CN113092933A - LSTM-based single-phase earth fault line selection method and system - Google Patents

LSTM-based single-phase earth fault line selection method and system Download PDF

Info

Publication number
CN113092933A
CN113092933A CN202110307110.2A CN202110307110A CN113092933A CN 113092933 A CN113092933 A CN 113092933A CN 202110307110 A CN202110307110 A CN 202110307110A CN 113092933 A CN113092933 A CN 113092933A
Authority
CN
China
Prior art keywords
lstm
line selection
fault line
earth fault
module
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
Application number
CN202110307110.2A
Other languages
Chinese (zh)
Inventor
邓长虹
刘正谊
应花梅
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Wuhan University WHU
Original Assignee
Wuhan University WHU
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Wuhan University WHU filed Critical Wuhan University WHU
Priority to CN202110307110.2A priority Critical patent/CN113092933A/en
Publication of CN113092933A publication Critical patent/CN113092933A/en
Pending legal-status Critical Current

Links

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
    • G01R31/08—Locating faults in cables, transmission lines, or networks
    • G01R31/081—Locating faults in cables, transmission lines, or networks according to type of conductors
    • G01R31/086—Locating faults in cables, transmission lines, or networks according to type of conductors in power transmission or distribution networks, i.e. with interconnected conductors
    • 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
    • G01R31/08—Locating faults in cables, transmission lines, or networks
    • G01R31/088—Aspects of digital computing
    • 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
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Computing Systems (AREA)
  • Computational Linguistics (AREA)
  • Biomedical Technology (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Biophysics (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

The invention discloses a single-phase earth fault line selection method and a single-phase earth fault line selection system based on LSTM, which are used for data acquisition and data processing, wherein the method comprises the steps of acquiring a plurality of conventional electrical quantities monitored by a power distribution network based on a scheduling platform, and performing data cleaning operation to obtain original characteristics; selecting characteristic quantities, including constructing a plurality of new characteristics according to the collected electrical quantities, obtaining the importance degree of each characteristic from the original characteristics and the new characteristics by adopting a plurality of characteristic quantity extraction algorithms, and selecting the characteristic quantities according to the importance degree; establishing an LSTM-based single-phase earth fault line selection model, which comprises an input layer, three hidden layers and an output layer, wherein the input layer is determined by characteristic quantity, the first two hidden layers are LSTM modules, the third hidden layer is a Dense module, and the output layer is a Dense module; and training the LSTM network and testing, wherein the method comprises the steps of training a single-phase earth fault line selection model based on the LSTM by using the characteristic quantity of sample data, and then obtaining an earth fault judgment result by using the trained single-phase earth fault line selection model.

