Single-phase earth fault identification method and system based on centralized feeder automation

Through centralized feeder automation technology, wavelet packet decomposition and convolutional neural network are used to identify feeder failures, solving the problem of single-phase grounding fault identification caused by distributed power supplies, and achieving efficient fault identification and management.

CN118937885APending Publication Date: 2024-11-12ELECTRIC POWER RES INST OF GUANGXI POWER GRID CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410862631.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-06-28
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

The randomness and volatility of distributed power supplies lead to instability of feeders, an increase in the incidence of single-phase grounding faults, diversified fault characteristic information, and difficulty in accurately identifying them.

Method used

Based on centralized feeder automation, wavelet packet decomposition and feature matrix sorting are obtained by obtaining fault power signals, and converted into image recognition tasks. Convolutional neural networks are used to identify fault types and phases, combined with thermal metal encoding and pixel normalization, and single-phase grounding faults are realized. Accurate identification.

Benefits of technology

It improves the accuracy and efficiency of single-phase grounding fault identification, reduces manual intervention, enhances the level of automated management of feeders, shortens fault detection and response time, and improves the reliability of power supply.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118937885A_ABST
    Figure CN118937885A_ABST
Patent Text Reader

Abstract

The invention discloses a single-phase earth fault identification method based on centralized feeder automation, and the method comprises the steps: obtaining a fault electric quantity signal, carrying out the first judgment according to the opening condition of a circuit breaker, carrying out the wavelet packet decomposition of the fault electric quantity signal if the fault is judged to be a permanent fault, and obtaining a characteristic frequency band of each electric quantity signal, sorting the characteristic frequency bands to obtain a characteristic matrix of the nodes, and sorting the characteristic matrix of the nodes to obtain a characteristic matrix of the feeder lines; carrying out pixel normalization on the feature matrix of the feeder line to obtain a grayscale image, and carrying out color enhancement on the grayscale image by utilizing hot metal coding to obtain a colored fault image; and inputting the fault image into a convolutional neural network model for image recognition to obtain a fault image label, and performing second judgment according to the fault image label to obtain a fault type and a fault phase, so that the single-phase earth fault can be accurately recognized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of power system feeder automation, and in particular to a single-phase grounding fault identification method and system based on centralized feeder automation. Background Art

[0002] In the context of building a new power system, a large number of distributed generation (DG) are connected to the feeder, and the feeder power supply mode is changed from centralized power supply to distributed power supply, promoting the green and economic transformation of the feeder. However, the randomness and volatility of DG lead to an increase in the instability factors of the feeder, and the incidence of single-phase grounding faults in the feeder increases. In addition, DG makes the weak fault feature information diversified, the fault boundary becomes more blurred, and the accurate identification of single-phase grounding faults becomes more difficult. The existing active feeder single-phase grounding fault identification method still has certain limitations. The implementation of single-phase grounding fault identification depends on feeder automation technology. Feeder Automation (FA) is a technical system that uses modern communication, control and protection technologies to automatically control feeders. It is mainly used on the power supply side of the feeder to realize automatic control and protection of the feeder and improve the reliability, safety and economy of the power grid.

[0003] At present, the method of identifying single-phase grounding faults based on the information upload of distribution terminal units is very popular. The implementation of this method depends on centralized feeder automation. Centralized feeder automation is based on communication networks, with centralized control as the core, and integrates multiple control modes such as current protection, distribution terminal remote control, and reclosing to complete the rapid identification of single-phase grounding faults. Therefore, it is necessary to provide a single-phase grounding fault identification solution based on centralized feeder automation. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is: the present invention solves the problem that the randomness and volatility of a large number of distributed power sources (DG) lead to an increase in unstable factors of the feeder, an increase in the incidence of single-phase grounding faults of the feeder, and DG makes weak fault characteristic information diversified, the fault boundary becomes more blurred, and the accurate identification of single-phase grounding faults becomes more difficult.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a single-phase grounding fault identification method based on centralized feeder automation, comprising:

[0008] Obtain a fault power signal, make a first judgment according to the tripping condition of the circuit breaker, and if it is judged to be a permanent fault, perform wavelet packet decomposition on the fault power signal to obtain a characteristic frequency band of each power signal, sort the characteristic frequency bands to obtain a characteristic matrix of the node, sort the characteristic matrix of the node to obtain a characteristic matrix of the feeder;

[0009] Normalizing the pixels of the characteristic matrix of the feeder to obtain a grayscale image, and enhancing the color of the grayscale image by using hot metal coding to obtain a color fault image;

[0010] The fault image is input into a convolutional neural network model for image recognition to obtain a fault image label, and a second judgment is performed based on the fault image label to obtain a fault type and a fault phase.

[0011] As a preferred solution of the single-phase ground fault identification method based on centralized feeder automation described in the present invention, the characteristic matrix of the feeder includes:

[0012] The fault power signal includes three-phase voltage, three-phase current and zero-sequence current;

[0013] Sort the characteristic frequency band of each electric quantity signal according to frequency to obtain a characteristic matrix of the electric quantity signal;

[0014] The characteristic matrix of each electrical quantity signal is sorted in the order of zero-sequence current, A-phase voltage, A-phase current, B-phase voltage, B-phase current, C-phase voltage, and C-phase current to obtain the characteristic matrix of the node;

[0015] The characteristic matrix of each node is sorted according to the degree of distance between the node and the head end of the feeder. If the node of the branch line is at the end, it is still sorted according to the degree of distance between the node and the head end of the branch line to obtain the characteristic matrix of the feeder.

[0016] As a preferred solution of the single-phase ground fault identification method based on centralized feeder automation described in the present invention, wherein: pixel normalization of the characteristic matrix of the feeder includes:

[0017] By identifying the maximum and minimum values ​​of each row in the feature matrix;

[0018] According to the ratio between the eigenvalue and the maximum value of the row, each eigenvalue is mapped to the set mapping range using linear scaling;

[0019] According to the result of linear scaling, the pixel value of the image is obtained;

[0020] Normalize the pixel values ​​to obtain a grayscale image.

[0021] As a preferred solution of the single-phase ground fault identification method based on centralized feeder automation of the present invention, the method of color enhancement of the grayscale image by using hot metal coding comprises:

[0022] The mapping range is set to 0~X i , let the range of pixel values ​​be N j =N1,N2…,N j , where i and j are natural numbers.

[0023] The red transformation function is expressed as:

[0024]

[0025] The green transformation function is expressed as:

[0026]

[0027] The blue transformation function is expressed as:

[0028]

[0029] Among them, f(x,y) is the pixel value at the pixel point (x,y), and x and y are the position information of the pixel point in the matrix or image.

[0030] As a preferred solution of the single-phase grounding fault identification method based on centralized feeder automation described in the present invention, wherein: inputting the fault image into the convolutional neural network model for image recognition includes:

[0031] The fault image is resized to a uniform size and input into the pre-trained ResNet18 model. The output dimension of the last layer of the ResNet18 model is modified according to the number of categories of the classification task.

[0032] Select some layer parameters of the ResNet18 model to freeze, train the unfrozen layer parameters, and obtain the first ResNet18 model;

[0033] The fault image is input into the first ResNet18 model for forward propagation and back propagation calculations, and the first ResNet18 model outputs a feature vector;

[0034] From the feature vector output by the first ResNet18 model, the category index corresponding to the vector with the largest probability value is obtained, and the category of the fault image is obtained.

[0035] As a preferred solution of the single-phase grounding fault identification method based on centralized feeder automation described in the present invention, the ResNet18 model includes:

[0036] Before pre-training, the number of neurons in the output layer of the convolutional neural network ResNet18 model is set to be the same as the total number of fault types.

