Single-phase grounding fault detection method and system for small-current grounding system, and storage medium

Abstract

The invention discloses a single-phase grounding fault detection method and system for a small-current grounding system, and a storage medium. The method comprises the steps: obtaining the zero-sequence current and zero-sequence voltage of a to-be-detected line in a to-be-detected region; if the zero-sequence voltage of the to-be-detected line is greater than a starting value or the zero-sequence voltage break variable is greater than a set value, entering a single-phase earth fault detection step, otherwise, replacing one line in the to-be-detected area as the to-be-detected line, and repeating the above steps until entering the single-phase earth fault detection step; the single-phase earth fault detection step comprises the following steps of: preprocessing zero-sequence current and zero-sequence voltage of a line to be detected to obtain preprocessed image data; detecting the preprocessed image data by using a trained single-phase earth fault detection model to obtain a single-phase earth fault detection result; according to the invention, the accuracy and anti-interference capability of single-phase grounding fault detection in the small-current grounding system can be improved.

Description

Technical Field

[0001] The present invention relates to the interdisciplinary technical field of artificial intelligence and power distribution network technology, and in particular to a single-phase grounding fault detection method, system and storage medium for a low-current grounding system. Background Art

[0002] my country's medium and low voltage distribution networks mainly operate in a low current grounding mode. Common low current grounding systems include three types: neutral point ungrounded, arc suppression coil grounded, and resistor grounded. After long-term practice and exploration, my country's distribution network currently mainly uses the first two grounding methods. Among these systems, single-phase grounding faults have the highest probability of occurrence, accounting for more than 80% of the total number of faults. However, when a fault occurs, it is affected by a variety of external factors, the fault current is small and the situation is complicated, which brings great challenges to fault line selection (detection of faulty lines).

[0003] In the line selection of small current grounding system, traditional methods include zero-sequence current amplitude comparison method, zero-sequence admittance method, etc., which are examples based on the relationship between zero-sequence voltage and current as the judgment criteria. However, these methods are more or less unable to identify the fault line; and the use of artificial intelligence methods to identify images can better collect information on zero-sequence current and zero-sequence voltage; with the rapid development of artificial intelligence technology, image classification, as one of the core tasks in the field of computer vision, has received extensive attention and research. Traditional image classification methods mainly rely on artificially designed features, such as color, texture, and shape, which are limited in processing complex and diverse images; the rise of deep learning, especially the introduction of convolutional neural networks (CNN), has brought revolutionary breakthroughs in image classification; by automatically extracting multi-level features, CNN can achieve high-precision classification on large-scale image data sets. In recent years, more and more people have used artificial intelligence methods for fault detection, but the activation function used in the network structure is not complete enough, and the accuracy of grounding fault detection in small current grounding systems is poor, and the performance in noise resistance is poor.

[0004] The information disclosed in this background technology section is only intended to enhance the understanding of the overall background of the invention and should not be regarded as an acknowledgement or any form of suggestion that the information constitutes the prior art already known to ordinary technicians in this field. Summary of the invention

[0005] The object of the present invention is to provide a single-phase grounding fault detection method, system and storage medium for a small current grounding system, which can improve the accuracy and anti-interference ability of single-phase grounding fault detection in the small current grounding system.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] In a first aspect, the present invention provides a single-phase grounding fault detection method for a small current grounding system, comprising:

[0008] Obtaining zero-sequence current and zero-sequence voltage of the line to be tested in the area to be tested;

[0009] If the zero-sequence voltage of the line to be tested is greater than the starting value, or the zero-sequence voltage mutation is greater than the set value, the single-phase grounding fault detection step is entered; otherwise, another line is changed in the area to be tested as the line to be tested and the above steps are repeated until the single-phase grounding fault detection step is entered;

[0010] The single-phase ground fault detection step comprises:

[0011] Preprocessing the zero-sequence current and zero-sequence voltage of the line to be tested to obtain preprocessed image data;

[0012] The preprocessed image data is detected using a trained single-phase grounding fault detection model to obtain a single-phase grounding fault detection result.