Description

LSTM-based single-phase earth fault line selection method and system
Technical Field
The invention belongs to the field of single-phase earth fault line selection of medium and low voltage power distribution networks, and particularly relates to a single-phase earth fault line selection method and system based on LSTM.
Background
Most of 3-66 kV power distribution networks in China are low-current grounding systems with neutral points not grounded or grounded through arc suppression coils. Statistical data show that in a medium and low voltage distribution system, the proportion of system faults caused by single-phase earth faults reaches 80%, the safe operation of the system is influenced, and equipment is damaged. In practical application, the line is selected by adopting a mode of combining a fault line selection device and a manual pull line, when the manual pull line is adopted, unnecessary short-time power failure of a non-fault line is caused, the power supply reliability is reduced, and the economic benefits of a power supply department and a user are influenced. Therefore, when single-phase grounding occurs, the fault line can be judged quickly and accurately, and then the fault line is processed to avoid the expansion of accidents.
When a single-phase earth fault occurs, the amplitude of the fault phase zero-sequence current is equal to the sum of the amplitudes of the non-fault phases, and the phase angles of the fault phase zero-sequence current and the non-fault phases are opposite. Therefore, amplitude and phase angle of zero sequence current are mostly adopted as characteristic quantities in the aspect of single-phase earth fault line selection, and then wavelet packet decomposition or neural network is utilized to carry out fault line selection.
With the continuous development of artificial intelligence algorithms, artificial intelligence techniques are used in various industries. The artificial intelligence technology is applied to the single-phase earth fault line selection technology, so that automation and intelligence of the power distribution network are facilitated, the accuracy of single-phase earth fault line selection is improved, and high-efficiency operation of the power distribution network is promoted.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides a single-phase earth fault line selection method and system based on LSTM. And conventional electrical quantities are collected based on the dispatching platform, and the electrical quantities are processed to form characteristic quantities. And establishing a fault line selection model by adopting the LSTM to realize single-phase earth fault line selection.
The invention adopts the technical scheme that a single-phase earth fault line selection method based on LSTM comprises the following steps:
step S1, data acquisition and data processing, including acquiring a plurality of conventional electrical quantities monitored by the power distribution network based on the dispatching platform, and performing data cleaning operation to obtain original characteristics;
s2, selecting characteristic quantities, including constructing a plurality of new characteristics according to the electric quantities acquired in the step 1, obtaining the importance degree of each characteristic from the original characteristics and the new characteristics by adopting a plurality of characteristic quantity extraction algorithms, and selecting the characteristic quantities according to the importance degrees; step S3, establishing a single-phase earth fault line selection model based on LSTM as follows,
the LSTM-based single-phase earth fault line selection model comprises 5 layers in total, wherein the input layer, the three hidden layers and the output layer are determined by characteristic quantity, the first two hidden layers are LSTM modules, the third hidden layer is a Dense module, the output layer is a Dense module, and a result for judging whether a line is in fault or not is output;
and S4, training and testing the LSTM network, including training the LSTM-based single-phase earth fault line selection model by using the characteristic quantity of the sample data, and then obtaining an earth fault judgment result by using the trained single-phase earth fault line selection model.
In addition, in step S1, based on the electric quantity collected by the dispatching platform including the current, active and reactive data before and after the single-phase fault, the method of lai-da is adopted to identify and reject abnormal data for each electric quantity characteristic cycle.
Furthermore, the step S2 is realized as follows,
the new characteristics constructed include the current difference Δ I, the active difference Δ P, the reactive difference Δ Q, and the voltage variation amplitude Δ U' before and after the fault, as calculated below,
ΔI＝Iaf-Ibf
ΔP＝Paf-Pbf
ΔQ＝Qaf-Qbf
Figure BDA0002987965750000021
in the formula, the subscript Iaf、Paf、QafIndicating current, active and reactive power before line fault, Ibf、Pbf、QbfRepresenting the current, active power and reactive power after the line fault;
