CN111965486B - Power distribution network fault identification method and system based on intelligent data fusion analysis - Google Patents
Power distribution network fault identification method and system based on intelligent data fusion analysis Download PDFInfo
- Publication number
- CN111965486B CN111965486B CN202010787350.2A CN202010787350A CN111965486B CN 111965486 B CN111965486 B CN 111965486B CN 202010787350 A CN202010787350 A CN 202010787350A CN 111965486 B CN111965486 B CN 111965486B
- Authority
- CN
- China
- Prior art keywords
- matrix
- power distribution
- distribution network
- fault
- state monitoring
- 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
- 238000009826 distribution Methods 0.000 title claims abstract description 84
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000004458 analytical method Methods 0.000 title claims abstract description 22
- 230000004927 fusion Effects 0.000 title claims abstract description 19
- 238000004364 calculation method Methods 0.000 claims abstract description 28
- 230000009467 reduction Effects 0.000 claims abstract description 15
- 239000011159 matrix material Substances 0.000 claims description 128
- 238000012544 monitoring process Methods 0.000 claims description 69
- 238000001514 detection method Methods 0.000 claims description 13
- 230000008569 process Effects 0.000 claims description 13
- 239000013598 vector Substances 0.000 claims description 6
- 238000004422 calculation algorithm Methods 0.000 abstract description 4
- 238000000354 decomposition reaction Methods 0.000 description 9
- 238000004590 computer program Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 6
- 238000005070 sampling Methods 0.000 description 6
- 230000009466 transformation Effects 0.000 description 6
- 238000003860 storage Methods 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000016507 interphase Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000739 chaotic effect Effects 0.000 description 1
- 239000003795 chemical substances by application Substances 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000011946 reduction process Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000017105 transposition Effects 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
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/088—Aspects of digital computing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
- Y04S10/52—Outage or fault management, e.g. fault detection or location
Abstract
The application discloses a power distribution network fault identification method and system based on intelligent data fusion analysis, wherein a sensor is used for acquiring various electrical characteristic quantities of acquisition nodes and carrying out fault identification, so that the reliability is high; meanwhile, the method sequentially carries out regional differential processing, expansion and dimension reduction on the acquired electrical characteristic quantity, determines outliers based on an LOF algorithm to predict whether faults occur, determines fault positions if faults occur, carries out wavelet singular entropy calculation on three-phase voltages of the fault positions, inputs calculation results into an SVM fault recognition model, obtains a power distribution network fault phase recognition result, does not need complex setting calculation, and is high in speed.
Description
Technical Field
The application relates to a power distribution network fault identification method and system based on intelligent data fusion analysis, and belongs to the technical field of power distribution network fault early warning analysis.
Background
The traditional medium-voltage distribution network mostly adopts a radiation type structure, and in normal operation, tide flows unidirectionally in the network, and the setting of a protection method is simpler. The power distribution network under the Internet of things has complex network structure and various fault characteristics, and if a single small sample electrical characteristic quantity is used as a fault criterion, the reliability is low; meanwhile, the fault cannot be processed in time after the fault is generated, and after the fault is detected, if the fault is judged, complex setting calculation is needed, so that the speed is low.
Disclosure of Invention
The application provides a power distribution network fault identification method and system based on intelligent data fusion analysis, which solve the problems of low reliability and low speed of the existing method.
In order to solve the technical problems, the application adopts the following technical scheme:
a power distribution network fault identification method based on intelligent data fusion analysis comprises the following steps:
acquiring various electrical characteristic quantities uploaded by acquisition nodes of a power distribution network for fault early warning;
performing regional differential processing on the electrical characteristic quantity according to a pre-constructed network association matrix of the power distribution network to obtain a single-period single-characteristic quantity state monitoring matrix;
expanding the single-period single-feature quantity state monitoring matrix to obtain a high-dimensional space-time state monitoring matrix;
sequentially performing dimension reduction and LOF detection on the high-dimensional space-time state monitoring matrix to obtain LOF values of all the acquisition nodes;
determining a fault position according to the LOF value and a preset rule;
and (3) performing wavelet singular entropy calculation on the three-phase voltage at the fault position, and inputting a calculation result into a pre-trained SVM fault recognition model to obtain a power distribution network fault recognition result.
The electrical characteristic amount includes an electrical characteristic amount of a current type and an electrical characteristic amount of a power type.
The power distribution network association matrix is constructed according to a power distribution network acquisition node network topology structure, and takes the area as a row, the acquisition nodes as columns and the association values of the acquisition nodes and the area as elements.
