Power distribution network high-resistance arc fault detection method

Abstract

The invention discloses a power distribution network high-resistance arc fault detection method, which comprises the following steps of: 1, sampling zero-sequence current of each feeder line of a power distribution network small-resistance grounding system, decomposing the zero-sequence current of each feeder line by using an MVMD algorithm, solving a corresponding frequency through Fourier transform, and taking a mode with the frequency of 45Hz-65Hz as a fault characteristic mode; 2, calculating a waveform similarity matrix and a waveform similarity coefficient of each feeder, if the waveform similarity coefficient is smaller than a threshold value, determining that the feeder corresponding to the minimum waveform similarity coefficient is a fault feeder, otherwise, determining that the feeder is a bus fault; 3, when the maximum amplitude of the electric charge quantity is greater than the sum of the amplitudes of the other electric charge quantities, determining that the corresponding feeder line is a fault feeder line, otherwise, determining that the bus has a fault; 4, when the maximum amplitude of the current change rate is greater than the sum of the amplitudes of other current change rates, determining that the feeder corresponding to the current change rate is a faulty feeder, otherwise, determining that the bus is faulty; 5, if the judgment results in the steps 2-4 are consistent, directly obtaining a fault; otherwise, voting to select a fault; and otherwise, returning to the steps 1-5, so as to cope with various high-resistance arc faults and detect the high-resistance arc faults.

Description

Technical Field

[0001] The invention belongs to the technical field of power cable fault detection, and in particular relates to a method for detecting high-resistance arc faults in a distribution network. Background Art

[0002] High-resistance arc fault is a common form of distribution network fault, accounting for about 5% to 20% of medium-voltage distribution network faults. For some grounding materials with very high impedance, such as soil, sand, asphalt, concrete, cement and branches, the grounding current amplitude is only a few amperes, most of which are accompanied by arcing. Traditional protection cannot guarantee reliability. The IEEE Relay Protection Committee reported that the current success rate of high-resistance fault detection above 1kΩ is still less than 20%. If the faulty feeder cannot be detected in time, it may cause large-scale power outages or forest fires, and there is a risk of electric shock. For example: in June 2016, a cable trench in a large city in my country exploded, causing two substations to burn, resulting in power outages for nearly 90,000 users. At the same time, it caused a loss of about 10% to Samsung Semiconductor Factory, nearly 10 billion won; in March 2020, a forest fire broke out in Xichang City, Sichuan Province, and 19 firefighters were killed. The direct cause was a line grounding fault; most of the serious fires in Australia, the United States and Brazil were confirmed to be caused by high-resistance faults. In addition, according to statistics, more than 8,000 people die from electric shock in my country every year, of which 10% are caused by line fall caused by high-resistance arc faults in the distribution network. For example, in April 2016, a 10kV wire fell to the ground in a certain province, causing three passers-by to be electrocuted to death.

[0003] Since the characteristics of high-resistance arc faults are very weak, existing protection devices cannot operate, causing relatively serious property losses and casualties. Therefore, it is of great significance to study high-resistance arc fault detection technology. Summary of the invention

[0004] In view of the deficiencies in the prior art, an object of the present invention is to provide a method for detecting high-resistance arc faults in a distribution network, which can cope with the diversity of high-resistance arc faults and detect high-resistance arc faults quickly and accurately.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A method for detecting high-resistance arc faults in a distribution network comprises the following steps:

[0007] Step 1: Sample the zero-sequence current of each feeder in the low-resistance grounding system of the distribution network, decompose the zero-sequence current of each feeder using the multivariate variational mode decomposition algorithm, perform Fourier transform on each decomposed mode, obtain the corresponding frequency, and select the mode in the frequency band of 45Hz to 65Hz as the fault characteristic mode x l ;

[0008] Step 2: Use the fault characteristic modes of each feeder to calculate the waveform similarity matrix D, and then use the waveform similarity matrix D to calculate the waveform similarity coefficient ρ of each feeder l , when the waveform similarity coefficient is less than the threshold value β, the feeder corresponding to the waveform similarity coefficient with the smallest value is determined to be the faulty feeder, otherwise, it is determined to be a busbar fault;