[0037] As a preferred solution of the single-phase grounding fault identification method based on centralized feeder automation described in the present invention, the second judgment according to the fault image label includes:

[0038] If the fault image label is a single-phase permanent ground fault, the distribution master station obtains the fault phase and fault location based on the fault image label, and sends a closing command to the circuit breaker and section switch at the head end of the feeder. The circuit breaker and section switch are closed, and the feeder resumes normal operation.

[0039] In a second aspect, the present invention provides a system for identifying single-phase grounding faults based on centralized feeder automation, comprising:

[0040] A data integration module is used to obtain a fault power signal, make a first judgment according to the opening condition of the circuit breaker, and if it is judged to be a permanent fault, perform wavelet packet decomposition on the fault power signal to obtain a characteristic frequency band of each power signal, sort the characteristic frequency bands to obtain a characteristic matrix of the node, sort the characteristic matrix of the node to obtain a characteristic matrix of the feeder;

[0041] A color enhancement module, used for performing pixel normalization on the characteristic matrix of the feeder to obtain a grayscale image, and performing color enhancement on the grayscale image using hot metal coding to obtain a color fault image;

[0042] The judgment module is used to input the fault image into a convolutional neural network model for image recognition to obtain a fault image label, and perform a second judgment based on the fault image label to obtain a fault type and a fault phase.

[0043] In a third aspect, the present invention provides a computing device, comprising:

[0044] Memory and processor;

[0045] The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the single-phase grounding fault identification method based on centralized feeder automation are implemented.

[0046] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the single-phase grounding fault identification method based on centralized feeder automation.

[0047] Beneficial effects of the present invention: The single-phase grounding fault identification method based on centralized feeder automation proposed in the present invention obtains the three-phase voltage, three-phase current and zero-sequence current information on the switches in each section after the fault, amplifies the fault characteristics contained in the electrical signal, and converts the single-phase grounding fault identification problem into an image recognition task, and identifies the single-phase grounding fault in a data-driven manner; wherein, the feeder fault information is fully extracted and the image expression is further enhanced. At the same time, the convolutional neural network ResNet18 with superior performance is adopted, the image recognition effect is excellent, and the single-phase grounding fault can be accurately identified. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0049] Figure 1 An overall flow chart of a single-phase grounding fault identification method based on centralized feeder automation provided by the present invention;

[0050] Figure 2 Representation of recognition accuracy after pre-training of a large number of fault images generated by a single-phase grounding fault recognition method based on centralized feeder automation provided by the present invention;

[0051] Figure 3 The invention provides a method for identifying a single-phase ground fault based on centralized feeder automation and a node sequence for constructing a characteristic matrix of a feeder. DETAILED DESCRIPTION

[0052] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0053] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0054] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0055] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0056] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0057] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0058] Example 1

[0059] Reference Figure 1 , as an embodiment of the present invention, provides a single-phase grounding fault identification method based on centralized feeder automation, comprising:

[0060] S100: Acquire a fault power signal, and make a first judgment according to the tripping condition of the circuit breaker. If it is judged to be a permanent fault, perform wavelet packet decomposition on the fault power signal to obtain a characteristic frequency band of each power signal, sort the characteristic frequency bands to obtain a characteristic matrix of the node, sort the characteristic matrix of the node to obtain a characteristic matrix of the feeder;

[0061] In the embodiment of the present application, the fault power signal includes three-phase voltage, three-phase current and zero-sequence current;

[0062] Sort the characteristic frequency band of each electric quantity signal according to frequency to obtain a characteristic matrix of the electric quantity signal;

[0063] The characteristic matrix of each electrical quantity signal is sorted in the order of zero-sequence current, A-phase voltage, A-phase current, B-phase voltage, B-phase current, C-phase voltage, and C-phase current to obtain the characteristic matrix of the node;

[0064] The characteristic matrix of each node is sorted according to the degree of distance between the node and the head end of the feeder. If the node of the branch line is at the end, it is still sorted according to the degree of distance between the node and the head end of the branch line to obtain the characteristic matrix of the feeder.

