CN112462193B - Automatic reclosing judgment method for power distribution network based on real-time fault filtering data - Google Patents
Automatic reclosing judgment method for power distribution network based on real-time fault filtering data Download PDFInfo
- Publication number
- CN112462193B CN112462193B CN202011223358.2A CN202011223358A CN112462193B CN 112462193 B CN112462193 B CN 112462193B CN 202011223358 A CN202011223358 A CN 202011223358A CN 112462193 B CN112462193 B CN 112462193B
- Authority
- CN
- China
- Prior art keywords
- wavelet
- fault
- function
- frequency
- space
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 19
- 238000001914 filtration Methods 0.000 title claims abstract description 7
- 238000004458 analytical method Methods 0.000 claims abstract description 35
- 238000012706 support-vector machine Methods 0.000 claims abstract description 32
- 230000009466 transformation Effects 0.000 claims abstract description 22
- 238000004088 simulation Methods 0.000 claims abstract description 20
- 238000012549 training Methods 0.000 claims abstract description 20
- 238000013145 classification model Methods 0.000 claims abstract description 14
- 238000000605 extraction Methods 0.000 claims abstract description 6
- 238000005094 computer simulation Methods 0.000 claims abstract description 4
- 230000006870 function Effects 0.000 claims description 85
- 230000001052 transient effect Effects 0.000 claims description 31
- 238000000354 decomposition reaction Methods 0.000 claims description 29
- 230000009977 dual effect Effects 0.000 claims description 9
- 238000013507 mapping Methods 0.000 claims description 9
- 238000005070 sampling Methods 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 9
- 230000008859 change Effects 0.000 claims description 7
- 238000001514 detection method Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 230000009467 reduction Effects 0.000 claims description 6
- 238000010586 diagram Methods 0.000 claims description 4
- 230000006835 compression Effects 0.000 claims description 3
- 238000007906 compression Methods 0.000 claims description 3
- 230000001939 inductive effect Effects 0.000 claims description 3
- 230000001788 irregular Effects 0.000 claims description 3
- 238000002372 labelling Methods 0.000 claims description 3
- 238000004091 panning Methods 0.000 claims description 3
- 230000035945 sensitivity Effects 0.000 claims description 3
- 230000003595 spectral effect Effects 0.000 claims description 3
- 230000001131 transforming effect Effects 0.000 claims description 3
- 238000013519 translation Methods 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims 1
- 230000008901 benefit Effects 0.000 description 6
- 239000000284 extract Substances 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 230000007704 transition Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000012850 discrimination method Methods 0.000 description 1
- 238000013095 identification testing Methods 0.000 description 1
- 238000009413 insulation Methods 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—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
- G01R31/50—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
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention relates to a method for judging automatic reclosing of a power distribution network based on real-time fault filtering data, and belongs to the technical field of power grids. The method comprises the following steps: s1: establishing a fault extraction model based on wavelet transformation; s2: establishing a fault comprehensive research-judgment-support vector machine classification model; s3: simulation modeling and analysis. The three-phase voltage, three-phase current and zero sequence current are extracted by wavelet transformation to serve as characteristic quantities, wavelet coefficient reconstruction after wavelet analysis is used as a data set of a support vector machine, algorithm model training data are derived from actual recording data, reliability and practicality of the model are improved, and recognition accuracy is close to 90%. After the algorithm model is built, the algorithm model is tested by using simulation data, the accuracy reaches 95%, and the feasibility of the algorithm model is further verified.
Description
Technical Field
The invention belongs to the technical field of power grids, and relates to an automatic reclosing judgment method for a power distribution network based on real-time fault filtering data.
Background
The distribution network is an intermediate network between the backbone network of the power system and the users and is positioned at the extreme end of the power grid, so that the requirements of the users on the power supply safety, the power quality and the like of the distribution network can be objectively reflected. Although the distribution automation and management level of China is continuously enhanced, the distribution network has wide radiation range, a plurality of branch lines and complex running environment, and has great challenges on the reliability and safety of power supply of the distribution network. Based on the relevant data statistics, about 85% of the outage in the power system is due to faults occurring in the distribution network, of which about 70% are single-phase earth faults.
After the power distribution network has permanent faults to trip the line, insulation cannot be recovered by itself, if automatic reclosing is thrown to the permanent fault point, accidents are enlarged, and personal safety accidents are possibly caused, so that instantaneous faults and the permanent faults must be identified on line in real time. The existing mature high-voltage transmission line fault identification method cannot be effectively applied to a distribution line, and has not been studied too much in the aspect of effectively identifying the type of the ground fault by deep mining fault characteristics based on fault recording data. The invention provides a classification and discrimination method of single-phase earth fault permanent and transient faults based on wavelet analysis and a support vector machine. And finally, simulating two faults according to an actual topological model of the power distribution network in an MATLAB/SIMULINK environment, and testing a trained model by simulating obtained data, wherein the result proves the feasibility of the algorithm.
Disclosure of Invention
Therefore, the invention aims to provide a method for judging automatic reclosing of a power distribution network based on real-time fault filtering data.
In order to achieve the above purpose, the present invention provides the following technical solutions:
a method for judging automatic reclosing of a power distribution network based on real-time fault filtering data comprises the following steps:
s1: establishing a fault extraction model based on wavelet transformation;
s2: establishing a fault comprehensive research-judgment-support vector machine classification model;
s3: simulation modeling and analysis.
