnosis result of a test sample according to a feature mapping distance corresponding to each type of support sample; in order to fully mine complementary information between PRPS and PRPD patterns and improve the accuracy of partial discharge fault diagnosis, the following formula is constructed:
E(i)＝E PRPD (i)+E PRPS (i) (9)
e (i) is the sum of the similarity of the test sample and the support set sample on the PRPD and PRPS patterns, and by calculating E (i), the misjudgment of the model on part of the samples under a single pattern can be avoided, and the diagnosis performance of the model is improved.
E max ＝Softmax{E(i)} (10)
Where Softmax is the maximum value between the calculated arrays. E (E) max Is the value of the maximum similarity. The output diagnosis result is the category in the support set corresponding to the maximum similarity value.
It should be appreciated that in other possible embodiments, the final diagnostic result may be determined using averaging, re-ordering, etc., in addition to summing. Or by using other existing mathematical means.
In summary, according to the depth twin network-based power equipment partial discharge fault diagnosis method provided by the invention, 5 types of standard samples are selected and similar calculation is performed based on depth characteristics, so that the partial discharge fault diagnosis result of the power equipment to be tested is determined, a large number of samples are not needed in the process, and high-precision diagnosis can be completed under the condition of few samples.
Example 2:
the present implementation provides a diagnostic system based on a power equipment partial discharge fault diagnosis method, which includes:
the sample acquisition module is used for acquiring the characteristic patterns of the power equipment to be tested and taking the characteristic patterns as test samples, and acquiring the characteristic patterns of the power equipment under various partial discharge faults and without partial discharge faults and taking the characteristic patterns as support set samples.
As in example 1 above, the phase profile PRPD and pulse sequence profile PRPS synergy may be selected as feature profiles to participate in training and calculation. In other possible embodiments, either or both may be selected or combined with other types of images.
And the depth feature acquisition module is used for inputting the feature maps of the test sample and the support set sample into two depth twin network models correspondingly to obtain respective corresponding depth features. Reference may be made in particular to the statements made in the context of the process in example 1.
And the distance calculation module is used for calculating a feature mapping distance based on the depth features of the test sample and the depth features corresponding to each type of support sample in the support set samples.
In this embodiment, the euclidean distance is selected to calculate the feature mapping distance, and in other possible embodiments, other similar functions may be selected to calculate the feature mapping distance.
And the diagnosis module is used for determining the partial discharge fault diagnosis result of the test sample according to the feature mapping distance corresponding to the test sample and each type of support sample. Reference may be made in particular to statements of the method content.
The specific implementation process of each unit module refers to the corresponding process of the method. It should be understood that, in the specific implementation process of the above unit module, reference is made to the method content, the present invention is not specifically described herein, and the division of the functional module unit is merely a division of a logic function, and there may be another division manner when actually implemented, for example, multiple units or components may be combined or may be integrated into another system, or some features may be omitted or not performed. Meanwhile, the integrated units can be realized in a hardware form or a software functional unit form.
Example 3:
the embodiment provides a terminal, which comprises one or more processors and a memory for storing one or more programs, wherein the processors call the programs in the memory to realize:
a method for diagnosing partial discharge faults of power equipment based on a deep twin network.
For example, the following steps are specifically performed:
s1: and acquiring a characteristic map of the electric equipment to be tested and taking the characteristic map as a test sample.
S2: and acquiring characteristic maps of the power equipment under various partial discharge faults and under no partial discharge faults, and taking the characteristic maps as a support set sample.
S3: and respectively inputting the characteristic maps of the test sample and the support set sample into two depth twin network models correspondingly to obtain respective corresponding depth characteristics.
S4: and calculating a feature mapping distance based on the depth features of the test sample and the depth features corresponding to each type of support sample in the support set samples.
S5: and determining a partial discharge fault diagnosis result of the test sample according to the feature mapping distance between the test sample and each type of support sample.
The terminal further includes: and the communication interface is used for communicating with external equipment and carrying out data interaction transmission.
