CN117269660A - Fault arc detection method and system based on variation coefficient difference algorithm - Google Patents

Abstract

The invention discloses a fault arc detection method and system based on a coefficient of variation differential algorithm. The method includes collecting the current signal of the line to be tested; obtaining a sample sequence of second-order differential value of the coefficient of variation according to the current signal; The differential value sample sequence is subjected to wavelet decomposition, and the decomposition results are denoised to obtain denoised coefficients; an adaptive threshold is calculated based on the denoised coefficients; arc fault detection is performed based on the adaptive threshold. The invention overcomes the low accuracy and many false positives and false positives caused by the inability of the traditional algorithm to adapt the threshold in real time during fault detection, thereby improving the reliability of detection.

Description

Fault arc detection method and system based on variation coefficient difference algorithm

Technical field

The invention relates to the technical field of arc detection, and in particular to a fault arc detection method and system based on a coefficient of variation differential algorithm.

Background technique

Arc faults caused by factors such as damage, aging, and loose connections of power distribution system lines can cause local high temperatures, which can easily lead to electrical fires or even explosions. Therefore, arc faults are one of the main causes of electrical fires. The core temperature of the arc can be as high as 5000K to 15000K, and once a breakdown point occurs, arcs will occur frequently. Therefore, fire accidents caused by electrical fires are generally more serious, and it is of great significance to avoid the occurrence of electrical fires as much as possible.

In order to prevent fires caused by arc faults, arc fault detection technology is usually used to detect arc faults. Most arc fault detectors currently on the market use arc fault detection methods based on feature vector threshold monitoring. However, the characteristic threshold set by the detector is generally a fixed threshold. When the characteristic threshold is set too high, detection failure will occur during arc fault detection; if the characteristic threshold is set too low, false alarms will occur. . It is obviously difficult to set appropriate characteristic thresholds for various types of loads in actual power consumption scenarios. Therefore, it is very appropriate and important to adopt an adaptive threshold strategy for the characteristic thresholds set by arc fault detectors.

In related technology, the patent application document with publication number CN109061414A proposes a combined fault detection method based on wavelet transform and singular value decomposition (SVD) to detect fault arc signals; this solution is suitable for DC systems, and the wavelet coefficients are analyzed from the perspective of singular values , inconsistent with the representation of the communication system.

In the patent application document with publication number CN112162172A, based on the statistical information of virtual energy indicators, periodic fluctuation indicators and dynamic thresholds, as well as fault identification methods and dynamic threshold update algorithms are established; this solution is based on the maximum value, minimum value, and mean value. Virtual energy indicators established by indicators such as variance and variance.

Contents of the invention

The technical problem to be solved by the present invention is how to improve the reliability of fault arc detection.

The present invention solves the above technical problems through the following technical means:

In a first aspect, the present invention proposes an arc fault detection method based on a coefficient of variation differential algorithm. The method includes:

Collect the current signal of the line under test;

According to the current signal, a sample sequence of second-order difference coefficients of variation values is obtained;

Perform wavelet decomposition on the second-order difference value sample sequence, and perform denoising processing on the decomposition results to obtain denoised coefficients;

Based on the denoised coefficients, calculate an adaptive threshold;

Arc fault detection is performed based on the adaptive threshold.

Further, obtaining a sample sequence of second-order difference coefficients of variation according to the current signal includes:

Extract the coefficient of variation of each half cycle of the current signal;

Based on the coefficient of variation of the current half-cycle and the coefficient of variation of the next half-cycle, calculate the first-order difference value of the current half-cycle;

Based on the first-order difference value of the current half-cycle and the first-order difference value of the next half-cycle, calculate the second-order difference value of the current half-cycle;

The second-order difference values of each half-cycle constitute the second-order difference value sample sequence.

Further, the second-order differential value sample sequence is subjected to wavelet decomposition, and the decomposition results are denoised to obtain denoised coefficients, including:

Perform wavelet decomposition on the second-order difference value sample sequence to obtain approximate coefficients and high-frequency detail coefficients;

The high-frequency detail coefficients are subjected to hard-threshold wavelet noise reduction processing to obtain denoised coefficients.