Optionally, the step S1 specifically includes the following steps:
s11: performing wavelet transformation;
psi (t) is a basic wavelet function, and a is a scale factor; b is a panning factor, then the basic definition of the wavelet transform is:
for any one function f (t) E L 2 The wavelet transform of (R) is defined as:
wherein a, b and t are continuous variables, known as continuous wavelet transforms; at psi a,b When (t) is a complex function, the complex conjugate function is used in the integrationCorresponding to f (t) ∈L 2 (R) in the family of functions ψ a,b Decomposition at (t); a. b determinationIs a spectral change of (2); changing b will change the signal region to be analyzed; changing a will affect the time and frequency resolution of the signal; the decomposition must satisfy the following permissible conditions:
or (b)
Wherein ψ (ω) is the fourier transform of ψ (t);
the discretization of the scale is achieved in the form of a power series, i.e. the scale parameter a isTo prevent loss of the detection signal, the translation parameter b is +.>As a sampling interval of the signal; wherein b 0 Uniform sampling interval at j=0; taking a 0 ＝2,b 0 =2, then ψ a,b (t) becomes:
the above is called psi j,k The discrete wavelet of (t);
decomposition tree of discrete wavelet analysis of the original signal T (n): CA (CA) k Low frequency coefficient, CD, representing decomposition of the kth layer k High frequency coefficients representing the decomposition of the k-th layer;
the signal component obtained after the single reconstruction of the decomposition coefficient of each layer is recorded as A i And D i The sum of the reconstructed signal components of the original signal T (n) is expressed as
S12: selecting a wavelet base;
the wavelet basis functions are: haar, dbN, coifN, bior, symN the wavelet basis function for ground fault line selection has been found to be dbN, coifN, symN; the orthogonality, the tight support and the sensitivity to irregular signals of dbN series wavelets are considered, so that the method is suitable for analyzing transient signals; wherein N represents the vanishing moment of the wavelet basis function, and vanishing moment characteristics are a key parameter for signal noise reduction, compression, singularity detection and other applications; the actual signal contains a lot of noise and combines the characteristics of the wavelet base function, and the mother wavelet with the noise reduction function is required to be selected, so db5 wavelet is selected for wavelet transformation; the sampling frequency of the actual transient wave recording indicator is 4096Hz, so that the power frequency component is not mixed into the high-frequency component, more interference factors are avoided, the number of decomposition layers is 4, and the frequency range of each layer is 0-1024Hz, 1024-2048Hz, 2048-3072Hz and 3072-4096Hz; and (3) carrying out wavelet analysis on the voltage or current of each fault line, carrying out coefficient reconstruction on the high-frequency part of each fault line to form a wavelet coefficient database, and carrying out labeling on each fault line to serve as a sample of a classification model so as to provide basic data and a label set for a subsequent classification model.
Optionally, the step S2 specifically includes the following steps:
s21: establishing a nonlinear support vector machine
Assume that a training data set t= { (x) on one feature space is given 1 ,y 1 ),(x 2 ,y 2 ),(x i ,y i )...(x N ,y N ) X, where x i ∈R n N is sample x i Dimension, y i Is x i N is the total number of samples, (x) i ,y i ) Referred to as sample points;
set the original space asx＝(x (1) ,x (2) ) T E chi, new space ∈χ>z＝(z (1) ,z (2) ) T E Z, defining a transformation from the original space to the new space:
z＝φ(x)＝((x (1) ) 2 ,(x (2) ) 2 ) T (6)
transformed z=phi (x), original spaceTransform into new space +.>The corresponding transformation of points in the original space into points in the new space, ellipses in the original space
ω 1 (x (1) ) 2 +ω 2 (x (2) ) 2 +b＝0 (7)
Transforming into straight lines in new space
ω 1 z (1) +ω 2 z (2) +b＝0 (8)
ω 1 、ω 2 Is the normal vector of the hyperplane, b is the intercept, and the straight line omega is in the new space after transformation 1 z (1) +ω 2 z (2) +b=0 can correctly separate the transformed positive and negative instance points; the nonlinear separable problem of the original space becomes the linear separable problem of the new space;
in the dual problem of the linear support vector machine, both the objective function and the decision function only involve the inner product between the input instance and the instance; inner product x in objective function of dual problem i ·x j Can be used with a kernel function K (x i ·x j )＝φ(x i )·φ(x j ) Instead, φ (x) is a mapping function; the objective function of the dual problem at this time becomes
The inner product in the classification decision function is replaced by a kernel function, the original input space is transformed into a new feature space through a mapping function phi, and when the mapping function is a nonlinear function, the learned support vector machine containing the kernel function is a nonlinear classification model;
s22: establishing a fault identification network;
firstly, training an algorithm model by using existing actual data, and testing the algorithm model by taking data obtained by actual line simulation as a test sample; three-phase voltage, three-phase current and zero sequence current are extracted as characteristic quantities, each characteristic quantity is subjected to db5 wavelet analysis pretreatment, 4 layers of high-frequency wavelet coefficients after decomposition and reconstruction are extracted as the input of a support vector machine, the output is two types, one type is a permanent fault, and the other type is a transient fault.