calculating the importance degree of each feature by adopting a Person correlation coefficient, recursive feature elimination and Person correlation coefficient methods for the constructed features and the original features, and selecting feature quantities; the finally selected characteristic quantities are a current difference value delta I, a voltage change amplitude delta U', a reactive power Q _ bf before grounding, an active power difference value delta P and a reactive power difference value delta Q.
In S3, a module in the Keras library is used to establish a single-phase ground fault line selection model, where the first and second hidden layers both use LSTM modules, the activation function is "relu", the third hidden layer and the output layer use density modules, the third hidden layer activation function is "tanh", and the output layer activation function is "sigmoid".
In S4, the characteristic amount data of each sample is subjected to z-score normalization processing, and then input to the LSTM-based single-phase ground fault line selection model.
And, for use in power distribution networks containing distributed power sources.
Moreover, the distributed power sources in the power distribution network are photovoltaic and/or energy storage power sources.
In another aspect, the present invention provides an LSTM-based single-phase ground fault line selection system, which is used to implement the LSTM-based single-phase ground fault line selection method.
And, including the following modules,
the first module is used for data acquisition and data processing, and comprises a dispatching platform, a first module and a second module, wherein the dispatching platform is used for acquiring a plurality of conventional electric quantities monitored by the power distribution network and performing data cleaning operation to obtain original characteristics;
the second module is used for selecting the characteristic quantity, constructing a plurality of new characteristics according to the electric quantity acquired by the first module, solving the importance degree of each characteristic by adopting a plurality of characteristic quantity extraction algorithms from the original characteristics and the new characteristics, and selecting the characteristic quantity according to the importance degree;
a third module for establishing an LSTM-based single-phase ground fault line selection model as follows,
the LSTM-based single-phase earth fault line selection model comprises 5 layers in total, wherein the input layer, the three hidden layers and the output layer are determined by characteristic quantity, the first two hidden layers are LSTM modules, the third hidden layer is a Dense module, the output layer is a Dense module, and a result for judging whether a line is in fault or not is output;
and the fourth module is used for training and testing the LSTM network, and comprises the steps of training a single-phase earth fault line selection model based on the LSTM by using the characteristic quantity of the sample data and then obtaining an earth fault judgment result by using the trained single-phase earth fault line selection model.
Further, a processor and a memory are included, the memory for storing program instructions, the processor for invoking the stored instructions in the memory to perform a LSTM based single phase ground fault line selection method as described above.
According to the invention, conventional electrical quantities are collected by the dispatching platform, the acquisition difficulty is low, the LSTM is used for describing the nonlinear relation between input and output, the operation is fast, the single-phase earth fault line can be effectively identified, and the reliability of the power distribution network system is improved.
The method is suitable for the power distribution network of the distributed power supply, is simple and convenient to realize, is real-time and efficient, and can support automatic and intelligent solution of line faults.
Drawings
FIG. 1 is a flow chart of a single-phase earth fault line selection method according to an embodiment of the present invention;
fig. 2 is a flow chart of feature quantity selection in the single-phase earth fault line selection method according to the embodiment of the invention;
fig. 3 is a schematic diagram of an LSTM structure in the single-phase ground fault line selection method according to the embodiment of the present invention.
Detailed Description
The present invention is further described with reference to the following drawings and examples, which are only used for more clearly illustrating the technical solutions of the present invention, and the protection scope of the present invention is not limited thereby.
The method comprises the steps of acquiring data before and after a single-phase earth fault based on a scheduling platform, analyzing the data, and mainly cleaning and standardizing the data; then, some new features are constructed according to the original features, and feature quantities are selected by fusing a plurality of feature selection methods, so that overfitting of the model caused by excessive input quantity is avoided; and then establishing an LSTM (Long Short-term memory, LSTM) model to realize single-phase earth fault line selection.