Further, the high-frequency detail coefficients are subjected to hard threshold wavelet denoising processing to obtain denoised coefficients, including:

When the absolute value of the high-frequency detail coefficient is greater than a given universal threshold λ, the coefficient of the high-frequency detail coefficient remains unchanged;

When the absolute value of the high-frequency detail coefficient is less than or equal to a given universal threshold λ, the coefficient of the high-frequency detail coefficient is set to zero;

Wherein, the general threshold σ is the standard deviation of the coefficients of the detail component, and n _D is the number of coefficients of the detail component.

Further, calculating an adaptive threshold based on the denoised coefficient includes:

Perform wavelet reconstruction on the denoised coefficients to obtain a discrete sequence consistent with the length of the current signal;

Use the information entropy of the discrete sequence as the arc fault factor;

Calculate the load adaptation factor based on the reconstructed coefficients after the noise reduction;

The adaptive threshold is calculated based on the arc fault factor, the load adaptation factor and the empirical self-correction factor.

Further, the formula of the arc fault factor is expressed as:

In the formula: AFF is the arc fault factor, Represents the i-th/> The value of the sequence, n _D represents/> The number of coefficients in the sequence,/> Represents the offline sequence.

Further, the formula of the load adaptation factor is expressed as:

In the formula: LAF is the load adaptation factor, _di is the coefficient reconstructed after noise reduction of the third layer detail coefficient of the current signal, i=1, 2,...,N.

Further, the formula of the adaptive threshold is expressed as:

thd＝k·AFF·LAF

In the formula: thd is the adaptive threshold, AFF is the arc fault factor, LAF is the load adaptation factor, and k is the empirical self-correction factor.

Further, the arc fault detection based on the adaptive threshold includes:

When the real-time threshold t of a detection period is greater than or equal to the adaptive threshold, it is determined that a series fault arc occurs;

When the real-time threshold t of one detection period is less than the adaptive threshold, it is determined that a series fault arc does not occur.

In a second aspect, the present invention also proposes an arc fault detection system based on the coefficient of variation differential algorithm. The system includes:

Acquisition module, used to collect the current signal of the line under test;

A feature extraction module, configured to obtain a sample sequence of second-order difference coefficients of variation according to the current signal;

A decomposition and noise reduction module, used to perform wavelet decomposition on the second-order differential value sample sequence, and perform denoising processing on the decomposition results to obtain denoised coefficients;

A threshold calculation module, configured to calculate an adaptive threshold based on the denoised coefficient;

A detection module configured to perform arc fault detection based on the adaptive threshold.

The advantages of the present invention are:

(1) The present invention extracts the second-order difference value sample sequence of the variation coefficient of the current signal as the fault characteristic value, and performs wavelet decomposition and noise reduction processing on the second-order difference value sample sequence to obtain the denoised coefficient, thereby according to the noise reduction The final coefficient determines the adaptive threshold, and finally uses the adaptive threshold to detect fault arc; the present invention adopts a technical means that combines the variation coefficient difference algorithm, adaptive threshold, wavelet decomposition and reconstruction and noise reduction to overcome the traditional algorithm in fault detection. The inability to adapt the threshold in real time causes low accuracy and many false negatives and false positives, thereby improving the reliability of detection.

Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.

Description of the drawings

Figure 1 is a schematic flow chart of the arc fault detection method based on the coefficient of variation differential algorithm proposed by the embodiment of the present invention;

Figure 2 is a schematic diagram of the principle of wavelet decomposition in an embodiment of the present invention;

Figure 3 is a schematic diagram of the principle of wavelet reconstruction in the embodiment of the present invention;

Figure 4 is a diagram showing the situation of detecting resistive load current by wavelet transform according to an embodiment of the present invention;

Figure 5 is a schematic diagram of the current signal and sequence of a resistive load under normal and arc conditions according to an embodiment of the present invention;

Figure 6 shows the arc fault factor AFF of the current signal of the resistive load under normal and arc conditions according to an embodiment of the present invention;

Figure 7 is the load adaptation factor LAF of the current signal of the resistive load under normal and arc conditions according to the embodiment of the present invention;

Figure 8 shows the adaptive threshold thd of the current signal of the resistive load under normal and arc conditions according to an embodiment of the present invention;

Figure 9 is a schematic diagram of the test platform shown in the embodiment of the present invention;

FIG. 10 is a schematic structural diagram of an arc fault detection system based on the coefficient of variation differential algorithm proposed by an embodiment of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are part of the present invention. Examples, not all examples. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts fall within the scope of protection of the present invention.