Optionally, the step S3 specifically includes the following steps:
s31: wavelet analysis of actual data waveforms
Extracting three-phase voltage, three-phase current and zero sequence current as fault characteristic signals to perform wavelet analysis, adopting db5 wavelet, wherein the number of decomposition layers is 4, extracting 4 layers of high-frequency coefficients, which are equivalent to 4 wavelet coefficient groups in each waveform, and respectively analyzing one permanent fault record data and one transient fault record data;
wavelet analysis is carried out on the A-phase voltage and the zero-sequence current of one permanent fault and one transient fault, wavelet coefficients of the two types of faults are compared, and the amplitude and the frequency of the occurrence of singular points are obviously different; the high-frequency coefficient of wavelet analysis of each characteristic signal is reserved, and a wavelet coefficient library is formed and used as the input of a support vector machine recognition model;
s32: permanent and transient fault identification simulation verification of support vector machine
What is needed to be done after the kernel function is selected is the selection of parameters, including the selection of the parameter sigma of the radial basis function and the selection of the positive parameter C, and then the proper parameters C and sigma are selected through simulation, so that higher classification accuracy is obtained; the whole simulation is realized in an MATLAB simulation environment, different parameters C and sigma are set, and the accuracy under different parameters is compared; according to the learning and classifying process of the SVM, setting the value of the parameter C as infinity in the simulating process, and taking training results obtained by different parameter values; the sigma value is fixed to be 0.1, and the positive parameter C takes training results with different values;
selecting sigma value to be 0.01, and selecting C value to be 100; simulating two faults according to an actual topology model of the power distribution network by applying MATLAB/SIMULINK environment, simulating permanent faults caused by disconnection of a certain phase of a three-phase circuit breaker, controlling the occurrence time of a fault module of the SIMULINK to simulate transient faults, and changing fault positions and fault samples; the power system model adopts a power end transformer 110/10Kv, a load end transformer 10/0.4Kv, and is built according to an actual distribution network circuit topological diagram, wherein the circuit comprises overhead lines, cables and a mixed circuit of the overhead lines and the cables; the output voltage of the three-phase voltage source is 110kV, the frequency is 50Hz, and the overhead part parameters in the line are respectively positive sequence impedance 0.03 Ω/Km, capacitance 15nF/Km and inductance 0.95mH/Km; zero sequence impedance is 0.9 ohm/Km, capacitance resistance is 22nF/Km, inductance is 6mH/Km; the parameters of the cable line are set to be positive sequence impedance 0.008 ohm/Km, capacitance impedance 6nF/Km and inductance 0.36mH/Km; zero sequence impedance 0.5 ohm/Km, capacitive reactance 8nF/Km, inductive reactance 2.1mH/Km.
The invention has the beneficial effects that:
1. performing fault identification before reclosing, and providing a wavelet analysis based on wave recording data and a support vector machine fault algorithm identification model;
2. the proposed algorithm model is verified through simulation, and the recognition effect is good.
Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention. The objects and other advantages of the invention may be realized and obtained by means of the instrumentalities and combinations particularly pointed out in the specification.
Drawings
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in the following preferred detail with reference to the accompanying drawings, in which:
FIG. 1 is a wavelet decomposition tree;
FIG. 2 is a non-linear support vector machine classification;
FIG. 3 is a topology of a nonlinear support vector machine;
FIG. 4 is a diagram of a failure recognition network
Fig. 5 is a line simulation model.
Detailed Description
Other advantages and effects of the present invention will become apparent to those skilled in the art from the following disclosure, which describes the embodiments of the present invention with reference to specific examples. The invention may be practiced or carried out in other embodiments that depart from the specific details, and the details of the present description may be modified or varied from the spirit and scope of the present invention. It should be noted that the illustrations provided in the following embodiments merely illustrate the basic idea of the present invention by way of illustration, and the following embodiments and features in the embodiments may be combined with each other without conflict.
Wherein the drawings are for illustrative purposes only and are shown in schematic, non-physical, and not intended to limit the invention; for the purpose of better illustrating embodiments of the invention, certain elements of the drawings may be omitted, enlarged or reduced and do not represent the size of the actual product; it will be appreciated by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted.
The same or similar reference numbers in the drawings of embodiments of the invention correspond to the same or similar components; in the description of the present invention, it should be understood that, if there are terms such as "upper", "lower", "left", "right", "front", "rear", etc., that indicate an azimuth or a positional relationship based on the azimuth or the positional relationship shown in the drawings, it is only for convenience of describing the present invention and simplifying the description, but not for indicating or suggesting that the referred device or element must have a specific azimuth, be constructed and operated in a specific azimuth, so that the terms describing the positional relationship in the drawings are merely for exemplary illustration and should not be construed as limiting the present invention, and that the specific meaning of the above terms may be understood by those of ordinary skill in the art according to the specific circumstances.
1. Wavelet transform-based fault extraction model
Current and voltage are the most fundamental electrical characteristics of a power distribution network. Then the three-phase voltage and the three-phase current are analyzed to be the quantities which can most directly reflect the fault characteristics of the distribution network when the distribution network breaks down. When a single-phase earth fault occurs in the power distribution network, the current can reflect the state change of the power distribution network, the zero-sequence current appears in the system after the fault occurs, and in the extremely short time when the earth fault occurs, the zero-sequence current contains a transient free oscillation component with a very large transient value, which is possibly many times larger than a stable value, so that the time domain and frequency domain characteristics of the zero-sequence current are necessary to be analyzed. The wavelet transformation has good time-frequency localization characteristics and has unique advantages for analyzing high-frequency mutation signals and low-frequency slowly-varying signals. In analysing the signal, different resolutions should be available at different frequencies. The frequency resolution should be higher at low frequencies, while the frequency resolution may be reduced at high frequency bands, such that the frequency resolution Δf varies with the frequency f. Wavelet transforms are having such characteristics. Because the three-phase voltage and the three-phase current are the characteristic quantities of direct projection faults and the zero-sequence current contains a lot of fault transient information, the invention extracts the three-phase voltage and the three-phase current and the zero-sequence current as fault characteristic signals to carry out wavelet analysis.
