CN114089218A - Power distribution network high-resistance grounding fault identification method, device, terminal and medium - Google Patents
Power distribution network high-resistance grounding fault identification method, device, terminal and medium Download PDFInfo
- Publication number
- CN114089218A CN114089218A CN202111217367.5A CN202111217367A CN114089218A CN 114089218 A CN114089218 A CN 114089218A CN 202111217367 A CN202111217367 A CN 202111217367A CN 114089218 A CN114089218 A CN 114089218A
- Authority
- CN
- China
- Prior art keywords
- fault
- ground fault
- distribution network
- resistance ground
- power distribution
- 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
- 238000000034 method Methods 0.000 title claims abstract description 81
- 238000009826 distribution Methods 0.000 title claims abstract description 67
- 238000012549 training Methods 0.000 claims abstract description 45
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 39
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 26
- 238000005070 sampling Methods 0.000 claims abstract description 23
- 230000009466 transformation Effects 0.000 claims abstract description 12
- 238000000605 extraction Methods 0.000 claims abstract description 7
- 238000007637 random forest analysis Methods 0.000 claims description 52
- 230000008569 process Effects 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 9
- 238000002372 labelling Methods 0.000 claims description 8
- 238000012216 screening Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 7
- 238000003066 decision tree Methods 0.000 description 6
- 238000012360 testing method Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 101100499229 Mus musculus Dhrsx gene Proteins 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 239000003990 capacitor Substances 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Images
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
- G01R31/52—Testing for short-circuits, leakage current or ground faults
-
- 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
- G01R31/58—Testing of lines, cables or conductors
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Testing Of Short-Circuits, Discontinuities, Leakage, Or Incorrect Line Connections (AREA)
Abstract
The application discloses a method, a device, a terminal and a medium for identifying a high-resistance ground fault of a power distribution network, the method for identifying the high-resistance ground fault of the power distribution network realizes feature extraction and feature screening by wavelet transformation and according to the sampling frequency of zero-sequence current data and combining with the number of target decomposition layers determined by the corresponding relation between the preset sampling frequency and the number of the target decomposition layers, the screened fault features are classified by a semi-supervised learning algorithm classifier, the high-resistance ground fault and a common disturbance event can be more reliably distinguished under the condition of a small amount of labeled training samples, and certain anti-noise capability is realized, so that the effective identification for the high-resistance ground fault is realized, and the technical problem of high-resistance ground fault identification difficulty in the prior art is solved.
Description
Technical Field
The application relates to the technical field of power distribution network fault identification, in particular to a method, a device, a terminal and a medium for identifying a high-resistance grounding fault of a power distribution network.
Background
With the development of urban power distribution networks, distribution lines become increasingly complex and the probability of faults is increased. High impedance ground fault (HIF) is a single-phase ground fault which is difficult to detect in a power distribution network, and due to large resistance of a transition resistor at the initial stage of the fault, the fault characteristic is weak, and the starting threshold of a traditional microcomputer protection device cannot be triggered. Although the low fault current of the high-resistance earth fault is less harmful to equipment, the fault increases the risk of mountain forest fire if existing for a long time and threatens the personal safety of residents.
The single-phase earth fault detection method can be generally divided into a transient state method, a steady state method and an artificial intelligence method. The existing single-phase ground fault detection method mainly aims at low-resistance ground faults, and the method is generally limited due to the characteristics of high resistance state, nonlinear distortion and the like of the high-resistance ground fault, so how to provide an effective method for identifying the high-resistance ground fault already becomes a technical problem to be solved urgently by technical personnel in the field.
Disclosure of Invention
The application provides a method, a device, a terminal and a medium for identifying a high-resistance ground fault of a power distribution network, which are used for solving the technical problem of high difficulty in identifying the high-resistance ground fault in the prior art.
The application provides a method for identifying a high-resistance grounding fault of a power distribution network in a first aspect, which comprises the following steps:
acquiring zero sequence current data of the power distribution network;
performing multi-layer decomposition on the zero-sequence current data in a wavelet transformation mode to obtain multiple groups of wavelet components;
extracting wavelet components corresponding to the target decomposition layer number as fault features according to the target decomposition layer number, wherein the target decomposition layer number is determined according to the sampling frequency of the zero-sequence current data and by combining the corresponding relation between the preset sampling frequency and the target decomposition layer number;
and inputting the fault characteristics into a high-resistance grounding fault recognition model to obtain a high-resistance grounding fault recognition result of the power distribution network through the operation of the high-resistance grounding fault recognition model, wherein the high-resistance grounding fault recognition model is a semi-supervised learning model obtained by training a preset fault characteristic sample.
Preferably, the performing multi-layer decomposition on the zero-sequence current data in a wavelet transform manner to obtain multiple groups of wavelet components further includes:
obtaining a wavelet threshold value according to the product of the maximum and minimum threshold values of the wavelet components of each group and the noise approximate variance;
and denoising each group of wavelet components by a threshold denoising method according to the wavelet threshold so as to extract fault features according to the denoised wavelet components.
