A method and system for testing series arc faults in AC and DC power supply circuits

Abstract

The present invention provides a method and system for testing a series arc fault in an AC / DC power supply circuit, and relates to the technical field of DC fault arc testing. The method comprises: decomposing and reconstructing a load loop current signal based on a reconstructed singular value algorithm to obtain a singular value reconstructed current signal; utilizing a fault mode decomposition method to decompose the reconstructed current signal into multiple mode components by selecting a mode and a center frequency, introducing a multi-mode arrangement algorithm to perform sequence quantization on the decomposed multiple mode components, performing feature selection, and determining a main feature component; inputting the main feature component into a classification neural network to obtain the type of arc fault; obtaining a fault traveling wave transmitted by a cable during a current arc fault, analyzing the natural frequency in the fault traveling wave, calculating the distance of the arc fault according to the natural frequency, and locating the position of the arc fault.

Description

Technical Field

[0001] The present disclosure relates to the technical field of DC fault arc testing, and in particular to a method and system for testing series arc faults in AC and DC power supply circuits. Background Art

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art.

[0003] Uninterruptible AC and DC power supplies need to cover relatively complex and changeable usage scenarios. Sometimes, battery packs are required to output DC power alone. For example, the full-capacity discharge of battery packs in substations below 10kV, with the battery pack and rectifier module working together to output AC power. For example, the control and protection power supply of a substation, which is composed of a battery pack and a UPS module, is used for power protection operations for important activities.

[0004] At the same time, fault testing of AC and DC power supply circuits is also crucial to the operation of the power grid. According to the connection relationship between the arc and the circuit when the arc fault occurs, the arc fault can be divided into many different types. For series arc faults, since the arc is connected in series in the circuit, the impedance load of the arc limits the current size, and the current in the circuit is less than the load current during normal operation. Ordinary overcurrent protection devices and fuses cannot identify arc faults and protect the circuit. At the same time, the normal current waveform of certain nonlinear loads and interference conditions also has the current characteristics of the arc, which increases the difficulty of identifying series arc faults. The occurrence of arc faults is not accompanied by leakage current. The residual current protector can only detect this leakage current and cannot identify the arc. Currently, there is no fault test for low-voltage AC series arcs.

[0005] The existing arc fault simulation detection is carried out by constructing an arc test mathematical model. However, there are assumptions and uncertainties in the process of establishing the mathematical model, which cannot fully reflect all the characteristics of the arc fault. In addition, for the circuit connected to the load, when an arc fault occurs in the transmission line, it is necessary to accurately calibrate the time for the fault traveling wave to reach the load to achieve the distance measurement of the arc fault. The existing method requires the calibration of the initial traveling wave, and the time to determine the arrival at the ranging point must be very precise, which is difficult. Summary of the invention

[0006] In order to solve the above problems, the present disclosure proposes a method and system for testing series arc faults in AC and DC power supply circuits. The main characteristic components of the arc fault in the current signal are extracted by reconstructing and modal decomposing the extracted current signal, and the type and location distance of the arc fault are determined by using the fault traveling wave during the arc fault.

[0007] According to some embodiments, the present disclosure adopts the following technical solutions:

[0008] A method for testing series arc faults in AC and DC power supply circuits, comprising:

[0009] Extracting the load loop current signal of the circuit;

[0010] Based on the reconstruction singular value algorithm, the load loop current signal is decomposed and reconstructed to obtain a singular value reconstructed current signal;

[0011] The fault mode decomposition method is used to decompose the reconstructed current signal into multiple modal components by selecting the mode and the center frequency. The multi-modal arrangement algorithm is introduced to quantize the decomposed multiple modal components, perform feature selection, and determine the main characteristic components.

[0012] The main characteristic components are input into a classification neural network to obtain the type of arc fault; the fault traveling wave transmitted by the cable during the current arc fault is obtained, the natural frequency of the fault traveling wave is analyzed, the distance of the arc fault is calculated according to the natural frequency, and the position of the arc fault is located.