The embodiment of the invention relates to a single-phase earth fault line selection method based on LSTM, which comprises the following specific steps:
s1, data acquisition and data processing: and acquiring some conventional electrical quantities monitored by the power distribution network based on the scheduling platform to form an original data set. And performing data cleaning operation on the data set.
The photovoltaic and energy storage distributed power supply system is suitable for a power distribution network containing photovoltaic and energy storage distributed power supplies. Firstly, data are collected according to a dispatching platform, wherein the data comprise current, active data and reactive data before and after single-phase faults, and the current, active data and reactive data serve as original data sets.
The original data set may contain abnormal data such as missing values and outliers, and needs to be cleaned along with the data. The collected data are from the outgoing line connected with any bus in the whole area and may be grounding data at any time, so that no time correlation exists among samples, and the samples with abnormal data are directly removed for simplifying the processing. The invention adopts Lauda rule to identify outliers, and the formula of the Lauda rule is as follows: yci＝|xi-x|＞3σ
In the formula, yciRepresenting abnormal data, xiDenotes the ith sample, x denotes the mean of the feature, and σ denotes the standard deviation of the feature.
In specific implementation, after data cleaning is performed on a certain original electrical quantity characteristic by adopting the Lauda rule, data cleaning is performed on the processed data aiming at another characteristic until all the characteristics are operated.
S2, selecting characteristic quantity: and constructing some new characteristics according to the collected electrical quantity, and selecting the characteristic quantity by adopting a Person correlation coefficient, recursive characteristic elimination and other characteristic quantity extraction algorithms in combination with all the characteristics.
And only active power, reactive power and current before and after the fault are acquired according to the data acquired by the dispatching platform. According to the existing feature extraction algorithm, the importance of the original features obtained by corresponding extraction on target variables is found to be low, so that the invention constructs some features based on the original features, namely a current difference value (delta I), an active difference value (delta P), a reactive difference value (delta Q) and a voltage change amplitude (delta U'), and the specific expressions are as follows:
ΔI＝Iaf-Ibf
ΔP＝Paf-Pbf
ΔQ＝Qaf-Qbf
Figure BDA0002987965750000041
in the formula, the subscript Iaf、Paf、QafIndicating current, active and reactive power before line fault, Ibf、Pbf、QbfRepresenting the current, active power and reactive power after a line fault.
The invention adopts a method of random forest, Person correlation coefficient and recursive characteristic elimination to select characteristic quantity. And then the final characteristic quantity is selected by integrating the results of the three characteristic selection algorithms. The results are shown in the following table:
Figure BDA0002987965750000042
Figure BDA0002987965750000051
combining the results of the three algorithms, the selected characteristic quantity is as follows: current difference (Δ I), active difference (Δ P), reactive difference (Δ Q) and reactive power Q before line faultaf. And substituting the sample into the model, and finding that the added voltage variation amplitude is beneficial to improving the accuracy of fault line selection, so that the finally selected characteristic quantities are a current difference value, an active difference value, a reactive difference value, reactive before grounding and the voltage variation amplitude.
Referring to fig. 2, features are selected herein in conjunction with a random forest, recursive feature elimination, and person correlation coefficient approach. The same weight is set for the result of each method, and the first 4 characteristics are selected as characteristic quantities. The feature quantity and the original feature quantity obtained by the feature extraction algorithm are substituted into the same fault line selection model, the obtained fault line selection accuracy is shown in the following table, the accuracy obtained by adopting the original feature after data cleaning is 0.818, and the accuracy obtained by adopting the feature extraction algorithm is 0.954. The selected characteristic quantity is effective. Most of researches in recent years adopt zero sequence components for research, and data is derived from data obtained by modeling simulation and has a gap with actual fault data. The power distribution network is also continuously improved, all lines are not provided with zero sequence acquisition transposes, and the method is used for carrying out data based on some data which can be acquired by the power distribution network and is more suitable for actual working conditions.
