Intelligent power distribution network high-resistance identification and fault accurate positioning method, medium and intelligent power distribution network high-resistance identification and fault accurate positioning system

Abstract

The invention provides an intelligent power distribution network high-resistance identification and fault accurate positioning method, medium and system, and the method comprises the steps: uploading the zero-sequence voltage and zero-sequence current signals of each node of an intelligent power distribution network to a main station, carrying out the mean interpolation, logarithmic transformation and normalization processing of abnormal data, and carrying out the recognition of the zero-sequence voltage and zero-sequence current signals of each node of the intelligent power distribution network; then, the sequential sequence signals are converted into images through the Gramer angle field transformation thought, spatial domain full information features are fused, a fault feature quantity space-time atlas sample library is established, convolution and activation function operation and pooling operation are conducted on input samples through a two-channel convolutional neural network model, back propagation is conducted on full connection layer output, and the fault feature quantity space-time atlas sample library is established; the method is used for calculating the gradient of a loss function relative to network parameters and updating the weight and bias of the network through the gradients so as to minimize the loss function, and through establishment of a two-channel convolutional neural network model, the method can perform accurate fault identification and accurate fault positioning on the high-resistance fault of the intelligent power distribution network, and has a good fault identification effect on the high-resistance fault of the intelligent power distribution network. And the influence of transition resistance is avoided.

Description

Technical Field

[0001] The present application relates to the technical field of fault detection in distribution networks, and more specifically, to a method, medium, computer program product, electronic device and system for high-resistance identification and precise fault location in intelligent distribution networks. Background Art

[0002] With the large-scale integration of new energy sources into the distribution network, the length and structure of its lines are constantly changing, which brings great challenges to the protection of the distribution network. Due to the complex structure and large scale of the distribution network, the probability of faults in its lines increases, and when the distribution network is grounded with high resistance, the grounding fault signal is weak, the fault characteristics are not obvious, and the transient process time is short, making it difficult to identify the fault. Due to the long-term existence of the fault, the insulation of the equipment is greatly damaged, which in turn causes metallic grounding or phase-to-phase short-circuit grounding, causing fires and large-scale power outages. Most of the existing newly built distribution network lines are cable-laid. Compared with overhead lines, the capacitance to the ground is larger, and the line parameters are asymmetric. Therefore, when a phase-to-ground fault occurs, the capacitive current is large and a large arc is formed at the grounding point. Generally, an arc suppression coil is used to compensate the capacitive current in the form of overcompensation. This makes it difficult for traditional relay protection to identify faults and fault locations by relying on simple three-phase voltage, three-phase current, and zero-sequence voltage and current. Therefore, it is particularly important to deeply study new methods for distribution network protection and fully explore the fault information uploaded by each sensor.

[0003] The traditional method is to use Fourier transform, short-time Fourier transform, wavelet decomposition and other technologies to decompose transient signals, extract the high-frequency components of the first half cycle after the fault occurs, and compare them with the signals before the fault to obtain the fault feature quantity and identify the fault. The above methods only consider the high-frequency or low-frequency components within a certain time period, and do not analyze the full-domain signal. Therefore, the fault characteristics are not obvious enough, and there is a possibility of misjudgment or missed judgment. Summary of the invention

[0004] The main purpose of the present application is to provide a method, medium, computer program product, electronic device and system for high-resistance identification and precise fault location of an intelligent distribution network, so as to at least solve the problem that the prior art only considers the high-frequency or low-frequency components within a certain time period for distribution network fault detection, and does not analyze the global signal. Therefore, the fault characteristics are not obvious enough, and there is a possibility of misjudgment or missed judgment.

[0005] In order to achieve the above-mentioned object, according to one aspect of the present application, a method for high-resistance identification and accurate fault location of a smart distribution network is provided, the method comprising:

[0006] Step 1: The wide-area measurement system uses the zero-sequence voltage and current signals with a unified time stamp uploaded to the master station by the intelligent terminal of the distribution network, and performs mean interpolation, logarithmic transformation and normalization on the abnormal data to label the data set as normal and fault data;

[0007] Step 2: Use the Gram angle field to convert the time series signal into an image, fuse the full information features in the spatial domain, and form a fault feature map through data preprocessing and spatiotemporal feature fusion;

[0008] Step 3: Construct a fault feature map sample library;

[0009] Step 4: Use a dual-channel convolutional neural network to build a distribution network fault identification and accurate fault location model;

[0010] Step 5: Use the PyTorch artificial intelligence deep learning framework to verify the model on the main site GPU server, with 70% of the sample library used as a training set and 30% as a test set.