[0013] Optionally, the preprocessing of the zero-sequence current and zero-sequence voltage of the line to be tested to obtain preprocessed image data includes:

[0014] A time window T is set, and a time domain image of the zero-sequence current and the zero-sequence voltage is generated based on the time window T and the zero-sequence current and the zero-sequence voltage of the line to be tested;

[0015] Normalizing the time domain images of the zero-sequence current and the zero-sequence voltage to obtain normalized time domain images of the zero-sequence current and the zero-sequence voltage;

[0016] The normalized time domain images of zero-sequence current and zero-sequence voltage are superimposed to obtain preprocessed image data.

[0017] Optionally, the time domain images of the zero-sequence current and the zero-sequence voltage are normalized, and the normalization formula is as follows:

[0018]

[0019] in, is the modulus value of the electrical quantity at the current moment in the normalized time domain image, and the electrical quantity is expressed as zero-sequence voltage or zero-sequence current; is the modulus value of the electrical quantity at the current moment of the time domain image; It is the maximum modulus of the electrical quantity within the time window T.

[0020] Optionally, the training step of the single-phase grounding fault detection model includes:

[0021] Acquire an image data set, wherein the image data set includes time domain images of zero-sequence current and voltage of a normal line and a fault line after preprocessing;

[0022] The image dataset is used as the training set for the single-phase ground fault detection model, and “line normal” and “line fault” are used as labels;

[0023] Based on the training set, the single-phase grounding fault detection model is trained to determine the optimal parameters of the single-phase grounding fault detection model;

[0024] According to the optimal parameters of the single-phase grounding fault detection model, a trained single-phase grounding fault detection model is obtained.

[0025] Optionally, the single-phase grounding fault detection model is obtained by improving the ResNet neural network using the Kolmogorov-Arnold theorem, including: a pooling layer, a first convolutional layer, a first residual module, a second residual module, a third residual module, a fourth residual module, a fully connected layer and a KAN classifier connected in sequence; the first residual module includes three first residual blocks connected in sequence; the first residual block includes a second convolutional layer, a KAN activation function layer, a third convolutional layer, a KAN activation function layer and a fourth convolutional layer connected in sequence. The second residual module includes four second residual blocks connected in sequence; the second residual block includes a fifth convolution layer, a KAN activation function layer, a sixth convolution layer, a KAN activation function layer and a seventh convolution layer connected in sequence; the third residual module includes six third residual blocks connected in sequence; the third residual block includes a fourth convolution layer, a KAN activation function layer, an eighth convolution layer, a KAN activation function layer and a ninth convolution layer connected in sequence; the fourth residual module includes three fourth residual blocks connected in sequence; the fourth residual block includes a seventh convolution layer, a KAN activation function layer, a tenth convolution layer, a KAN activation function layer and an eleventh convolution layer connected in sequence.

[0026] Optionally, the first convolution layer includes 64 7x7 convolution kernels; the second convolution layer includes 64 1x1 convolution kernels; the third convolution layer includes 64 3x3 convolution kernels; the fourth convolution layer includes 256 1x1 convolution kernels; the fifth convolution layer includes 128 1x1 convolution kernels; the sixth convolution layer includes 128 3x3 convolution kernels; the seventh convolution layer includes 512 1x1 convolution kernels; the eighth convolution layer includes 256 3x3 convolution kernels; the ninth convolution layer includes 1024 1x1 convolution kernels; the tenth convolution layer includes 512 3x3 convolution kernels; the eleventh convolution layer includes 2048 1x1 convolution kernels.

[0027] Optionally, the KAN activation function layer adopts a KAN activation function, which is expressed as follows:

[0028]

[0029] in, and These are all trainable parameters; is the SiLu activation function, x is the object; is the intermediate function; is some kind of spline activation function.

[0030] Optionally, the startup value is set to 0.1 times the rated voltage.