[0009] Step 3: Calculate the charge Q l , expressed as:

[0010] Q l =∫x l dt

[0011] In the formula, t represents time, x l represents the fault characteristic mode of feeder l;

[0012] The fault characteristic mode x l The accumulated product of the time t is the charge Q l , when the maximum charge magnitude is greater than the sum of the remaining charge magnitudes, the feeder corresponding to the maximum charge magnitude is determined to be a faulty feeder, otherwise, it is determined to be a busbar fault;

[0013] Step 4: Calculate the current change rate V l , expressed as:

[0014]

[0015] In the formula, t represents time, x l represents the fault characteristic mode of feeder l;

[0016] The fault characteristic mode x l The current change rate V is obtained by accumulating the derivative with respect to time t. l , when the maximum amplitude of the current change rate is greater than the sum of the amplitudes of the other current change rates, the feeder corresponding to the maximum amplitude of the current change rate is determined to be a faulty feeder, otherwise, it is determined to be a busbar fault;

[0017] Step 5: When the determination results of Step 2 to Step 4 are consistent, the final faulty feeder is directly obtained; otherwise, a voting method is adopted to obtain the final faulty feeder according to the principle of minority obeys majority; if the faulty feeder still cannot be determined, return to Step 1 to Step 5.

[0018] Furthermore, the specific process of step 1 is as follows:

[0019] Step 1.1: Number each feeder l in the low-resistance grounding system, expressed as:

[0020] l=1,2,…,n

[0021] Where n is the total number of feeders in the low-resistance grounding system, n>3;

[0022] Step 1.2: Collect the zero-sequence current ZSC of each feeder in the data window l , where the data window length is from 0.04s before the fault to 0.05s after the fault;

[0023] Step 1.3: Use the multivariate variational mode decomposition algorithm to calculate the zero-sequence current ZSC of each feeder. l Decompose, perform Fourier transform on the decomposed mode, obtain the corresponding frequency, and select the mode in the frequency band of 45Hz to 65Hz as the fault characteristic mode x l .

[0024] Furthermore, the specific process of step 2 is as follows:

[0025] Step 2.1, define =1,2,…,n, and have the same value range as l, use ρ l represents the fault characteristic mode x of feeder l l The fault characteristic mode x of feeder h h The cosine similarity of the waveform is calculated by calculating the waveform similarity matrix D, which is expressed as:

[0026]

[0027] Step 2.2: Calculate the waveform similarity coefficient ρ of each feeder l , expressed as:

[0028]

[0029] Step 2.3, set a threshold β of the waveform similarity coefficient. When the calculated waveform similarity coefficient is less than β, the feeder corresponding to the waveform similarity coefficient with the smallest value is determined to be a faulty feeder. Otherwise, it is determined to be a busbar fault.

[0030] Furthermore, the fault characteristic mode x of the feeder line l in step 2.1 is l The fault characteristic mode x of feeder h h The cosine similarity ρ l It is expressed as:

[0031]

[0032] Where U is x l and x h The sampling point.

[0033] Furthermore, in step 2.3, the threshold β is set to 0.8.

[0034] Compared with the prior art, the present invention has the following technical effects:

[0035] 1) The present invention can cope with the diversity of high-resistance arc faults in a low-resistance grounding system, and quickly and accurately detect high-resistance arc faults, thereby facilitating timely action by the protection device and reducing safety accidents caused by high-resistance arc faults.

[0036] 2) Filling the market gap: Although the relay protection problem of high-voltage transmission lines has been properly solved, there has been no good solution for the fault handling of distribution networks, and the domestic and foreign markets have a huge demand for this technology. Therefore, the present invention develops a high-resistance arc fault detection method for a low-resistance grounding system of a 110kV substation, which can fill the gap in the domestic market; at the same time, it can gradually get rid of the technical domination of other countries and improve the development and progress of the domestic medium and low voltage distribution network fault diagnosis industry.