According to a pre-constructed network association matrix of the power distribution network, carrying out regional differential processing on the electrical characteristic quantity to obtain a single-period single-characteristic quantity state monitoring matrix, wherein the specific process is as follows:
calculating a regional differential matrix of each electrical characteristic quantity according to a column matrix formed by the network association matrix of the power distribution network and the electrical characteristic quantity uploaded by each acquisition node;
and obtaining a single-period single-feature-quantity state monitoring matrix according to the network association matrix of the power distribution network and the regional differential matrix of each electrical feature quantity.
The single-period single-feature quantity state monitoring matrix formula is as follows:
Ri＝AT i
Ci＝A T Ri
wherein A is a network association matrix of a power distribution network, and T is i And (3) a column matrix formed by the electrical characteristic quantities uploaded by the ith acquisition node, wherein Ri is a regional differential matrix of each electrical characteristic quantity, T is a transpose, and Ci is a single-period single-characteristic-quantity state monitoring matrix.
Expanding a single-period single-feature quantity state monitoring matrix to obtain a high-dimensional space-time state monitoring matrix, wherein the specific process is as follows:
expanding the single-period single-feature quantity state monitoring matrix into a single-period multi-feature quantity state monitoring matrix in space;
expanding the single-period multi-feature-quantity state monitoring matrix into a multi-period multi-feature-quantity high-dimensional space-time state monitoring matrix on a time sequence.
The process of performing dimension reduction treatment on the high-dimension space-time state monitoring matrix comprises the following steps:
calculating Euclidean distance among elements of the high-dimensional space-time state monitoring matrix to obtain a similarity matrix among the elements;
calculating a center-to-center inner product matrix according to the similarity matrix;
solving the first two characteristic roots of the central inner product matrix and orthogonalization characteristic vectors corresponding to the characteristic roots, and using the matrix formed by the orthogonalization characteristic vectors as the representation of the high-dimensional space-time state monitoring matrix in a two-dimensional space.
The calculation formula of the LOF value of the acquisition node is as follows:
wherein LOF k (p) LOF value of acquisition node p, N k (p) is the K neighborhood of acquisition node p, o is other acquisition nodes in the K domain of acquisition node p, lrd k (o)、lrd k (p) acquisition respectivelyLocal reachable densities of nodes o, p.
The fault position is determined according to the LOF value, and the specific process is as follows:
responding to the condition that the LOF value of the generalized node does not exceed a setting value, wherein the acquisition node with the maximum LOF value has a sensor fault; the generalized node is a collection node of the whole power distribution network;
in response to the LOF value of the generalized node exceeding a set value, no acquisition node sensor failure occurring, and only one acquisition node LOF exceeding a LOF threshold, a terminal connected to the acquisition node fails.
A power distribution network fault identification system based on intelligent data fusion analysis comprises,
the feature quantity acquisition module is used for: acquiring various electrical characteristic quantities uploaded by acquisition nodes of a power distribution network for fault early warning;
region differential processing module: performing regional differential processing on the electrical characteristic quantity according to a pre-constructed network association matrix of the power distribution network to obtain a single-period single-characteristic quantity state monitoring matrix;
and an expansion module: expanding the single-period single-feature quantity state monitoring matrix to obtain a high-dimensional space-time state monitoring matrix;
the dimension reduction and LOF detection module: sequentially performing dimension reduction and LOF detection on the high-dimensional space-time state monitoring matrix to obtain LOF values of all the acquisition nodes;
a fault location determination module: determining a fault location according to the LOF value;
fault phase identification module: and carrying out wavelet singular entropy calculation on the three-phase voltage at the fault position, and inputting a calculation result into a pre-trained SVM fault recognition model to obtain the fault phase of the power distribution network.
The application has the beneficial effects that:
the application adopts various electrical characteristic quantities to perform fault early warning, and has high reliability; meanwhile, the application sequentially carries out regional differential processing, expansion and dimension reduction on the electrical characteristic quantity, carries out fault early warning and determines the fault position based on an LOF algorithm, carries out wavelet singular entropy calculation on the three-phase voltage of the fault position, inputs the calculation result into an SVM fault recognition model, and obtains the fault recognition result of the power distribution network, without complex setting calculation, and has high speed.
Drawings
FIG. 1 is a flow chart of the method of the present application.
Detailed Description
The application is further described below with reference to the accompanying drawings. The following examples are only for more clearly illustrating the technical aspects of the present application, and are not intended to limit the scope of the present application.
As shown in fig. 1, a power distribution network fault identification method based on intelligent data fusion analysis includes the following steps:
step 1, acquiring various electrical characteristic quantities which are uploaded by acquisition nodes of a power distribution network and used for fault early warning.
And taking a power frequency period as a unit of single-period time, sampling each acquisition node for 32 times in a single period, wherein the types of sampling information are current, power and the like, and uploading the sampling information with a unified time mark after sampling every 30 periods.