The memory may comprise high-speed RAM memory, and may also include a non-volatile defibrillator, such as at least one disk memory.
If the memory, processor, and communication interface are implemented independently, the memory, processor, and communication interface may be interconnected and communicate with each other via a bus. The bus may be an industry standard architecture bus, an external device interconnect bus, or an extended industry standard architecture bus, among others. The buses may be classified as address buses, data buses, control buses, etc.
Alternatively, in a specific implementation, if the memory, the processor, and the communication interface are integrated on a chip, the memory, the processor, or the communication interface may perform communication with each other through the internal interface.
It should be understood that the above steps are specific to the corresponding steps of the above method.
In an embodiment of the present invention, the processor may be a central processing unit (Central Processing Unit, CPU), which may also be other general purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), off-the-shelf Programmable gate arrays (FPGAs) or other Programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The memory may include read only memory and random access memory and provide instructions and data to the processor. A portion of the memory may also include non-volatile random access memory. For example, the memory may also store information of the device type.
Example 4:
an embodiment of the present invention provides a readable storage medium storing a computer program that is called by a processor to implement:
a method for diagnosing partial discharge faults of power equipment based on a deep twin network.
For example, the following steps are specifically performed:
s1: and acquiring a characteristic map of the electric equipment to be tested and taking the characteristic map as a test sample.
S2: and acquiring characteristic maps of the power equipment under various partial discharge faults and under no partial discharge faults, and taking the characteristic maps as a support set sample.
S3: and respectively inputting the characteristic maps of the test sample and the support set sample into two depth twin network models correspondingly to obtain respective corresponding depth characteristics.
S4: and calculating a feature mapping distance based on the depth features of the test sample and the depth features corresponding to each type of support sample in the support set samples.
S5: and determining a partial discharge fault diagnosis result of the test sample according to the feature mapping distance between the test sample and each type of support sample.
It should be understood that the above steps are specific to the corresponding steps of the above method.
The readable storage medium is a computer readable storage medium, which may be an internal storage unit of the controller according to any one of the foregoing embodiments, for example, a hard disk or a memory of the controller. The readable storage medium may also be an external storage device of the controller, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card) or the like, which are provided on the controller. Further, the readable storage medium may also include both an internal storage unit and an external storage device of the controller. The readable storage medium is used to store the computer program and other programs and data required by the controller. The readable storage medium may also be used to temporarily store data that has been output or is to be output.
Based on such understanding, the technical solution of the present invention is essentially or a part contributing to the prior art, or all or part of the technical solution may be embodied in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned readable storage medium includes: a U-disk, a removable hard disk, a Read-only Memory (ROM), a random access Memory (RAM, randomAccess Memory), a magnetic disk, an optical disk, or other various media capable of storing program codes.
It should be emphasized that the examples described herein are illustrative rather than limiting, and that this invention is not limited to the examples described in the specific embodiments, but is capable of other embodiments in accordance with the teachings of the present invention, as long as they do not depart from the spirit and scope of the invention, whether modified or substituted, and still fall within the scope of the invention.
Claims (8)
1. A power equipment partial discharge fault diagnosis method based on a deep twin network is characterized in that: the method comprises the following steps:
step 1: acquiring a characteristic spectrum of the power equipment to be tested and taking the characteristic spectrum as a test sample, and acquiring the characteristic spectrum of the power equipment under various partial discharge faults and without partial discharge faults and taking the characteristic spectrum as a support set sample;
wherein the characteristic map comprises partial discharge characteristic information; the characteristic spectrum is a combination of a pulse sequence spectrum PRPS and a phase spectrum PRPD;
step 2: inputting the characteristic maps of the test sample and the support set sample into two depth twin network models correspondingly to obtain respective corresponding depth characteristics;
step 3: calculating a feature mapping distance based on the depth features of the test sample and the depth features corresponding to each type of support sample in the support set sample;
step 4: determining a partial discharge fault diagnosis result of a test sample according to a feature mapping distance corresponding to each type of support sample, wherein the specific process is as follows:
firstly, calculating the sum of similar distances between a test sample and each type of support sample on a PRPD map and a PRPS map;
E(i)＝E PRPD (i)+E PRPS (i)
e (i) is the sum of the similar distances of the test sample and the class i support sample on the PRPD and PRPS profiles; e (E) PRPD (i) Representing the similar distance on the PRPD pattern between the test sample and the support sample of type i, E PSPD (i) Representing the similar distance between the test sample and the class i support sample on the PRPS profile;
then, the maximum value E in the sum of the similar distances is selected max The type of the selected support sample corresponds to the partial discharge fault diagnosis result of the test sample;
E max ＝Softmax{E(i)}
wherein Softmax is a function of the maximum;
the larger the feature mapping distance is, the closer the test sample is to the corresponding support sample.