[0013] According to some embodiments, the present disclosure adopts the following technical solutions:

[0014] A series arc fault testing system for AC and DC power supply circuits, comprising:

[0015] A data acquisition module, used for extracting a load loop current signal of a circuit;

[0016] A signal reconstruction module, used for decomposing and reconstructing the load loop current signal based on a reconstruction singular value algorithm to obtain a singular value reconstructed current signal;

[0017] The modal feature extraction module is used to use the fault modal decomposition method to decompose the reconstructed current signal into multiple modal components by selecting the mode and the center frequency, introduce the multi-modal arrangement algorithm to sequence quantize the decomposed multiple modal components, perform feature selection, and determine the main characteristic components;

[0018] The classification and distance measurement module is used to input the main characteristic components into the classification neural network to obtain the type of arc fault; obtain the fault traveling wave transmitted by the cable during the current arc fault, analyze the natural frequency of the fault traveling wave, calculate the distance of the arc fault according to the natural frequency, and locate the arc fault position.

[0019] According to some embodiments, the present disclosure adopts the following technical solutions:

[0020] A non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, a method for testing series arc faults in AC and DC power supply circuits is implemented.

[0021] According to some embodiments, the present disclosure adopts the following technical solutions:

[0022] An electronic device comprises: a processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory so that the electronic device executes the method for testing series arc faults in AC and DC power supply circuits.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] The present invention discloses a method for testing series arc faults in AC and DC power supply circuits. The method constructs a trajectory matrix by increasing the delay step of data points between rows and combines a singular value energy difference spectrum method and a singular value energy mean method to obtain a reconstructed singular value algorithm of the effective order of singular values. This method greatly improves the operation speed while retaining a good denoising effect.

[0025] The present invention discloses a method for testing series arc faults in AC and DC power supply circuits. A multi-modal arrangement algorithm is introduced to perform sequence quantization on the decomposed multiple modal components, and feature selection is performed to determine the main feature components. The components can automatically divide the signal frequency band into different frequency ranges, which means that higher frequency components contain more fault information, thereby effectively avoiding the misjudgment problem caused by using components with fixed frequency bands.

[0026] The present invention discloses a method for testing arc faults in series on AC and DC power supply circuits. The method uses the natural frequency to measure the arc fault distance of cables in a DC distribution network, and describes the transmission process of a fault traveling wave after an arc fault occurs on a DC transmission line. The method can accurately measure the arc fault distance and locate the fault. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The accompanying drawings constituting a part of the present disclosure are used to provide a further understanding of the present disclosure. The illustrative embodiments of the present disclosure and their descriptions are used to explain the present disclosure and do not constitute an improper limitation on the present disclosure.

[0028] Figure 1 A flowchart for implementing the method of the embodiment of the present disclosure;

[0029] Figure 2 The present invention is a flowchart of series arc fault detection based on a reconstructed singular value algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The present disclosure is further described below in conjunction with the accompanying drawings and embodiments.

[0031] It should be noted that the following detailed descriptions are all illustrative and are intended to provide further explanation of the present disclosure. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present disclosure belongs.

[0032] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present disclosure. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0033] Example 1

[0034] In one embodiment of the present disclosure, a method for testing a series arc fault in an AC / DC power supply circuit is provided, comprising:

[0035] Step 1: extract the load loop current signal of the circuit;

[0036] Step 2: Decompose and reconstruct the load loop current signal based on a reconstructed singular value algorithm to obtain a singular value reconstructed current signal;

[0037] Step 3: Using the fault mode decomposition method, the reconstructed current signal is decomposed into multiple modal components by selecting the mode and the center frequency. The multi-modal arrangement algorithm is introduced to sequence the decomposed multiple modal components, perform feature selection, and determine the main feature components.

[0038] Step 4: Input the main characteristic components into the classification neural network to obtain the arc fault type; obtain the fault traveling wave transmitted by the cable during the current arc fault, analyze the natural frequency of the fault traveling wave, calculate the distance of the arc fault according to the natural frequency, and locate the arc fault position.