Preferably, the threshold denoising method specifically includes: a hard threshold denoising method or a soft threshold denoising method.
Preferably, the denoising processing on each group of wavelet components by a threshold denoising method according to the wavelet threshold to obtain a plurality of groups of wavelet components further includes:
and identifying and counting peak and valley points in each group of wavelet components through a peak and valley identification algorithm, and executing subsequent steps when the number of the peak and valley points reaches a preset number threshold value.
Preferably, the high-resistance ground fault identification model is a semi-supervised random forest model obtained by training a preset fault characteristic sample.
Preferably, the obtaining process of the high-resistance ground fault identification model specifically includes:
acquiring a first sample set and a second sample set, wherein the first sample set and the second sample set are equal and both contain labeled fault feature samples;
respectively taking the first sample set and the second sample set as input quantities of two random forest classifiers, and carrying out preliminary training on the two random forest classifiers;
acquiring a third sample set and a fourth sample set, wherein the third sample set and the fourth sample set are equal and all contain unlabeled fault feature samples;
respectively taking the third sample set and the fourth sample set as input quantities of two preliminarily trained random forest classifiers, and performing identification and labeling through operation of the two random forest classifiers;
updating the sample sets of the two random forest classifiers by adding the labeling data output by one random forest classifier to the sample set of the other random forest classifier, and performing iterative training on the two random forest classifiers through the updated sample sets until all the unlabeled fault feature samples are identified and labeled.
Preferably, after the fault characteristics are input to a high-resistance ground fault identification model to obtain a high-resistance ground fault identification result of the power distribution network through operation of the high-resistance ground fault identification model, the method further includes:
and based on the two random forest classifiers in the high-resistance ground fault recognition model, when the high-resistance ground fault recognition results output by the two random forest classifiers are inconsistent, taking the high-resistance ground fault recognition result output by the random forest classifier with a high weight parameter as a final recognition result according to the weight parameters of the two random forest classifiers.
This application second aspect provides a distribution network high resistance ground fault recognition device, includes:
the zero sequence current acquisition unit is used for acquiring zero sequence current data of the power distribution network;
the wavelet transformation processing unit is used for carrying out multilayer decomposition on the zero sequence current data in a wavelet transformation mode to obtain a plurality of groups of wavelet components;
the fault feature extraction unit is used for extracting wavelet components corresponding to the target decomposition layer number as fault features according to the target decomposition layer number, wherein the target decomposition layer number is determined according to the sampling frequency of the zero-sequence current data and by combining the corresponding relation between the preset sampling frequency and the target decomposition layer number;
and the fault identification result obtaining unit is used for inputting the fault characteristics into a high-resistance ground fault identification model so as to obtain a high-resistance ground fault identification result of the power distribution network through the operation of the high-resistance ground fault identification model, wherein the high-resistance ground fault identification model is a semi-supervised learning model obtained by training a preset fault characteristic sample.
The third aspect of the present application provides a distribution network high resistance ground fault identification terminal, including: a memory and a processor;
the memory is used for storing program codes, and the program codes correspond to the identification method for the high-resistance grounding fault of the power distribution network provided by the first aspect of the application;
the processor is configured to execute the program code.
A fourth aspect of the present application provides a computer-readable storage medium, in which program codes corresponding to the method for identifying a high impedance-to-ground fault of a power distribution network according to the first aspect of the present application are stored.
According to the technical scheme, the embodiment of the application has the following advantages:
according to the method for identifying the high-resistance ground fault of the power distribution network, the number of target decomposition layers determined by combining the preset corresponding relation between the sampling frequency and the number of target decomposition layers is combined through wavelet transformation and according to the sampling frequency of zero-sequence current data, feature extraction and feature screening are achieved, the screened fault features are classified through a semi-supervised learning algorithm classifier, the high-resistance ground fault and a common disturbance event can be more reliably distinguished under the condition that a small number of labeled training samples exist, and certain anti-noise capacity is achieved, so that effective identification of the high-resistance ground fault is achieved, and the technical problem that the high-resistance ground fault identification difficulty is large in the prior art is solved.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without inventive exercise.
Fig. 1 is a schematic flowchart of a first embodiment of a method for identifying a high-resistance ground fault of a power distribution network according to the present application.
Fig. 2 is a schematic diagram of discrete wavelet transform.
Fig. 3 is a schematic flowchart of a second embodiment of a method for identifying a high impedance-to-ground fault of a power distribution network according to the present application.
Fig. 4 is an overall flow logic block diagram of a power distribution network high resistance ground fault identification method provided by the present application.
Fig. 5 is a schematic topology diagram of a 10kV resonant grounded power distribution network provided in the present application.
Fig. 6 is a schematic diagram illustrating comparison of trigger accuracy rates of fault feature extraction steps before and after denoising processing.
Fig. 7 is a waveform diagram illustrating marking of peaks and valleys based on a 10dB noise waveform.
Fig. 8 is a schematic structural diagram of an embodiment of a high-resistance ground fault identification device for a power distribution network according to the present application.