[0037] 3) Improve the sense of gain in electricity use: After further promotion in China, the present invention can greatly reduce the troubleshooting time and the number of fault shutdowns, saving a lot of human resources. It is estimated that the troubleshooting time will be reduced by 400 hours in the past three years; for users, the fault power outage time is greatly shortened, which improves the user's sense of gain in electricity use.

[0038] 4) Improve the scientific and technological innovation capabilities of enterprises: During the project development period, a patent development license for "New Method for Detecting High-Resistance Arc Faults in Low-Resistance Grounding Systems" can be signed with the enterprise to develop a high-resistance arc fault detection device. The project team will conduct production, learning and research cooperation with the enterprise to improve the enterprise's scientific and technological innovation capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 : Flow chart of the method for detecting high-resistance arc fault in distribution network of the present invention;

[0040] Figure 2 : Topological structure diagram of the low-resistance grounding system of the distribution network of the present invention;

[0041] Figure 3 : Zero-sequence equivalent network diagram when a high-resistance fault occurs in the low-resistance grounding system of the distribution network of the present invention;

[0042] Figure 4 : 10kV radial low resistance grounding system model in the embodiment of the present invention;

[0043] Figure 5 : In the embodiment of the present invention, 41 The current waveform of each feeder when grounded through a 2000Ω grounding resistor for 0.5s;

[0044] Figure 6 : In the embodiment of the present invention, 41When the grounding is performed through a 2000Ω grounding resistor for 0.5s, the current waveform of each feeder is decomposed by the MVMD algorithm to obtain the fault characteristic mode waveform;

[0045] Figure 7 : Emanuel arc model simulation in an embodiment of the present invention. DETAILED DESCRIPTION

[0046] The specific contents of the present invention are further explained in detail below in conjunction with embodiments.

[0047] like Figure 1 As shown, a method for detecting high-resistance arc faults in a distribution network comprises the following steps:

[0048] Step 1: Sample the zero-sequence current of each feeder in the low-resistance grounding system of the distribution network, decompose the zero-sequence current of each feeder using the Multivariate Variational Mode Decomposition (MVMD) algorithm, perform Fourier transform on each decomposed mode, obtain the corresponding frequency, and select the mode in the frequency band of 45Hz to 65Hz as the fault characteristic mode x. l , the specific process is:

[0049] Step 1.1: Number each feeder l in the low-resistance grounding system, expressed as:

[0050] l=1,2,…,n

[0051] Where n is the total number of feeders in the low-resistance grounding system, n>3;

[0052] Step 1.2: Collect the zero-sequence current ZSC of each feeder in the data window l , where the data window length is from 0.04s before the fault to 0.05s after the fault;

[0053] Step 1.3: Use the multivariate variational mode decomposition algorithm to calculate the zero-sequence current ZSC of each feeder. l Decompose, perform Fourier transform on the decomposed mode, obtain the corresponding frequency, and select the mode in the frequency band of 45Hz to 65Hz as the fault characteristic mode x l ;

[0054] Step 2: Use the fault characteristic modes of each feeder to calculate the waveform similarity matrix D, and then use the waveform similarity matrix D to calculate the waveform similarity coefficient ρ of each feeder l When the waveform similarity coefficient is less than the threshold value of 0.8, the feeder corresponding to the waveform similarity coefficient with the smallest value is determined to be the faulty feeder. Otherwise, it is determined to be a busbar fault. The specific process is as follows:

[0055] Step 2.1, define =1,2,…,n, and have the same value range as l, use ρ l represents the fault characteristic mode x of feeder l l The fault characteristic mode x of feeder h h The cosine similarity of the waveform is calculated by calculating the waveform similarity matrix D, which is expressed as:

[0056]

[0057] Taking signals s1 and s2 as an example, the calculation formula of cosine similarity is expressed as:

[0058]

[0059] Where U is the sampling point of signals s1 and s2;

[0060] Therefore, the fault characteristic mode x of feeder l can be derived l The fault characteristic mode x of feeder h h The cosine similarity ρ l It is expressed as:

[0061]

[0062] Where U is x l and x h The sampling points;

[0063] Step 2.2: Calculate the waveform similarity coefficient ρ of each feeder l , expressed as:

[0064]

[0065] Step 2.3, the threshold of the waveform similarity coefficient is set to 0.8. When the calculated waveform similarity coefficient is less than 0.8, the feeder corresponding to the waveform similarity coefficient with the smallest value is determined to be the faulty feeder. Otherwise, it is determined to be a busbar fault.