1.1 wavelet transform principle
The basic idea of wavelet analysis is to represent or approximate a certain signal or function with a family of functions, called wavelet function systems, which are constructed by shifting and scaling the different dimensions of the basic wavelet function.
Psi (t) is a basic wavelet function, and a is a scale factor; b is a panning factor, then the basic definition of the wavelet transform is:
thus for any one function f (t) ∈L 2 The wavelet transform of (R) is defined as:
where a, b and t are continuous variables, and thus the formula is also known as continuous wavelet transform. At psi a,b When (t) is a complex function, the complex conjugate function is used in the integrationIt corresponds to f (t) ∈L 2 (R) in the family of functions ψ a,b Decomposition at (t). a. b decision->Is a spectral change of (2); changing b will change the signal region to be analyzed; changing a will affect the time and frequency resolution of the signal. The decomposition must satisfy the following permissible conditions:
or (b)
Where ψ (ω) is the fourier transform of ψ (t).
Since the actual collected data are all discrete data, a and b are required to be discretized, and therefore discrete wavelet transformation is adopted for actual signal processing. Typically, the discretization of the scale is implemented in the form of a power series, i.e. the scale parameter a isTo prevent loss of the detection signal, the translation parameter b is +.>As the sampling interval of the signal. Wherein b 0 A uniform sampling interval at j=0. In practical application, a is generally taken 0 ＝2,b 0 =2, then ψ a,b (t) becomes:
the above is called psi j,k A discrete wavelet of (t).
The decomposition tree of the discrete wavelet analysis of the original signal T (n) is shown in fig. 1: CA in the figure k Low frequency coefficient, CD, representing decomposition of the kth layer k Representing the high frequency coefficients of the k-th layer decomposition.
The signal component obtained after the single reconstruction of the decomposition coefficient of each layer is recorded as A i And D i The sum of the reconstructed signal components of the original signal T (n) is expressed as
1.2 selection of wavelet basis (mother wavelet)
The wavelet basis functions commonly used are mainly: haar, dbN, coifN, bior, symN, et al, have demonstrated dbN, coifN, symN wavelet basis functions suitable for ground fault line selection. The method is suitable for analyzing transient signals in consideration of orthogonality, tight support and sensitivity to irregular signals of dbN series wavelets. Wherein N represents the vanishing moment of the wavelet basis function, and vanishing moment characteristic is a key parameter for signal noise reduction, compression, singularity detection and other applications. The actual signal contains much noise and combines the characteristics of the wavelet base function, and the parent wavelet with the noise reduction function is required to be selected, so that the db5 wavelet is selected for wavelet transformation. The sampling frequency of the actual transient wave recording indicator is 4096Hz, so that the power frequency component is not mixed into the high-frequency component, more interference factors are avoided, the number of decomposition layers is 4, and the frequency range of each layer is 1024-2048Hz, 512-1024Hz, 256-512Hz and 128-256Hz. And (3) carrying out wavelet analysis on the voltage or current of each fault line, carrying out coefficient reconstruction on the high-frequency part of each fault line to form a wavelet coefficient database, and carrying out labeling on each fault line to serve as a sample of a classification model so as to provide basic data and a label set for a subsequent classification model.
2. Comprehensive fault research and judgment-support vector machine classification model
The nonlinear support vector machine maps the data of the original space to the high-dimensional feature space through one transformation based on the kernel space theory, so that the linear partitionable or approximate linear partable of the sample in the high-dimensional space is realized. The essence is also to convert the non-linear problem into a linear problem. The invention selects three-phase voltage and current and zero sequence current as characteristic quantities, after wavelet analysis, the fault characteristics can be reflected in different decomposition layers, so that the randomness of the selected abnormal signals exists, the relation between input and output can not be reflected by one characteristic, and the characteristics of the signals also have randomness, so that the classification problem of the invention is nonlinear classification. For training samples, after sample data are analyzed, the output is of two types, one type is permanent faults, the other type is transient faults, and the support vector machine is a nonlinear two-class model.
2.1 nonlinear support vector machine
As shown in fig. 2, assume that training data set t= { (x) on a given one feature space 1 ,y 1 ),(x 2 ,y 2 ),(x i ,y i )...(x N ,y N ) X, where x i ∈R n N is sample x i Dimension, y i Is x i N is the total number of samples, (x) i ,y i ) Referred to as sample points.
Set the original space asx＝(x (1) ,x (2) ) T E chi, new space ∈χ>z＝(z (1) ,z (2) ) T E Z, defining a transformation from the original space to the new space:
z＝φ(x)＝((x (1) ) 2 ,(x (2) ) 2 ) T (6)
transformed z=phi (x), original spaceTransform into new space +.>The corresponding transformation of points in the original space into points in the new space, ellipses in the original space
ω 1 (x (1) ) 2 +ω 2 (x (2) ) 2 +b＝0 (7)
Transforming into straight lines in new space
ω 1 z (1) +ω 2 z (2) +b＝0 (8)
ω 1 、ω 2 Is the normal vector of the hyperplane, b is the intercept, and the straight line omega is in the new space after transformation 1 z (1) +ω 2 z (2) +b=0 can correctly separate the transformed positive and negative instance points. Thus, the nonlinear separable problem of the original space becomes the linear separable problem of the new space.