2. The method according to claim 1, characterized in that: the depth characteristics of the test sample and the support sample in step 2 are respectively expressed as:
wherein O is t In order to test the corresponding depth features of the sample,for testing the sample pulse sequence profile +.>Corresponding output result,/->To test sample phase pattern/>Outputting a corresponding result; o (O) s (i) For supporting the corresponding depth features of the sample +.>Pulse sequence profile for supporting samples +.>Corresponding output result,/->Phase profile for supporting sample->Corresponding output results, i corresponds to the type of support sample;
the feature mapping distances of the test sample and the support sample in the step 3 include similar distances on the PRPD map and similar distances on the PRPS map, expressed as:
wherein E is PRPD (i) Representing the similar distance on the PRPD pattern between the test sample and the support sample of type i, E PSPD (i) Representing the similar distance on the PRPS profile between the test sample and the class i support sample.
3. The method according to claim 1, characterized in that: the similar distance is a Euclidean distance.
4. The method according to claim 1, characterized in that: the support set sample includes: metal particle discharge, levitation discharge, corona discharge, and insulating gap discharge, and normal partial discharge-free type 5 samples.
5. The method according to claim 1, characterized in that: the depth twin network model is a VGG-16 model and consists of a multi-layer convolution layer, a Sigmoid activation function, a Mean-Pooling layer and a full connection layer.
6. A system based on the method of any one of claims 1-5, characterized in that: comprising the following steps:
the sample acquisition module is used for acquiring a characteristic spectrum of the power equipment to be tested and taking the characteristic spectrum as a test sample, and acquiring the characteristic spectrum of the power equipment under various partial discharge faults and without the partial discharge faults and taking the characteristic spectrum as a support set sample;
wherein the characteristic map comprises partial discharge characteristic information;
the depth feature acquisition module is used for inputting the feature maps of the test sample and the support set sample into two depth twin network models correspondingly to obtain corresponding depth features;
the distance calculation module is used for calculating a feature mapping distance based on the depth features of the test sample and the depth features corresponding to each type of support sample in the support set samples;
the diagnosis module is used for determining a partial discharge fault diagnosis result of the test sample according to the feature mapping distance corresponding to the test sample and each type of support sample;
the larger the feature mapping distance is, the closer the test sample is to the corresponding support sample.
7. A terminal, characterized by: a memory comprising one or more processors and one or more programs stored therein, the processors invoking the programs in the memory to implement:
the method of any one of claims 1-5.