The distribution network has single-phase short circuit grounding, two-phase interphase short circuit, two-phase short circuit grounding, acquisition node sensor faults, terminal faults and the like, and the corresponding fault characteristic quantities of different fault types are different. Aiming at fault types to be identified, only part of sampling information is used for subsequent fault identification by combining the actual fault conditions of the low-voltage distribution network, the correlation of a state detection matrix and the efficiency of data fusion among different characteristic quantities, wherein the selected electric characteristic quantities comprise electric characteristic quantities of current types and electric characteristic quantities of power types, in particular three-phase current, negative sequence current, zero sequence current and zero sequence active and reactive power.
And 2, carrying out regional differential processing on the electrical characteristic quantity according to the pre-selected constructed network association matrix of the power distribution network to obtain a single-period single-characteristic quantity state monitoring matrix.
The power distribution network association matrix is constructed according to the network topology structure of the power distribution network acquisition nodes, the power distribution network association matrix can reflect the topology structure of the power distribution network, and the construction process is as follows: firstly, numbering acquisition nodes in a power distribution network; and numbering the areas between the nodes, and finally, generating a network association matrix A of the power distribution network by taking the areas as rows, the acquisition nodes as columns and the association values (shown in a table 1) of the acquisition nodes and the areas as elements. The numbering of the nodes and the areas in the power distribution network has no substantial effect on the network association matrix, so that specific rules do not need to be followed when the nodes and the areas in the power distribution network are numbered.
Table 1 correlation value table
| Correlation value | Association relation |
| 0 | Nodes not at both ends of the area |
| 1 | The node is located at the regional current/power outflow section |
| -1 | The node is located at the regional current/power inflow section |
Since most of the power distribution network time is in a normal running state, most of the power distribution network state monitoring data are normal data; in addition, even if the power distribution network is in a fault state, the difference of the original state monitoring data of the fault node and the normal node is not particularly remarkable, and the fault identification is not facilitated. For this reason, the original state monitoring data needs to be subjected to regional differential processing to enhance the difference between the fault node and the normal node, and the specific processing steps are as follows:
1) And calculating a regional differential matrix of each electrical characteristic quantity according to a column matrix formed by the network association matrix of the power distribution network and the electrical characteristic quantity uploaded by each acquisition node.
Ri＝AT i
Wherein Ri is a regional differential matrix of each electrical characteristic quantity, T i And (3) a column matrix formed by the electrical characteristic quantities uploaded by the ith acquisition node, wherein T is a transposition.
2) And obtaining a single-period single-feature-quantity state monitoring matrix according to the network association matrix of the power distribution network and the regional differential matrix of each electrical feature quantity.
Ci＝A T Ri
Wherein Ci is a single-period single-feature quantity state monitoring matrix.
And step 3, expanding the single-period single-feature-quantity state monitoring matrix (period expansion and feature quantity expansion) to obtain a high-dimensional space-time state monitoring matrix.
The specific process is as follows:
1) Expanding the single-period single-feature quantity state monitoring matrix into a single-period multi-feature quantity state monitoring matrix W in space i″ ＝[C 1 C 2 C 3 …C n ]；
2) Expanding the single-period multi-feature-quantity state monitoring matrix into a multi-period multi-feature-quantity high-dimensional space-time state monitoring matrix W= [ W ] on a time sequence 1 W 2 W 3 …W n ]。
The time length of the state monitoring matrix with single-period and multiple characteristic quantities is one power frequency period, the sensor of the acquisition node in the time period performs 32 times of sampling, and each high-dimensional space-time state monitoring matrix is formed by expanding 30 state monitoring matrices with single-period and multiple characteristic quantities.
And 4, sequentially performing dimension reduction and LOF detection on the high-dimensional space-time state monitoring matrix to obtain LOF values of all the acquisition nodes.
The multidimensional scale analysis is to perform dimension reduction processing on the high-dimensional space-time state monitoring matrix, and the similarity degree of each node is represented according to the relative distance of each node. When the power distribution network normally operates, the inter-row difference degree of the constructed high-dimensional space-time state monitoring matrix is small, and outliers are not present. When the power distribution network is in fault, the difference between the fault node and the non-fault node is large, and a small number of nodes deviate from normal nodes to form outliers.
The multi-dimensional dimension reduction process for the high-dimensional space-time state monitoring matrix is as follows:
1) Calculating Euclidean distance among elements of the high-dimensional space-time state monitoring matrix to obtain a similarity matrix D among the elements;
wherein d i″′j″′ X is an element in the similarity matrix D i″′k 、x j″′k The kth element of the ith and jth rows, respectively, of the high-dimensional spatiotemporal state monitoring matrix.