In the dual problem of the linear support vector machine, both the objective function and the decision function involve only the inner product between the input instances. Inner product x in objective function of dual problem i ·x j Can be used with a kernel function K (x i ·x j )＝φ(x i )·φ(x j ) Instead, φ (x) is a mapping function. The objective function of the dual problem at this time becomes
Similarly, the inner product of the classification decision function can be replaced by a kernel function, the original input space is transformed into a new feature space through the mapping function phi, and when the mapping function is a nonlinear function, the learned support vector machine containing the kernel function is a nonlinear classification model, as shown in fig. 3. Currently, widely used kernel functions are linear kernel functions, polynomial kernel functions, radial Basis Function (RBF), sigmoid functions, and the like. The RBF kernel function has the advantage that the dimension of the high-dimensional feature space is infinite, any sample is linearly separable after being mapped from the low-dimensional space to the feature space, and the effect is good, so the RBF kernel function is adopted in the invention.
2.2 Fault identification network
As shown in fig. 4, the algorithm model is first trained using existing actual data, and the actual line simulation data is used as a test sample to test the algorithm model. The analysis and the extraction of three-phase voltage, three-phase current and zero sequence current are known as characteristic quantities, each characteristic quantity is preprocessed through db5 wavelet analysis, 4 layers of high-frequency wavelet coefficients after decomposition and reconstruction are extracted as the input and the output of the support vector machine, and the output is of two types, namely permanent faults and transient faults.
3 simulation modeling and analysis
The data used for the algorithm model training is from the data collected by the transient fault wave recording indicator at the fault moment of the 10kV power distribution network in certain city, and the data of the single-phase ground fault after certain screening is primarily classified into two major types of permanent faults and transient faults according to the reported information of the transient fault wave recording indicator.
3.1 wavelet analysis of actual data waveforms
The invention extracts three-phase voltage, three-phase current and zero sequence current as fault characteristic signals to carry out wavelet analysis, adopts db5 wavelet, has 4 layers of decomposition layers, extracts 4 layers of high-frequency coefficients, is equivalent to 4 wavelet coefficient groups in each waveform, and respectively analyzes one permanent and transient fault recording data
Wavelet analysis is carried out on the A-phase voltage and the zero-sequence current of one permanent fault and one transient fault, wavelet coefficients of the two types of faults are compared, the frequency of the amplitude and the frequency of the occurrence of singular points are obviously different, for example, when the A-phase voltage is compared, the difference between the two phases can be well reflected by the d2 second-layer wavelet coefficient, and the invention proves that the characteristic extraction is feasible by utilizing the wavelet analysis. After a large amount of analysis is carried out on the actual data, the number of high-frequency layers of the characteristic signals of the permanent faults and the transient faults under the single-phase grounding condition influenced by different grounding modes, the operation working conditions and different grounding fault types is uncertain, so that the high-frequency coefficients of the wavelet analysis of each characteristic signal are reserved, and a wavelet coefficient library is formed to serve as the input of a support vector machine identification model.
3.2 permanent and transient failure recognition simulation verification of support vector machine
What follows after the kernel function has been selected is the selection of parameters, including the selection of the parameter sigma of the radial basis function and the selection of the positive parameter C, and the selection of the appropriate parameters C and sigma by simulation, resulting in a higher classification accuracy. The whole simulation is realized in a MATLAB simulation environment, different parameters C and sigma are set, and the accuracy under different parameters is compared. According to the learning classification process of the SVM, when the introduction parameter is infinite, the accuracy is high, so that the value of the parameter C is set to be infinite in the simulation process, and training results obtained by taking different parameter values are shown in a table 1. The training results when the positive parameter C takes different values are shown in table 2, with sigma value fixed at 0.1.
Table 1 Sigma-different training results
TABLE 2 training results for different positive parameters C
It can be seen from tables 1 and 2 that the training results and the training accuracy are different when different parameters are selected, so that the training time and the accuracy are considered when the gaussian radial basis function is applied as the kernel function of the support vector machine to select the parameters C and sigma. After analyzing the rules, selecting sigma value to be 0.01 and C value to be 100 from the total consideration. Two faults are simulated according to an actual topological model of the power distribution network by applying MATLAB/SIMULINK environment, permanent faults caused by disconnection are simulated by using whether a certain phase of a three-phase circuit breaker is disconnected or not, the occurrence time of a fault module of the SIMULINK is controlled to simulate transient faults, and the fault position and a fault sample are changed. The power system model adopts a power source end transformer 110/10Kv, a load end transformer 10/0.4Kv, and is built according to an actual distribution network circuit topological diagram, and the circuit comprises overhead lines, cables and a mixed circuit of the overhead lines and the cables. The output voltage of the three-phase voltage source is 110kV, the frequency is 50Hz, and the overhead part parameters in the line are respectively positive sequence impedance 0.03 Ω/Km, capacitance 15nF/Km and inductance 0.95mH/Km; zero sequence impedance is 0.9 ohm/Km, capacitance resistance is 22nF/Km, inductance is 6mH/Km; the parameters of the cable line are set to be positive sequence impedance 0.008 ohm/Km, capacitance impedance 6nF/Km and inductance 0.36mH/Km; zero sequence impedance 0.5 ohm/Km, capacitive reactance 8nF/Km, inductive reactance 2.1mH/Km.
As shown in fig. 5, a test sample of the failure recognition network: the 2 fault types (permanent and transient faults) of the line are taken, the grounding fault transition resistance is taken as 10Ω and 500Ω, the fault position is taken as 30% and 60% of the line, the fault initial phase angle is taken as 0, 45 and 90, and the partial fault identification test results are shown in table 3.
TABLE 3 test sample fault type identification results
As shown in the table, the input characteristic quantity of the fault identification network is better corresponding to the fault type, and the correct identification rate of the SVM fault identification network is 95%.