8. A readable storage medium, characterized by: a computer program is stored, which is called by a processor to implement:
the method of any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110854734.6A CN113655348B (en) | 2021-07-28 | 2021-07-28 | Power equipment partial discharge fault diagnosis method, system terminal and readable storage medium based on deep twin network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110854734.6A CN113655348B (en) | 2021-07-28 | 2021-07-28 | Power equipment partial discharge fault diagnosis method, system terminal and readable storage medium based on deep twin network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113655348A CN113655348A (en) | 2021-11-16 |
| CN113655348B true CN113655348B (en) | 2023-12-08 |
Family
ID=78478839
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110854734.6A Active CN113655348B (en) | 2021-07-28 | 2021-07-28 | Power equipment partial discharge fault diagnosis method, system terminal and readable storage medium based on deep twin network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113655348B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114049507A (en) * | 2021-11-19 | 2022-02-15 | 国网湖南省电力有限公司 | Distribution network line insulator defect identification method, equipment and medium based on twin network |
| CN114295942B (en) * | 2021-11-29 | 2024-03-15 | 国网北京市电力公司 | Power cable fault diagnosis system, power cable fault determination method and device |
| CN114236329A (en) * | 2021-11-30 | 2022-03-25 | 国网北京市电力公司 | Method and device for detecting running state of high-voltage cable |
| CN114520781B (en) * | 2022-04-21 | 2022-09-30 | 中国科学技术大学 | Network technology and protocol testing platform based on digital twinning and testing method thereof |
Citations (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014101579A1 (en) * | 2012-12-28 | 2014-07-03 | 国家电网公司 | Method for identifying a transformer local discharge mode based on a singular value decomposition algorithm |
| CN104198898A (en) * | 2014-08-04 | 2014-12-10 | 西安交通大学 | Local discharge development process diagnosis method based on pulse-train analysis |
| WO2015070513A1 (en) * | 2013-11-14 | 2015-05-21 | 国家电网公司 | Pattern recognition method for partial discharge of three-phase in one enclosure type ultrahigh voltage gis |
| CN105044566A (en) * | 2015-06-25 | 2015-11-11 | 国家电网公司 | GIS partial discharge fault detection method based on characteristic ultrahigh frequency signal |
| CN106556781A (en) * | 2016-11-10 | 2017-04-05 | 华乘电气科技（上海）股份有限公司 | Shelf depreciation defect image diagnostic method and system based on deep learning |
| CN107449994A (en) * | 2017-07-04 | 2017-12-08 | 国网江苏省电力公司电力科学研究院 | Partial discharge method for diagnosing faults based on CNN DBN networks |
| CN108596277A (en) * | 2018-05-10 | 2018-09-28 | 腾讯科技（深圳）有限公司 | A kind of testing vehicle register identification method, apparatus and storage medium |
| CN108805200A (en) * | 2018-06-08 | 2018-11-13 | 中国矿业大学 | Optical remote sensing scene classification method and device based on the twin residual error network of depth |
| CN109188211A (en) * | 2018-07-30 | 2019-01-11 | 上海交通大学 | A kind of high voltage equipment insulation method for diagnosing faults and system |
| CN109376990A (en) * | 2018-09-14 | 2019-02-22 | 中国电力科学研究院有限公司 | A kind of method and system for the critical clearing time determining electric system based on Siamese network model |
| CN109766921A (en) * | 2018-12-19 | 2019-05-17 | 合肥工业大学 | A kind of vibration data Fault Classification based on depth domain-adaptive |
| CN110412431A (en) * | 2019-08-05 | 2019-11-05 | 国网湖南省电力有限公司 | A kind of diagnostic method and diagnostic system of the shelf depreciation defect type of power equipment |
| CN110533654A (en) * | 2019-08-30 | 2019-12-03 | 北京明略软件系统有限公司 | The method for detecting abnormality and device of components |
| CN110648320A (en) * | 2019-09-19 | 2020-01-03 | 京东方科技集团股份有限公司 | Bone age acquisition method and system, server, computer device and medium |