[0065] In another possible implementation, the characteristic frequency band of each electric quantity signal can also be sorted from low to high and from top to bottom according to the contribution of the frequency component to the signal, the rate of change of the frequency component, and the stability of the frequency component, so as to obtain a characteristic matrix of the signal; the characteristic matrix of each electric quantity signal can also be sorted from high to low and from large to small according to the type and importance of the load connected to the node, the current passing through the node, and the voltage level or voltage deviation of the node, so as to obtain a characteristic matrix of the node; the characteristic matrix of each node is sorted according to the failure probability of the node, the response time and action speed of the protection device on the node, and the consistency of the electrical parameters, so as to obtain a characteristic matrix of the feeder;

[0066] Specifically, the characteristic frequency band of each electric quantity signal is sorted from low to high from top to bottom according to the frequency to obtain the characteristic matrix of the signal; the characteristic matrix of each electric quantity signal is sorted from top to bottom in the order of zero-sequence current, A-phase voltage, A-phase current, B-phase voltage, B-phase current, C-phase voltage, and C-phase current to obtain the characteristic matrix of the node; the characteristic matrix of each node is sorted from near to far from top to bottom according to the degree of distance of the node from the head end of the feeder to obtain the characteristic matrix of the feeder, wherein the node of the branch line is arranged at the end, and the order is still arranged from near to far from top to bottom according to the degree of distance of the node from the head end of the branch line. If other arrangement methods are selected, consistency must be ensured. The present invention preferably uses the sorting method of the invention, and the three-phase voltage, three-phase current and zero-sequence current information are the electric quantity information when a fault occurs, and the length of the collected information is 0.04s;

[0067] Specifically, the feeder is segmented through nodes, loads or branch lines are drawn from the nodes, a circuit breaker or a segment switch is set between every two nodes, and a contact switch is set at the end of the feeder. Each switch is equipped with a distribution terminal unit, and each distribution terminal unit is connected to the distribution master station through a communication network. At least one node in the feeder is connected to the distributed power source. When the feeder operates normally, the segment switch and the outgoing circuit breaker are closed;

[0068] The first judgment that the interconnecting switch is disconnected includes that when a fault occurs on the feeder, each distribution terminal unit collects the three-phase voltage, three-phase current and zero-sequence current information on the corresponding switch, and the section switch closest to the upstream of the fault point trips. The section switch is reclosed. If a transient fault occurs, the section switch recloses successfully and the feeder resumes normal operation; if a permanent fault occurs, the reclosing fails, and then the outgoing circuit breaker trips, and the previously collected power information is uploaded to the distribution master station.

[0069] It should be noted that by setting up section switches and connecting switches in the feeder, the fault area can be quickly isolated when a fault occurs, reducing the scope of power outage and improving the power supply reliability of the entire power grid. When a feeder fault occurs, the outgoing circuit breaker will be opened. The voltage and current information collected by the distribution terminal unit can quickly locate the fault point and isolate it to reduce the impact of the fault on the power grid. Connecting distributed power sources to the feeder can improve the energy utilization efficiency of the power grid. At the same time, through automated control, the load and power supply of the power grid can be better balanced and the operating status of the power grid can be optimized.

[0070] It should also be noted that by sorting the feeders from the beginning to the end, the feeders can be analyzed hierarchically, which facilitates the identification and understanding of the grid structure and the path of signal propagation. Sorting the nodes according to how far they are from the beginning helps to quickly locate the fault, especially when a fault occurs, the affected area can be quickly identified. Putting the branch line nodes at the end can determine the priority of handling faults or maintenance work based on the topological structure and operating status of the feeders. By sorting the zero-sequence current and the voltage and current of each phase, the load distribution in the grid can be identified, which helps to optimize the load management of the grid. This sorting method provides a structured and systematic analysis method for feeder operation and maintenance, fault diagnosis, load management and grid planning, which helps to improve the operating efficiency and reliability of the grid.