Detailed Description
The embodiment of the application provides a method, a device, a terminal and a medium for identifying a high-resistance ground fault of a power distribution network, and is used for solving the technical problem that the identification difficulty of the high-resistance ground fault is high in the prior art.
In order to make the objects, features and advantages of the present invention more apparent and understandable, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application, and it is apparent that the embodiments described below are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
Referring to fig. 1, a method for identifying a high-resistance ground fault of a power distribution network according to a first embodiment of the present application includes:
And 102, performing multi-layer decomposition on the zero-sequence current data in a wavelet transform mode to obtain multiple groups of wavelet components.
It should be noted that, based on the obtained zero sequence current data, the zero sequence current data is subjected to multi-layer decomposition in a wavelet transform processing manner to obtain multiple groups of wavelet components, and the number of decomposed layers is preferably 6.
According to the discrete wavelet transform diagram shown in fig. 2, the DWT passes through multiple layers of low-pass and high-pass filters to obtain multiple levels of approximate components and detail components of the original signal, respectively, as shown in the figure. In the figure, LF denotes a low-pass filter, HF denotes a high-pass filter, AiRepresenting the high-scale low-frequency component of the original signal, D, as an approximation componentiThe detail component represents the low-scale high-frequency information of the original signal.
And 103, extracting wavelet components corresponding to the target decomposition layer number as fault characteristics according to the target decomposition layer number.
The target decomposition layer number is determined according to the sampling frequency of the zero sequence current data and by combining the corresponding relation between the preset sampling frequency and the target decomposition layer number.
It should be noted that the high-impedance ground fault and the disturbance event have a relatively obvious characteristic difference between 0 and 312.5Hz, taking a sampling frequency of 10kHz as an example, under the condition that the sampling frequency is 10kHz, the detail components D5(156.25 to 312.5Hz) and D6(78.125 to 156.25Hz) of the 5 th layer and the 6 th layer are generally selected as fault characteristics.
And 104, inputting the fault characteristics into the high-resistance grounding fault identification model to obtain a high-resistance grounding fault identification result of the power distribution network through the operation of the high-resistance grounding fault identification model.
The high-resistance grounding fault identification model is a semi-supervised learning model obtained by training a preset fault characteristic sample.
It should be noted that, next, the fault features obtained in the previous step are input into a high-resistance ground fault identification model for identification, and a trained semi-supervised model classifier is used to classify the fault features to distinguish a high-resistance ground fault from a non-high-resistance ground fault, whereas the high-resistance ground fault identification model of the present embodiment is a semi-supervised learning model obtained by training a preset fault feature sample, and the fault feature sample is extracted from the zero-sequence current sample based on a preset zero-sequence current sample in a manner similar to steps 102 to 103 in the present embodiment.
The above content is a detailed description of a first embodiment of the power distribution network high-resistance ground fault identification method provided by the present application, and the power distribution network high-resistance ground fault identification method provided by this embodiment implements feature extraction and feature screening by wavelet transformation and according to the sampling frequency of zero-sequence current data, in combination with a target decomposition layer number determined by a preset corresponding relationship between the sampling frequency and the target decomposition layer number, and the screened fault features are classified by a semi-supervised learning algorithm classifier, so that a high-resistance ground fault and a common disturbance event can be more reliably distinguished under the condition of a small number of labeled training samples, and a certain anti-noise capability is provided, thereby implementing effective identification for a high-resistance ground fault and solving the technical problem of high-resistance ground fault identification difficulty in the prior art.
The following is a detailed description of a second embodiment of a method for identifying a high impedance-to-ground fault in a power distribution network, which is provided by the present application on the basis of the first embodiment.
Referring to fig. 3 and 4, a method for identifying a high impedance ground fault of a power distribution network according to a second embodiment of the present application includes:
further, step 102 in the first embodiment may be followed by the following steps:
1001, obtaining wavelet threshold values according to products of maximum and minimum threshold values of all groups of wavelet components and noise approximate variances;
and 1002, denoising each group of wavelet components by a threshold denoising method according to the wavelet threshold so as to extract fault features according to the denoised wavelet components.
It should be noted that, the high-resistance ground fault current is weak in amplitude and is easily interfered by noise. In order to further improve the reliability of fault identification, noise filtering can be performed before the fault features are extracted.
The equipment acquisition signal is composed of an original signal and a noise signal in a superposition mode, approximate components and detail components are obtained through wavelet transformation, each component is still composed of the original signal and the noise signal, most of noise is high-frequency components, and the noise is located in the detail components of the wavelet transformation. The modulus of the original signal is much larger than that of the noise signal, and the actual noise is compounded by a plurality of sources and can be regarded as the sum of variables of different independent probability distributions. According to the central limit theorem, the noise distribution will approach a gaussian distribution. Thus, according to the 3 σ i principle of Gaussian distribution, the noise will be distributed mainly within the interval of [ -3 σ i,3 σ i ]. In this embodiment, the data of each layer component in the interval [ -3 σ i,3 σ i ] is preferably processed by using the minimum maximum threshold, and the denoised component of each layer is reconstructed to obtain the denoised signal. Where σ i represents the approximate variance of the noise.