| CN111612763A (en) * | 2020-05-20 | 2020-09-01 | 重庆邮电大学 | Mobile phone screen defect detection method, device and system, computer equipment and medium |
| CN111613339A (en) * | 2020-05-15 | 2020-09-01 | 山东大学 | Similar medical record searching method and system based on deep learning |
| CN112070104A (en) * | 2020-07-10 | 2020-12-11 | 国网浙江平阳县供电有限责任公司 | Main transformer partial discharge identification method |
| CN112418451A (en) * | 2020-11-05 | 2021-02-26 | 武汉大学 | Transformer fault diagnosis positioning system based on digital twinning |
| CN112686093A (en) * | 2020-12-02 | 2021-04-20 | 重庆邮电大学 | Fusion partial discharge type identification method based on DS evidence theory |
| CN112949402A (en) * | 2021-02-01 | 2021-06-11 | 南京航空航天大学 | Fault diagnosis method for planetary gear box under minimum fault sample size |
| CN113064032A (en) * | 2021-03-26 | 2021-07-02 | 云南电网有限责任公司电力科学研究院 | Partial discharge mode identification method based on map features and information fusion |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080036487A1 (en) * | 2006-08-09 | 2008-02-14 | Arm Limited | Integrated circuit wearout detection |
-
2021
- 2021-07-28 CN CN202110854734.6A patent/CN113655348B/en active Active
Patent Citations (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014101579A1 (en) * | 2012-12-28 | 2014-07-03 | 国家电网公司 | Method for identifying a transformer local discharge mode based on a singular value decomposition algorithm |
| WO2015070513A1 (en) * | 2013-11-14 | 2015-05-21 | 国家电网公司 | Pattern recognition method for partial discharge of three-phase in one enclosure type ultrahigh voltage gis |
| CN104198898A (en) * | 2014-08-04 | 2014-12-10 | 西安交通大学 | Local discharge development process diagnosis method based on pulse-train analysis |
| CN105044566A (en) * | 2015-06-25 | 2015-11-11 | 国家电网公司 | GIS partial discharge fault detection method based on characteristic ultrahigh frequency signal |
| CN106556781A (en) * | 2016-11-10 | 2017-04-05 | 华乘电气科技（上海）股份有限公司 | Shelf depreciation defect image diagnostic method and system based on deep learning |
| CN107449994A (en) * | 2017-07-04 | 2017-12-08 | 国网江苏省电力公司电力科学研究院 | Partial discharge method for diagnosing faults based on CNN DBN networks |
| CN108596277A (en) * | 2018-05-10 | 2018-09-28 | 腾讯科技（深圳）有限公司 | A kind of testing vehicle register identification method, apparatus and storage medium |
| CN108805200A (en) * | 2018-06-08 | 2018-11-13 | 中国矿业大学 | Optical remote sensing scene classification method and device based on the twin residual error network of depth |
| CN109188211A (en) * | 2018-07-30 | 2019-01-11 | 上海交通大学 | A kind of high voltage equipment insulation method for diagnosing faults and system |
| CN109376990A (en) * | 2018-09-14 | 2019-02-22 | 中国电力科学研究院有限公司 | A kind of method and system for the critical clearing time determining electric system based on Siamese network model |
| CN109766921A (en) * | 2018-12-19 | 2019-05-17 | 合肥工业大学 | A kind of vibration data Fault Classification based on depth domain-adaptive |
| CN110412431A (en) * | 2019-08-05 | 2019-11-05 | 国网湖南省电力有限公司 | A kind of diagnostic method and diagnostic system of the shelf depreciation defect type of power equipment |
| CN110533654A (en) * | 2019-08-30 | 2019-12-03 | 北京明略软件系统有限公司 | The method for detecting abnormality and device of components |
| CN110648320A (en) * | 2019-09-19 | 2020-01-03 | 京东方科技集团股份有限公司 | Bone age acquisition method and system, server, computer device and medium |
| CN111613339A (en) * | 2020-05-15 | 2020-09-01 | 山东大学 | Similar medical record searching method and system based on deep learning |
| CN111612763A (en) * | 2020-05-20 | 2020-09-01 | 重庆邮电大学 | Mobile phone screen defect detection method, device and system, computer equipment and medium |
| CN112070104A (en) * | 2020-07-10 | 2020-12-11 | 国网浙江平阳县供电有限责任公司 | Main transformer partial discharge identification method |
| CN112418451A (en) * | 2020-11-05 | 2021-02-26 | 武汉大学 | Transformer fault diagnosis positioning system based on digital twinning |