As shown in Figure 1, the first embodiment of the present invention proposes a fault arc detection method based on the coefficient of variation differential algorithm. The method includes the following steps:

S10. Collect the current signal of the line under test;

It should be noted that after initializing the device, when obtaining the real-time current signal of the device, the current signal in any load circuit can be obtained through an external circuit as the signal to be measured.

S20. According to the current signal, obtain a sample sequence of second-order difference values of the coefficient of variation;

S30. Perform wavelet decomposition on the second-order difference value sample sequence, and perform denoising processing on the decomposition results to obtain denoised coefficients;

S40. Calculate the adaptive threshold based on the denoised coefficient;

S50. Perform arc fault detection based on the adaptive threshold.

This embodiment is aimed at the AC system, starting from the correlation between the discreteness and variability reflected by the wavelet coefficient and the fault arc, using the variation coefficient to represent the fault arc, using the variation coefficient difference algorithm, adaptive threshold, wavelet decomposition and reconstruction and The technical means combined with noise reduction overcomes the low accuracy and many false negatives and false positives caused by the inability of the traditional algorithm to adapt the threshold in real time during fault detection, thus improving the reliability of detection.

In one embodiment, the step S20: Obtaining a sample sequence of second-order difference coefficients of variation according to the current signal includes the following steps:

S21. Extract the coefficient of variation of each half cycle of the current signal;

Specifically, assume that A consecutive periods of equally spaced sampling value detection are performed, and each period is composed of B sampling points. Calculate the number of points in each half cycle of the signal under test The formula is expressed as:/> Among them, F _s is the sampling frequency, and f is 50Hz.

Calculate the coefficient of variation of the mth half-cycle Among them,/> is the standard deviation of the coefficient for each half period,/> is the average value of the absolute value of each half-cycle coefficient, x(i) is the wavelet coefficient, and N is the total number of wavelet coefficients.

S22. Based on the coefficient of variation of the current half period and the coefficient of variation of the next half period, calculate the first-order difference value of the current half period;

Specifically, the first-order difference value of the coefficient of variation of the m-th half period C·V ⁽¹⁾ (m)=|C·V(m+1)-C·V(m)|.

S23. Calculate the second-order difference value of the current half-cycle based on the first-order difference value of the current half-cycle and the first-order difference value of the next half-cycle;

Specifically, the second-order difference value of the coefficient of variation of the m-th half period C·V ⁽²⁾ (m)=|C·V ⁽¹⁾ (m+1)-C·V ⁽¹⁾ (m)|.

S24. Constitute the second-order difference value sample sequence of each half-cycle into the second-order difference value sample sequence, which is represented by D ⁽²⁾ .

It should be noted that the differential coefficient of variation algorithm (DACV) is a low-voltage series fault arc detection algorithm based on arc electrical signals and differential algorithms. In order to avoid the potential damaging effects of pre-processing methods such as noise reduction on the source signal and maintain the purity of the source signal, DACV chooses to give priority to directly extracting fault features from the collected source signal, and then reuse noise reduction and other signals in subsequent processing. Processing techniques for data enhancement. In this embodiment, DACV needs to consider the magnitude of C·V and its first-order and second-order difference values as a criterion for determining whether a series fault arc occurs.

It should be noted that the Daubechies wavelet family has a large number of vanishing moments and can extract local signal disturbances with high quality. By comparing the performance of various wavelets in extracting fault arcs, this embodiment gives priority to the tightly supported and orthogonal db4 wavelet. In order to study the ability of wavelet transform to represent fault arcs, this embodiment conducted the following experiments. Figure 4 shows the detection of resistive load current by wavelet transform. d3 is the high-frequency detailed signal obtained after wavelet decomposition and reconstruction. coefficient. The principle process of wavelet decomposition and reconstruction is shown in Figure 2 and Figure 3 respectively.