The above formula is a calculation formula of the maximum minimum threshold λ, and represents that when the sampling point N is greater than 32, the threshold is calculated by the above formula, otherwise, the threshold is 0.
σi＝median(sj(k))/0.6745(k＝0,1,...,N-1)
In the formula, sj(k) Data points representing a certain layer component, mean represents the median of the acquired data, and N represents the number of sampling points.
In the application of the wavelet threshold denoising algorithm, not only the distribution range of noise but also the influence of the variance σ i on the threshold need to be considered, where σ i of each layer component is multiplied by the minimum and maximum threshold λ of each layer to obtain a new threshold, i.e., the wavelet threshold λ' mentioned in this embodiment.
Further, the threshold denoising method specifically includes: a hard threshold denoising method or a soft threshold denoising method.
When a threshold denoising method is implemented for denoising based on the wavelet threshold obtained by calculation, the method that can be adopted includes: a hard threshold denoising method or a soft threshold denoising method, wherein the hard threshold denoising method obtains a threshold lambda ' according to calculation, a point with a modulus smaller than lambda ' in a wavelet component to be denoised is set to be zero, and a point larger than lambda ' is reserved; the soft threshold denoising method subtracts lambda ' from the point of the component with the modulus larger than lambda ', and zeros the point with the modulus smaller than lambda '. Compared with a hard threshold denoising method, the soft threshold denoising method can remove noise more smoothly and ensure the waveform restoring degree, so the soft threshold denoising method is preferably used for denoising.
Further, according to the wavelet threshold, denoising each group of wavelet components by a threshold denoising method, and after obtaining the plurality of groups of wavelet components, the method further comprises:
Further, the high-resistance grounding fault identification model is a semi-supervised random forest model obtained by training preset fault characteristic samples.
It should be noted that the high impedance ground fault identification model is a model obtained by training a preset fault characteristic sample through a semi-supervised random forest algorithm, and the training process may refer to one of the following examples:
(1) the samples to be classified and evaluated are saved as a sample library S, and the Gini index G (S) is calculated:
in the formula piIs the proportion of the sample labeled i in the sample library S, pi′Is the proportion of samples labeled non-i in the sample library S, and n represents the total number of labels in the sample library. Obviously, the smaller the G (S) is, the larger the proportion of the sample labeled i in the sample library is.
(2) The sample has K characteristic samples, and a certain characteristic is set as K, so that the sample labeled as i is divided according to the K characteristic samples
And calculating the kini index G (S, k) for feature k:
where | S | and
represents the corresponding number of samples, where | S | represents the number of all samples.
(3) And (3) randomly extracting m samples in a place-back mode, randomly selecting h features (h < < K), constructing branch of the decision tree by the feature with the minimum Gini index in the h features, and traversing the features to generate m decision trees.
(4) And (4) repeating the steps (2) to (3), and if the labels under the branches of the single decision tree are consistent or K characteristics are traversed, the single decision tree is successfully constructed. And when all the decision trees are constructed, establishing a random forest, and determining the labels of the input samples according to the classification voting results of all the decision trees in the forest.
Further, the obtaining process of the high-resistance ground fault identification model specifically includes:
acquiring a first sample set and a second sample set, wherein the first sample set and the second sample set are equal and both contain labeled fault feature samples;
respectively taking the first sample set and the second sample set as input quantities of two random forest classifiers, and carrying out preliminary training on the two random forest classifiers;
acquiring a third sample set and a fourth sample set, wherein the third sample set and the fourth sample set are equal and all contain unlabeled fault feature samples;
respectively taking the third sample set and the fourth sample set as input quantities of two random forest classifiers after preliminary training, and carrying out identification and labeling through the operation of the two random forest classifiers;
updating the sample sets of the two random forest classifiers by adding the labeling data output by one random forest classifier to the sample set of the other random forest classifier, and performing iterative training on the two random forest classifiers through the updated sample sets until all the unlabeled fault feature samples are identified and labeled.
It should be noted that, supervised learning requires a large amount of labeled data to train the classifier. However, a large amount of data collected by the power distribution network lacks clear labels, and only the labels can be observed manually, which consumes a large amount of manpower and material resources and is not favorable for fully utilizing the collected label-free samples. In order to effectively utilize unlabeled samples, the present embodiment adopts a collaborative training scheme, and combines with a semi-supervised learning mechanism, so as to improve training efficiency and further ensure recognition accuracy, the collaborative training mode provided by the present embodiment forms a semi-supervised classifier by 2 Random Forest classifiers (Random Forest 1 and Random Forest 2, hereinafter referred to as RF1 and RF2), and the specific steps are as follows:
1) processing the tagged zero sequence current data by wavelet threshold denoising, extracting related wavelet coefficients for training random forests, constructing two equal tag sample sets X1 and X2, and respectively training two different random forest classifiers;
2) labeling unlabelled samples in the unlabelled sample set Y by using a trained random forest classifier, adding the labeled samples serving as new samples into an original training set to construct a new training set, namely labeling elements in Y by using RF1 and adding the labeled data to X2 serving as a new training set X'2Elements in Y are labeled with RF2 and labeled data is added to X1 as a new training set X'1。
3) And training the two random forest classifiers again by using the new training set, and repeating the operation until all the unlabeled samples are labeled with the same label by the two random forests, finishing learning the unlabeled data, finishing the training of the two random forest classifiers, and obtaining a final high-resistance grounding fault recognition model.
