Power distribution network online early warning method and system based on improved SVDD

Through the improved SVDD method and fuzzy membership function, a super-sphere model of the distribution network was constructed, which solved the problem of inaccurate boundary identification caused by insufficient data in the distribution network, achieved more accurate decision-making boundary identification and fault warning, and improved distribution The operational state perception capability of the power grid.

CN119025971APending Publication Date: 2024-11-26STATE GRID BEIJING ELECTRIC POWER CO +1
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411145689.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-20
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

Insufficient data in the distribution network under abnormal operating state leads to inaccurate boundary identification. The existing technology has problems such as large data calculation volume, strong subjectivity and immutable threshold, making it difficult to effectively perceive abnormal changes in the distribution network operating state.

Method used

Using the improved support vector data description (SVDD) method, the distribution network power measurement data is divided by a fuzzy membership function, mapped to the high-dimensional feature space, and constructed a target hyperspheric model to identify decision boundaries, and Determine whether to issue early warning information based on the positional relationship between the operating point to be detected and the decision boundary.

Benefits of technology

It improves the accuracy of identification of decision-making boundaries, can promptly determine whether the operating point exceeds the boundaries or there is a breakthrough trend, provides effective online warning, and enhances the ability to prevent distribution network faults.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119025971A_ABST
    Figure CN119025971A_ABST
Patent Text Reader

Abstract

The invention discloses a power distribution network online early warning method and system based on improved SVDD, and relates to the technical field of power distribution network abnormity early warning, and the method comprises the steps: carrying out the positive and negative sample division of a power measurement data set of a power distribution network, and obtaining a target sample; based on a fuzzy membership function, calculating a membership value of each sample in the target samples; based on a support vector data description method, mapping a target sample to a high-dimensional feature space; based on the membership value of each sample, constructing a target hypersphere model in the high-dimensional feature space, and taking the boundary of the target hypersphere model as the decision boundary of the power distribution network; and based on the position relation between the to-be-detected operation point and the power distribution network decision boundary, whether early warning information is sent is judged. The technical problem that early warning boundary recognition of a traditional classifier is limited due to insufficient abnormal operation sample data in the prior art is relieved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of abnormal early warning of distribution network, and in particular to an online early warning method and system for distribution network based on improved SVDD. Background Art

[0002] At present, the research on fault self-prevention of distribution network is gradually improving and perfecting. A model for sensing the operating status of distribution network and combining the operating trend for online early warning has been established, as well as an evaluation index system for online early warning model. However, these index evaluation systems have problems such as large amount of data calculation, subjectivity and immutable thresholds. In addition, in the context of new power systems, the operating conditions of distribution networks change frequently, and the ability to perceive whether their operating status is abnormal needs to be improved. The early warning boundary can reduce the subjectivity of the indicator threshold to a certain extent, shorten the data calculation time, and more intuitively reflect the operating status of the distribution network. Therefore, it is particularly important to build an online early warning model that can sense the operating status of the distribution network.

[0003] The power measurement data collected by the SCADA and PMU of the distribution network include node voltage amplitude, node voltage phase angle, line current amplitude, node active power and reactive power, line active power and reactive power, etc. These data need to be normalized. In the daily operation of the distribution network, most of them operate in normal state, and there are few abnormal operation states. Therefore, the sample data of abnormal operation state that can be collected is relatively small, which greatly limits the application of traditional classifiers in early warning boundary identification. Summary of the invention

[0004] The purpose of the present invention is to provide a distribution network online early warning method and system based on improved SVDD in order to solve at least one of the above technical problems.

[0005] In the first aspect, an embodiment of the present invention provides an online early warning method for a distribution network based on an improved SVDD, comprising: dividing the power measurement data set of the distribution network into positive and negative samples to obtain target samples; the target samples include positive samples and negative samples; based on a fuzzy membership function, calculating the membership value of each sample in the target samples; based on a method of support vector data description, mapping the target samples to a high-dimensional feature space; based on the membership value of each sample, constructing a target hypersphere model in the high-dimensional feature space, and using the boundary of the target hypersphere model as a distribution network decision boundary; based on the positional relationship between the operating point to be detected and the distribution network decision boundary, determining whether to issue an early warning message.