2) Calculating a center-to-center inner product matrix according to the similarity matrix;
3) Solving the first two feature roots of the central inner transformation product matrix and orthogonalization feature vectors corresponding to the feature roots, and enabling the two feature roots to be in a same shape as the central inner transformation product matrixMatrix x= (X) composed of orthogonalized eigenvectors (1) ,x (2) ) As a representation of the high-dimensional space-time state monitoring matrix in two-dimensional space; wherein x is (i′) The orthogonalization feature vector corresponding to the ith feature root is represented by T, which is the transpose.
The outlier recognition method based on the LOF detection algorithm can be used for recognizing outliers of the set with different density dispersion conditions, and has high sensitivity to outliers. Therefore, fault early warning can be carried out according to the LOF value, and the fault position can be further determined. When the power distribution network normally operates, no outlier exists, and the LOF value of each node is approximately 1; when the distribution network fails, the LOF value of the outlier is large, and the LOF value of the normal node is still maintained to be about 1. Therefore, fault localization can be further performed according to the magnitude of the LOF value. The procedure for LOF detection was as follows:
1) Obtaining the distance between the acquisition node p and the nearest node, i.e. K distance K dist (p)；
2) Calculating a K neighborhood of the acquisition node p;
N k (p)＝{q∈N/{p}|dist(p,q)≤K dist (p)}
wherein dist (p, q) is the spatial distance between the acquisition node p and the acquisition node q; n represents the whole collection node level, and q E N/{ p } means that q is all nodes except p.
3) Determining local reach distances of nodes k (p,o)：
reach-dist k (p,o)＝max{K dist (o)-dist(p,o)}
Wherein dist (p, o) is the spatial distance between acquisition node p and acquisition node o, K dist (o) is the distance between the acquisition node o and the nearest node;
4) Calculating local reachable density of each node;
wherein o is the K field N of the acquisition node p k Other acquisition nodes within (p), lrd k (p) local reachable density of acquisition node p
5) Solving local abnormal factors of each acquisition node object, namely LOF values;
wherein LOF k (p) LOF value of acquisition node p, N k (p) K neighborhood of acquisition node lrd k (o) is the local reachable density of acquisition node o.
Step 5, determining a fault position according to the LOF value; the method comprises the following steps: and determining outliers according to the LOF value, judging whether faults occur, and further determining the fault position if the faults occur.
Faults mainly include acquisition node sensor faults and terminal faults.
Acquisition node sensor failure: and taking the whole power distribution network part as an acquisition node, namely a generalized node, and under the condition of single type (power and sensor) fault, if and only if a power system fault occurs in an area in the generalized node of the power distribution network, the LOF value of the generalized node (the LOF value of the generalized node is calculated in the same way as other single nodes) exceeds a setting value. In addition, when the sensor of the acquisition node fails at the non-boundary node of the power distribution network, the LOF value of the failure node is the largest, so that whether the sensor failure exists at the acquisition node can be judged according to whether the LOF value of the generalized node exceeds a setting value. And responding to the condition that the LOF value of the generalized node does not exceed the setting value, and the acquisition node with the maximum LOF value is subjected to sensor faults.
Terminal (i.e., power end) failure: only one of the nearest acquisition nodes connected with the terminal is provided, so that if the LOF value of the generalized node exceeds a setting value, no acquisition node sensor fault occurs, and only one acquisition node LOF exceeds an LOF threshold (i.e. is large), the terminal fault connected with the acquisition node can be judged.
And 6, performing wavelet singular entropy calculation on the three-phase voltage at the fault position, and inputting a calculation result into a pre-trained SVM fault recognition model to obtain a recognition result of the fault phase of the power distribution network.
After wavelet transformation is carried out on the three-phase voltage, the fault phases under different faults are not sufficiently identified by means of wavelet time-frequency characteristics, the wavelet coefficient matrix after wavelet transformation is considered, singular value decomposition is carried out, finally the decomposed diagonal matrix is processed into a probability distribution sequence, the wavelet singular entropy value is calculated, and the wavelet singular entropy value of the sequence is used for reflecting the fault phases of the three-phase voltage signals.
The wavelet coefficient low-frequency component and the wavelet coefficient high-frequency component after m-layer decomposition of the signal U (n) are A respectively m (n) and B m (n). The wavelet low frequency energy and the wavelet high frequency energy corresponding to the single scale are E respectively L ＝||A m (n)|| 2 And E is H ＝||B m (n)|| 2 (m＝1，2，…，M，Where M is the number of wavelet decomposition layers). The m components of the signal U (n) may form an m n wavelet coefficient matrix W which may be expressed as a singular value decomposition
W＝USV T
Wherein U and V are m×m and n×n order orthogonal matrices, respectively, S is a generalized diagonal matrix, and diagonal elements thereof are arranged in descending order.