Further, the method for identifying the high-resistance ground fault of the power distribution network further comprises the following steps of inputting the fault characteristics into the high-resistance ground fault identification model, and obtaining the high-resistance ground fault identification result of the power distribution network through the operation of the high-resistance ground fault identification model:
based on two random forest classifiers in the high-resistance ground fault recognition model, when the high-resistance ground fault recognition results output by the two random forest classifiers are inconsistent, the high-resistance ground fault recognition result output by the random forest classifier with a high weight parameter is taken as a final recognition result according to the weight parameters of the two random forest classifiers.
It can be understood that after training is completed, the two random forest classifiers can obtain a determined weight, and when the classification results of the two random forests are different, the classification result with the larger weight can be preferentially selected as the final result.
In order to verify the reliability of the algorithm provided by the embodiment, a resonant grounded power distribution network is established based on PSCAD/EMTDC simulation software, as shown in fig. 5, the sampling rate is 10kHz, the compensation degree is 8%, and the equivalent arc suppression coil inductance value is 0.4223H. And carrying out grounding arc modeling by using a parallel Emanuel model.
The zero sequence current is used as a fault data set, noise is added to form a noise data set with the signal-to-noise ratio of 3dB to 30dB, then the noise is filtered through a wavelet threshold denoising algorithm to form a denoising data set, the noise data set and the denoising data set are tested through a peak and trough starting algorithm, and the test result is shown in fig. 6. When the signal-to-noise ratio is reduced to be below 15dB, the starting accuracy of the noise data set is obviously reduced, the de-noising data set keeps higher accuracy, and therefore the accuracy of the starting algorithm can be effectively improved by the wavelet threshold de-noising algorithm.
For the example of a 10dB noise waveform, the partially marked peak-to-valley waveform is shown in FIG. 7. Wherein (a) - (c) in fig. 7 respectively represent zero sequence currents of the line when high-resistance ground fault, load switching and capacitor switching occur, and red dots in the graph represent peak and valley points detected by the algorithm. The original waveform, the noise waveform with the signal-to-noise ratio of 10dB and the waveform after the wavelet threshold denoising are longitudinally compared, and the starting algorithm can detect 12 peak valley points in 8 periods under three conditions and accords with the starting conditions. However, due to the influence of noise, the peak-valley point detected by the 10dB noise waveform is partially lost, and the waveform denoised by the wavelet threshold value does not have the undetected peak-valley point. Therefore, in order to improve the accuracy of the starting algorithm, the collected zero sequence current needs to be subjected to noise filtering by a wavelet threshold denoising algorithm, then whether a fault or disturbance is possible is judged by the wave crest and trough starting algorithm, and if the starting is successful, the subsequent fault identification algorithm is adopted for further judgment.
Meanwhile, in order to verify that the recognition algorithm provided by the embodiment can obtain higher accuracy and generalization capability than supervised learning under fewer label samples, a supervised algorithm is selected for comparison. The current was added to gaussian white noise to construct noise waveforms with signal to noise ratios of 7dB, 10dB, 20dB and 30dB, which were 1700 sets with the original waveform.
1) 238 groups of noise waveforms and corresponding 7dB, 10dB, 20dB and 30dB noise waveforms 952 groups are selected from the noise-free waveforms 340 group, 1190 group data is used as a training set, and the rest 510 groups are used as a test set. The algorithm provided by the invention deletes 595 groups of data labels in the training set as unlabeled data and trains the unlabeled data together with labeled data, while the training sets used by the other two methods are both labeled, and the classification test result is shown in table 1, and the method of the embodiment is higher than the traditional supervision algorithm.
2) 102 sets of noise waveforms and their corresponding sets of 7dB, 10dB, 20dB, and 30dB noise waveforms 416 are selected from the set of noise-free waveforms 340, and 510 sets of data are used as a training set, and the remaining set 1190 is used as a test set. The algorithm still uses half of the label data in the training set, but still keeps the accuracy rate close to that of the experiment (1), and the accuracy rate of other methods is obviously reduced along with the reduction of the label data in the training set;
therefore, the semi-supervised algorithm based on the collaborative training is superior to the compared supervised algorithm in fewer label training sets, because the collaborative training learns the non-label data through the two classifiers, a more accurate classification effect is obtained, but the learning difficulty of the classifiers is increased by using excessive non-label data training, and the accuracy is reduced.