[0031] In a second aspect, the present invention provides a single-phase grounding fault detection system for a small current grounding system, comprising:

[0032] A data acquisition module, used for acquiring the zero-sequence current and zero-sequence voltage of the line to be tested in the area to be tested;

[0033] A judgment module is used to enter the single-phase grounding fault detection step if the zero-sequence voltage of the line to be tested is greater than the starting value, or the zero-sequence voltage mutation amount is greater than the set value; otherwise, another line in the area to be tested is used as the line to be tested and the above steps are repeated until the single-phase grounding fault detection step is entered;

[0034] A single-phase grounding fault detection module is connected to the data acquisition module and the judgment module respectively, and is used to preprocess the zero-sequence current and zero-sequence voltage of the line to be tested to obtain preprocessed image data; and is used to use the trained single-phase grounding fault detection model to detect the preprocessed image data to obtain a single-phase grounding fault detection result.

[0035] In a third aspect, the present invention provides a computer-readable storage medium storing a computer program, which, when executed, implements the single-phase grounding fault detection method for a small current grounding system described in the first aspect of claim 1.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] The present invention provides a single-phase grounding fault detection method, system and storage medium for a small current grounding system. The method first determines whether the line to be tested needs to be detected for faults based on the acquired zero-sequence current and zero-sequence voltage of the line to be tested, and then performs single-phase grounding fault detection or reselects the line as the line to be tested based on the judgment result; wherein the single-phase grounding fault detection uses a trained single-phase grounding fault detection model to accurately identify whether a single-phase grounding fault exists, obtains a detection result, improves the accuracy of fault detection, and has anti-interference ability as proved by experimental results;

[0038] The present invention provides a single-phase grounding fault detection method for a small current grounding system. An image data set is used as a training set for a single-phase grounding fault detection model, wherein the image data set includes time domain images of zero-sequence current and voltage preprocessed for a normal line and a fault line, and the time domain images are obtained by normalizing and superimposing the zero-sequence current time domain image and the zero-sequence voltage time domain image, so that the trained single-phase grounding fault detection model has good anti-interference ability;

[0039] The present invention provides a single-phase grounding fault detection method for a small current grounding system. The single-phase grounding fault detection model is obtained by improving the ResNet neural network based on the Kolmogorov-Arnold theorem, and the fault detection accuracy is improved compared with the original convolutional neural network (VGG16 network, AlexNet network, ResNet network, DenseNet network); and the single-phase grounding fault detection model is used for fault detection in a noisy environment. The detection accuracy is still high, indicating that the method has good anti-interference ability;

[0040] The present invention provides a single-phase grounding fault detection method for a small current grounding system. The single-phase grounding fault detection model improves the activation function in the ResNet network so that the activation function becomes more complete. When the activation function becomes more complex, the accuracy is higher than that of the original network, thereby increasing the accuracy of single-phase grounding fault detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 The figure is a schematic flow chart of a single-phase ground fault detection method in an embodiment of the present invention;

[0042] Figure 2 The figure shows an example of image data after preprocessing of a fault line in an embodiment of the present invention;

[0043] Figure 3 The figure shows an example of image data after normal line preprocessing in one embodiment of the present invention;

[0044] Figure 4The figure shows a R-KAN network structure diagram in one embodiment of the present invention;

[0045] Figure 5 The figure shows a distribution network simulation model diagram after adding DG in one embodiment of the present invention;

[0046] Figure 6 The figure is a schematic diagram of the change of loss value during the R-KAN network training process in an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0048] Example 1

[0049] like Figure 1 As shown, the embodiment of the present invention introduces a single-phase grounding fault detection method for a small current grounding system, comprising the following steps:

[0050] S1: Obtain zero-sequence current and zero-sequence voltage of the line to be tested in the area to be tested;

[0051] S2: If the zero-sequence voltage of the line to be tested is greater than the starting value, or the zero-sequence voltage mutation is greater than the set value, the single-phase grounding fault detection step is entered; otherwise, another line is changed in the area to be tested as the line to be tested and the above steps are repeated until the single-phase grounding fault detection step is entered;

[0052] The single-phase ground fault detection step comprises:

[0053] S3: preprocessing the zero-sequence current and zero-sequence voltage of the line to be tested to obtain preprocessed image data;

[0054] S4: Detect the preprocessed image data using the trained single-phase grounding fault detection model to obtain a single-phase grounding fault detection result.