[0039] As an embodiment, the specific implementation process of a method for testing a series arc fault in an AC / DC power supply circuit disclosed in the present invention is as follows:

[0040] Step 1: Extract the load loop current signal of the circuit;

[0041] Load loop current signal to the circuit x i The current transformer is used to obtain the current signal in the loop, and the voltage signal output by the secondary side is filtered out by a high-pass filter to remove the DC and low-frequency components, and finally enters MATLAB for analysis through the acquisition card.

[0042] Step 2: Decompose and reconstruct the load loop current signal based on the reconstructed singular value algorithm to obtain a singular value reconstructed current signal;

[0043] The load loop current signal is decomposed and reconstructed based on the reconstruction singular value algorithm, including:

[0044] Based on the load loop current signal, a current signal trajectory matrix is ​​constructed, and the current signal is subjected to singular value decomposition. The original current signal is decomposed into multiple component signals, which are then superimposed. Each singular value represents a component. By selecting valid singular values ​​and setting the remaining singular values ​​to zero, the noise decomposition and removal of the original current signal is completed.

[0045] Specifically, singular value decomposition (SVD) is a matrix orthogonalization decomposition method, so the first step in performing singular value decomposition on a one-dimensional signal is to construct a trajectory matrix. x i The constructible trajectory matrix A is as follows:

[0046]

[0047] in, m、n are the number of rows and columns of the matrix respectively, and τ is the delay step size of the trajectory matrix.

[0048] When τ = 1, the trajectory matrix is ​​called a Hankel matrix. When τ = n, the trajectory matrix is ​​called a periodic truncation matrix. A ∈ R m × n When , there exists an orthogonal matrix U∈ R m × m and the orthogonal matrix V∈ R n × n So that:

[0049]

[0050] in, , , r For the matrix A The rank of is 0, which is a zero matrix. l = min(m,n) , there are diagonal elements σ1≥σ2≥…≥σ r > σ r + 1 = … = σ l = 0, this l The diagonal elements are collectively called the matrix A The singular values ​​of . U For the matrix A The left singular vector matrix of , V is the right singular vector matrix.

[0051] After removing the singular values ​​equal to zero, In short:

[0052]

[0053] in, u i and v i The matrices are U , V No. i vectors.

[0054] Singular value decomposition is to decompose the original signal into a series of component signals and superimpose them. Each singular value represents a component. Therefore, by selecting valid singular values ​​and setting the remaining singular values ​​to zero, the noise in the original signal can be removed.

[0055] Further, the steps of singular value decomposition are: Since the matrix U and A They are all orthogonal matrices, so we can get:

[0056]

[0057] Among them, the matrix A T The eigenvector matrix of is the left singular vector matrix of matrix A. T The eigenvalue of λ i and the singular values ​​of the matrix A σ i The relationship is,

[0058]

[0059] At present, the current signal is constructed into a trajectory matrix A for singular value reconstruction and decomposition, which is a denoising process. The present invention improves the decomposition process by increasing the delay step length of the trajectory matrix. τ To reduce the number of lines m Reaching the reduction matrix AA T The scale of the delay step reduces the operation time, but at the same time it also reduces the number of singular values, making the selection of the effective order of singular values ​​more difficult and the denoising effect worse. τ The selection of needs to comprehensively consider the operation time and denoising effect. Secondly, the number of columns in the trajectory matrix n The change in the overall computing time does not increase significantly, and n When n is half the signal length, the data correlation between adjacent rows is the highest and the signal separation capability is the strongest. Therefore, n = 512 is finally selected. τ= 8, making m = 65. The new trajectory matrix is,

[0060]

[0061] In addition, the position where the singular value sequence changes is often the dividing point between the useful signal and the noise signal singular value. Considering that the noise is always much larger than other signals under different currents, the singular value energy difference spectrum {bi} is constructed, and its calculation formula is:

[0062]

[0063] Since the larger the signal amplitude, the larger the corresponding singular value energy, we only need to set the first y singular values ​​of the singular value sequence to zero to remove the noise. Secondly, for random noise with large amplitude, considering that the existence of arc signal makes the singular value energy in the fault state greater than the singular value energy in the normal state, we use 1.6 times the mean of the remaining singular value energy as the threshold k, and the expression is:

[0064]

[0065] The remaining singular values ​​with energy greater than the threshold k are set to zero, so that random noise with large amplitude can be removed. The threshold is set so that in normal state, the threshold is small, and noise is better removed. In fault state, the threshold is large, and arc characteristics are better preserved.