[0066] When the busbar fails, the waveform similarity coefficient of each feeder is close to 0.9 or even greater than 0.9, because the waveforms of each feeder are very similar. Therefore, in order to distinguish between busbar faults and feeder faults, the threshold of the waveform similarity coefficient is set to 0.8. At the same time, the experimental simulation data and the field recording data are used for verification and correction.

[0067] Step 3: Calculate the charge Q l , expressed as:

[0068] Q l =∫x l dt (5)

[0069] In the formula, t represents time, x lrepresents the fault characteristic mode of feeder l;

[0070] The fault characteristic mode x l The accumulated product of the time t is the charge Q l , for the calculated charge Q l Perform amplitude comparison. When the maximum charge amplitude is greater than the sum of the remaining charge amplitudes, the feeder corresponding to the maximum charge amplitude is determined to be a faulty feeder. Otherwise, it is determined to be a busbar fault.

[0071] Step 4: Calculate the current change rate V l , expressed as:

[0072]

[0073] In the formula, t represents time, x l represents the fault characteristic mode of feeder l;

[0074] The fault characteristic mode x l The current change rate V is obtained by accumulating the derivative with respect to time t. l , for the calculated current change rate V l Perform amplitude comparison. When the maximum amplitude of the current change rate is greater than the sum of the amplitudes of the other current change rates, the feeder corresponding to the maximum amplitude of the current change rate is determined to be a faulty feeder. Otherwise, it is determined to be a busbar fault.

[0075] Step 5: When the judgment results of steps 2 to 4 are consistent, the final faulty feeder is directly obtained; otherwise, a voting method is adopted to obtain the final faulty feeder according to the principle of "the minority obeys the majority"; if the faulty feeder cannot be determined, return to step 1, move the data window for resampling, and repeat steps 1 to 5.

[0076] The method for detecting high-resistance arc faults in a distribution network proposed in this embodiment calculates the steady-state fault characteristics of a low-resistance grounding system as follows: Figure 2 Taking the topological structure of the low-resistance grounding system of the distribution network shown in FIG as an example, when an arc high-resistance fault occurs, the feeder inductance and feeder resistance can be ignored. Therefore, the zero-sequence equivalent network of the low-resistance grounding system is as follows: Figure 3 As shown, there are n feeders in total. Assuming that the nth feeder is a faulty feeder, C 0k is the zero-sequence capacitance of the kth feeder to ground, k = 1, 2, ..., n, R is equal to 3 times the grounding resistance R g , R N Equal to 3 times R aN ,i 0kz is the zero-sequence current of the kth feeder, i 0fz with i 0Nz They represent the fault point current and neutral point current respectively, u 0fzand u0(t2) represent the zero-sequence voltage at the fault point and the zero-sequence voltage at the busbar, respectively. Figure 3 , write the fault point current i 0fz The expression of is:

[0077]

[0078] Let u 0fz =U mz sin(ω0t2+θ z ), U mz and θ z are the phase voltage amplitude and initial phase angle before the fault, ω0 represents the power frequency angular frequency, t2 represents the time sampling point, and the bus zero-sequence voltage u0(t2) is expressed as:

[0079]

[0080] The neutral point current i 0Nz It is expressed as:

[0081]

[0082] The healthy feeder zero sequence current i 0kz It is expressed as:

[0083]

[0084] Where, k = 1, 2, ..., n-1;

[0085] The fault feeder zero sequence current i 0nz It is expressed as:

[0086]

[0087] In the formula, C 0n represents the capacitance of the nth feeder to ground. The above formulas (7) to (11) involve λ and δ z Respectively expressed as:

[0088]

[0089] The neutral point current steady-state component i 0Nz_t2 It is expressed as:

[0090]