Figure BDA0002987965750000052
S3, establishing an LSTM model: and establishing an LSTM-based single-phase earth fault line selection model by utilizing the Keras encapsulated module.
And constructing an LSTM model by using a Keras library, wherein the LSTM model comprises an input layer, a hidden layer and an output layer. The hidden layer comprises three layers, the first two layers are LSTM (long short term memory) modules, the third layer is a Dense (full connection layer) module, the activation function of the LSTM module is 'relu (linear rectification function)', the activation function of the Dense module is 'tanh (hyperbolic tangent function)', the number of neurons is 100, 100 and 200 respectively, wherein the return _ sequences (whether the hidden state is output or not) of the first hidden layer is equal to True, and the return _ sequences of the second hidden layer is False. Adding a Dropout block prevents overfitting, with the drop rate set to 0.1. The output layer is a density module, and since this is a binary problem, the activation function is "sigmoid (S-type function)". In particular, the number of iterations may be set to 150, and a binary loss function "binary _ cross _ may be selected to compile the model internally.
The design of the model structure can avoid overfitting to a certain extent and accelerate the running speed. The setting of each parameter is determined according to the calculation example, and the accuracy of the LSTM fault line selection model is guaranteed. The hidden layer is set to two layers to better fit the nonlinear relation of input and output.
FIG. 3 is a LSTM fault line selection model established by a keras-based encapsulation module, wherein x1～x5Representing the selected characteristic quantities, namely a current difference value, an active difference value, a reactive before grounding and a voltage change amplitude respectively; the number in each block represents the number of neurons for that block. The arrow between two modules represents the calculation of one module as input for the next module. Y represents the final output and represents the result of determining whether the line is faulty. Characteristic quantity x of input layer1、x2、x3、x4、x5The outputs of the LSTM module 11, the LSTM module 12, the LSTM module 13, the LSTM module 14, and the LSTM module 15 respectively input into the hidden layer 1 are respectively connected to the LSTM module 21, the LSTM module 22, the LSTM module 23, the LSTM module 24, and the LSTM module 25 of the hidden layer 2, the output of the LSTM module 21 is connected to the LSTM module 22, the output of the LSTM module 22 is connected to the LSTM module 23, the output of the LSTM module 23 is connected to the LSTM module 24, the output of the LSTM module 24 is connected to the LSTM module 25, the output of the LSTM module 25 is connected to the density module 31 in the hidden layer 3, the output of the density module 31 is connected to the Dropout layer, the output of the Dropout layer is connected to the density module 41 in the output layer, and the output of the density module 41 in the output layer is the final output Y.
S4, data set division and normalization processing: the method comprises the steps of training an LSTM-based single-phase earth fault line selection model by utilizing the characteristic quantity of sample data, and then obtaining an earth fault judgment result by using the trained single-phase earth fault line selection model.
In the embodiment, for the sake of convenience of verification, the data set formed by the feature quantities of each sample obtained in step S1 is first divided into a training set and a test set, and then 20% of the data in the training set is divided into a verification set. The data was z-score normalized.
All sample data are divided into a training set and a testing set according to a certain proportion, and the proportion of the grounding samples in the two data sets is the same. Since the number of samples is small in this embodiment, the ratio of the two is 9: 1. and when data processing is carried out, the training set and the verification set are uniformly regarded as the training set, and the mean value mu and the standard deviation sigma of the training set are solved. The training set is z-score normalized and the test set is processed using the μ and σ of the training set.
Figure BDA0002987965750000061
Figure BDA0002987965750000062
Where μ and σ refer to the mean and standard deviation, x, of the test set data prior to processingtrainAnd xtestInput data for the training and test sets before processing, xnew_trainAnd xnew_testInput data for the processed training set and test set.