[0011] Optionally, in step 1, an exponentially weighted moving average method is used to solve the problem of data frame loss and reduce the impact of noise in the data.

[0012] Optionally, in step 2, the time series X = {x1, x2, …, x n}, first normalize to the interval [-1,1], calculation formula:

[0013]

[0014] in, is the normalized time series data point;

[0015] Secondly, convert the Cartesian coordinates of the time series data into polar coordinates and calculate the formula:

[0016]

[0017] αi is the polar angle, ri is the polar diameter;

[0018] Finally, the correlation of time series is mined according to the polar coordinate form, and the Gram angle and field G are defined. ASF G ADF , the calculation formula is:

[0019]

[0020] Optionally, in step 3, a fault feature map sample library is constructed according to the Gram angle sum field GASF and Gram difference field GADF matrices of the time series data, and is input into a dual-channel convolutional neural network for fault identification and fault location.

[0021] Optionally, the fault identification channel in step 4 includes two convolutional layers and two pooling layers, where the cross entropy loss function is:

[0022]

[0023] yc is a Boolean variable, indicating the logic state; c is the fault area;

[0024] The method selects 13 areas in the smart distribution network topology, where the normal state is represented by 14. The output of the last fully connected layer of the fault identification channel is used as the input of softmax. The softmax activation function can process the input data into a probability distribution, that is, the 13 fault areas and the normal state are output in the form of probability; the output of the softmax function is:

[0025]

[0026] Optionally, the fault location channel in step 4 includes three convolutional layers and three pooling layers, where the loss function is:

[0027]

[0028] Where N is the number of samples, Y t is the predicted value, is the actual value of the sample;

[0029] For the distribution network high-resistance precise positioning model, the fault location is:

[0030]

[0031] w i and x i are the weights and outputs of the last layer of neurons, respectively, and b i is the bias term, n l is the number of neurons in the last layer.

[0032] According to another aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute any one of the methods described.

[0033] According to another aspect of the present application, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, any one of the methods described above is implemented.

[0034] According to another aspect of the present application, an electronic device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to execute any one of the methods described through the computer program.

[0035] According to another aspect of the present application, a system for high-resistance identification and precise fault location in a smart distribution network is provided, the system comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include methods for executing any one of the methods described.

[0036] By applying the technical solution of the present application, the Gram angular field is used to convert the time series signal into a fused spatial domain full information feature map, and a dual-channel convolutional neural network model is used to mine the feature map information of the input samples. The convolutional neural network is calculated through the input layer, convolution layer, pooling layer and fully connected layer, combined with the fault identification and fault location dual-channel model, to achieve the detection and location of distribution network faults. The inventive method is not easily affected by transition resistance, thereby solving the problem that the prior art only considers the high-frequency or low-frequency components within a certain time period for distribution network fault detection, and does not analyze the global signal. Therefore, the fault characteristics are not obvious enough, and there is a possibility of misjudgment or missed judgment. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The drawings constituting part of the present application are used to provide a further understanding of the present application. The exemplary embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0038] Figure 1 This is a topological diagram of a 10kV distribution network containing distributed power sources;

[0039] Figure 2 It is a flow chart for fault identification and fault location;

[0040] Figure 3 It is a dual-channel convolutional neural network fault identification and fault location model. DETAILED DESCRIPTION

[0041] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0042] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0043] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0044] As introduced in the background technology, the traditional method is to use Fourier transform, short-time Fourier transform, wavelet decomposition and other technologies to decompose the transient signal, extract the high-frequency component of the first half cycle after the fault occurs, and compare it with the signal before the fault to obtain the fault characteristic quantity and identify the fault. The above methods only consider the high-frequency or low-frequency components in a certain time period, and do not analyze the global signal. Therefore, the fault characteristics are not obvious enough, and there is a possibility of misjudgment and missed judgment. In order to solve the problem that the prior art only considers the high-frequency or low-frequency components in a certain time period for distribution network fault detection, and does not analyze the global signal, so the fault characteristics are not obvious enough, and there is a possibility of misjudgment and missed judgment, the embodiments of the present application provide a method, medium, computer program product, electronic device and system for high-resistance identification and accurate fault location of an intelligent distribution network.

[0045] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0046] The present application provides a method for high-resistance identification and fault accurate location of a smart distribution network, the method comprising:

[0047] Step 1: The wide-area measurement system uses the zero-sequence voltage and current signals with a unified time stamp uploaded to the master station by the intelligent terminal of the distribution network, and performs mean interpolation, logarithmic transformation and normalization on the abnormal data to label the data set as normal and fault data;

[0048] Step 2: Use the Gram angle field to convert the time series signal into an image, fuse the full information features in the spatial domain, and form a fault feature map through data preprocessing and spatiotemporal feature fusion;

[0049] Step 3: Construct a fault feature map sample library;

[0050] Step 4: Use a dual-channel convolutional neural network to build a distribution network fault identification and accurate fault location model;

[0051] Step 5: Use the PyTorch artificial intelligence deep learning framework to verify the model on the main site GPU server, with 70% of the sample library used as a training set and 30% as a test set.