[0066] Step 3: Using the fault mode decomposition method, the reconstructed current signal is decomposed into multiple modal components by selecting the mode and the center frequency. The multi-modal arrangement algorithm is introduced to sequence the decomposed multiple modal components, perform feature selection, and determine the main characteristic components;

[0067] Among them, the fault modal decomposition method includes: analyzing the reconstructed current signal, calculating its unilateral spectrum value, setting the operator, allocating the center band of the analytical signal to the corresponding baseband, solving the gradient information of the analytical signal, and introducing a variety of operators and penalty factors by calculating the frequency limit of the analytical signal. The mode and center frequency are continuously updated to meet the iteration accuracy and obtain multiple modal components.

[0068] Specifically, the fault modal decomposition uses the variational modal decomposition algorithm, which decomposes the complex signal into multiple modal components by selecting the mode and center frequency, while keeping each component as monotonous as possible, making the decomposed signal easier to interpret and analyze. In addition, the algorithm has a strong anti-interference ability, can overcome the influence of high-frequency noise, and balance the error between the time domain signal and the corresponding frequency domain signal. The solution process is as follows:

[0069] 1) Reconstructed current analytical signal obtained by Hilbert transform X(t), analyze the signal and calculate its unilateral spectrum value, and the operator Multiply, and then adjust the center band of the analytical signal to the corresponding baseband.

[0070]

[0071] 2) Solve the norm L of the signal gradient information 2 , and the frequency limits of the decomposed modal components (IMFs) are calculated simultaneously.

[0072]

[0073] in, is the Dirac function, represents the decomposed IMF components, Represents the center frequency of each part, K is the number of modes, k is the current mode number between [1,K].

[0074] 3) Introducing Lagrange multipliers , signal reconstruction operator S ( t ) and penalty factor Expand the Lagrangian expression and rebuild the constraints.

[0075]

[0076] 4) Continuously update the pattern set and center frequency , until the iteration accuracy is met and multiple modal components are obtained.

[0077] Furthermore, a multi-scale permutation entropy algorithm that can characterize signal diversity and complexity is introduced to quantify the sequence of fault mode decomposition and calculate the entropy value of each state. The larger the entropy value, the greater the disorder of the sequence and the more complex the state. The absolute value of the multi-scale permutation entropy difference before and after the series arc occurs is used to measure the deviation of each component fault state from the normal state. The main characteristic components are selected based on the deviation from the normal state:

[0078] Normalizing and removing the periodicity of the time series of length L yields an approximate series, expressed as:

[0079]

[0080] in,[ L / b ] indicates L / b Round off, b is the sampling step size. When b=1, the approximate sequence represents the initialization sequence.

[0081] Then, the obtained approximate sequences are arranged in ascending order to obtain sequence combinations at different scales.

[0082] Use the probability value to calculate the permutation entropy value at each scale. Combine the permutation entropy values ​​at different scales to obtain the multi-scale permutation entropy value of the entire approximate sequence. Calculate the entropy value of the permutation sequence at each scale until the permutation entropy of the approximate sequence at all scales is calculated. The multi-scale permutation entropy value and absolute difference of the IMF under the DC fault arc state and the stable operation state are displayed. The absolute value of the multi-scale permutation entropy difference of each IMF component before and after the series arc occurs in the DC circuit is used to measure the deviation of the fault state of each IMF component from the normal state. The larger the value, the more fault information the IMF component contains and the greater the degree of deviation from stable operation. Based on this, the first two values ​​with the largest values ​​can be used as the main components for feature extraction.

[0083] Step 4: Input the main characteristic components into the classification neural network to obtain the arc fault type; obtain the fault traveling wave transmitted by the cable during the current arc fault, analyze the natural frequency of the fault traveling wave, calculate the distance of the arc fault according to the natural frequency, and locate the arc fault position.