[0071] S200: normalizing the pixels of the feature matrix of the feeder to obtain a grayscale image, and performing color enhancement on the grayscale image using hot metal coding to obtain a color fault image;

[0072] In the embodiment of the present application, by identifying the maximum and minimum values ​​of each row in the feature matrix;

[0073] According to the ratio between the eigenvalue and the maximum value of the row, each eigenvalue is mapped to the set mapping range using linear scaling;

[0074] According to the result of linear scaling, the pixel value of the image is obtained;

[0075] Normalize the pixel values ​​to obtain a grayscale image;

[0076] Exemplarily, all elements are mapped to 0-255, the maximum and minimum values ​​of each row in the feature matrix are identified, and each eigenvalue is mapped to the range of 0-255 using a linear scaling method based on the ratio between the eigenvalue and the maximum value of the row. Specifically, the element of a row in the feature matrix minus the minimum value is divided by the maximum value minus the minimum value and multiplied by 255 to obtain the linearly scaled value. After completing the linear scaling, the calculated result is rounded to the nearest integer.

[0077] It should be noted that the original electrical characteristic data is converted into a grayscale image so that the distribution and characteristics of the data can be intuitively displayed in the visual image.

[0078] In the embodiment of the present application, the mapping range is set to 0~X i , let the range of pixel values ​​be N j =N1,N2…,N j , where i and j are natural numbers.

[0079] The red transformation function is expressed as:

[0080]

[0081] The green transformation function is expressed as:

[0082]

[0083] The blue transformation function is expressed as:

[0084]

[0085] Among them, f(x,y) is the pixel value at the pixel point (x,y), and x and y are the position information of the pixel point in the matrix or image.

[0086] For example, all elements are mapped to 0 to 255.

[0087] The red transformation function is expressed as:

[0088]

[0089] The green transformation function is expressed as:

[0090]

[0091] The blue transformation function is expressed as:

[0092]

[0093] Among them, f(x,y) is the pixel value at the pixel point (x,y), and x and y are the position information of the pixel point in the matrix or image.

[0094] S300: inputting the fault image into the convolutional neural network model for image recognition to obtain a fault image label, and performing a second judgment based on the fault image label to obtain a fault type and a fault phase;

[0095] In the embodiment of the present application, the fault image is adjusted to a uniform size and input into a pre-trained ResNet18 model, and the output dimension of the last layer of the ResNet18 model is modified according to the number of categories of the classification task;

[0096] Select some layer parameters of the ResNet18 model to freeze, train the unfrozen layer parameters, and obtain the first ResNet18 model;

[0097] The fault image is input into the first ResNet18 model for forward propagation and back propagation calculations, and the first ResNet18 model outputs a feature vector;

[0098] From the feature vector output by the first ResNet18 model, the category index corresponding to the vector with the largest probability value is obtained, and the category of the fault image is obtained.

[0099] Exemplarily, the fault image is adjusted to 400*400*3 and input into the pre-trained ResNet18 model. The number of categories of the classification task is 10. The output dimension of the last layer of the ResNet18 model is changed to 10, the parameters of the first 7 layers are frozen, and the parameters of the last 11 layers are trained. The parameters are updated through the back propagation algorithm, and the first ResNet18 model is obtained as the trained ResNet18 model. The resized fault image is input into the trained ResNet18 model, and forward propagation calculation is performed to obtain the output vector. If the output vector has the same degree of overlap with the feature vector recognizable by the trained ResNet18 model, the index corresponding to this element is the category of the fault image.

[0100] Specifically, each fault type uses a coding identifier, and the fault image label output by the convolutional neural network ResNet18 is the coding identifier of the fault type, so the distribution master station can determine the type of fault that has occurred through the output coding identifier.

[0101] In an embodiment of the present application, the number of output layer neurons of the convolutional neural network ResNet18 model is set to be the same as the total number of fault types before pre-training.

[0102] It should be noted that directly setting the number of neurons in the output layer to the total number of fault types makes the output dimension of the model match the number of categories of the classification task, which helps the model to better learn the characteristics of specific tasks. When the data set is small or the number of categories is small, reducing the number of neurons in the output layer can reduce the complexity of the model, thereby reducing the risk of overfitting and reducing the number of model parameters. This can reduce the computational burden during training to a certain extent and speed up training.