Conclusion 4
The invention establishes an algorithm identification model, can quickly and accurately identify two fault types, is not influenced by transition resistance, fault position and fault initial phase angle, and can effectively avoid switching-on failure caused by permanent faults in line tripping caused by single-phase grounding faults in a power distribution network. The invention extracts three-phase voltage, three-phase current and zero sequence current by wavelet transformation as characteristic quantities, and reconstructs wavelet coefficients after wavelet analysis as a data set of a support vector machine, and algorithm model training data are derived from actual recording data, so that the reliability and practicality of the model are improved, and the recognition accuracy is close to 90%. After the algorithm model is built, the algorithm model is tested by using simulation data, the accuracy reaches 95%, and the feasibility of the algorithm model is further verified.
Finally, it is noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications and equivalents may be made thereto without departing from the spirit and scope of the present invention, which is intended to be covered by the claims of the present invention.
Claims (1)
1. A method for judging automatic reclosing of a power distribution network based on real-time fault filtering data is characterized by comprising the following steps of: the method comprises the following steps:
s1: establishing a fault extraction model based on wavelet transformation;
s2: establishing a fault comprehensive research-judgment-support vector machine classification model;
s3: simulation modeling and analysis;
the step S1 specifically comprises the following steps:
s11: performing wavelet transformation;
psi (t) is a basic wavelet function, and a is a scale factor; b is a panning factor, then the basic definition of the wavelet transform is:
for any one function f (t) E L 2 The wavelet transform of (R) is defined as:
wherein a, b and t are continuous variables, known as continuous wavelet transforms; at psi a,b When (t) is a complex function, the complex conjugate function is used in the integrationCorresponding to f (t) ∈L 2 (R) in the family of functions ψ a,b Decomposition at (t); a. b decision->Is a spectral change of (2); changing b will change the signal region to be analyzed; changing a will affect the time and frequency resolution of the signal; the decomposition must satisfy the following permissible conditions:
or (b)
Wherein ψ (ω) is the fourier transform of ψ (t);
the discretization of the scale is achieved in the form of a power series, i.e. the scale parameter a isTo prevent loss of the detection signal, the translation parameter b is +.>As a sampling interval of the signal; wherein b 0 Uniform sampling interval at j=0; taking a 0 ＝2,b 0 =2, then ψ a,b (t) becomes:
the above is called psi j,k The discrete wavelet of (t);
decomposition tree of discrete wavelet analysis of the original signal T (n): CA (CA) k Low frequency coefficient, CD, representing decomposition of the kth layer k High frequency coefficients representing the decomposition of the k-th layer;
the signal component obtained after the single reconstruction of the decomposition coefficient of each layer is recorded as A i And D i The sum of the reconstructed signal components of the original signal T (n) is expressed as
S12: selecting a wavelet base;
the wavelet basis functions are: haar, dbN, coifN, bior, symN the wavelet basis function for ground fault line selection has been found to be dbN, coifN, symN; the orthogonality, the tight support and the sensitivity to irregular signals of dbN series wavelets are considered, so that the method is suitable for analyzing transient signals; wherein N represents the vanishing moment of the wavelet basis function, vanishing moment characteristics are a key parameter for signal noise reduction, compression and singularity detection applications; the actual signal contains a lot of noise and combines the characteristics of the wavelet base function, and the mother wavelet with the noise reduction function is required to be selected, so db5 wavelet is selected for wavelet transformation; the sampling frequency of the actual transient wave recording indicator is 4096Hz, so that the power frequency component is not mixed into the high-frequency component, more interference factors are avoided, the number of decomposition layers is 4, and the frequency range of each layer is 0-1024Hz, 1024-2048Hz, 2048-3072Hz and 3072-4096Hz; for the voltage or current of each fault line, carrying out wavelet analysis on the fault line, carrying out coefficient reconstruction on a high-frequency part of the fault line to form a wavelet coefficient database, and carrying out labeling on the fault line to serve as a sample of a classification model to provide basic data and a label set of the fault line for a subsequent classification model;
the step S2 specifically comprises the following steps:
s21: establishing a nonlinear support vector machine
Assume that a training data set t= { (x) on one feature space is given 1 ,y 1 ),(x 2 ,y 2 ),(x i ,y i )...(x N ,y N ) X, where x i ∈R n N is sample x i Dimension, y i Is x i N is the total number of samples, (x) i ,y i ) Referred to as sample points;
set the original space asx＝(x (1) ,x (2) ) T E chi, new space ∈χ>z＝(z (1) ,z (2) ) T E Z, defining a transformation from the original space to the new space:
z＝φ(x)＝((x (1) ) 2 ,(x (2) ) 2 ) T (6)
transformed z=phi (x), original spaceTransform into new space +.>The corresponding transformation of points in the original space into points in the new space, ellipses in the original space
ω 1 (x (1) ) 2 +ω 2 (x (2) ) 2 +b＝0 (7)
Transforming into straight lines in new space
ω 1 z (1) +ω 2 z (2) +b＝0 (8)
ω 1 、ω 2 Is the normal vector of the hyperplane, b is the intercept, and the straight line omega is in the new space after transformation 1 z (1) +ω 2 z (2) +b=0 can correctly separate the transformed positive and negative instance points; the nonlinear separable problem of the original space becomes the linear separable problem of the new space;