[0006] Further, based on the fuzzy membership function, the membership value of each sample in the target sample is calculated, including:

[0007]

[0008] Among them, ρ i is the relative density of the ith positive sample, z i is the membership value of the i-th positive sample, ρ max and ρ min are the maximum and minimum values ​​of the relative density of positive samples, respectively.

[0009] Furthermore, the target hypersphere model comprises:

[0010]

[0011] Among them, x i represents positive samples, N1 represents the number of positive samples, z i represents the membership value of the positive sample, x l represents negative samples, N2 represents the number of negative samples, z l represents the membership value of the negative sample, ξ i and l is the relaxation factor; C1 represents the error penalty coefficient for adjusting the number of samples inside the hypersphere and the algorithm complexity; C2 represents the error penalty coefficient for adjusting the number of samples outside the hypersphere and the algorithm complexity; is a nonlinear mapping of x, γ represents a Gaussian kernel parameter, F represents the high-order feature space, R represents the radius of the hypersphere, and a represents the center of the hypersphere.

[0012] Furthermore, based on the positional relationship between the operating point to be detected and the decision boundary of the distribution network, it is determined whether to issue a warning message, including: determining whether the operating point to be detected is outside the decision boundary of the distribution network; if so, issuing a warning message; if not, determining whether the operating point to be detected has a trend of breaking through the decision boundary of the distribution network, and if so, issuing a warning message.

[0013] Further, determining whether the operating point to be detected is outside the decision boundary of the distribution network includes: calculating a target distance between the operating point to be detected and the center of the target hypersphere model; determining whether the target distance is greater than the radius of the target hypersphere model; if so, determining that the operating point to be detected is outside the decision boundary of the distribution network.

[0014] In the second aspect, a distribution network online early warning system based on improved SVDD is also provided, including: a division module, a calculation module, a mapping module, a construction module and an early warning module; wherein the division module is used to divide the power measurement data set of the distribution network into positive and negative samples to obtain target samples; the target samples include positive samples and negative samples; the calculation module is used to calculate the membership value of each sample in the target sample based on a fuzzy membership function; the mapping module is used to map the target sample to a high-dimensional feature space based on a support vector data description method; the construction module is used to construct a target hypersphere model in the high-dimensional feature space based on the membership value of each sample, and use the boundary of the target hypersphere model as the distribution network decision boundary; the early warning module is used to determine whether to issue an early warning message based on the positional relationship between the operating point to be detected and the distribution network decision boundary.

[0015] Furthermore, the target hypersphere model comprises:

[0016]

[0017] Among them, x i represents positive samples, N1 represents the number of positive samples, z i represents the membership value of the positive sample, x l represents negative samples, N2 represents the number of negative samples, z l represents the membership value of the negative sample, ξ i and l is the relaxation factor; C1 represents the error penalty coefficient for adjusting the number of samples inside the hypersphere and the algorithm complexity; C2 represents the error penalty coefficient for adjusting the number of samples outside the hypersphere and the algorithm complexity; is a nonlinear mapping of x, γ represents a Gaussian kernel parameter, F represents the high-order feature space, R represents the radius of the hypersphere, and a represents the center of the hypersphere.

[0018] Furthermore, the early warning module is also used to: determine whether the operating point to be detected is outside the decision boundary of the distribution network; if so, issue an early warning message; if not, determine whether the operating point to be detected has a trend of breaking through the decision boundary of the distribution network, and if so, issue an early warning message.

[0019] In a third aspect, an embodiment of the present invention further provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method described in the first aspect when executing the computer program.

[0020] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, and when the computer instructions are executed by a processor, the method described in the first aspect above is implemented.