[0052] The zero-sequence voltage and zero-sequence current signals of each node in the smart distribution network are uploaded to the main station, and the abnormal data are processed by mean interpolation, logarithmic transformation and normalization. The idea of ​​Gram angular field transformation is then used to convert the time series signal into an image, and the full information features in the spatial domain are integrated. A sample library of spatiotemporal graphs of fault feature quantities is established. A dual-channel convolutional neural network model is used to perform convolution, activation function operations and pooling operations on the input samples, and the output of the fully connected layer is back-propagated to calculate the gradient of the loss function relative to the network parameters, and these gradients are used to update the weights and biases of the network, thereby minimizing the loss function. Through the establishment of a dual-channel convolutional neural network model, the present application can accurately identify and precisely locate high-resistance faults in the smart distribution network, and is not easily affected by transition resistance.

[0053] Specifically, Figure 1 As shown, firstly, step S1 is executed to collect data of the distribution network, and abnormal processing is performed on the frame loss data and the data affected by noise, and the exponential weighted moving average method is used to solve the problem of data frame loss and reduce the noise impact in the data. In step S1, the data of each node of the distribution network is collected. Figure 2 As shown, the acquisition device can be an intelligent terminal device, but is not limited to this device, and can also be a synchronous measurement device. Any device that can measure the voltage and current phase angle of the distribution network node is within the protection scope of the present invention. The synchronization of data adopts Beidou timing, which is not limited to the Beidou timing system, and can also adopt GPS timing.

[0054] like Figure 1 As shown, in step S2, by simulating the high-resistance fault scenarios in each section and the normal operation scenarios of the system, the zero-sequence voltage and zero-sequence current time series signals of each state are extracted, the time series signals are converted into images using the Gram angle field, the full information features in the spatial domain are fused, and the fault feature quantity map is formed through data preprocessing and spatiotemporal feature fusion.

[0055] like Figure 1 As shown, in step S3, a distribution network fault identification model and a fault precise location model are constructed based on a full-information spatiotemporal feature graph convolutional neural network.

[0056] like Figure 1 As shown, in step S4, the fault identification model sets the fault area Z4 line length to a high-resistance grounding fault at 50%, such as Figure 2As shown in the figure, the fault identification model inputs the Gram angular field feature map through two convolutions and two pooling operations. The tensor size of convolution layer 1 is 3*4*32, the pooling layer is 1*3, the tensor size of convolution layer 2 is 1*6*48, and the pooling layer is 1*3. Finally, the output result of the fully connected layer is input into the softmax activation function, and the fault area and fault status results are processed into a probability distribution form, which can accurately identify the distribution network fault and the fault area Z4.

[0057] In one embodiment of the present application, in step 1, an exponentially weighted moving average method is used to solve the problem of data frame loss and reduce the impact of noise in the data.

[0058] In one embodiment of the present application, in step 2, the time series X = {x1, x2, ..., x n}, first normalize to the interval [-1,1], calculation formula:

[0059]

[0060] in, is the normalized time series data point;

[0061] Secondly, convert the Cartesian coordinates of the time series data into polar coordinates and calculate the formula:

[0062]

[0063] αi is the polar angle, ri is the polar diameter;

[0064] Finally, the correlation of time series is mined according to the polar coordinate form, and the Gram angle and field G are defined. ASF G ADF , the calculation formula is:

[0065]

[0066] In one embodiment of the present application, in step 3, a fault feature map sample library is constructed based on the Gram angle sum field GASF and Gram difference field GADF matrices of the time series data, and is input into a dual-channel convolutional neural network for fault identification and fault location.