It can be found from Figure 4 that the places with obvious fluctuations in the high-frequency detail coefficient d3 correspond to the stage of arc fault occurrence. In other words, d3 can accurately capture fault arc. When an arc fault occurs, some high-frequency detail coefficients are significantly improved in value compared to normal working conditions. It can be expected that the fault arc indication features extracted based on the high-frequency detail coefficient d3 should have a good representation effect on the fault arc, but further analysis is still needed.

Considering that the existing fault-generating devices and test platforms on the market are difficult to stably conduct frequent tests, this embodiment also builds an experimental platform that complies with the American standard UL 1699 and the national standard GB14287.4, as shown in Figure 9, including a power supply, a circuit breaker, Current transformers, data acquisition cards, PCs, fault arc generators and various experimental loads.

Data acquisition cards and current transformers are used to collect relevant current data. The sampling frequency Fs is 50KHz, and different types of loads are taken into consideration. The frequency band of the n-th detail signal is (Fs/2n+1, Fs/2n), and correspondingly the frequency band of the n-th approximate signal is (0, Fs/2n+1). Collect current samples under normal and arc conditions, save them on the PC and perform preprocessing and implement SAF detection strategies.

It should be noted that the current data under different working conditions (normal operation and arc fault) and different loads are obtained by adjusting the clamping gap of the fault arc generator to control the generation of arc and conducting experiments by connecting different types of electrical appliances. .

Furthermore, the series fault arc condition used in this embodiment is four peaks occurring within two basic periods. 2 basic cycles correspond to 4 half-cycles, that is, 4 C·V samples, 3 C·V ⁽¹⁾ , and 2 C·V ⁽²⁾ . In order to adapt the conditions to the variation coefficient difference algorithm described in this application, this application inputs four consecutive periods of the original signal to obtain 6 C·V ⁽²⁾ samples, and assumes that the fault condition corresponds to the occurrence of C·V, At least four peak events that occur jointly between C·V ⁽¹⁾ and C·V ⁽²⁾ .

Further, based on the resistive load of this application, the current signals and sequences C·V, C·V ⁽¹⁾ , C·V ⁽²⁾ of the resistive load under normal and arc conditions are shown in Figure 5.

In one embodiment, S30: Perform wavelet decomposition on the second-order differential value sample sequence, and perform denoising processing on the decomposition results to obtain denoised coefficients, including the following steps:

S31. Perform wavelet decomposition on the second-order difference value sample sequence to obtain approximate coefficients and high-frequency detail coefficients;

S32. Perform hard threshold wavelet denoising processing on the high-frequency detail coefficients to obtain denoised coefficients.

It should be noted that if the current signal is directly decomposed by wavelet, some details will be lost, which is not conducive to extracting the coefficient of variation of each half cycle, because noise and fault arcs are mostly reflected in high-frequency detail coefficients, and environmental noise can be ignored. Therefore, this embodiment gives priority to directly extracting the coefficient of variation of each half cycle from the sample sequence.

In one embodiment, step S32: performing hard-threshold wavelet noise reduction processing on the high-frequency detail coefficients to obtain denoised coefficients includes the following steps:

S321. When the absolute value of the high-frequency detail coefficient is greater than the given universal threshold λ, the coefficient of the high-frequency detail coefficient remains unchanged;

S322. When the absolute value of the high-frequency detail coefficient is less than or equal to the given universal threshold λ, the coefficient of the high-frequency detail coefficient is set to zero;

Wherein, the general threshold σ is the standard deviation of the coefficients of the detail component, and n _D is the number of coefficients of the detail component.

Further, the hard threshold function:

Among them, x(i) represents the wavelet coefficient, i=1,...,N.

This embodiment performs wavelet hard threshold denoising processing with universal thresholds on high-frequency detail coefficients to obtain a set of coefficients that are approximately "noise-free" and can shield most of the interference caused by the external environment.

In an embodiment, step S40: Calculate an adaptive threshold based on the denoised coefficients, specifically includes the following steps:

S41. Perform wavelet reconstruction on the denoised coefficients to obtain a discrete sequence consistent with the length of the current signal, recorded as

S42. Use the information entropy of the discrete sequence as the arc fault factor;

S43. Calculate the load adaptation factor based on the reconstructed coefficients after denoising;

S44. Calculate the adaptive threshold according to the arc fault factor, the load adaptation factor and the empirical self-correction factor.