[0055] The present embodiment provides a single-phase grounding fault detection method for a small current grounding system, which can first determine whether the line to be tested needs to be detected for faults based on the acquired zero-sequence current and zero-sequence voltage of the line to be tested, and then perform single-phase grounding fault detection or reselect the line as the line to be tested based on the judgment result; wherein the single-phase grounding fault detection is to use a trained single-phase grounding fault detection model to accurately identify whether a single-phase grounding fault exists, obtain a fault detection result, and improve the fault line selection, that is, the accuracy and anti-interference ability of fault detection.

[0056] In this embodiment, the zero-sequence current and zero-sequence voltage of the line to be tested in the test area are obtained in step S1, and are collected in real time by the built-in sensors (zero-sequence current transformer and zero-sequence voltage transformer) of the microcomputer protection device; the microcomputer protection device is a key equipment in the modern power system. It combines computer technology, signal processing technology and automatic control technology, and has the characteristics of high precision, high reliability and intelligence. Since the device is a prior art and is not an innovation of the present invention, it will not be described in detail here.

[0057] In this embodiment, if the zero-sequence voltage of the line to be tested is greater than the starting value, or the zero-sequence voltage mutation amount is greater than the set value in step S2, the single-phase grounding fault detection step is entered, otherwise, another line is changed in the test area as the line to be tested and the above steps are repeated until the single-phase grounding fault detection step is entered, which specifically includes:

[0058] Based on the zero-sequence voltage U0 obtained in real time in step S1, it is judged whether the line to be tested needs to enter the single-phase grounding fault detection; if necessary, it enters the single-phase grounding fault detection steps S3-S4; if not, another line is selected as the line to be tested, and the process returns to step S1;

[0059] The judgment basis is: zero sequence voltage U0> starting value U set , or zero-sequence voltage mutation △U0> preset value ε;

[0060] Specifically, U set The value is 0.1 times the rated voltage, △U0 is the zero-sequence voltage at the current moment minus the zero-sequence voltage at the previous moment, and ε is 0.5-1V.

[0061] In this embodiment, in step S3, the zero-sequence current and zero-sequence voltage of the line to be tested are preprocessed to obtain preprocessed image data. The specific steps are as follows:

[0062] S31: setting a time window T, and generating a time domain image of the zero-sequence current and the zero-sequence voltage based on the time window T and the zero-sequence current and the zero-sequence voltage of the line to be tested;

[0063] Specifically, Figure 2-3 As shown in the figure, the time window T is set to 0.06 seconds; based on the zero-sequence current and zero-sequence voltage collected by the protection device, the time domain images of the zero-sequence current and zero-sequence voltage are generated by using data processing software or algorithms according to the setting of the time window T and the requirements of the sampling frequency. The time domain image usually uses time as the horizontal axis and the zero-sequence current or zero-sequence voltage as the vertical axis to display the change trend of the data in the form of a curve chart.

[0064] S32: normalizing the time domain images of the zero-sequence current and the zero-sequence voltage to obtain normalized time domain images of the zero-sequence current and the zero-sequence voltage;

[0065] Specifically, the normalization formula is as follows:

[0066]

[0067] in, is the modulus value of the electrical quantity at the current moment in the normalized time domain image, and the electrical quantity is expressed as zero-sequence voltage or zero-sequence current; is the modulus value of the electrical quantity at the current moment of the time domain image; It is the maximum modulus of the electrical quantity within the time window T.

[0068] S33: Superimpose the normalized time domain images of zero-sequence current and zero-sequence voltage to obtain preprocessed image data, such as Figure 2 and Figure 3 Shown are the image data of the faulty line and the normal line after preprocessing.