[0021] The present invention provides an online early warning method and system for a distribution network based on improved SVDD, which solves the problem of inaccurate boundary identification caused by insufficient abnormal operating state data and improves the accuracy of decision boundary identification; after obtaining the online early warning boundary, it is determined whether to give an alarm signal based on the distance relationship between the operating point and the center of the hypersphere and the movement trend of the operating point toward the boundary, which has certain practical value for the self-prevention of faults in the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings:

[0023] Figure 1 A flow chart of an online early warning method for a distribution network based on improved SVDD provided by an embodiment of the present invention;

[0024] Figure 2 A schematic diagram of a topological structure of a power distribution network system provided by an embodiment of the present invention;

[0025] Figure 3 A predicted power value curve diagram of wind power, photovoltaic power and load provided in an embodiment of the present invention;

[0026] Figure 4 A data distribution diagram in a two-dimensional coordinate system provided by an embodiment of the present invention;

[0027] Figure 5 An operation trend diagram of an operating point in the time period of 00:00-11:00 provided by an embodiment of the present invention;

[0028] Figure 6 An operation trend diagram of an operating point in the time period of 11:00-12:30 provided by an embodiment of the present invention;

[0029] Figure 7 An operation trend diagram of an operation point in the time period of 12:30-14:30 provided by an embodiment of the present invention;

[0030] Figure 8 An operation trend diagram of an operating point in the time period of 16:30-24:00 provided by an embodiment of the present invention;

[0031] Fig. 9A schematic diagram of an online early warning system for a distribution network based on an improved SVDD is provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0032] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that the embodiments and features in the embodiments of the present invention can be combined with each other without conflict.

[0033] The following detailed description is an exemplary description, which is intended to provide further detailed description of the present invention. Unless otherwise specified, all technical terms used in the present invention have the same meaning as those generally understood by those skilled in the art to which the present invention belongs. The terms used in the present invention are only for describing specific embodiments, and are not intended to limit the exemplary embodiments according to the present invention.

[0034] Embodiment 1

[0035] Figure 1 FIG. 1 is a flow chart of an online early warning method for a distribution network based on an improved SVDD according to an embodiment of the present invention. Figure 1 As shown, the method specifically comprises the following steps:

[0036] Step S102, dividing the power measurement data set of the distribution network into positive and negative samples to obtain target samples; the target samples include positive samples and negative samples.

[0037] For example, positive and negative samples are divided, positive samples are marked as +1, and negative samples are marked as -1.

[0038] Step S104, calculating the membership value of each sample in the target sample based on the fuzzy membership function.

[0039] Step S106, mapping the target sample to a high-dimensional feature space based on a method of support vector data description.

[0040] Step S108, based on the membership value of each sample, construct a target hypersphere model in the high-dimensional feature space, and use the boundary of the target hypersphere model as the distribution network decision boundary.

[0041] Step S110, judging whether to issue warning information based on the position relationship between the operating point to be detected and the decision boundary of the distribution network.

[0042] In the embodiment of the present invention, the basic idea of ​​support vector data description (SVDD) is to map the target sample to a high-dimensional feature space, and construct a hypersphere with the smallest volume that can contain all or most of the target class samples in this space. The samples inside and on the surface of the hypersphere are positive samples, and the samples outside the hypersphere are negative samples.

[0043] For a data set x i =1,2,....N, the slack variable ζ and penalty function C are introduced into the SVDD model, and its specific mathematical function can be expressed as:

[0044]

[0045] Among them, a is the center of the hypersphere, R is its radius, ζ is to adapt to potential training errors, and C is the weight parameter for adjusting training errors.

[0046] The constraints are:

[0047] ||φ(x i -a)|| 2 ≤R 2 +ζ i , i ≥0,i=1,2....,N (2)

[0048] Where, φ: R D →H is the characteristic map of Hilbert space (RKHS).

[0049] The obtained Lagrange function is as follows:

[0050]

[0051] L to R, a, ζ i Taking partial derivatives and setting them to zero, we get:

[0052]

[0053] The problem now turns into:

[0054]

[0055] The constraints are:

[0056]

[0057] Among them, K(·) is the introduced kernel function.

[0058] Solve equation (6) to calculate the Lagrange multiplication factor α i , the decision function at this time is:

[0059] f(x)=sign(R 2 -||φ(x)-a|| 2 ) (7)

[0060] The radius R of the hypersphere model is:

[0061]

[0062] Among them, x sv is the support vector, and the value of the center a of the hypersphere is obtained by solving. At this time, the samples can be classified according to the distance between the sample point and the center of the hypersphere:

[0063] When R 2 >||xa|| 2 →When α=0, it means that the distance from the sample to be tested to the center of the sphere is less than the radius of the hypersphere, and the sample point is inside the hypersphere. The sample is a reserved vector;

[0064] When R 2 =||xa|| 2 →When 0<α<C, it means that the distance from the sample to be tested to the center of the sphere is exactly equal to the radius of the hypersphere, and the sample point is on the surface of the hypersphere. The sample is a boundary support vector;

[0065] When R 2 <||xa|| 2 →When α=C, it means that the distance from the sample to be tested to the center of the sphere is greater than the radius of the hypersphere, and the sample point is outside the hypersphere. The sample is the error support vector.