After the wavelet coefficient matrix W of the three-phase voltage signal is decomposed by singular values, the diagonal matrix can well reflect the time-frequency distribution characteristics of fault signals. Since elements close to zero hardly reflect the characteristics of the fault signal, it is negligible in order to reduce the amount of calculation.
Information entropy is a measure of the amount of information and is also a measure of the amount of system information. If the system is ordered, the system entropy is lower; if the system is chaotic, the system entropy is high. Therefore, the order degree of the system can be well quantified and counted by utilizing the information entropy theory. After singular value decomposition is carried out on the three-phase voltage, if diagonal elements are more average, the system fault characteristics are more obvious, and the system entropy is larger; otherwise, the less the signal fault amount is, or no fault signal is included, the smaller the system entropy value is. Wavelet singular entropy values are products based on a combination of wavelet transforms, singular value decomposition, and information entropy. Carrying out singular value decomposition on a coefficient matrix of the three-phase voltage after wavelet transformation, wherein the singular value can reflect the basic characteristics of the original three-phase voltage signal; then the statistical characteristic of the information entropy is utilized to process the diagonal matrix after decomposition into a probability distribution sequence and calculate the wavelet singular entropy value thereof, and the wavelet singular entropy value is used as the fault characteristic quantity to reflect the fault phase of the system
According to definition of information entropy, it is assumed that characteristic value x of signal source j The probability of (2) is P j ＝P{X＝x j } (j=1, 2,.,. And L), andthe information entropy is:
according to wavelet analysis and singular value decomposition theory, the wavelet singular entropy calculation formula is as follows:
where λi represents the ith element value, K and L are the number of diagonal elements, here the first 9 valid diagonal elements are selected.
The SVM support vector machine can better solve the nonlinear problem, and is particularly suitable for small sample analysis. The basic principle is that the optimal classification surface is searched to maximize the classification gap of the linearly separable samples. For the linear inseparable problem, a kernel function is generally introduced, a low-dimensional nonlinear inseparable sample is mapped to a certain high-dimensional linear inseparable space, an optimal classification surface is sought in the transformed high-dimensional space, and linear classification after nonlinear transformation is realized by introducing the kernel function, but the computational complexity is not increased. The wavelet singular entropy of the three-phase voltage signal is taken as a fault characteristic quantity, a certain number of samples are obtained through a large number of simulations to form a characteristic sample space, and an SVM fault recognition model is established to rapidly identify fault phases of a fault area. And an SVM classification model is established according to the existing data, and the accuracy rate can reach 99.67%.
Take ground fault as an example: and (3) performing wavelet singular entropy calculation on the three-phase voltage at the fault position, inputting a calculation result into a pre-trained SVM fault recognition model to obtain a zero-sequence voltage low-frequency signal energy value, and taking the zero-sequence voltage low-frequency signal energy reflecting ground fault information as an electrical characteristic quantity for judging whether a ground fault occurs or not, wherein when the ground fault occurs, the zero-sequence voltage low-frequency signal energy exceeds an energy threshold (namely, is quite large). As shown in table 2, when the fault type contains G, that is, a ground fault occurs, the zero sequence voltage low frequency signal energy is large; when no ground fault occurs, the zero sequence voltage low frequency signal energy is small.
TABLE 2 zero sequence voltage first layer low frequency energy for different fault types
| Fault type | Zero sequence voltage first layer low frequency energy |
| AG | 3.1275×10 10 |
| ABG | 6.9872×10 10 |
| AB | 1.6589×10 -3 |
| ABC | 3.9873×10 -3 |
The method is based on a high-dimensional space-time state monitoring matrix formed by fusing a plurality of characteristic quantity data when the identification is carried out, only the LOF value of the local abnormal factor of the fault node is reduced when a scene with zero part of characteristic quantity (zero sequence component when the interphase short circuit occurs) appears in actual use, and the fault can still be reliably identified.
The method adopts various electrical characteristic quantities to perform fault early warning, and has high reliability; meanwhile, the application sequentially carries out regional differential processing, expansion and dimension reduction on the electrical characteristic quantity, carries out fault early warning and determines the fault position based on an LOF algorithm, carries out wavelet singular entropy calculation on the three-phase voltage of the fault position, inputs the calculation result into an SVM fault recognition model, and obtains the fault recognition result of the power distribution network, without complex setting calculation, and has high speed.