[0067] In one embodiment of the present application, the fault identification channel in step 4 includes two convolutional layers and two pooling layers, wherein the cross entropy loss function is:

[0068]

[0069] yc is a Boolean variable, indicating the logic state; c is the fault area;

[0070] The method selects 13 areas in the smart distribution network topology, where the normal state is represented by 14. The output of the last fully connected layer of the fault identification channel is used as the input of softmax. The softmax activation function can process the input data into a probability distribution, that is, the 13 fault areas and the normal state are output in the form of probability; the output of the softmax function is:

[0071]

[0072] like Figure 1 As shown, in step S4, the Gram angular field feature map is input into the fault location model through three convolution and pooling operations. The fault location channel includes three convolution layers and three pooling layers, where the convolution 1 tensor size is 3*4*32, the pooling layer 1 size is 1*3, the convolution layer 2 tensor size is 1*6*48, the pooling layer 2 size is 1*3, the convolution layer 3 tensor size is 2*4*64, and the pooling layer 3 size is 1*3. The loss function is:

[0073]

[0074] Where N is the number of samples, Y t is the predicted value, is the actual value of the sample;

[0075] The high-resistance fault at 50% of the fault line length is identified, and combined with the output results of the fault identification convolutional neural network, the high-resistance fault at 50% of the line length in the Z4 area of ​​the distribution network is accurately identified and located.

[0076] For the distribution network high-resistance precise positioning model, the fault location is:

[0077]

[0078] w i and x i are the weights and outputs of the last layer of neurons, respectively, and b i is the bias term, n l is the number of neurons in the last layer.

[0079] Combining the output results of the dual-channel convolutional neural network for fault identification and fault location, high-resistance faults in the distribution network can be accurately identified and located.

[0080] Dual-channel convolutional neural network fault identification and fault location model Figure 3 As shown, no further details will be given here.

[0081] The present application also provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute any one of the methods described.

[0082] The present application also provides a computer program product, including a computer program, which implements any of the methods described above when executed by a processor.

[0083] The present application also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute any one of the methods described through the computer program.

[0084] The present application also provides a system for high-resistance identification and precise fault location in a smart distribution network, the system comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include methods for executing any one of the methods described.

[0085] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0086] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0087] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes 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 generate 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.

[0088] 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.

[0089] 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.

[0090] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0091] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0092] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0093] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0094] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for high resistance identification and fault accurate location of intelligent distribution network, characterized in that: include: Step 1: The wide-area measurement system uses the zero-sequence voltage and current signals with a unified time stamp uploaded to the master station by the intelligent terminal of the distribution network, and performs mean interpolation, logarithmic transformation and normalization on the abnormal data to label the data set as normal and fault data; Step 2: Use the Gram angle field to convert the time series signal into an image, fuse the full information features in the spatial domain, and form a fault feature map through data preprocessing and spatiotemporal feature fusion; Step 3: Construct a fault feature map sample library; Step 4: Use a dual-channel convolutional neural network to build a distribution network fault identification and accurate fault location model; Step 5: Use the PyTorch artificial intelligence deep learning framework to verify the model on the main site GPU server, with 70% of the sample library used as a training set and 30% as a test set.

2. The method according to claim 1, characterized in that In step 1, the exponentially weighted moving average method is used to solve the problem of data frame loss and reduce the impact of noise in the data.

3. The method according to claim 1, characterized in that: In step 2, the time series X = {x1, x2, …, x n }, first normalize to the interval [-1,1], calculation formula: in, is the normalized time series data point; Secondly, convert the Cartesian coordinates of the time series data into polar coordinates and calculate the formula: αi is the polar angle, ri is the polar diameter; Finally, the correlation of time series is mined according to the polar coordinate form, and the Gram angle and field G are defined. ASF G ADF , the calculation formula is:

4. The method according to claim 1, characterized in that In step 3, a fault feature map sample library is constructed based on the Gram angle sum field GASF and Gram difference field GADF matrices of the time series data, and input into a dual-channel convolutional neural network for fault identification and fault location.

5. The method according to claim 1, characterized in that The fault identification channel in step 4 contains two convolutional layers and two pooling layers, where the cross entropy loss function is: yc is a Boolean variable, indicating the logic state; c is the fault area; The method selects 13 areas in the smart distribution network topology, where the normal state is represented by 14. The output of the last fully connected layer of the fault identification channel is used as the input of softmax. The soft ma activation function can process the input data into a probability distribution, that is, the 13 fault areas and the normal state are output in the form of probability; the output of the softmax function is:

6. The method according to claim 1, characterized in that The fault location channel in step 4 contains three convolutional layers and three pooling layers, where the loss function is: Where N is the number of samples, Y t is the predicted value, is the actual value of the sample; For the distribution network high-resistance precise positioning model, the fault location is: w i and x i are the weights and outputs of the last layer of neurons, respectively, and b i is the bias term, n l is the number of neurons in the last layer.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 6.

8. A computer program product, characterized in that The method comprises a computer program, which implements the method according to any one of claims 1 to 6 when being executed by a processor.

9. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 6 through the computer program.

10. A smart distribution network high resistance identification and fault accurate positioning system, characterized in that: include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include methods for executing any one of claims 1 to 6.

Images