The network is then trained and tested: and (5) training the network by substituting the processed training set data into the model, and testing the effect of the model by substituting the test set data into the trained network. In practical application, according to the data collected in real time, the feature quantity is extracted in the same manner as the steps S1-S2, and then the trained model can be input to obtain the fault judgment result.
The embodiment sets 20% of the training set as the verification set, and substitutes the rest samples into the model to train the network, and in the process of training the model, the effect of the model is verified by the data in the verification set. After the model training is finished, the data of the test set is substituted into the network, due to the sigmoid (S-shaped function) function used by the output layer, within 0-1 of the output value, according to the characteristics of the function, the fault line selection result of the test set can be obtained by considering the output value smaller than 0.5 as 0 and the output value greater than or equal to 0.5 as 1.
The selection and evaluation indexes are as follows,
Figure BDA0002987965750000071
Figure BDA0002987965750000072
wherein accuracy represents precision, yiThe actual value is represented by a value that is,
Figure BDA0002987965750000073
indicating the predicted value. N is a radical ofiDenotes the ith sample, m denotes the total number of samples, and i denotes the sample number.
According to the evaluation index, the accuracy of the embodiment can reach 95.4 percent through calculation.
The embodiment provides an LSTM-based single-phase earth fault line selection method which can be used for fault line selection after a single-phase earth fault occurs. The data source and the scheduling platform and the data of the actual operation working condition enable the result of the method to be more persuasive. And constructing new features on the basis of the original features, and combining the results of various algorithms to select feature quantities. The LSTM algorithm constructs a nonlinear relation between input and output, and according to results, the model has high accuracy in the aspect of single-phase fault line selection.
In specific implementation, a person skilled in the art can implement the automatic operation process by using a computer software technology, and a system device for implementing the method, such as a computer-readable storage medium storing a corresponding computer program according to the technical solution of the present invention and a computer device including a corresponding computer program for operating the computer program, should also be within the scope of the present invention.
In some possible embodiments, a single-phase ground fault line selection system based on LSTM is provided, which includes a first module configured to collect and process data, including collecting a plurality of conventional electrical quantities monitored by a distribution network based on a scheduling platform, and performing data cleaning operation to obtain original characteristics;
the second module is used for selecting the characteristic quantity, constructing a plurality of new characteristics according to the electric quantity acquired by the first module, solving the importance degree of each characteristic by adopting a plurality of characteristic quantity extraction algorithms from the original characteristics and the new characteristics, and selecting the characteristic quantity according to the importance degree;
a third module for establishing an LSTM-based single-phase ground fault line selection model as follows,
the LSTM-based single-phase earth fault line selection model comprises 5 layers in total, wherein the input layer, the three hidden layers and the output layer are determined by characteristic quantity, the first two hidden layers are LSTM modules, the third hidden layer is a Dense module, the output layer is a Dense module, and a result for judging whether a line is in fault or not is output;
and the fourth module is used for training and testing the LSTM network, and comprises the steps of training a single-phase earth fault line selection model based on the LSTM by using the characteristic quantity of the sample data and then obtaining an earth fault judgment result by using the trained single-phase earth fault line selection model.
In some possible embodiments, an LSTM-based single-phase ground fault line selection system is provided that includes a processor and a memory, the memory storing program instructions, the processor being configured to invoke the stored instructions in the memory to perform an LSTM-based single-phase ground fault line selection method as described above.
In some possible embodiments, an LSTM-based single-phase ground fault line selection system is provided, including a readable storage medium having stored thereon a computer program that, when executed, implements an LSTM-based single-phase ground fault line selection method as described above.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.