A power distribution network fault identification system based on intelligent data fusion analysis comprises,
the feature quantity acquisition module is used for: acquiring various electrical characteristic quantities uploaded by acquisition nodes of a power distribution network for fault early warning;
region differential processing module: performing regional differential processing on the electrical characteristic quantity according to a pre-constructed network association matrix of the power distribution network to obtain a single-period single-characteristic quantity state monitoring matrix;
and an expansion module: expanding the single-period single-feature quantity state monitoring matrix to obtain a high-dimensional space-time state monitoring matrix;
dimension reduction LOF detection module: performing dimension reduction and LOF detection on the high-dimensional space-time state monitoring matrix in sequence to obtain LOF values of all the acquisition nodes, and performing detection and early warning on faults;
a fault location determination module: determining a fault location according to the LOF value;
and an identification module: and carrying out wavelet singular entropy calculation on the three-phase voltage at the fault position, and inputting a calculation result into a pre-trained SVM fault recognition model to obtain a power distribution network fault recognition result.
A computer readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by a computing device, cause the computing device to perform a power distribution network fault warning method.
A computing device comprising one or more processors, memory, and one or more programs, wherein one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs comprising instructions for performing a power distribution network fault warning method.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The foregoing is illustrative of the present application and is not to be construed as limiting thereof, but rather as providing for the use of additional embodiments and advantages of all such modifications, equivalents, improvements and similar to the present application are intended to be included within the scope of the present application as defined by the appended claims.
Claims (9)
1. The power distribution network fault identification method based on intelligent data fusion analysis is characterized by comprising the following steps of:
acquiring various electrical characteristic quantities uploaded by acquisition nodes of a power distribution network for fault identification;
performing regional differential processing on the electrical characteristic quantity according to a pre-constructed network association matrix of the power distribution network to obtain a single-period single-characteristic quantity state monitoring matrix;
expanding the single-period single-feature quantity state monitoring matrix to obtain a high-dimensional space-time state monitoring matrix;
sequentially performing dimension reduction and LOF detection on the high-dimensional space-time state monitoring matrix to obtain LOF values of all the acquisition nodes;
the fault position is determined according to the LOF value, and the specific process is as follows:
responding to the condition that the LOF value of the generalized node does not exceed a setting value, wherein the acquisition node with the maximum LOF value has a sensor fault; the generalized node is a collection node of the whole power distribution network; responsive to the LOF value of the generalized node exceeding a set value, no acquisition node sensor failure occurring, and only one acquisition node LOF exceeding an LOF threshold, a terminal failure connected to the acquisition node;
and (3) performing wavelet singular entropy calculation on the three-phase voltage at the fault position, and inputting a calculation result into a pre-trained SVM fault recognition model to obtain a power distribution network fault recognition result.
2. The power distribution network fault identification method based on intelligent data fusion analysis according to claim 1, wherein the method comprises the following steps: the electrical characteristic amount includes an electrical characteristic amount of a current type and an electrical characteristic amount of a power type.
3. The power distribution network fault identification method based on intelligent data fusion analysis according to claim 1, wherein the method comprises the following steps: the power distribution network association matrix is constructed according to a power distribution network acquisition node network topology structure, and takes the area as a row, the acquisition nodes as columns and the association values of the acquisition nodes and the area as elements.
4. The power distribution network fault identification method based on intelligent data fusion analysis according to claim 1, wherein the method comprises the following steps: according to a pre-constructed network association matrix of the power distribution network, carrying out regional differential processing on the electrical characteristic quantity to obtain a single-period single-characteristic quantity state monitoring matrix, wherein the specific process is as follows:
calculating a regional differential matrix of each electrical characteristic quantity according to a column matrix formed by the network association matrix of the power distribution network and the electrical characteristic quantity uploaded by each acquisition node;
and obtaining a single-period single-feature-quantity state monitoring matrix according to the network association matrix of the power distribution network and the regional differential matrix of each electrical feature quantity.
5. The power distribution network fault identification method based on intelligent data fusion analysis according to claim 4, wherein the method comprises the following steps: the single-period single-feature quantity state monitoring matrix formula is as follows:
Ri＝AT i
Ci＝A T Ri
wherein A is a network association matrix of a power distribution network, and T is i And (3) a column matrix formed by the electrical characteristic quantities uploaded by the ith acquisition node, wherein Ri is a regional differential matrix of each electrical characteristic quantity, T is a transpose, and Ci is a single-period single-characteristic-quantity state monitoring matrix.
6. The power distribution network fault identification method based on intelligent data fusion analysis according to claim 1, wherein the method comprises the following steps: expanding a single-period single-feature quantity state monitoring matrix to obtain a high-dimensional space-time state monitoring matrix, wherein the specific process is as follows:
expanding the single-period single-feature quantity state monitoring matrix into a single-period multi-feature quantity state monitoring matrix in space;
expanding the single-period multi-feature-quantity state monitoring matrix into a multi-period multi-feature-quantity high-dimensional space-time state monitoring matrix on a time sequence.