[0091] The busbar zero-sequence voltage steady-state component u 0_t2 It is expressed as:

[0092]

[0093] The steady-state component of the zero-sequence current of the healthy feeder is expressed as:

[0094]

[0095] The fault feeder zero-sequence current steady-state component i 0nz_t2 It is expressed as:

[0096]

[0097] This embodiment analyzes the fault characteristics of the low-resistance grounding system by taking into account the actual parameters of the distribution network. When the ground capacitance current ranges from 10A to 200A, C 0sum The value range is 1.84×10 -6 F~3.68×10 -5 F; at the same time, the maximum capacitance current of a single feeder to ground does not exceed 50A, then C 0k The maximum value is 9.193×10 -6 F; Analyze the steady-state component i of zero-sequence current of fault feeder in actual distribution network parameters 0nz_t2 The steady-state component i of the zero-sequence current of the healthy feeder 0kz_t2 Phase and amplitude differences:

[0098] For the phase difference, the actual parameters of the distribution network can be used to solve 0.017≤C 0sum R N ω0≤0.35, and then it is deduced that the approximate range of λ is 0.97°~19.29°, then i 0nz_t2 with i 0kz_t2 The phase difference is 90.97°~119.29°, and this feature is independent of the grounding resistance. Therefore, the zero-sequence current steady-state component of the fault feeder and the zero-sequence current steady-state component of the sound feeder have a high degree of phase differentiation;

[0099] For the amplitude difference, the amplitude difference between the steady-state component of the neutral point current and the steady-state component of the sound feeder zero-sequence current is R N U mz C 0k ω0 and In general, And this feature is independent of the grounding resistance. Therefore, the amplitude of the zero-sequence current steady-state component of the fault feeder is highly distinguishable from that of the sound feeder.

[0100] In order to verify the effectiveness of the distribution network high-resistance arc fault detection method proposed in this embodiment, the following is established: Figure 4 The 10kV radial low-resistance grounding system model shown in Figure 1 has a total of 4 feeders, among which the distribution network parameters are shown in Table 1 and Figure 4As shown in the figure, M1~M6 represent zero-sequence current transformers, l represents overhead lines or cables, l1, l2, l3, and l4 are feeder 1, feeder 2, feeder 3, and feeder 4 respectively. When an arc fault occurs, the current is severely distorted and has an indirect arcing phenomenon. The Emanuel arc model is used to simulate high-resistance arc faults. The specific parameters are shown in Table 2 and Figure 7 As shown;

[0101] Table 1 Distribution network parameters

[0102]

[0103]

[0104] Table 2 Parameters of three types of arc

[0105]

[0106] Figure 5 In the 10kV radial low resistance grounding system model, 41 The current waveforms of each feeder when grounded for 0.5s through a 2000Ω grounding resistor are observed. Figure 5 In (a), (b), (c) and (d), it can be concluded that 0.5s after the fault occurs, the waveform of the faulty feeder l4 is different from that of other feeders in amplitude and phase, among which the amplitude difference is more obvious; Figure 6 In the 10kV radial low resistance grounding system model, 41 The fault characteristic mode waveforms obtained by decomposing the current waveforms of each feeder through a 2000Ω grounding resistor for 0.5s are observed. Figure 6 From (a), (b), (c) and (d), it can be concluded that 0.5s after the fault occurs, the fault characteristic mode waveform of the faulty feeder l4 has obvious differences in amplitude and phase with the fault characteristic mode waveforms of other feeders. Therefore, these differences can be used to construct the high-resistance arc fault detection criterion.