[0069] In this embodiment, the training steps of the single-phase grounding fault detection model in step S4 are as follows:

[0070] Acquire an image data set, wherein the image data set includes time domain images of zero-sequence current and voltage of a normal line and a fault line after preprocessing;

[0071] The image dataset is used as the training set of the single-phase grounding fault detection model, and "line normal" and "line fault" are used as labels; "line fault" means that there is a single-phase grounding fault in the line;

[0072] Based on the training set, the single-phase grounding fault detection model is trained to determine the optimal parameters of the single-phase grounding fault detection model;

[0073] According to the optimal parameters of the single-phase grounding fault detection model, a trained single-phase grounding fault detection model is obtained.

[0074] Specifically, the single-phase grounding fault detection model (R-KAN network, ResNet-Kolmogorov–ArnoldNetworks network) is obtained by improving the ResNet neural network using the Kolmogorov–Arnold theorem;

[0075] The Kolmogorov-Arnold theorem states that if is a multivariate continuous function on a bounded domain, then It can be written as a combination of a finite number of single-variable continuous functions and addition operations of binary functions.

[0076] Furthermore, for a smooth function ;

[0077]

[0078] in, is a function, and ; is a mapping function, , represents the set of real numbers, and n, p, and q represent sequence numbers.

[0079] Specifically, Figure 4 As shown, the single-phase grounding fault detection model, that is, the structure of the R-KAN network includes:

[0080] The pooling layer, the first convolutional layer (conv1), the first residual module, the second residual module, the third residual module, the fourth residual module, the fully connected layer and the KAN classifier are connected in sequence;

[0081] The first residual module includes three first residual blocks connected in sequence; the first residual block includes a second convolutional layer (conv2), a KAN activation function layer (KAN), a third convolutional layer (conv3), a KAN activation function layer (KAN) and a fourth convolutional layer (conv4) connected in sequence.

[0082] The second residual module includes four second residual blocks connected in sequence; the second residual block includes a fifth convolution layer (conv5), a KAN activation function layer, a sixth convolution layer (conv6), a KAN activation function layer and a seventh convolution layer (conv7) connected in sequence;

[0083] A third residual module, comprising six third residual blocks connected in sequence; the third residual block comprises a fourth convolutional layer (conv4), a KAN activation function layer, an eighth convolutional layer (conv8), a KAN activation function layer and a ninth convolutional layer (conv9) connected in sequence;

[0084] The fourth residual module includes three sequentially connected fourth residual blocks; the fourth residual block includes a seventh convolution layer (conv7), a KAN activation function layer, a tenth convolution layer (conv10), a KAN activation function layer and an eleventh convolution layer (conv11) which are sequentially connected.

[0085] Specifically, the first convolution layer includes 64 7x7 convolution kernels; the second convolution layer includes 64 1x1 convolution kernels; the third convolution layer includes 64 3x3 convolution kernels; the fourth convolution layer includes 256 1x1 convolution kernels; the fifth convolution layer includes 128 1x1 convolution kernels; the sixth convolution layer includes 128 3x3 convolution kernels; the seventh convolution layer includes 512 1x1 convolution kernels; the eighth convolution layer includes 256 3x3 convolution kernels; the ninth convolution layer includes 1024 1x1 convolution kernels; the tenth convolution layer includes 512 3x3 convolution kernels; the eleventh convolution layer includes 2048 1x1 convolution kernels.

[0086] In this embodiment, the above-mentioned Kolmogorov-Arnold theorem (KAN theorem) is used to change the formula of the forward transfer of the neural network to:

[0087]

[0088] in, for Gradually nested to sign in Result; is the mapping function for the kth formula nesting, is the input of the network, For network output, The essence of The weight bias operation in is combined with the activation function. It is called KAN activation function.

[0089] The activation function formula is as follows:

[0090]

[0091] in, and These are all trainable parameters; is the SiLu activation function, x is the object; is the intermediate function; is some kind of spline activation function.