[0066] The present invention selects Gaussian kernel function K(·) as the mapping transformation function. The positive sample relative density ρ of the kernel density estimation function i for:

[0067]

[0068]

[0069]

[0070] Where: γ is the Gaussian kernel parameter; ω is the weight; s is the kernel density estimation smoothing parameter. i When the larger the positive sample point x i The more densely populated the area.

[0071] However, in order to ensure the accuracy of decision boundary fitting and reduce the influence of bad measurement data on the training results of SVDD, the present invention introduces a fuzzy membership function to improve the SVDD algorithm.

[0072] Specifically, the fuzzy membership function is used to calculate the fuzzy membership value corresponding to each sample, and the importance of each sample is judged according to the size of the sample fuzzy membership value, wherein the importance of the sample is directly proportional to the size of its fuzzy membership function value.

[0073] Specifically, the membership value includes:

[0074]

[0075] Among them, ρ i is the relative density of the ith positive sample, z i is the membership value of the i-th positive sample, ρ max and ρ min are the maximum and minimum values ​​of the relative density of positive samples, respectively.

[0076] According to formula (12), the membership value corresponding to each sample can be calculated one by one, thereby reflecting the importance of each sample. In this way, when fitting the decision boundary, samples with high relative density are more likely to be surrounded by the hypersphere.

[0077] Specifically, the target hypersphere model includes:

[0078]

[0079] Among them, x i represents positive samples, N1 represents the number of positive samples, z i represents the membership value of the positive sample, x l represents negative samples, N2 represents the number of negative samples, z l represents the membership value of the negative sample, ξ i and l is the relaxation factor; C1 represents the error penalty coefficient for adjusting the number of samples inside the hypersphere and the algorithm complexity; C2 represents the error penalty coefficient for adjusting the number of samples outside the hypersphere and the algorithm complexity; is the nonlinear mapping of x, γ represents the Gaussian kernel parameter, F represents the high-order feature space, R represents the radius of the hypersphere, and a represents the center of the hypersphere.

[0080] Specifically, step S110 also includes the following steps:

[0081] Determine whether the operating point to be detected is outside the decision boundary of the distribution network;

[0082] If so, an early warning message is issued;

[0083] If not, it is determined whether the operating point to be detected has a trend of breaking through the decision boundary of the distribution network. If so, an early warning message is issued.

[0084] Among them, judging whether the operating point to be detected is outside the decision boundary of the distribution network includes:

[0085] Calculate the target distance between the running point to be detected and the center of the target hypersphere model;

[0086] Determine whether the target distance is greater than the radius of the target hypersphere model;

[0087] If yes, it is determined that the operating point to be detected is outside the decision boundary of the distribution network.

[0088] Optionally, judging whether the operating point to be detected has a trend of breaking through the decision boundary of the distribution network includes:

[0089] Periodically detecting a target distance between the operating point and the center of a target hypersphere model;

[0090] Determine whether the target distance has a trend of increasing periodically;

[0091] If yes, it is determined that the operating point to be detected has a tendency to break through the decision boundary of the distribution network.

[0092] In an optional implementation manner provided by an embodiment of the present invention, after the distribution network decision boundary fitting is completed, the distance r from the operating point u to the sphere center a can be calculated by the following expression:

[0093]

[0094] If the calculated r value is greater than the radius R of the hypersphere, it means that the operating point is outside the decision boundary. The system will issue an alarm signal and take relevant measures to avoid the occurrence of faults. When the calculated r value is less than the radius of the hypersphere, but the operating point has a tendency to break through the decision boundary, the system will also issue an alarm signal. In this way, timely measures can be taken before a fault occurs to ensure the stable operation of the power grid.