7. The power distribution network fault identification method based on intelligent data fusion analysis according to claim 1, wherein the method comprises the following steps: the process of performing dimension reduction treatment on the high-dimension space-time state monitoring matrix comprises the following steps:
calculating Euclidean distance among elements of the high-dimensional space-time state monitoring matrix to obtain a similarity matrix among the elements;
calculating a center-to-center inner product matrix according to the similarity matrix;
solving the first two characteristic roots of the central inner product matrix and orthogonalization characteristic vectors corresponding to the characteristic roots, and using the matrix formed by the orthogonalization characteristic vectors as the representation of the high-dimensional space-time state monitoring matrix in a two-dimensional space.
8. The power distribution network fault identification method based on intelligent data fusion analysis according to claim 1, wherein the method comprises the following steps: the calculation formula of the LOF value of the acquisition node is as follows:
wherein LOF k (p) LOF value of acquisition node p, N k (p) is the K neighborhood of acquisition node p, o is other acquisition nodes in the K domain of acquisition node p, lrd k (o)、lrd k (p) is the local reachable density of the acquisition nodes o, p, respectively.
9. A power distribution network fault identification system based on intelligent data fusion analysis is characterized in that: comprising the steps of (a) a step of,
the feature quantity acquisition module is used for: acquiring various electrical characteristic quantities uploaded by acquisition nodes of a power distribution network for fault early warning;
region differential processing module: performing regional differential processing on the electrical characteristic quantity according to a pre-constructed network association matrix of the power distribution network to obtain a single-period single-characteristic quantity state monitoring matrix;
and an expansion module: expanding the single-period single-feature quantity state monitoring matrix to obtain a high-dimensional space-time state monitoring matrix;
the dimension reduction and LOF detection module: sequentially performing dimension reduction and LOF detection on the high-dimensional space-time state monitoring matrix to obtain LOF values of all the acquisition nodes;
a fault location determination module: the fault position is determined according to the LOF value, and the specific process is as follows:
responding to the condition that the LOF value of the generalized node does not exceed a setting value, wherein the acquisition node with the maximum LOF value has a sensor fault; the generalized node is a collection node of the whole power distribution network; responsive to the LOF value of the generalized node exceeding a set value, no acquisition node sensor failure occurring, and only one acquisition node LOF exceeding an LOF threshold, a terminal failure connected to the acquisition node;
fault phase identification module: and carrying out wavelet singular entropy calculation on the three-phase voltage at the fault position, and inputting a calculation result into a pre-trained SVM fault recognition model to obtain the fault phase of the power distribution network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010787350.2A CN111965486B (en) | 2020-08-07 | 2020-08-07 | Power distribution network fault identification method and system based on intelligent data fusion analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010787350.2A CN111965486B (en) | 2020-08-07 | 2020-08-07 | Power distribution network fault identification method and system based on intelligent data fusion analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111965486A CN111965486A (en) | 2020-11-20 |
| CN111965486B true CN111965486B (en) | 2023-12-15 |
Family
ID=73365206
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010787350.2A Active CN111965486B (en) | 2020-08-07 | 2020-08-07 | Power distribution network fault identification method and system based on intelligent data fusion analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111965486B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113702760B (en) * | 2021-08-26 | 2023-08-25 | 济南大学 | Method and system for identifying transverse faults and ferromagnetic resonance states of distribution line |
| CN114048815B (en) * | 2021-11-12 | 2023-11-28 | 国网江苏省电力有限公司常州供电分公司 | Power grid operation information sensing system and sensing method based on plant side |
| CN114325225A (en) * | 2021-12-16 | 2022-04-12 | 国网江苏省电力有限公司盐城供电分公司 | Method for early warning and positioning faults of decentralized low-voltage power distribution network |
| CN116244964B (en) * | 2023-03-27 | 2023-10-24 | 国网河南省电力公司电力科学研究院 | Power distribution network storm disaster power failure prediction method based on numerical simulation and SVD model |
| CN117458722B (en) * | 2023-12-26 | 2024-03-08 | 西安民为电力科技有限公司 | Data monitoring method and system based on electric power energy management system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104615244A (en) * | 2015-01-23 | 2015-05-13 | 深圳大学 | Automatic gesture recognizing method and system |