[0092] In this embodiment, the model training verification hardware and software environment of the patent of the present invention is: 13th Gen intelcore(TM) i5-13600KF processor, NVIDIA 4070 ti super (16G), memory size is 32G, python3.11, CUDN121, pytorch2.2.1.

[0093] Example 2

[0094] Based on the single-phase grounding fault detection method for a small current grounding system provided in Example 1, this embodiment provides a calculation example, which is introduced as follows:

[0095] This embodiment uses PSCAD / EMTDC to simulate the system and build a simulation model. The simulation structure diagram is as follows: Figure 5 The figure shows the distribution network simulation model after adding DG (Distributed Generation). The simulation model has four feeders with lengths of 4km, 5km, 4km, and 5km respectively. The remaining parameters are set as follows: the system power supply voltage is 10kV, the frequency is 50Hz, the number of connected DGs is 2, one of which is a wind turbine model and the other is a photovoltaic model, and the inductance value Lg of the arc suppression coil is 0.3H. The simulation sampling frequency is 40MHz, and the time when the fault starts is 0.8s. The time window of the neural network training and validation set is set to 0.005s. The first feeder segmentation is set to 1.5km, 1.5km, and 1km, and the other feeders are set in half. The line segment close to the busbar is an overhead line, and the one far away is a cable line. The line parameters are shown in Table 1:

[0096] Table 1 Line parameters

[0097]

[0098] The R-KAN neural network is first trained with parameters, and the loss value changes during the training process are as follows: Figure 6 , as the number of iterations increases, the loss value gradually decreases and eventually approaches 0.

[0099] To verify the R-KAN neural network, you need to build a test set. The method to build the test set is as follows:

[0100] Multiple sets of zero-sequence voltage and zero-sequence current data are obtained through simulation, and the superposition waveforms of zero-sequence voltage and zero-sequence current are made. The composition of the training set is shown in Table 2. In Tables 2 and 3, (a, b km) means that the access point is at feeder a, b kilometers away from the first section.

[0101] Define the judgment accuracy of the data set It is the ratio of the number of correct samples to the total number of samples.

[0102] Table 2 Ground fault parameters of the test data set

[0103]

[0104] Verification test 1 (comparison of accuracy of different convolutional neural networks):

[0105] By changing the grounding resistance value and position of the fault point, a test data set is prepared as shown in Table 2. The test data set is put into the VGG16 network, AlexNet network, ResNet network, DenseNet network and R-KAN network respectively. The detection accuracy results are shown in Table 3:

[0106] Table 3 Detection accuracy of different types of convolutional neural networks

[0107]

[0108] From the above table, we can see that the detection accuracy of R-KAN reaches 100%, which is higher than the accuracy of VGG16 network, AlexNet network, ResNet network and DenseNet network. R-KAN neural network, i.e., the single-phase grounding fault detection model proposed by the present invention, effectively improves the accuracy of single-phase grounding fault detection in small current grounding system.

[0109] Verification test 2 (accuracy comparison of noisy data set):

[0110] In active distribution networks, power electronic devices such as inverters will introduce noise, which will interfere with the voltage and current of the system, which may affect the accuracy of fault identification. Therefore, it is necessary to verify the accuracy of fault identification on lines affected by noise. To this end, 20Hz Gaussian white noise is added to the original zero-sequence voltage and zero-sequence current waveforms to generate a noise data set. The test data sets are placed in the VGG16 network, AlexNet network, ResNet network, DenseNet network and R-KAN network respectively, and the detection accuracy results are shown in Table 4;

[0111] Table 4 Detection accuracy of different types of convolutional neural networks on the noise set

[0112]

[0113] From the table above, we can see that R-KAN has the best anti-interference ability. Even under the influence of Gaussian white noise, the detection accuracy is still 99.11%.

[0114] Verification test 3 (field recording verification test):

[0115] The data after the fault occurred on site was extracted through the recording software, and the zero-sequence voltage and zero-sequence current superposition diagram was made, and then put into the R-KAN network for on-site recording verification test. After field verification, R-KAN's accuracy rate for detecting small current single-phase grounding faults is 100%.