[0095] Embodiment 2

[0096] Figure 2 is a schematic diagram of a topological structure of a distribution network system according to an embodiment of the present invention. Figure 2 Taking the system shown as an example, the effectiveness of an online early warning method for a distribution network based on improved SVDD provided by an embodiment of the present invention is verified.

[0097] Verify that the environment is set up as: Figure 2 For the improved IEEE33 node distribution system diagram, the changes in wind power, photovoltaic power and load adopt the photovoltaic output curve, wind power output curve and load forecast curve of a typical day in a certain area for 24 hours. Specifically, wind turbines are connected at nodes 14 and 30, and distributed photovoltaics are connected at nodes 7 and 19. Figure 3 A predicted power value curve chart of wind power, photovoltaic power and load provided in an embodiment of the present invention.

[0098] The parameter settings of distributed power supply are shown in Table 1:

[0099] Table 1: Parameters of distributed power supply

[0100]

[0101] S1: According to the basic idea of ​​support vector data description, the positive samples in the sample set data are assigned labels of +1 and the negative samples are assigned labels of -1, and the decision boundary is accurately fitted. It should be noted that in the example setting of the present invention, the data that does not meet the operating status of safe, reliable and economical power supply is negative sample data, and the rest is positive sample data.

[0102] The initial sample set is the data of a certain area in the past two months. The time interval between sampling points is 15 minutes. There are 5126 positive sample vectors and 634 negative sample vectors in the sample set.

[0103] When considering positive and negative samples, we are solving an optimization problem:

[0104]

[0105] Where: x i represents a positive sample, x l Represents negative samples, both dimensions are q, x i ∈R q , i=1,2,…,N1,x l ∈R q ,l=1,2,…,N2;ξ i and l is the relaxation factor; C represents the error penalty coefficient for adjusting the number of samples inside and outside the hypersphere and the complexity of the algorithm; It is a nonlinear mapping of x, responsible for mapping the samples to the high-dimensional kernel feature space F.

[0106] The online warning boundary obtained by fitting with the improved SVDD algorithm is actually a multi-dimensional hypersphere, which is not conducive to observing the operating trend of each operating point. Therefore, the present invention considers reducing the sample to two dimensions and obtaining the warning boundary in a two-dimensional coordinate system, such as Figure 4 As shown, the red line is the warning boundary.

[0107] like Figure 4 As shown, there are still some negative samples inside the red line and some positive samples outside it. This is due to the influence of some bad data on boundary recognition. To avoid this, the present invention improves the SVDD algorithm by using the membership function so that samples with high relative density are surrounded by a hypersphere, thereby solving the problem of inaccurate boundary recognition caused by bad data.

[0108] S2: Use the fuzzy membership function to judge the importance of samples according to their corresponding membership values, and reduce the interference of individual extreme samples on boundary identification. It should be noted that:

[0109] The penalty coefficient C and Gaussian kernel parameter γ need to be continuously optimized to make the boundary fitting effect as optimal as possible. The specific implementation method is: fix a certain parameter value and continuously iterate until another parameter value corresponding to the highest accuracy is found, and finally the optimal value C = 10, γ = 0.01 is obtained.

[0110] Considering the membership value of the sample, formula (14) is transformed into:

[0111]

[0113] Introduce the Lagrange multiplier α to solve equation (6), let α i ' =y i α i , that is, for the positive sample α i ' =α i , negative sample α i ' =-α i , the dual expression is as follows:

[0114]

[0115] The method of the present invention is applied to the distribution network boundary identification model, and the accuracy of decision boundary identification and model training time of the SVDD method and the improved SVDD method are compared and analyzed. See Table 2. The improved SVDD algorithm is superior in both boundary identification accuracy and model training time, which proves that the method used in the present invention is effective and improves the accuracy of the boundary identification model to a certain extent.

[0116] Table 2

[0117]

[0118] S3: Propose the judgment basis of the online early warning method, and judge whether to issue an alarm signal according to the movement trend and operation rate of the operating point towards the boundary. It should be noted that: observe the trend and rate of the operating point moving towards the boundary at different times in the future. If the operating point has a trend of breaking through the boundary, the system will issue an alarm signal.