TABLE 1 results of experimental comparison
The above content is a detailed description of a second embodiment of the method for identifying a high impedance ground fault of a power distribution network provided by the present application. The following is a description of an embodiment of a high impedance ground fault identification apparatus for a power distribution network according to the present application.
Referring to fig. 8, a third embodiment of the present application provides a device for identifying a high impedance ground fault of a power distribution network, including:
a zero-sequence current obtaining unit 201, configured to obtain zero-sequence current data of the power distribution network;
the wavelet transform processing unit 202 is configured to perform multi-layer decomposition on the zero-sequence current data in a wavelet transform manner to obtain multiple groups of wavelet components;
the fault feature extraction unit 203 is configured to extract wavelet components corresponding to the target decomposition layer number as fault features according to the target decomposition layer number, where the target decomposition layer number is determined according to the sampling frequency of the zero-sequence current data and by combining a preset corresponding relationship between the sampling frequency and the target decomposition layer number;
the fault identification result obtaining unit 204 is configured to input the fault characteristics to the high-resistance ground fault identification model, so as to obtain a high-resistance ground fault identification result of the power distribution network through operation of the high-resistance ground fault identification model, where the high-resistance ground fault identification model is a semi-supervised learning model obtained through training of preset fault characteristic samples.
In addition, this application fourth embodiment provides a distribution network high resistance ground fault identification terminal, includes: a memory and a processor;
the memory is used for storing program codes, and the program codes correspond to a high-resistance grounding fault identification method of the power distribution network provided by the first embodiment or the second embodiment of the application;
the processor is used for executing the program codes.
A fifth embodiment of the present application provides a computer-readable storage medium, in which program codes corresponding to a method for identifying a high impedance-to-ground fault in a power distribution network as provided in the first embodiment or the second embodiment of the present application are stored.
It can be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the terminal, the apparatus and the unit described above may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the several embodiments provided in the present application, it should be understood that the disclosed terminal, apparatus and method may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
The terms "first," "second," "third," "fourth," and the like in the description of the application and the above-described figures, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the application described herein are, for example, capable of operation in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The above embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions in the embodiments of the present application.
Claims (10)
1. A method for identifying a high-resistance grounding fault of a power distribution network is characterized by comprising the following steps:
acquiring zero sequence current data of the power distribution network;
performing multi-layer decomposition on the zero-sequence current data in a wavelet transformation mode to obtain multiple groups of wavelet components;
extracting wavelet components corresponding to the target decomposition layer number as fault features according to the target decomposition layer number, wherein the target decomposition layer number is determined according to the sampling frequency of the zero-sequence current data and by combining the corresponding relation between the preset sampling frequency and the target decomposition layer number;
and inputting the fault characteristics into a high-resistance grounding fault recognition model to obtain a high-resistance grounding fault recognition result of the power distribution network through the operation of the high-resistance grounding fault recognition model, wherein the high-resistance grounding fault recognition model is a semi-supervised learning model obtained by training a preset fault characteristic sample.
2. The method for identifying the high-resistance ground fault of the power distribution network according to claim 1, wherein the step of performing multi-layer decomposition on the zero-sequence current data in a wavelet transformation manner to obtain multiple groups of wavelet components further comprises:
obtaining a wavelet threshold value according to the product of the maximum and minimum threshold values of the wavelet components of each group and the noise approximate variance;
and denoising each group of wavelet components by a threshold denoising method according to the wavelet threshold so as to extract fault features according to the denoised wavelet components.
3. The method for identifying the high-resistance ground fault of the power distribution network according to claim 2, wherein the threshold denoising method specifically comprises the following steps: a hard threshold denoising method or a soft threshold denoising method.
4. The method for identifying the high-resistance ground fault of the power distribution network according to claim 2, wherein the denoising processing is performed on each group of wavelet components by a threshold denoising method according to the wavelet threshold, and after obtaining multiple groups of wavelet components, the method further comprises:
and identifying and counting peak and valley points in each group of wavelet components through a peak and valley identification algorithm, and executing subsequent steps when the number of the peak and valley points reaches a preset number threshold value.
5. The method for identifying the high-resistance ground fault of the power distribution network according to claim 1, wherein the high-resistance ground fault identification model is a semi-supervised random forest model obtained by training preset fault feature samples.
6. The method for identifying the high-resistance ground fault of the power distribution network according to claim 5, wherein the obtaining process of the high-resistance ground fault identification model specifically comprises the following steps:
acquiring a first sample set and a second sample set, wherein the first sample set and the second sample set are equal and both contain labeled fault feature samples;
respectively taking the first sample set and the second sample set as input quantities of two random forest classifiers, and carrying out preliminary training on the two random forest classifiers;
acquiring a third sample set and a fourth sample set, wherein the third sample set and the fourth sample set are equal and all contain unlabeled fault feature samples;
respectively taking the third sample set and the fourth sample set as input quantities of two preliminarily trained random forest classifiers, and performing identification and labeling through operation of the two random forest classifiers;
updating the sample sets of the two random forest classifiers by adding the labeling data output by one random forest classifier to the sample set of the other random forest classifier, and performing iterative training on the two random forest classifiers through the updated sample sets until all the unlabeled fault feature samples are identified and labeled.