in the dual problem of the linear support vector machine, both the objective function and the decision function involve only the inner product between the input instance and the instance; inner product x in objective function of dual problem i ·x j By a kernel function K (x i ·x j )＝φ(x i )·φ(x j ) Instead, φ (x) is a mapping function; the objective function of the dual problem at this time becomes
The inner product in the classification decision function is replaced by a kernel function, the original input space is transformed into a new feature space through a mapping function phi, and when the mapping function is a nonlinear function, the learned support vector machine containing the kernel function is a nonlinear classification model;
s22: establishing a fault identification network;
firstly, training a support vector machine classification model by using existing actual data, and testing an algorithm model by taking data obtained by actual line simulation as a test sample; extracting three-phase voltage, three-phase current and zero sequence current as characteristic quantities, preprocessing each characteristic quantity through db5 wavelet analysis, extracting 4 layers of high-frequency wavelet coefficients after decomposition and reconstruction as the input of a support vector machine, and outputting two types of the high-frequency wavelet coefficients, wherein one type is a permanent fault and the other type is a transient fault;
the step S3 specifically comprises the following steps:
s31: wavelet analysis of actual data waveforms
Extracting three-phase voltage, three-phase current and zero sequence current as fault characteristic signals to perform wavelet analysis, adopting db5 wavelet, wherein the number of decomposition layers is 4, extracting 4 layers of high-frequency coefficients, which are equivalent to 4 wavelet coefficient groups in each waveform, and respectively analyzing one permanent fault record data and one transient fault record data;
wavelet analysis is carried out on the A-phase voltage and the zero-sequence current of one permanent fault and one transient fault, wavelet coefficients of the two types of faults are compared, and the amplitude and the frequency of the occurrence of singular points are obviously different; the high-frequency coefficient of wavelet analysis of each characteristic signal is reserved, and a wavelet coefficient library is formed and used as the input of a support vector machine recognition model;
s32: permanent and transient fault identification simulation verification of support vector machine
What is needed to be done after the kernel function is selected is the selection of parameters, including the selection of the parameter sigma of the radial basis function and the selection of the positive parameter C, and then the proper parameters C and sigma are selected through simulation, so that higher classification accuracy is obtained; the whole simulation is realized in an MATLAB simulation environment, different parameters C and sigma are set, and the accuracy under different parameters is compared; according to the learning and classifying process of the SVM, setting the value of the parameter C as infinity in the simulating process, and taking training results obtained by different parameter values; the sigma value is fixed to be 0.1, and the positive parameter C takes training results with different values;
selecting sigma value to be 0.01, and selecting C value to be 100; simulating two faults according to an actual topology model of the power distribution network by applying MATLAB/SIMULINK environment, simulating permanent faults caused by disconnection of a certain phase of a three-phase circuit breaker, controlling the occurrence time of a fault module of the SIMULINK to simulate transient faults, and changing fault positions and fault samples; the power system model adopts a power end transformer 110/10Kv, a load end transformer 10/0.4Kv, and is built according to an actual distribution network circuit topological diagram, wherein the circuit comprises overhead lines, cables and a mixed circuit of the overhead lines and the cables; the output voltage of the three-phase voltage source is 110kV, the frequency is 50Hz, and the overhead part parameters in the line are respectively positive sequence impedance 0.03 Ω/Km, capacitance 15nF/Km and inductance 0.95mH/Km; zero sequence impedance is 0.9 ohm/Km, capacitance resistance is 22nF/Km, inductance is 6mH/Km; the parameters of the cable line are set to be positive sequence impedance 0.008 ohm/Km, capacitance impedance 6nF/Km and inductance 0.36mH/Km; zero sequence impedance 0.5 ohm/Km, capacitive reactance 8nF/Km, inductive reactance 2.1mH/Km.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011223358.2A CN112462193B (en) | 2020-11-05 | 2020-11-05 | Automatic reclosing judgment method for power distribution network based on real-time fault filtering data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011223358.2A CN112462193B (en) | 2020-11-05 | 2020-11-05 | Automatic reclosing judgment method for power distribution network based on real-time fault filtering data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112462193A CN112462193A (en) | 2021-03-09 |
| CN112462193B true CN112462193B (en) | 2023-10-24 |
Family
ID=74825100
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011223358.2A Active CN112462193B (en) | 2020-11-05 | 2020-11-05 | Automatic reclosing judgment method for power distribution network based on real-time fault filtering data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112462193B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113985733B (en) * | 2021-10-26 | 2023-11-17 | 云南电网有限责任公司电力科学研究院 | Power distribution network fault identification method based on self-adaptive probability learning |
| CN114742295A (en) * | 2022-04-11 | 2022-07-12 | 四川大学 | Equipment power utilization fault prediction method based on wavelet transform technology |
| CN116008731B (en) * | 2023-02-15 | 2023-08-25 | 重庆大学 | Power distribution network high-resistance fault identification method and device and electronic equipment |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TW201303319A (en) * | 2011-07-06 | 2013-01-16 | Univ Nat Taiwan | Method and system for fault detection, identification and location in high-voltage power transmission networks |
| CN103136587A (en) * | 2013-03-07 | 2013-06-05 | 武汉大学 | Power distribution network operating state classification recognition method based on support vector machine |