In one embodiment, the formula of the arc fault factor is expressed as:

In the formula: AFF is the arc fault factor, Represents the i-th/> The value of the sequence, n _D represents/> The number of coefficients in the sequence,/> Represents the offline sequence.

It should be noted that AFF represents The average uncertainty level of the sequence is a characteristic quantity that has a certain correlation with arc faults.

In one embodiment, the formula of the load adaptation factor is expressed as:

In the formula: LAF is the load adaptation factor, _di is the coefficient reconstructed after noise reduction of the third layer detail coefficient of the current signal, i=1, 2,...,n.

In one embodiment, the formula of the adaptive threshold is expressed as:

thd＝k·AFF·LAF

In the formula: thd is the adaptive threshold, AFF is the arc fault factor, LAF is the load adaptation factor, and k is the empirical self-correction factor.

This embodiment can combine the load adaptation factor and the fault arc factor to obtain fault differentiation conditions that can adapt to different load conditions and avoid load interference with the detection of fault arc.

It should be noted that the characterization of AFF and LAF under resistive load is shown in Figure 6 and Figure 7. The reference empirical self-correction factor k=10 ^-6 is given. Therefore, the adaptive threshold can be calculated according to the formula thd=k·AFF·LAF. The characterization of thd under resistive load is shown in Figure 8.

In an embodiment, the step S50: performing arc fault detection based on the adaptive threshold includes the following steps:

S51. When the real-time threshold t of a detection period is greater than or equal to the adaptive threshold, it is determined that a series fault arc has occurred;

S52. When the real-time threshold t of a detection period is less than the adaptive threshold, it is determined that no series fault arc has occurred.

It should be noted that the series arc fault (SAF) detection strategy based on adaptive threshold setting in this embodiment is that when the t value of a detection period is greater than or equal to the given threshold thd, SAF is set to 1, otherwise SAF is set to 0. The formula of the detection strategy is expressed as: t _i represents the t value of the i-th detection period.

In addition, as shown in Figure 10, the second embodiment of the present invention proposes an arc fault detection system based on the coefficient of variation differential algorithm. The system includes:

The collection module 10 is used to collect the current signal of the line to be tested;

The feature extraction module 20 is used to obtain a sample sequence of second-order difference coefficients of variation according to the current signal;

The decomposition and noise reduction module 30 is used to perform wavelet decomposition on the second-order differential value sample sequence, and perform denoising processing on the decomposition results to obtain denoised coefficients;

A threshold calculation module 40, configured to calculate an adaptive threshold based on the denoised coefficient;

The detection module 50 is configured to perform arc fault detection based on the adaptive threshold.

This embodiment uses a technical means that combines the variation coefficient difference algorithm, adaptive threshold, wavelet decomposition and reconstruction, and noise reduction to overcome the low accuracy, missed reports, and false positives caused by the inability of the traditional algorithm to adapt the threshold in real time during fault detection. Report too many situations, thereby improving the reliability of detection.

In one embodiment, the feature extraction module 20 specifically includes:

A coefficient of variation extraction unit, used to extract the coefficient of variation of each half cycle of the current signal;

The first-order difference calculation unit is used to calculate the first-order difference value of the current half-cycle based on the coefficient of variation of the current half-cycle and the coefficient of variation of the next half-cycle;

A second-order difference calculation unit is used to calculate the second-order difference value of the current half-cycle based on the first-order difference value of the current half-cycle and the first-order difference value of the next half-cycle;

A sample sequence construction unit is used to construct the second-order difference value sample sequence from the second-order difference values of each half period.

In one embodiment, the decomposition and noise reduction module 30 specifically includes:

A decomposition unit, used to perform wavelet decomposition on the second-order difference value sample sequence to obtain approximate coefficients and high-frequency detail coefficients;

A noise reduction unit is used to perform hard threshold wavelet noise reduction processing on the high-frequency detail coefficients to obtain denoised coefficients.