7. The method for identifying the high-resistance ground fault of the power distribution network according to claim 6, wherein the step of inputting the fault characteristics into a high-resistance ground fault identification model to obtain a high-resistance ground fault identification result of the power distribution network through operation of the high-resistance ground fault identification model further comprises:
and based on the two random forest classifiers in the high-resistance ground fault recognition model, when the high-resistance ground fault recognition results output by the two random forest classifiers are inconsistent, taking the high-resistance ground fault recognition result output by the random forest classifier with a high weight parameter as a final recognition result according to the weight parameters of the two random forest classifiers.
8. The utility model provides a distribution network high resistance ground fault recognition device which characterized in that includes:
the zero sequence current acquisition unit is used for acquiring zero sequence current data of the power distribution network;
the wavelet transformation processing unit is used for carrying out multilayer decomposition on the zero sequence current data in a wavelet transformation mode to obtain a plurality of groups of wavelet components;
the fault feature extraction unit is used for extracting wavelet components corresponding to the target decomposition layer number as fault features according to the target decomposition layer number, wherein the target decomposition layer number is determined according to the sampling frequency of the zero-sequence current data and by combining the corresponding relation between the preset sampling frequency and the target decomposition layer number;
and the fault identification result obtaining unit is used for inputting the fault characteristics into a high-resistance ground fault identification model so as to obtain a high-resistance ground fault identification result of the power distribution network through the operation of the high-resistance ground fault identification model, wherein the high-resistance ground fault identification model is a semi-supervised learning model obtained by training a preset fault characteristic sample.
9. The utility model provides a distribution network high resistance ground fault discernment terminal which characterized in that includes: a memory and a processor;
the memory is used for storing program codes, and the program codes correspond to the identification method of the high-resistance earth fault of the power distribution network according to any one of claims 1 to 7;
the processor is configured to execute the program code.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored therein program code corresponding to a method for identifying a high impedance-to-ground fault of a power distribution network according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111217367.5A CN114089218A (en) | 2021-10-19 | 2021-10-19 | Power distribution network high-resistance grounding fault identification method, device, terminal and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111217367.5A CN114089218A (en) | 2021-10-19 | 2021-10-19 | Power distribution network high-resistance grounding fault identification method, device, terminal and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114089218A true CN114089218A (en) | 2022-02-25 |
Family
ID=80297224
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111217367.5A Pending CN114089218A (en) | 2021-10-19 | 2021-10-19 | Power distribution network high-resistance grounding fault identification method, device, terminal and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114089218A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114742295A (en) * | 2022-04-11 | 2022-07-12 | 四川大学 | Equipment power utilization fault prediction method based on wavelet transform technology |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109613402A (en) * | 2019-02-14 | 2019-04-12 | 福州大学 | Power distribution network high resistance earthing fault detection method based on wavelet transformation and neural network |
| CN111170103A (en) * | 2019-12-27 | 2020-05-19 | 猫岐智能科技（上海）有限公司 | Equipment fault identification method |
| CN111487508A (en) * | 2020-06-01 | 2020-08-04 | 南方电网调峰调频发电有限公司西部检修试验分公司 | Single-phase high-resistance grounding fault identification device for power distribution network |
| CN111796166A (en) * | 2020-08-27 | 2020-10-20 | 广东电网有限责任公司电力调度控制中心 | Power distribution network single-phase high-resistance earth fault line selection method, system and equipment |
| CN111833275A (en) * | 2020-07-20 | 2020-10-27 | 山东师范大学 | Image denoising method based on low-rank analysis |
| CN112016473A (en) * | 2020-08-31 | 2020-12-01 | 国网安徽省电力有限公司电力科学研究院 | Power distribution network high-resistance grounding fault diagnosis method based on semi-supervised learning and attention mechanism |
| CN112014773A (en) * | 2020-09-04 | 2020-12-01 | 内蒙古电力（集团）有限责任公司呼和浩特供电局 | Method for detecting early fault of small current grounding system cable |
| CN112748359A (en) * | 2019-10-30 | 2021-05-04 | 中国电力科学研究院有限公司 | Power distribution network ground fault identification method and system based on random forest |
| CN113447766A (en) * | 2021-08-17 | 2021-09-28 | 广东电网有限责任公司东莞供电局 | Method, device, equipment and storage medium for detecting high-resistance ground fault |
| WO2021190377A1 (en) * | 2020-03-27 | 2021-09-30 | 华为技术有限公司 | Ppg signal obtaining method and apparatus, terminal device, and storage medium |