[0084] Specifically, the classification model is a softmax classification neural network formed by combining a BP neural network and a softmax function, including an input layer, a hidden layer and an output layer. During training, the proportional conjugate gradient method is selected for optimization, and the cross entropy loss function is combined to obtain the optimal classification neural network model.

[0085] The gradient descent algorithm selects the proportional conjugate gradient method, which can largely avoid falling into local

[0086] The optimal solution and overfitting phenomenon, fast convergence speed, high stability, and good local convergence; the loss function adopts the cross entropy loss function,

[0087]

[0088] In the formula, Y represents the actual result, Represents the predicted result.

[0089] The classification model outputs arc faults, and further implements fault location using a traveling wave method.

[0090] Specifically, after an arc fault occurs, the fault traveling wave is reflected back and forth between the two ends of the system and the fault point. The traveling wave has an inherent frequency that can be used to calculate the fault distance.

[0091] For the two fault conditions of series arc and parallel arc, the reflection coefficient of the fault traveling wave at the fault point is different. When an arc fault occurs at a certain point in the DC transmission line, the transient traveling wave caused by the fault is transmitted along the transmission line and reflected back and forth between the fault point and the system terminal to form a transient fault traveling wave. The frequency spectrum of the traveling wave in the system consists of a fundamental component and a series of harmonic components, which are called natural frequencies. Among them, the low-frequency fundamental component has the largest amplitude, which is called the main component of the natural frequency, and the amplitude of other harmonic components decreases with increasing frequency.

[0092] If a fault occurs on the line at a distance of Q from the power source, it is assumed that the impedance to the ground on the power source side is approximately 0. The fault traveling wave is transmitted from the fault point to the power source side, and negative total reflection will occur on the power source side. If the fault side is a short circuit fault, the impedance to the ground is also approximately 0, and the traveling wave propagating to the fault point also undergoes negative total reflection. The traveling wave repeatedly travels back and forth between the fault point and the power source side with a period of T=2Q / v (v is the speed of the traveling wave); if the fault side is a line break fault, the impedance to the ground is infinite, and the traveling wave undergoes positive total reflection at the fault point. At this time, the oscillation period of the voltage is T=4Q / v. In this way, the fault distance can be calculated based on the oscillation period (or frequency). But in fact, the fault traveling wave may not necessarily undergo total reflection at both ends of the transmission line, and there is a certain mathematical relationship between its frequency and the fault distance and line terminal conditions.

[0093] After a fault occurs on the transmission line, the initial traveling wave generated at the fault point will be transmitted along the transmission line and reflected at the power supply side and the fault point. Therefore, the fault traveling wave signal has a natural frequency. The natural frequency satisfies the equation in the Laplace domain:

[0094]

[0095] Where T is the time required for the traveling wave to propagate from one end of the line to the other end. If the line is of finite length d and the speed of the traveling wave is v, then T = d / v. In the case of high frequencies, the speed is calculated by the following formula:

[0096]

[0097] and is the line terminal reflection coefficient in the pull domain, which is determined by the impedance at the boundary and satisfies the following formula:

[0098]

[0099] According to the definition of Laplace transform, assume the transform coefficients. Assume that the frequencies of the components of the natural frequency of the traveling wave are f n It is known that the imaginary part of the transformation coefficient s is already determined, and it is substituted into the domain to satisfy the equation. Then when the reflection coefficient and When it is a complex number, it can be converted into exponential form using Euler's formula, that is,

[0100]

[0101] in and are the reflection angles of the traveling wave on the power supply side and the fault side, respectively, and -π / 2< + <π / 2.

[0102] Finally, we get the solution:

[0103]

[0104] Therefore, the natural frequency of the fault traveling wave is f n The size depends not only on the boundary conditions between the line head end and the fault point, but also on the fault distance. The size of the natural frequency is inversely proportional to the fault distance. Therefore, if the line boundary conditions and the size of the natural frequency are known, the fault distance can be calculated based on them.