Claims (10)

1. A single-phase earth fault line selection method based on LSTM is characterized by comprising the following steps:
step S1, data acquisition and data processing, including acquiring a plurality of conventional electrical quantities monitored by the power distribution network based on the dispatching platform, and performing data cleaning operation to obtain original characteristics;
s2, selecting characteristic quantities, including constructing a plurality of new characteristics according to the electric quantities acquired in the step 1, obtaining the importance degree of each characteristic from the original characteristics and the new characteristics by adopting a plurality of characteristic quantity extraction algorithms, and selecting the characteristic quantities according to the importance degrees;
step S3, establishing a single-phase earth fault line selection model based on LSTM as follows,
the LSTM-based single-phase earth fault line selection model comprises 5 layers in total, wherein the input layer, the three hidden layers and the output layer are determined by characteristic quantity, the first two hidden layers are LSTM modules, the third hidden layer is a Dense module, the output layer is a Dense module, and a result for judging whether a line is in fault or not is output;
and S4, training and testing the LSTM network, including training the LSTM-based single-phase earth fault line selection model by using the characteristic quantity of the sample data, and then obtaining an earth fault judgment result by using the trained single-phase earth fault line selection model.
2. The LSTM-based single-phase ground fault line selection method of claim 1, wherein: in step S1, the electrical quantities collected based on the scheduling platform include current, active data and reactive data before and after a single-phase fault, and abnormal data are identified and removed for each electrical quantity characteristic cycle by using the laiida method.
3. The LSTM-based single-phase ground fault line selection method of claim 1, wherein: the implementation of step S2 is as follows,
the new characteristics constructed include the current difference Δ I, the active difference Δ P, the reactive difference Δ Q, and the voltage variation amplitude Δ U' before and after the fault, as calculated below,
ΔI＝Iaf-Ibf
ΔP＝Paf-Pbf
ΔQ＝Qaf-Qbf
Figure FDA0002987965740000011
in the formula, the subscript Iaf、Paf、QafIndicating current, active and reactive power before line fault, Ibf、Pbf、QbfRepresenting the current, active power and reactive power after the line fault;
calculating the importance degree of each feature by adopting a Person correlation coefficient, recursive feature elimination and Person correlation coefficient methods for the constructed features and the original features, and selecting feature quantities; the finally selected characteristic quantities are a current difference value delta I, a voltage change amplitude delta U', a reactive power Q _ bf before grounding, an active power difference value delta P and a reactive power difference value delta Q.
4. The LSTM-based single-phase ground fault line selection method of claim 1, wherein: in S3, a single-phase earth fault line selection model is established by using modules in a Keras library, wherein LSTM modules are adopted for a first layer and a second layer of hidden layers, an activation function is 'relu', Dense modules are adopted for a third layer of hidden layers and an output layer, an activation function of the third layer of hidden layers is 'tanh', and an activation function of the output layer is 'sigmoid'.
5. The LSTM-based single-phase ground fault line selection method of claim 1, wherein: in S4, the characteristic quantity data of each sample is subjected to z-score normalization processing, and then input to the LSTM-based single-phase ground fault line selection model.
6. An LSTM-based single-phase ground fault line selection method according to claim 1, 2, 3, 4 or 5, wherein: for use in a power distribution network containing distributed power sources.
7. The LSTM-based single-phase ground fault line selection method of claim 6, wherein: and the distributed power supply in the power distribution network is a photovoltaic and/or energy storage power supply.
8. A single-phase earth fault route selection system based on LSTM, its characterized in that: for implementing an LSTM-based single-phase ground-fault line selection method according to any of claims 1-7.
9. The LSTM-based single-phase ground-fault line selection system of claim 8, wherein: comprises the following modules which are used for realizing the functions of the system,
the first module is used for data acquisition and data processing, and comprises a dispatching platform, a first module and a second module, wherein the dispatching platform is used for acquiring a plurality of conventional electric quantities monitored by the power distribution network and performing data cleaning operation to obtain original characteristics;
the second module is used for selecting the characteristic quantity, constructing a plurality of new characteristics according to the electric quantity acquired by the first module, solving the importance degree of each characteristic by adopting a plurality of characteristic quantity extraction algorithms from the original characteristics and the new characteristics, and selecting the characteristic quantity according to the importance degree;
a third module for establishing an LSTM-based single-phase ground fault line selection model as follows,
the LSTM-based single-phase earth fault line selection model comprises 5 layers in total, wherein the input layer, the three hidden layers and the output layer are determined by characteristic quantity, the first two hidden layers are LSTM modules, the third hidden layer is a Dense module, the output layer is a Dense module, and a result for judging whether a line is in fault or not is output;
and the fourth module is used for training and testing the LSTM network, and comprises the steps of training a single-phase earth fault line selection model based on the LSTM by using the characteristic quantity of the sample data and then obtaining an earth fault judgment result by using the trained single-phase earth fault line selection model.
10. The LSTM-based single-phase ground-fault line selection system of claim 9, wherein: comprising a processor and a memory, the memory being adapted to store program instructions, the processor being adapted to invoke the stored instructions in the memory to perform a LSTM-based single-phase ground fault line selection method as claimed in any one of claims 1 to 7.
CN202110307110.2A 2021-03-23 2021-03-23 LSTM-based single-phase earth fault line selection method and system Pending CN113092933A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110307110.2A CN113092933A (en) 2021-03-23 2021-03-23 LSTM-based single-phase earth fault line selection method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110307110.2A CN113092933A (en) 2021-03-23 2021-03-23 LSTM-based single-phase earth fault line selection method and system