-
2021
- 2021-10-19 CN CN202111217367.5A patent/CN114089218A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109613402A (en) * | 2019-02-14 | 2019-04-12 | 福州大学 | Power distribution network high resistance earthing fault detection method based on wavelet transformation and neural network |
| CN112748359A (en) * | 2019-10-30 | 2021-05-04 | 中国电力科学研究院有限公司 | Power distribution network ground fault identification method and system based on random forest |
| CN111170103A (en) * | 2019-12-27 | 2020-05-19 | 猫岐智能科技（上海）有限公司 | Equipment fault identification method |
| WO2021190377A1 (en) * | 2020-03-27 | 2021-09-30 | 华为技术有限公司 | Ppg signal obtaining method and apparatus, terminal device, and storage medium |
| CN111487508A (en) * | 2020-06-01 | 2020-08-04 | 南方电网调峰调频发电有限公司西部检修试验分公司 | Single-phase high-resistance grounding fault identification device for power distribution network |
| CN111833275A (en) * | 2020-07-20 | 2020-10-27 | 山东师范大学 | Image denoising method based on low-rank analysis |
| CN111796166A (en) * | 2020-08-27 | 2020-10-20 | 广东电网有限责任公司电力调度控制中心 | Power distribution network single-phase high-resistance earth fault line selection method, system and equipment |
| CN112016473A (en) * | 2020-08-31 | 2020-12-01 | 国网安徽省电力有限公司电力科学研究院 | Power distribution network high-resistance grounding fault diagnosis method based on semi-supervised learning and attention mechanism |
| CN112014773A (en) * | 2020-09-04 | 2020-12-01 | 内蒙古电力（集团）有限责任公司呼和浩特供电局 | Method for detecting early fault of small current grounding system cable |
| CN113447766A (en) * | 2021-08-17 | 2021-09-28 | 广东电网有限责任公司东莞供电局 | Method, device, equipment and storage medium for detecting high-resistance ground fault |
Non-Patent Citations (4)
| Title |
|---|
| BAQUI I ET AL.: "High impedance fault detection methodology using wavelet transform and artificial neural Networks", ELECTRIC POWER SYSTEM ＲESEARCH, vol. 81, no. 7, pages 1236 * |
| 何仁君 等: "小波法识别配电网线路高阻故障", 电气开关, vol. 56, no. 02, pages 25 - 28 * |
| 白浩 等: "人工智能在配电网高阻接地故障检测中的应用及展望", 南方电网技术, vol. 13, no. 2, pages 34 * |
| 赵海青 等: "基于小波变换的非对称高阻接地保护启动元件的研究", 福建电力与电工, vol. 23, no. 3, pages 2 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114742295A (en) * | 2022-04-11 | 2022-07-12 | 四川大学 | Equipment power utilization fault prediction method based on wavelet transform technology |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Jin et al. | A novel adaptive EEMD method for switchgear partial discharge signal denoising | |
| WO2021068454A1 (en) | Method for identifying energy of micro-energy device on basis of bp neural network | |
| CN108008244A (en) | A kind of small current grounding fault progressive classifying identification method at many levels | |
| CN111460956B (en) | Unbalanced electrocardiogram sample classification method based on data enhancement and loss weighting | |
| Al-Salman et al. | Detection of EEG K-complexes using fractal dimension of time frequency images technique coupled with undirected graph features | |
| Si et al. | A novel adaptive wavelet threshold estimation based on hybrid particle swarm optimization for partial discharge signal denoising | |
| Robertson et al. | Wavelets and power system transients: feature detection and classification | |
| CN114089218A (en) | Power distribution network high-resistance grounding fault identification method, device, terminal and medium | |
| CN113447766A (en) | Method, device, equipment and storage medium for detecting high-resistance ground fault | |
| CN112486137A (en) | Method and system for constructing fault feature library of active power distribution network and fault diagnosis method | |
| CN116838955A (en) | Two-stage oil and gas pipeline line interference identification method | |
| Asman et al. | Identification of transient overvoltage using discrete wavelet transform with minimised border distortion effect and support vector machine | |
| CN114355100A (en) | Power distribution network single-phase arc light earth fault line selection method, system, equipment and medium | |
| CN111898446A (en) | Single-phase earth fault studying and judging method based on multi-algorithm normalization analysis | |
| Geethanjali | Combined wavelet transfoms and neural network (WNN) based fault detection and classification in transmission lines | |
| Yu et al. | GAN-based semi-supervised learning method for identification of the faulty feeder in resonant grounding distribution networks | |
| Guo et al. | High impedance fault semi-supervised detection of distribution networks based on tri-training and support vector machine | |
| CN113533904B (en) | Method, device, equipment and medium for detecting high-resistance ground fault of power distribution network | |
| CN115936196A (en) | Monthly rainfall model prediction method based on time sequence convolution network | |
| CN116318925A (en) | Multi-CNN fusion intrusion detection method, system, medium, equipment and terminal | |
| Sima et al. | A framework for automatically cleansing overvoltage data measured from transmission and distribution systems | |
| CN112213561B (en) | Measurement data preprocessing method and device for leading load parameter noise identification | |
| CN114755529A (en) | Multi-feature fusion single-phase earth fault type identification method based on deep learning | |
| CN114415056A (en) | Distributed power supply island fault detection method | |
| Chen et al. | Power system transient modelling and classification |
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 |