[0105] Example 2

[0106] In one embodiment of the present disclosure, a system for testing series arc faults in AC and DC power supply circuits is provided, comprising:

[0107] A data acquisition module, used for extracting a load loop current signal of a circuit;

[0108] A signal reconstruction module, used for decomposing and reconstructing the load loop current signal based on a reconstruction singular value algorithm to obtain a singular value reconstructed current signal;

[0109] The modal feature extraction module is used to use the fault modal decomposition method to decompose the reconstructed current signal into multiple modal components by selecting the mode and the center frequency, introduce the multi-modal arrangement algorithm to sequence quantize the decomposed multiple modal components, perform feature selection, and determine the main characteristic components;

[0110] The classification and distance measurement module is used to input the main characteristic components into the classification neural network to obtain the type of arc fault; obtain the fault traveling wave transmitted by the cable during the current arc fault, analyze the natural frequency of the fault traveling wave, calculate the distance of the arc fault according to the natural frequency, and locate the arc fault.

[0111] Example 3

[0112] In one embodiment of the present disclosure, a non-transitory computer-readable storage medium is provided, wherein the non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the method for testing series arc faults in AC and DC power supply circuits is implemented.

[0113] Example 4

[0114] In one embodiment of the present disclosure, an electronic device is provided, including: a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes the method for testing series arc faults in AC and DC power supply circuits.

[0115] The present disclosure 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 disclosure. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to 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.

[0116] 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 in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0117] Although the above describes the specific implementation methods of the present disclosure in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present disclosure. Technical personnel in the relevant field should understand that on the basis of the technical solution of the present disclosure, various modifications or variations that can be made by those skilled in the art without creative work are still within the scope of protection of the present disclosure.

Claims

1. A method for testing series arc faults in AC and DC power supply circuits, characterized in that: include: Extracting the load loop current signal of the circuit; Based on the reconstruction singular value algorithm, the load loop current signal is decomposed and reconstructed to obtain a singular value reconstructed current signal; Based on the load loop current signal, a current signal trajectory matrix is ​​constructed, and the current signal is subjected to singular value decomposition, and the original current signal is decomposed into multiple component signals, which are then superimposed. Each singular value represents a component. By selecting valid singular values ​​and setting the remaining singular values ​​to zero, the noise decomposition and removal of the original current signal is completed. Constructing a current signal trajectory matrix, including: selecting the trajectory matrix, reducing the number of rows by increasing the delay step of the trajectory matrix to reduce the size of the trajectory matrix, and the selection of the delay step needs to comprehensively consider the operation time and the denoising effect; Among them, the delay step size of the trajectory matrix is ​​increased to reduce the number of rows to reduce the matrix AA T The scale, delay step length τ The selection of needs to comprehensively consider the operation time and denoising effect, and the number of columns of the trajectory matrix n The change in the overall computing time does not increase significantly, and n When it is half of the signal length, the data correlation between adjacent rows is the highest, and the signal separation ability is the strongest. The new trajectory matrix is, in, The fault mode decomposition method is used to decompose the reconstructed current signal into multiple modal components by selecting the mode and the center frequency. The multi-modal arrangement algorithm is introduced to quantize the multiple modal components, perform feature selection, and determine the main feature components. The fault mode decomposition method includes: analyzing the reconstructed current signal, calculating its unilateral spectrum value, setting the operator, allocating the center band of the analytical signal to the corresponding baseband, solving the gradient information of the analytical signal, calculating the frequency limit of the analytical signal, introducing multiple operators and penalty factors, continuously updating the mode and center frequency, directly meeting the iteration accuracy, and obtaining multiple modal components; The multi-scale permutation entropy algorithm that can characterize the diversity and complexity of signals is introduced to quantify the sequence of fault mode decomposition and calculate the entropy value of each state. The larger the entropy value, the greater the disorder of the sequence and the more complex the state. The absolute value of the multi-scale permutation entropy difference before and after the series arc occurs is used to measure the deviation of each component fault state from the normal state, and the main characteristic components are selected based on the deviation from the normal state. The main characteristic components are input into a classification neural network to obtain the type of arc fault; the fault traveling wave transmitted by the cable during the current arc fault is obtained, the natural frequency of the fault traveling wave is analyzed, the distance of the arc fault is calculated according to the natural frequency, and the position of the arc fault is located.