[0119] Figure 5 This is the moving trajectory diagram of the operating point in the time period of 00:00-11:00. Figure 6 This is the moving trajectory diagram of the operating point from 11:00 to 12:30. Figure 7 This is the movement trajectory diagram for the time period of 12:30-14:30. Figure 8 This is the movement trajectory diagram during the time period of 16:30-24:00. The red arrow indicates the general movement trend of the operating point, and the black line indicates the specific position change curve of the operating point.

[0120] observe Figure 5 , 6 , 7, 8. If the distribution network fails to make timely preventive control when there is a trend of breaking through the decision boundary at the operating point, the system will send out alarm signals in the three time periods of 10:00-12:00, 12:45-15:00 and 20:30-22:45, which is not conducive to the safe, economical and reliable operation of the distribution network. Therefore, the distribution network needs to monitor the operation status of the distribution network at 10:00, send out alarm signals when necessary, and take corresponding preventive control measures in time.

[0121] Embodiment 3

[0122] Fig. 9 FIG. 1 is a schematic diagram of an online early warning system for a distribution network based on an improved SVDD according to an embodiment of the present invention. Fig. 9 As shown, the system includes: a division module 10, a calculation module 20, a mapping module 30, a construction module 40 and an early warning module 50.

[0123] Specifically, the division module 10 is used to divide the power measurement data set of the distribution network into positive and negative samples to obtain target samples; the target samples include positive samples and negative samples.

[0124] The calculation module 20 is used to calculate the membership value of each sample in the target sample based on the fuzzy membership function.

[0125] The mapping module 30 is used to map the target sample to a high-dimensional feature space based on a method of supporting vector data description.

[0126] The construction module 40 is used to construct a target hypersphere model in the high-dimensional feature space based on the membership value of each sample, and use the boundary of the target hypersphere model as the distribution network decision boundary.

[0127] The early warning module 50 is used to determine whether to issue an early warning message based on the position relationship between the operating point to be detected and the decision boundary of the distribution network.

[0128] Specifically, the membership value includes:

[0129]

[0130] Among them, ρ i is the relative density of the ith positive sample, z i is the membership value of the i-th positive sample, ρ max and ρ min are the maximum and minimum values ​​of the relative density of positive samples, respectively.

[0131] Specifically, the target hypersphere model includes:

[0132]

[0133] Among them, x i represents positive samples, N1 represents the number of positive samples, z i represents the membership value of the positive sample, x l represents negative samples, N2 represents the number of negative samples, z l represents the membership value of the negative sample, ξ i and l is the relaxation factor; C1 represents the error penalty coefficient for adjusting the number of samples inside the hypersphere and the algorithm complexity; C2 represents the error penalty coefficient for adjusting the number of samples outside the hypersphere and the algorithm complexity; is the nonlinear mapping of x, γ represents the Gaussian kernel parameter, F represents the high-order feature space, R represents the radius of the hypersphere, and a represents the center of the hypersphere.

[0134] Specifically, the early warning module 50 is further used for:

[0135] Determine whether the operating point to be detected is outside the decision boundary of the distribution network;

[0136] If so, an early warning message is issued;

[0137] If not, it is determined whether the operating point to be detected has a trend of breaking through the decision boundary of the distribution network. If so, an early warning message is issued.

[0138] An embodiment of the present invention further provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method in the above-mentioned embodiment 1 when executing the computer program.

[0139] An embodiment of the present invention further provides a computer-readable storage medium, which stores computer instructions. When the computer instructions are executed by a processor, the method in the above-mentioned embodiment 1 is implemented.

[0140] It is known from common technical knowledge that the present invention can be implemented by other embodiments that do not deviate from its spirit or essential features. Therefore, the above disclosed embodiments are only illustrative in all respects and are not exclusive. All changes within the scope of the present invention or within the scope equivalent to the present invention are included in the present invention.

[0141] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0142] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0143] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0144] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0145] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A distribution network online early warning method for improving SVDD, characterized in that: include: The power measurement data set of the distribution network is divided into positive and negative samples to obtain the target samples; The target samples include positive samples and negative samples; Based on the fuzzy membership function, calculating the membership value of each sample in the target sample; Based on the method of support vector data description, the target sample is mapped to a high-dimensional feature space; Based on the membership value of each sample, a target hypersphere model is constructed in the high-dimensional feature space, and the boundary of the target hypersphere model is used as a distribution network decision boundary; Based on the positional relationship between the operating point to be detected and the decision boundary of the distribution network, it is determined whether to issue an early warning message.