[0103] In the embodiment of the present application, if the fault image label is a single-phase permanent ground fault, the power distribution master station obtains the fault phase and fault position according to the fault image label, and sends a closing command to the circuit breaker and section switch at the head end of the feeder, the circuit breaker and section switch are closed, and the feeder resumes normal operation;

[0104] Specifically, all fault types are represented by labels, and the fault type can be determined by the output label. If it is determined to be a single-phase permanent grounding fault, the fault phase and fault location are determined based on the fault image label. The distribution master station will send a closing command to the circuit breaker and section switch at the head end of the feeder. The circuit breaker and section switch are closed, and the feeder resumes normal operation.

[0105] If it is determined to be another type of fault such as a phase-to-phase short circuit fault, the fault phase and fault location are determined based on the fault image label, and the power distribution master station does not send any command to the feeder terminal device.

[0106] After determining that a single-phase permanent grounding fault has occurred and the faulty phase has occurred, the distribution master station will send a closing command to the circuit breaker and section switch at the head end of the feeder to control the closing of the circuit breaker and section switch at the head end, so that the feeder can operate with a single-phase grounding fault for a short time of 1 to 2 hours, thereby improving power supply reliability.

[0107] If it is determined that no single-phase grounding fault has occurred, the distribution master station will not issue any control commands temporarily, keep the feeder head end circuit breaker and section switch open, and wait for the distribution master station to carry out the next step of fault location, removal and restoration.

[0108] It should be noted that through automated image recognition and fault judgment, the need for manual intervention is reduced, the level of automated management of feeders is improved, and the use of deep learning models to quickly and accurately identify fault types can shorten fault detection and response time and speed up fault handling. The model can determine the fault phase and fault location based on the fault image label, which helps to accurately locate the fault point and provide accurate information for fault handling. For single-phase permanent grounding faults, by controlling the closing of the circuit breaker at the head end of the feeder and isolating the fault, unnecessary power outages can be avoided and the continuity and reliability of power supply can be improved.

[0109] The above is a schematic scheme of a single-phase ground fault identification method based on centralized feeder automation in this embodiment. It should be noted that the technical scheme of the device for single-phase ground fault identification based on centralized feeder automation and the technical scheme of the single-phase ground fault identification method based on centralized feeder automation belong to the same concept, and the details not described in detail in the technical scheme of the device for single-phase ground fault identification based on centralized feeder automation in this embodiment can be found in the description of the technical scheme of the single-phase ground fault identification method based on centralized feeder automation.

[0110] The device for identifying single-phase grounding faults based on centralized feeder automation in this embodiment includes:

[0111] A data integration module is used to obtain a fault power signal, make a first judgment according to the opening condition of the circuit breaker, and if it is judged to be a permanent fault, perform wavelet packet decomposition on the fault power signal to obtain a characteristic frequency band of each power signal, sort the characteristic frequency bands to obtain a characteristic matrix of the node, sort the characteristic matrix of the node to obtain a characteristic matrix of the feeder;

[0112] A color enhancement module, used for performing pixel normalization on the characteristic matrix of the feeder to obtain a grayscale image, and performing color enhancement on the grayscale image using hot metal coding to obtain a color fault image;

[0113] The judgment module is used to input the fault image into a convolutional neural network model for image recognition to obtain a fault image label, and perform a second judgment based on the fault image label to obtain a fault type and a fault phase.

[0114] This embodiment further provides a computing device, which is applicable to the identification of single-phase grounding faults based on centralized feeder automation, including:

[0115] Memory and processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the single-phase grounding fault identification method based on centralized feeder automation as proposed in the above embodiment.

[0116] This embodiment further provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for identifying a single-phase grounding fault based on centralized feeder automation as proposed in the above embodiment is implemented.