In one embodiment, the noise reduction unit is specifically configured to perform the following steps:

When the absolute value of the high-frequency detail coefficient is greater than a given universal threshold λ, the coefficient of the high-frequency detail coefficient remains unchanged;

When the absolute value of the high-frequency detail coefficient is less than or equal to a given universal threshold λ, the coefficient of the high-frequency detail coefficient is set to zero;

Wherein, the general threshold σ is the standard deviation of the coefficients of the detail component, and n _D is the number of coefficients of the detail component.

In one embodiment, the threshold calculation module 40 is specifically used to:

A discrete sequence acquisition unit, configured to perform wavelet reconstruction on the denoised coefficients to obtain a discrete sequence consistent with the length of the current signal;

An arc fault factor calculation unit is used to use the information entropy of the discrete sequence as the arc fault factor;

A load adaptation factor calculation unit, configured to calculate the load adaptation factor based on the reconstructed coefficients of the denoised coefficients;

An adaptive threshold calculation unit is configured to calculate the adaptive threshold based on the arc fault factor, the load adaptation factor and the empirical self-correction factor.

In one embodiment, the formula of the arc fault factor is expressed as:

In the formula: AFF is the arc fault factor, Represents the i-th/> The value of the sequence, n _D represents/> The number of coefficients in the sequence,/> Represents the offline sequence.

In one embodiment, the formula of the load adaptation factor is expressed as:

In the formula: LAF is the load adaptation factor, _di is the coefficient reconstructed after noise reduction of the third layer detail coefficient of the current signal, i=1, 2,...,n.

In one embodiment, the formula of the adaptive threshold is expressed as:

thd＝k·AFF·LAF

In the formula: thd is the adaptive threshold, AFF is the arc fault factor, LAF is the load adaptation factor, and k is the empirical self-correction factor.

In one embodiment, the detection module 50 is specifically configured to perform the following steps:

When the t value of a detection period is greater than or equal to the adaptive threshold, it is determined that a series fault arc occurs;

When the t value of one detection period is less than the adaptive threshold, it is determined that no series fault arc occurs.

In this embodiment, the coefficient of variation (C·V) of each half cycle and its first-order and second-order difference values are extracted from the current signal to be measured as fault indication features to detect fault arcs; through wavelet decomposition, noise reduction, reconstruction, etc. The arc fault factor AFF, the load adaptation factor LAF and the empirical self-correction factor k are obtained through experimental data to determine the real-time adaptive threshold thd, and then the SAF detection strategy is designed and applied to the detection period to be tested; overcoming the problem of traditional algorithms in fault The inability to adapt the threshold in real time during detection results in low accuracy and many false negatives and false positives, thereby improving the reliability of detection; it can protect users' lives and property and improve detection efficiency and rapid response compared to other detection methods. and reduced computational complexity.

It should be noted that other embodiments of the arc fault detection system based on the coefficient of variation differential algorithm of the present invention or implementation methods may refer to the above method embodiments, and no redundancy is provided here.

In the description of this specification, reference to the terms "one embodiment," "some embodiments," "an example," "specific examples," or "some examples" or the like means that specific features are described in connection with the embodiment or example. , structures, materials or features are included in at least one embodiment or example of the invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

In addition, the terms “first” and “second” are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the quantity of indicated technical features. Therefore, features defined as "first" and "second" may explicitly or implicitly include at least one of these features. In the description of the present invention, "plurality" means at least two, such as two, three, etc., unless otherwise expressly and specifically limited.

Although the embodiments of the present invention have been shown and described above, it can be understood that the above-mentioned embodiments are illustrative and should not be construed as limitations of the present invention. Those of ordinary skill in the art can make modifications to the above-mentioned embodiments within the scope of the present invention. The embodiments are subject to changes, modifications, substitutions and variations.