[0116] In summary, this patent improves the ResNet neural network into an R-KAN neural network by using the Kolmogorov-Arnold theorem, which improves the accuracy of single-phase grounding fault line detection in low-current grounding systems compared to the original convolutional neural networks (VGG16 network, AlexNet network, ResNet network, DenseNet network). The R-KAN neural network has better anti-interference ability than the original convolutional neural network. In addition, the R-KAN network has been verified by field recording data, and its fault detection accuracy is 100%. The new artificial intelligence network proposed in this patent improves the accuracy of single-phase grounding fault line detection in low-current grounding systems.

[0117] Example 3

[0118] This embodiment provides a single-phase grounding fault detection system for a small current grounding system, including:

[0119] A data acquisition module, used for acquiring the zero-sequence current and zero-sequence voltage of the line to be tested in the area to be tested;

[0120] A judgment module is used to enter the single-phase grounding fault detection step if the zero-sequence voltage of the line to be tested is greater than the starting value, or the zero-sequence voltage mutation amount is greater than the set value; otherwise, another line in the area to be tested is used as the line to be tested and the above steps are repeated until the single-phase grounding fault detection step is entered;

[0121] A single-phase grounding fault detection module is connected to the data acquisition module and the judgment module respectively, and is used to preprocess the zero-sequence current and zero-sequence voltage of the line to be tested to obtain preprocessed image data; and is used to use the trained single-phase grounding fault detection model to detect the preprocessed image data to obtain a single-phase grounding fault detection result.

[0122] Example 4

[0123] This embodiment provides a computer-readable storage medium storing a computer program, which, when executed, implements the single-phase grounding fault detection method for a small current grounding system described in claim embodiment 1.

[0124] 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 combining software and hardware. Moreover, the present application may adopt 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 codes.

[0125] 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.

[0126] 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.

[0127] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the enlightenment of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the purpose of the present invention and the claims, which all fall within the protection of the present invention.

Claims

1. A single-phase grounding fault detection method for a small current grounding system, characterized in that: include: Obtaining zero-sequence current and zero-sequence voltage of the line to be tested in the area to be tested; If the zero-sequence voltage of the line to be tested is greater than the starting value, or the zero-sequence voltage mutation is greater than the set value, the single-phase grounding fault detection step is entered; otherwise, another line is changed in the area to be tested as the line to be tested and the above steps are repeated until the single-phase grounding fault detection step is entered; The single-phase ground fault detection step comprises: Preprocessing the zero-sequence current and zero-sequence voltage of the line to be tested to obtain preprocessed image data; The preprocessed image data is detected using a trained single-phase grounding fault detection model to obtain a single-phase grounding fault detection result.

2. The single-phase grounding fault detection method for a small current grounding system according to claim 1, characterized in that: The preprocessing of the zero-sequence current and zero-sequence voltage of the line to be tested to obtain preprocessed image data includes: A time window T is set, and a time domain image of the zero-sequence current and the zero-sequence voltage is generated based on the time window T and the zero-sequence current and the zero-sequence voltage of the line to be tested; Normalizing the time domain images of the zero-sequence current and the zero-sequence voltage to obtain normalized time domain images of the zero-sequence current and the zero-sequence voltage; The normalized time domain images of zero-sequence current and zero-sequence voltage are superimposed to obtain preprocessed image data.

3. The single-phase grounding fault detection method for a small current grounding system according to claim 2, characterized in that: The time domain images of the zero-sequence current and the zero-sequence voltage are normalized, and the normalization formula is as follows: ； in, is the modulus value of the electrical quantity at the current moment in the normalized time domain image, and the electrical quantity is expressed as zero-sequence voltage or zero-sequence current; is the modulus value of the electrical quantity at the current moment of the time domain image; It is the maximum modulus of the electrical quantity within the time window T.