[0117] The storage medium proposed in this embodiment and the single-phase grounding fault identification method based on centralized feeder automation proposed in the above embodiment belong to the same inventive concept. The technical details not described in detail in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0118] Through the above description of the implementation methods, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ReadOnly, Memory, ROM), random access memory (RandomAccess Memory, RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform the methods of various embodiments of the present invention.

[0119] Example 2

[0120] Referring to Table 1, Figure 2-3 As shown, this embodiment is different from the first embodiment and provides a verification test of a single-phase grounding fault identification method based on centralized feeder automation to verify and illustrate the technical effects adopted in this method.

[0121] A large number of fault images are used to train the convolutional neural network Resnet18. The recognition accuracy of the generated fault images after pre-training reaches 99.0654%. Figure 2 As shown, the fault type coding is shown in Table 1.

[0122] Table 1 Fault type code

[0123]

[0124]

[0125] When a single-phase grounding fault of phase A occurs on the feeder, the FTU on all switches starts the sampling function with a sampling frequency of 10kHz and a sampling length of 0.04s. The three-phase voltage, three-phase current and zero-sequence current flowing through the switch are collected, and each electrical quantity signal contains 400 sampling points. Then the adjacent section switch trips, and after the reclosing fails, the outgoing circuit breaker opens and uploads the electrical quantity signal to the main distribution station.

[0126] The distribution master station uses wavelet packet transform to perform three-layer wavelet packet decomposition on seven electrical signals, namely, phase A voltage, phase B voltage, phase C voltage, phase A current, phase B current, phase C current and zero-sequence current. Each electrical signal is decomposed into 8 characteristic frequency bands. The 8 characteristic frequency bands of each electrical signal are arranged from low to high from top to bottom according to the frequency to obtain the characteristic matrix of the electrical signal. At this time, the characteristic matrix size of the electrical signal is 8*400; the characteristic matrix of each electrical signal is arranged from top to bottom in the order of zero-sequence current, phase A voltage, phase A current, phase B voltage, phase B current, phase C voltage and phase C current to obtain the characteristic matrix of the node. At this time, the size of the characteristic matrix of the node is 56*400; the characteristic matrix of each node is arranged from near to far from top to bottom according to the degree of distance from the node to the head end of the feeder to obtain the characteristic matrix of the feeder. At this time, the size of the characteristic matrix of the feeder is 392*400. Among them, the nodes of the branch line are arranged at the end, and the order is still arranged from near to far from top to bottom according to the degree of distance from the node to the head end of the branch line. The order of the nodes is as follows: Figure 3 As shown;

[0127] The characteristic matrix of the feeder is pixel-normalized, and the matrix elements are converted into pixel elements of 0 to 255 to obtain a grayscale image. The grayscale image is color-enhanced using hot metal coding to obtain a color fault image. After obtaining the fault image, the size of the fault image is adjusted to [400 400].

[0128] The fault image is input into the convolutional neural network ResNet18. After classification and recognition by the convolutional neural network ResNet18, the fault image label is output, which is [1,0,0,0,0,0,0,0,0,0]. Therefore, this fault will be judged as a single-phase grounding fault of phase A. The circuit breaker and the section switch at the head end of the feeder receive the closing command, and the feeder resumes normal operation.

[0129] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A single-phase grounding fault identification method based on centralized feeder automation, characterized in that: include: Obtain a fault power signal, make a first judgment according to the tripping condition of the circuit breaker, and if it is judged to be a permanent fault, perform wavelet packet decomposition on the fault power signal to obtain a characteristic frequency band of each power signal, sort the characteristic frequency bands to obtain a characteristic matrix of the node, sort the characteristic matrix of the node to obtain a characteristic matrix of the feeder; Normalizing the pixels of the characteristic matrix of the feeder to obtain a grayscale image, and enhancing the color of the grayscale image by using hot metal coding to obtain a color fault image; The fault image is input into a convolutional neural network model for image recognition to obtain a fault image label, and a second judgment is performed based on the fault image label to obtain a fault type and a fault phase.