Publications (1)

Publication Number Publication Date
CN113092933A true CN113092933A (en) 2021-07-09

Family

ID=76669210

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110307110.2A Pending CN113092933A (en) 2021-03-23 2021-03-23 LSTM-based single-phase earth fault line selection method and system

Country Status (1)

Country Link
CN (1) CN113092933A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113884818A (en) * 2021-10-27 2022-01-04 国网江苏省电力有限公司徐州供电分公司 LSTM-based power distribution network fault traveling wave arrival time accurate estimation method
CN114910793A (en) * 2022-04-24 2022-08-16 广东工业大学 SOH estimation method for series battery pack of energy storage power station

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110082640A (en) * 2019-05-16 2019-08-02 国网安徽省电力有限公司 A kind of distribution singlephase earth fault discrimination method based on long memory network in short-term
CN110646708A (en) * 2019-09-27 2020-01-03 中国矿业大学 10kV single-core cable early state identification method based on double-layer long-and-short-term memory network
CN111881971A (en) * 2020-07-24 2020-11-03 成都理工大学 Power transmission line fault type identification method based on deep learning LSTM model
CN112327101A (en) * 2020-10-30 2021-02-05 国网上海市电力公司 Power distribution network reliability detection method and system based on long-time and short-time memory neural network

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110082640A (en) * 2019-05-16 2019-08-02 国网安徽省电力有限公司 A kind of distribution singlephase earth fault discrimination method based on long memory network in short-term
CN110646708A (en) * 2019-09-27 2020-01-03 中国矿业大学 10kV single-core cable early state identification method based on double-layer long-and-short-term memory network
CN111881971A (en) * 2020-07-24 2020-11-03 成都理工大学 Power transmission line fault type identification method based on deep learning LSTM model
CN112327101A (en) * 2020-10-30 2021-02-05 国网上海市电力公司 Power distribution network reliability detection method and system based on long-time and short-time memory neural network

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
NA QU等: "Fault Detection on Insulated Overhead Conductors Based on DWT-LSTM and Partial Discharge", 《IEEE ACCESS》 *
刘正谊等: "基于随机森林算法的线路接地在线诊断模型", 《内蒙古电力技术》 *
翟二杰等: "基于VMD-LSTM 的小电流接地系统故障选线方法", 《电工电能新技术》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113884818A (en) * 2021-10-27 2022-01-04 国网江苏省电力有限公司徐州供电分公司 LSTM-based power distribution network fault traveling wave arrival time accurate estimation method
CN113884818B (en) * 2021-10-27 2024-04-09 国网江苏省电力有限公司徐州供电分公司 Method for accurately estimating arrival time of fault traveling wave of power distribution network based on LSTM
CN114910793A (en) * 2022-04-24 2022-08-16 广东工业大学 SOH estimation method for series battery pack of energy storage power station

Similar Documents

Publication Publication Date Title
CN110082640A (en) A kind of distribution singlephase earth fault discrimination method based on long memory network in short-term
CN108667005B (en) Power grid static and dynamic combination vulnerability assessment method considering new energy influence
Reche et al. Data mining-based method to reduce multiple estimation for fault location in radial distribution systems
EP2580696A2 (en) Detecting state estimation network model data errors
CN113092933A (en) LSTM-based single-phase earth fault line selection method and system
CN109102146B (en) Electric power system risk assessment acceleration method based on multi-parameter linear programming
CN107292481B (en) Power grid key node evaluation method based on node importance
CN109428327B (en) Power grid key branch and leading stable mode identification method and system based on response
CN113922412B (en) New energy multi-station short-circuit ratio panoramic evaluation method, system, storage medium and computing equipment
CN110687393A (en) Valve short-circuit protection fault positioning method based on VMD-SVD-FCM
CN114006413B (en) Power system transient stability control method and system based on graph neural network
CN108267673B (en) Distribution network fault line selection big data dimension reduction method and device
CN112288326A (en) Fault scene set reduction method suitable for toughness evaluation of power transmission system
CN115693661A (en) Cascading failure risk key line identification method based on graph neural network
CN111654392A (en) Low-voltage distribution network topology identification method and system based on mutual information
CN110389268B (en) Online monitoring system of electric power system
CN113937764A (en) Low-voltage distribution network high-frequency measurement data processing and topology identification method
CN107462810B (en) Fault section positioning method suitable for active power distribution network
CN111062569A (en) Low-current fault discrimination method based on BP neural network
Koley et al. Artificial neural network based protection scheme for one conductor open faults in six phase transmission line
Babayomi et al. Intelligent fault diagnosis in a power distribution network
CN115954956A (en) Method and system for evaluating access capacity of distributed power supply of power distribution network
CN113092934B (en) Single-phase earth fault judgment method and system based on clustering and LSTM
CN116131313A (en) Explanatory analysis method for association relation between characteristic quantity and transient power angle stability
CN113627655B (en) Method and device for simulating and predicting pre-disaster fault scene of power distribution network

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20210709