Claims (10)

1. A fault arc detection method based on the coefficient of variation differential algorithm, characterized in that the method includes: Collect the current signal of the line under test; According to the current signal, a sample sequence of second-order difference coefficients of variation values is obtained; Perform wavelet decomposition on the second-order difference value sample sequence, and perform denoising processing on the decomposition results to obtain denoised coefficients; Based on the denoised coefficients, calculate an adaptive threshold; Arc fault detection is performed based on the adaptive threshold. 2. The arc fault detection method based on the variation coefficient difference algorithm according to claim 1, characterized in that, according to the current signal, obtaining a sequence of second-order difference coefficient of variation value samples includes: Extract the coefficient of variation of each half cycle of the current signal; Based on the coefficient of variation of the current half-cycle and the coefficient of variation of the next half-cycle, calculate the first-order difference value of the current half-cycle; Based on the first-order difference value of the current half-cycle and the first-order difference value of the next half-cycle, calculate the second-order difference value of the current half-cycle; The second-order difference values of each half-cycle constitute the second-order difference value sample sequence. 3. The fault arc detection method based on the variation coefficient difference algorithm as claimed in claim 1, characterized in that the second-order difference value sample sequence is subjected to wavelet decomposition, and the decomposition result is subjected to noise reduction processing to obtain a reduced noise level. Post-noise coefficients, including: Perform wavelet decomposition on the second-order difference value sample sequence to obtain approximate coefficients and high-frequency detail coefficients; The high-frequency detail coefficients are subjected to hard-threshold wavelet noise reduction processing to obtain denoised coefficients. 4. The fault arc detection method based on the coefficient of variation differential algorithm as claimed in claim 3, characterized in that the high-frequency detail coefficients are subjected to hard threshold wavelet denoising processing to obtain denoised coefficients, including: When the absolute value of the high-frequency detail coefficient is greater than a given universal threshold λ, the coefficient of the high-frequency detail coefficient remains unchanged; When the absolute value of the high-frequency detail coefficient is less than or equal to a given universal threshold λ, the coefficient of the high-frequency detail coefficient is set to zero; Wherein, the general threshold σ is the standard deviation of the coefficients of the detail component, and n _D is the number of coefficients of the detail component. 5. The arc fault detection method based on the coefficient of variation differential algorithm as claimed in claim 1, wherein the step of calculating an adaptive threshold based on the noise reduction coefficient includes: Perform wavelet reconstruction on the denoised coefficients to obtain a discrete sequence consistent with the length of the current signal; Use the information entropy of the discrete sequence as the arc fault factor; Calculate the load adaptation factor based on the reconstructed coefficients after the noise reduction; The adaptive threshold is calculated based on the arc fault factor, the load adaptation factor and the empirical self-correction factor. 6. The fault arc detection method based on the coefficient of variation differential algorithm as claimed in claim 5, characterized in that the formula of the arc fault factor is expressed as: In the formula: AFF is the arc fault factor, Represents the i-th/> The value of the sequence, n _D represents/> The number of coefficients in the sequence,/> Represents the offline sequence. 7. The fault arc detection method based on the coefficient of variation differential algorithm as claimed in claim 5, wherein the formula of the load adaptation factor is expressed as: In the formula: LAF is the load adaptation factor, d _i is the coefficient reconstructed after noise reduction of the third layer detail coefficient of the current signal, i=1, 2,...,N. 8. The fault arc detection method based on the variation coefficient difference algorithm as claimed in claim 5, characterized in that the formula of the adaptive threshold is expressed as: thd＝k·AFF·LAF In the formula: thd is the adaptive threshold, AFF is the arc fault factor, LAF is the load adaptation factor, and k is the empirical self-correction factor. 9. The arc fault detection method based on the coefficient of variation differential algorithm as claimed in claim 1, wherein the arc fault detection based on the adaptive threshold includes: When the real-time threshold t of a detection period is greater than or equal to the adaptive threshold, it is determined that a series fault arc occurs; When the real-time threshold t of one detection period is less than the adaptive threshold, it is determined that a series fault arc does not occur. 10. A fault arc detection system based on the coefficient of variation differential algorithm, characterized in that the system includes: Acquisition module, used to collect the current signal of the line under test; A feature extraction module, configured to obtain a sample sequence of second-order difference coefficients of variation according to the current signal; A decomposition and noise reduction module, used to perform wavelet decomposition on the second-order differential value sample sequence, and perform denoising processing on the decomposition results to obtain denoised coefficients; A threshold calculation module, configured to calculate an adaptive threshold based on the denoised coefficient; A detection module configured to perform arc fault detection based on the adaptive threshold.