2. The single-phase grounding fault identification method based on centralized feeder automation as claimed in claim 1, characterized in that: The characteristic matrix of the feeder includes: The fault power signal includes three-phase voltage, three-phase current and zero-sequence current; Sort the characteristic frequency band of each electric quantity signal according to frequency to obtain a characteristic matrix of the electric quantity signal; The characteristic matrix of each electrical quantity signal is sorted in the order of zero-sequence current, A-phase voltage, A-phase current, B-phase voltage, B-phase current, C-phase voltage, and C-phase current to obtain the characteristic matrix of the node; The characteristic matrix of each node is sorted according to the degree of distance between the node and the head end of the feeder. If the node of the branch line is at the end, it is still sorted according to the degree of distance between the node and the head end of the branch line to obtain the characteristic matrix of the feeder.

3. The single-phase grounding fault identification method based on centralized feeder automation as claimed in claim 2, characterized in that: Pixel normalization of the feature matrix of the feed line includes: By identifying the maximum and minimum values ​​of each row in the feature matrix; According to the ratio between the eigenvalue and the maximum value of the row, each eigenvalue is mapped to the set mapping range using linear scaling; According to the result of linear scaling, the pixel value of the image is obtained; Normalize the pixel values ​​to obtain a grayscale image.

4. The single-phase grounding fault identification method based on centralized feeder automation as claimed in claim 3, characterized in that: Using hot metal coding to perform color enhancement on the grayscale image includes: The mapping range is set to 0~X i , let the range of pixel values ​​be N j =N1,N2…,N j , where i and j are natural numbers. The red transformation function is expressed as: The green transformation function is expressed as: The blue transformation function is expressed as: Among them, f(x,y) is the pixel value at the pixel point (x,y), and x and y are the position information of the pixel point in the matrix or image.

5. The single-phase grounding fault identification method based on centralized feeder automation as claimed in claim 4, characterized in that: Inputting the fault image into the convolutional neural network model for image recognition includes: The fault image is resized to a uniform size and input into the pre-trained ResNet18 model. The output dimension of the last layer of the ResNet18 model is modified according to the number of categories of the classification task. Select some layer parameters of the ResNet18 model to freeze, train the unfrozen layer parameters, and obtain the first ResNet18 model; The fault image is input into the first ResNet18 model for forward propagation and back propagation calculations, and the first ResNet18 model outputs a feature vector; From the feature vector output by the first ResNet18 model, the category index corresponding to the vector with the largest probability value is obtained, and the category of the fault image is obtained.

6. The single-phase grounding fault identification method based on centralized feeder automation according to claim 1 or 5, characterized in that: The ResNet18 model includes: Before pre-training, the number of neurons in the output layer of the convolutional neural network ResNet18 model is set to be the same as the total number of fault types.

7. The single-phase grounding fault identification method based on centralized feeder automation as claimed in claim 6, characterized in that: Performing a second judgment according to the fault image label includes: If the fault image label is a single-phase permanent ground fault, the distribution master station obtains the fault phase and fault location based on the fault image label, and sends a closing command to the circuit breaker and section switch at the head end of the feeder. The circuit breaker and section switch are closed, and the feeder resumes normal operation.

8. A system for identifying single-phase grounding faults based on centralized feeder automation, characterized in that: include: A data integration module is used to obtain a fault power signal, make a first judgment according to the opening condition of the circuit breaker, and if it is judged to be a permanent fault, perform wavelet packet decomposition on the fault power signal to obtain a characteristic frequency band of each power signal, sort the characteristic frequency bands to obtain a characteristic matrix of the node, sort the characteristic matrix of the node to obtain a characteristic matrix of the feeder; A color enhancement module, used for performing pixel normalization on the characteristic matrix of the feeder to obtain a grayscale image, and performing color enhancement on the grayscale image using hot metal coding to obtain a color fault image; The judgment module is used to input the fault image into a convolutional neural network model for image recognition to obtain a fault image label, and perform a second judgment based on the fault image label to obtain a fault type and a fault phase.

9. An electronic device, characterized in that: The device comprises: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.