| CN104462762A (en) * | 2014-11-04 | 2015-03-25 | 西南交通大学 | Fuzzy fault classification method of electric transmission line |
| CN104714171A (en) * | 2015-04-06 | 2015-06-17 | 长沙学院 | Switching circuit fault classifying method based on wavelet transform and ICA feature extraction |
| CN105093066A (en) * | 2015-08-12 | 2015-11-25 | 华北电力大学 | Line fault judgment method based on wavelet analysis and support vector machine |
| CN106249101A (en) * | 2016-06-30 | 2016-12-21 | 湖南大学 | A kind of intelligent distribution network fault identification method |
| CN109241944A (en) * | 2018-09-07 | 2019-01-18 | 福州大学 | A kind of distribution network failure recognition methods based on improvement multi-category support vector machines |
| CN111339872A (en) * | 2020-02-18 | 2020-06-26 | 国网信通亿力科技有限责任公司 | Power grid fault classification method based on classification model |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200225655A1 (en) * | 2016-05-09 | 2020-07-16 | Strong Force Iot Portfolio 2016, Llc | Methods, systems, kits and apparatuses for monitoring and managing industrial settings in an industrial internet of things data collection environment |
-
2020
- 2020-11-05 CN CN202011223358.2A patent/CN112462193B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TW201303319A (en) * | 2011-07-06 | 2013-01-16 | Univ Nat Taiwan | Method and system for fault detection, identification and location in high-voltage power transmission networks |
| CN103136587A (en) * | 2013-03-07 | 2013-06-05 | 武汉大学 | Power distribution network operating state classification recognition method based on support vector machine |
| CN104462762A (en) * | 2014-11-04 | 2015-03-25 | 西南交通大学 | Fuzzy fault classification method of electric transmission line |
| CN104714171A (en) * | 2015-04-06 | 2015-06-17 | 长沙学院 | Switching circuit fault classifying method based on wavelet transform and ICA feature extraction |
| CN105093066A (en) * | 2015-08-12 | 2015-11-25 | 华北电力大学 | Line fault judgment method based on wavelet analysis and support vector machine |
| CN106249101A (en) * | 2016-06-30 | 2016-12-21 | 湖南大学 | A kind of intelligent distribution network fault identification method |
| CN109241944A (en) * | 2018-09-07 | 2019-01-18 | 福州大学 | A kind of distribution network failure recognition methods based on improvement multi-category support vector machines |
| CN111339872A (en) * | 2020-02-18 | 2020-06-26 | 国网信通亿力科技有限责任公司 | Power grid fault classification method based on classification model |
Non-Patent Citations (2)
| Title |
|---|
| A Wavelet-Based Fault Diagnosis Approach for Permanent Magnet Synchronous Motors;Mehrdad Heydarzadeh 等;《IEEE Transactions on Energy Conversion》;第34卷(第2期);第761-772页 * |
| 高压输电线路自适应重合闸研究与开发;陈诚;《中国优秀硕士学位论文全文数据库工程科技Ⅱ辑》(第3期);第C042-2667页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112462193A (en) | 2021-03-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112462193B (en) | Automatic reclosing judgment method for power distribution network based on real-time fault filtering data | |
| Wang et al. | ArcNet: Series AC arc fault detection based on raw current and convolutional neural network | |
| Duan et al. | Identification of partial discharge defects based on deep learning method | |
| CN109444656B (en) | Online diagnosis method for deformation position of transformer winding | |
| CN109307824B (en) | Clustering-based power distribution network single-phase earth fault section positioning method | |
| Gu et al. | High impedance fault detection in overhead distribution feeders using a DSP-based feeder terminal unit | |
| CN107132443B (en) | A kind of three-level STATCOM intelligent failure diagnosis method | |
| CN111723684B (en) | Identification method for transient overvoltage type in offshore wind farm | |
| Koley et al. | Wavelet-aided SVM tool for impulse fault identification in transformers | |
| Song et al. | A negative selection algorithm-based identification framework for distribution network faults with high resistance | |
| Gomes et al. | High-impedance faults in power distribution systems: A narrative of the field’s developments | |
| Li et al. | A fault pattern and convolutional neural network based single-phase earth fault identification method for distribution network | |
| CN111999591B (en) | Method for identifying abnormal state of primary equipment of power distribution network | |
| Wang et al. | Stator single-line-to-ground fault protection for bus-connected powerformers based on S-transform and bagging ensemble learning | |
| Dey et al. | Rough-granular approach for impulse fault classification of transformers using cross-wavelet transform | |
| Li et al. | Ferromagnetic resonance over-voltage identification method based on Gram angle field | |
| Zhang et al. | Incipient fault identification of distribution networks based on feature matching of power disturbance data | |
| CN114742295A (en) | Equipment power utilization fault prediction method based on wavelet transform technology | |
| CN111896890B (en) | Micro-grid line fault diagnosis method and system based on Hilbert-Huang transform | |
| Couto et al. | High impedance fault detection on microgrids considering the impact of VSC based generation | |
| Zhou et al. | Method of cable incipient faults detection and identification based on wavelet transform and gray correlation analysis | |
| Baran et al. | A classifier for distribution feeder overcurrent analysis | |
| Kumar et al. | Classification of PD faults using features extraction and K-means clustering techniques | |
| CN114755529A (en) | Multi-feature fusion single-phase earth fault type identification method based on deep learning | |
| Alsafasfeh | Pattern recognition for fault detection, classification, and localization in electrical power systems |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240805 Address after: No. 1-7, Building 12, No.16 North Huafu Avenue, Jiulongpo District, Chongqing, with a budget of 400000 RMB Patentee after: Chongqing Xiteng Electromechanical Equipment Co.,Ltd. Country or region after: China Address before: 400065 Chongqing Nan'an District huangjuezhen pass Chongwen Road No. 2 Patentee before: CHONGQING University OF POSTS AND TELECOMMUNICATIONS Country or region before: China |
|
| TR01 | Transfer of patent right |