2. A method for testing series arc faults in AC and DC power supply circuits as claimed in claim 1, characterized in that: The classification model is a softmax classification neural network formed by combining the BP neural network and the softmax function, including an input layer, a hidden layer, and an output layer. During training, the proportional conjugate gradient method is selected for optimization, combined with the cross entropy loss function to obtain the optimal classification neural network model.

3. A method for testing series arc faults in AC and DC power supply circuits as claimed in claim 1, characterized in that: When an arc fault occurs at a certain point in a DC transmission line, the transient traveling wave caused by the fault is transmitted along the transmission line and reflected back and forth between the fault point and the load terminal to form a transient fault traveling wave. The fault traveling wave contains a natural frequency. A mathematical model of fault distance measurement using the natural frequency method is constructed, and the fault distance is calculated based on the known line boundary conditions and the size of the natural frequency.

4. A series arc fault test system for AC and DC power supply circuits, characterized in that: include: A data acquisition module, used for extracting a load loop current signal of a circuit; A signal reconstruction module, used for decomposing and reconstructing the load loop current signal based on a reconstruction singular value algorithm to obtain a singular value reconstructed current signal; Based on the load loop current signal, a current signal trajectory matrix is ​​constructed, and the current signal is subjected to singular value decomposition, and the original current signal is decomposed into multiple component signals, which are then superimposed. Each singular value represents a component. By selecting valid singular values ​​and setting the remaining singular values ​​to zero, the noise decomposition and removal of the original current signal is completed. Constructing a current signal trajectory matrix, including: selecting the trajectory matrix, reducing the number of rows by increasing the delay step of the trajectory matrix to reduce the size of the trajectory matrix, and the selection of the delay step needs to comprehensively consider the operation time and the denoising effect; Among them, the delay step size of the trajectory matrix is ​​increased to reduce the number of rows to reduce the matrix AA T The scale, delay step length τ The selection of needs to comprehensively consider the operation time and denoising effect, and the number of columns of the trajectory matrix n The change in the overall computing time does not increase significantly, and n When it is half of the signal length, the data correlation between adjacent rows is the highest, and the signal separation ability is the strongest. The new trajectory matrix is, in, is a discrete signal; The modal feature extraction module is used to use the fault modal decomposition method to decompose the reconstructed current signal into multiple modal components by selecting the mode and the center frequency, introduce the multi-modal arrangement algorithm to sequence quantize the decomposed multiple modal components, perform feature selection, and determine the main characteristic components; The fault mode decomposition method includes: analyzing the reconstructed current signal, calculating its unilateral spectrum value, setting the operator, allocating the center band of the analytical signal to the corresponding baseband, solving the gradient information of the analytical signal, calculating the frequency limit of the analytical signal, introducing multiple operators and penalty factors, continuously updating the mode and center frequency, directly meeting the iteration accuracy, and obtaining multiple modal components; The multi-scale permutation entropy algorithm that can characterize the diversity and complexity of signals is introduced to quantify the sequence of fault mode decomposition and calculate the entropy value of each state. The larger the entropy value, the greater the disorder of the sequence and the more complex the state. The absolute value of the multi-scale permutation entropy difference before and after the series arc occurs is used to measure the deviation of each component fault state from the normal state, and the main characteristic components are selected based on the deviation from the normal state. The classification and distance measurement module is used to input the main characteristic components into the classification neural network to obtain the type of arc fault; obtain the fault traveling wave transmitted by the cable during the current arc fault, analyze the natural frequency of the fault traveling wave, calculate the distance of the arc fault according to the natural frequency, and locate the arc fault position.

5. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by the processor, a method for testing series arc faults in AC and DC power supply circuits as described in any one of claims 1 to 3 is implemented.

6. An electronic device, characterized in that: include: A processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to implement a method for testing series arc faults in AC and DC power supply circuits as described in any one of claims 1 to 3.

Images