4. The single-phase grounding fault detection method for a small current grounding system according to claim 3, characterized in that: The training steps of the single-phase ground fault detection model include: Acquire an image data set, wherein the image data set includes time domain images of zero-sequence current and voltage of a normal line and a fault line after preprocessing; The image dataset is used as the training set for the single-phase ground fault detection model, and "line normal" and "line fault" are used as labels; Based on the training set, the single-phase grounding fault detection model is trained to determine the optimal parameters of the single-phase grounding fault detection model; According to the optimal parameters of the single-phase grounding fault detection model, a trained single-phase grounding fault detection model is obtained.

5. The single-phase grounding fault detection method for a small current grounding system according to claim 4, characterized in that: The single-phase grounding fault detection model is obtained by improving the ResNet neural network using the Kolmogorov-Arnold theorem, and includes: a pooling layer, a first convolution layer, a first residual module, a second residual module, a third residual module, a fourth residual module, a fully connected layer and a KAN classifier connected in sequence; the first residual module includes three first residual blocks connected in sequence; the first residual block includes a second convolution layer, a KAN activation function layer, a third convolution layer, a KAN activation function layer and a fourth convolution layer connected in sequence. The second residual module includes four second residual blocks connected in sequence; the second residual block includes a fifth convolution layer, a KAN activation function layer, a sixth convolution layer, a KAN activation function layer and a seventh convolution layer connected in sequence; the third residual module includes six third residual blocks connected in sequence; the third residual block includes a fourth convolution layer, a KAN activation function layer, an eighth convolution layer, a KAN activation function layer and a ninth convolution layer connected in sequence; the fourth residual module includes three fourth residual blocks connected in sequence; the fourth residual block includes a seventh convolution layer, a KAN activation function layer, a tenth convolution layer, a KAN activation function layer and an eleventh convolution layer connected in sequence.

6. The single-phase grounding fault detection method for a small current grounding system according to claim 5, characterized in that: The first convolution layer includes 64 7x7 convolution kernels; the second convolution layer includes 64 1x1 convolution kernels; the third convolution layer includes 64 3x3 convolution kernels; the fourth convolution layer includes 256 1x1 convolution kernels; the fifth convolution layer includes 128 1x1 convolution kernels; the sixth convolution layer includes 128 3x3 convolution kernels; the seventh convolution layer includes 512 1x1 convolution kernels; the eighth convolution layer includes 256 3x3 convolution kernels; the ninth convolution layer includes 1024 1x1 convolution kernels; the tenth convolution layer includes 512 3x3 convolution kernels; the eleventh convolution layer includes 2048 1x1 convolution kernels.

7. The single-phase grounding fault detection method for a small current grounding system according to claim 6, characterized in that: The KAN activation function layer adopts the KAN activation function, and the formula is as follows: ； in, and These are all trainable parameters; is the SiLu activation function, x is the object; is the intermediate function; is some kind of spline activation function.

8. The single-phase grounding fault detection method for a small current grounding system according to claim 1, characterized in that: The startup value is set to 0.1 times the rated voltage.

9. A single-phase grounding fault detection system for a small current grounding system, characterized in that: include: A data acquisition module, used for acquiring the zero-sequence current and zero-sequence voltage of the line to be tested in the area to be tested; A judgment module is used to enter the single-phase grounding fault detection step if the zero-sequence voltage of the line to be tested is greater than the starting value, or the zero-sequence voltage mutation amount is greater than the set value; otherwise, another line in the area to be tested is used as the line to be tested and the above steps are repeated until the single-phase grounding fault detection step is entered; A single-phase grounding fault detection module, which is connected to the data acquisition module and the judgment module respectively, and is used to pre-process the zero-sequence current and zero-sequence voltage of the line to be tested to obtain pre-processed image data; It is used to detect the preprocessed image data using the trained single-phase grounding fault detection model to obtain a single-phase grounding fault detection result.

10. A computer-readable storage medium, characterized in that: A computer program is stored therein, and when the computer program is executed, the single-phase grounding fault detection method for a small current grounding system according to any one of claims 1 to 8 is implemented.

Images