2. The method according to claim 1, characterized in that: Calculating the membership value of each sample in the target sample based on the fuzzy membership function includes: Among them, ρ i is the relative density of the ith positive sample, z i is the membership value of the i-th positive sample, ρ max and ρ min are the maximum and minimum values ​​of the relative density of positive samples, respectively.

3. The method according to claim 1, characterized in that: The target hypersphere model comprises: Among them, x i represents positive samples, N1 represents the number of positive samples, z i represents the membership value of the positive sample, x l represents negative samples, N2 represents the number of negative samples, z l represents the membership value of the negative sample, ξ i and l is the relaxation factor; C1 represents the error penalty coefficient for adjusting the number of samples inside the hypersphere and the algorithm complexity; C2 represents the error penalty coefficient for adjusting the number of samples outside the hypersphere and the algorithm complexity; is a nonlinear mapping of x, γ represents a Gaussian kernel parameter, F represents the high-order feature space, R represents the radius of the hypersphere, and a represents the center of the hypersphere.

4. The method according to claim 1, characterized in that: Based on the position relationship between the operating point to be detected and the decision boundary of the distribution network, determining whether to issue warning information includes: Determining whether the operating point to be detected is outside the decision boundary of the distribution network; If so, an early warning message is issued; If not, it is determined whether the operating point to be detected has a trend of breaking through the decision boundary of the distribution network. If so, an early warning message is issued.

5. The method according to claim 4, characterized in that: Determining whether the operating point to be detected is outside the distribution network decision boundary includes: Calculating a target distance between the operating point to be detected and the center of the target hypersphere model; Determining whether the target distance is greater than the radius of the target hypersphere model; If yes, it is determined that the operating point to be detected is outside the decision boundary of the power distribution network.

6. A distribution network online early warning system based on improved SVDD, characterized in that: include: Division module, calculation module, mapping module, construction module and warning module; among them, The division module is used to divide the power measurement data set of the distribution network into positive and negative samples to obtain target samples; the target samples include positive samples and negative samples; The calculation module is used to calculate the membership value of each sample in the target sample based on the fuzzy membership function; The mapping module is used to map the target sample to a high-dimensional feature space based on a method of supporting vector data description; The construction module is used to construct a target hypersphere model in the high-dimensional feature space based on the membership value of each sample, and use the boundary of the target hypersphere model as the distribution network decision boundary; The early warning module is used to determine whether to issue early warning information based on the position relationship between the operating point to be detected and the decision boundary of the distribution network.

7. The system according to claim 6, characterized in that: The target hypersphere model comprises: Among them, x i represents positive samples, N1 represents the number of positive samples, z i represents the membership value of the positive sample, x l represents negative samples, N2 represents the number of negative samples, z l represents the membership value of the negative sample, ξ i and l is the relaxation factor; C1 represents the error penalty coefficient for adjusting the number of samples inside the hypersphere and the algorithm complexity; C2 represents the error penalty coefficient for adjusting the number of samples outside the hypersphere and the algorithm complexity; is a nonlinear mapping of x, γ represents a Gaussian kernel parameter, F represents the high-order feature space, R represents the radius of the hypersphere, and a represents the center of the hypersphere.

8. The system according to claim 6, characterized in that: The early warning module is also used for: Determining whether the operating point to be detected is outside the decision boundary of the distribution network; If so, an early warning message is issued; If not, it is determined whether the operating point to be detected has a trend of breaking through the decision boundary of the distribution network. If so, an early warning message is issued.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 5 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and when the computer instructions are executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Citation Information

Patent Citations

  • A distribution network district risk early warning method and system based on support vector data description

    CN109670676A

  • High-voltage cable defect identification method based on multivariate fuzzy support vector data description

    CN109978068A

  • IFCM-BB-based transformer fault diagnosis method

    CN113887623A

  • Pesticide spraying management system using multiple drones and operation method therof

    KR102650604B1