| CN105322519A (en) * | 2015-11-02 | 2016-02-10 | 湖南大学 | Big data fusion analysis and running state monitoring method for intelligent power distribution network |
| CN106093703A (en) * | 2016-06-07 | 2016-11-09 | 湖南大学 | The identification of a kind of intelligent distribution network fault and localization method |
| CN106124929A (en) * | 2016-06-30 | 2016-11-16 | 湖南大学 | A kind of power distribution network physical fault and information fault identification method |
| CN106250905A (en) * | 2016-07-08 | 2016-12-21 | 复旦大学 | A kind of real time energy consumption method for detecting abnormality of combination colleges and universities building structure feature |
| CN106249101A (en) * | 2016-06-30 | 2016-12-21 | 湖南大学 | A kind of intelligent distribution network fault identification method |
-
2020
- 2020-08-07 CN CN202010787350.2A patent/CN111965486B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104615244A (en) * | 2015-01-23 | 2015-05-13 | 深圳大学 | Automatic gesture recognizing method and system |
| CN105322519A (en) * | 2015-11-02 | 2016-02-10 | 湖南大学 | Big data fusion analysis and running state monitoring method for intelligent power distribution network |
| CN106093703A (en) * | 2016-06-07 | 2016-11-09 | 湖南大学 | The identification of a kind of intelligent distribution network fault and localization method |
| CN106124929A (en) * | 2016-06-30 | 2016-11-16 | 湖南大学 | A kind of power distribution network physical fault and information fault identification method |
| CN106249101A (en) * | 2016-06-30 | 2016-12-21 | 湖南大学 | A kind of intelligent distribution network fault identification method |
| CN106250905A (en) * | 2016-07-08 | 2016-12-21 | 复旦大学 | A kind of real time energy consumption method for detecting abnormality of combination colleges and universities building structure feature |
Non-Patent Citations (4)
| Title |
|---|
| 基于LOF和SVM 的智能配电网故障辨识方法;胡伟;《电力自动化设备》;7-12页 * |
| 基于LOF和SVM的智能配电网故障辨识方法;胡伟;《电力自动化设备》;20160630;第7-12页 * |
| 基于智能配电网大数据分析的状态监测与故障处理方法;赵庆周;《电网技术》;774-780页 * |
| 胡伟.基于LOF和SVM的智能配电网故障辨识方法.《电力自动化设备》.2016, * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111965486A (en) | 2020-11-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111965486B (en) | Power distribution network fault identification method and system based on intelligent data fusion analysis | |
| Charalampidis | A modified k-means algorithm for circular invariant clustering | |
| Du et al. | Shape recognition based on radial basis probabilistic neural network and application to plant species identification | |
| CN109815934B (en) | Multi-scale OGLPE (one glass solution) feature extraction method applied to inverter faults | |
| Liu et al. | Short-term local prediction of wind speed and wind power based on singular spectrum analysis and locality-sensitive hashing | |
| Mukherjee et al. | Application of principal component analysis for fault classification in transmission line with ratio-based method and probabilistic neural network: a comparative analysis | |
| CN108764541B (en) | Wind energy prediction method combining space characteristic and error processing | |
| CN111275069B (en) | Non-invasive load monitoring method | |
| CN110503113B (en) | Image saliency target detection method based on low-rank matrix recovery | |
| CN116756594A (en) | Method, system, equipment and medium for detecting abnormal points of power grid data | |
| Krawczyk et al. | Experiments on distance measures for combining one-class classifiers | |
| Ren et al. | A comprehensive study of sparse codes on abnormality detection | |
| He et al. | Partial discharge pattern recognition algorithm based on sparse self-coding and extreme learning machine | |
| Tito et al. | Image segmentation-based event detection for non-intrusive load monitoring using Gramian Angular Summation Field | |
| Guo et al. | Image registration method based on improved SIFT algorithm and essential matrix estimation | |
| Zhou et al. | Method of Series Arc Fault Detection Based on Phase Space Reconstruction and Convolutional Neural Network | |
| CN111626329A (en) | Insulation pipe bus fault diagnosis method based on LDA optimization multi-scale texture features | |
| Delmas et al. | Asymptotic performance analysis of subspace adaptive algorithms introduced in the neural network literature | |
| Mahmud et al. | Fault classification in transmission line using wavelet features and fuzzy-knn | |
| Wang et al. | A Novel Feature Extraction Method for Soft Faults in Nonlinear Analog Circuits Based on LMD-GFD and KPCA | |
| CN107122755A (en) | Pedestrian's similarity calculating method and device | |
| TIAN et al. | A method for power system static stability situation assessment based on scale invariant feature transform | |
| Yang et al. | Extraction of symptom for on-line diagnosis of power equipment based on method of time series analysis | |
| Liu et al. | New media interaction in art design based on deep learning binocular stereo vision | |
| FUa et al. | Research on Transformer Fault Diagnosis in Extreme Weather Based on MDT-AdaBoost Algorithm |
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 |