| CN112686093A (en) * | 2020-12-02 | 2021-04-20 | 重庆邮电大学 | Fusion partial discharge type identification method based on DS evidence theory |
| CN112949402A (en) * | 2021-02-01 | 2021-06-11 | 南京航空航天大学 | Fault diagnosis method for planetary gear box under minimum fault sample size |
| CN113064032A (en) * | 2021-03-26 | 2021-07-02 | 云南电网有限责任公司电力科学研究院 | Partial discharge mode identification method based on map features and information fusion |
Non-Patent Citations (6)
| Title |
|---|
| Understanding of Incipient discharges in Transformer Insulation by reconstruction of Digital Twins for the discharges using Generative Adversarial Networks;Mahidhar, GDP 等;《IEEE》;全文 * |
| VMD-PE协同SNN的输电线路故障辨识方法;付华;金岑;;电子测量与仪器学报(第06期);全文 * |
| 付华 ; 金岑 ; .VMD-PE协同SNN的输电线路故障辨识方法.电子测量与仪器学报.2020,(第06期),全文. * |
| 基于变分贝叶斯自编码器的局部放电数据匹配方法;宋辉;代杰杰;张卫东;毕凯;罗林根;盛戈;江秀臣;;中国电机工程学报(第19期);全文 * |
| 基于深度稀疏降噪自编码网络的局部放电模式识别;郗晓光;何金;曹梦;陈荣;宋晓博;李苏雅;;电气自动化(第04期);全文 * |
| 小样本条件下基于卷积孪生网络的变压器故障诊断;朱瑞金等;电力系统及其自动化学报;第33卷(第1期);64-69、84 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113655348A (en) | 2021-11-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113655348B (en) | Power equipment partial discharge fault diagnosis method, system terminal and readable storage medium based on deep twin network | |
| Bessous et al. | Diagnosis of bearing defects in induction motors using discrete wavelet transform | |
| CN112034310A (en) | Partial discharge defect diagnosis method and system for combined electrical appliance | |
| CN113358993B (en) | Online fault diagnosis method and system for multi-level converter IGBT | |
| CN115908142B (en) | Visual identification-based damage inspection method for tiny contact net parts | |
| CN112419268A (en) | Method, device, equipment and medium for detecting image defects of power transmission line | |
| CN113409284A (en) | Circuit board fault detection method, device, equipment and storage medium | |
| CN109557434B (en) | Partial discharge signal identification method under strong background noise based on composite dictionary sparse representation classification | |
| CN112308148A (en) | Defect category identification and twin neural network training method, device and storage medium | |
| Sun et al. | Partial discharge pattern recognition of transformers based on the gray-level co-occurrence matrix of optimal parameters | |
| CN112146882B (en) | Bearing fault diagnosis method based on transfer learning vibration signal image recognition | |
| Ma et al. | Fractal‐based autonomous partial discharge pattern recognition method for MV motors | |
| CN114325433A (en) | Lithium ion battery fault detection method and system based on electrochemical impedance spectrum test | |
| Yu et al. | Design of the transmission line inspection system based on UAV | |
| CN112881879A (en) | High-voltage cable terminal partial discharge mode identification method, device and equipment | |
| Wu et al. | Hilbert ID considering multi-window feature extraction for transformer deep vision fault positioning | |
| CN111597096B (en) | Benchmark test method, system and terminal equipment | |
| CN113313709A (en) | Method and device for detecting defects of industrial parts | |
| CN111695501A (en) | Equipment soft fault detection method based on operating system kernel calling data | |
| CN117435964B (en) | Equipment state evaluation method and system based on data diagnostic instrument | |
| CN117074866A (en) | Single-phase double-split parallel cable fault diagnosis method, device, equipment and medium | |
| TWI715428B (en) | Identification method, identification device and identification system of cell image | |
| Li et al. | Fault Diagnosis of Transformer Windings Based on Decision Tree and Fully Connected Neural Network. Energies 2021, 14, 1531 | |
| CN117133312A (en) | Method and device for monitoring running state of equipment in machine room, electronic equipment and storage medium | |
| CN117825876A (en) | Cable fault feature extraction method and device combining feature fragments and wavelet packets |
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 |