[0107] When the feeder l 41When grounded for 0.5s through a 2000Ω grounding resistor, the fault judgment process of the distribution network high-resistance arc fault detection method proposed in the embodiment is analyzed. When a fault occurs, the MVMD algorithm is first used to extract the fault characteristic modes of feeder 1, feeder 2, feeder 3, and feeder 4, and then the waveform similarity coefficients of feeder 1, feeder 2, feeder 3, and feeder 4 are calculated to be 0.7194, 0.7969, 0.7326, and 0.3086, respectively. There is a waveform similarity coefficient less than 0.8, so the feeder corresponding to the minimum waveform similarity coefficient of 0.3086 can be judged as the fault feeder, so feeder 4 is judged to be the fault feeder; then the charge amounts of feeder 1, feeder 2, feeder 3, and feeder 4 are calculated to be -0.1257, 0.0783, -0.0231, and -0.3086, respectively. 12.5040, there is a maximum charge amplitude 12.5040 that is greater than the sum of the other charge amplitudes, and the feeder corresponding to the maximum charge amplitude 12.5040 is determined to be the faulty feeder, that is, feeder 4 is the faulty feeder; then the current change rates of feeder 1, feeder 2, feeder 3, and feeder 4 are calculated to be 0.7782, 0.1708, 0.2676, and 2.8147, respectively, and there is a maximum current change rate amplitude 2.8147 that is greater than the sum of the other current change rate amplitudes, and the feeder corresponding to the maximum current change rate amplitude 2.8147 is determined to be the faulty feeder, so feeder 4 can be determined to be the faulty feeder; finally, the fault determination results based on the waveform similarity coefficient criterion, the charge criterion, and the current change rate criterion are consistent, so the final fault is feeder 4.

[0108] When the bus is grounded through a 2000Ω grounding resistor for 0.5s, the fault determination process of the distribution network high-resistance arc fault detection method proposed in this embodiment is analyzed. After the fault occurs, the MVMD algorithm is first used to extract the fault characteristic modes of feeder 1, feeder 2, feeder 3, and feeder 4; then the waveform similarity coefficients of feeder 1, feeder 2, feeder 3, and feeder 4 are calculated to be 0.9293, 0.9143, 0.9230, and 0.9271, respectively. Since the waveform similarity coefficients are all greater than 0.8, it is determined to be a bus fault; then the charge amounts of feeder 1, feeder 2, feeder 3, and feeder 4 are calculated to be -0.74 62, -0.5786, -0.5092, -0.0580. Since there is no maximum charge amplitude greater than the sum of the remaining charge amplitudes, it is determined to be a bus fault. Then the current change rates of feeder 1, feeder 2, feeder 3, and feeder 4 are calculated to be -0.2780, -0.1981, -0.3545, and -0.0390, respectively. Since there is no maximum current change rate amplitude greater than the sum of the remaining current change rate amplitudes, it is determined to be a bus fault. Finally, according to the consistent judgment results of the waveform similarity coefficient criterion, the charge criterion, and the current change rate criterion, the final fault is the bus.

[0109] also, Figure 4When other feeders and busbars are grounded with different grounding resistances for 0.5s, the fault judgment results are shown in Table 3; Figure 7 Under the Emanuel arc model simulation shown, the high-resistance arc fault judgment results of the feeder and bus are shown in Table 4;

[0110] Table 3 Fault location determination results under different grounding resistances

[0111]

[0112]

[0113] Table 4 High resistance arc fault determination results

[0114]

[0115] It can be seen from Table 3 and Table 4 that the distribution network high-resistance arc fault detection method proposed in this embodiment can cope with the diversity of grounding faults and accurately and quickly detect high-resistance arc faults.

Claims

1. A method for detecting high-resistance arc faults in a distribution network, characterized in that: The steps include: Step 1: Sample the zero-sequence current of each feeder in the low-resistance grounding system of the distribution network, decompose the zero-sequence current of each feeder using the multivariate variational mode decomposition algorithm, perform Fourier transform on each decomposed mode, obtain the corresponding frequency, and select the mode in the frequency band of 45Hz to 65Hz as the fault characteristic mode x l ; Step 2: Use the fault characteristic modes of each feeder to calculate the similarity matrix D, and then use the similarity matrix D to calculate the waveform similarity coefficient ρ of each feeder l , when the waveform similarity coefficient is less than the threshold value β, the feeder corresponding to the waveform similarity coefficient with the smallest value is determined to be the faulty feeder, otherwise, it is determined to be a busbar fault; Step 3: Calculate the charge Q l , expressed as: Q l =∫x l dt In the formula, t represents time, x l represents the fault characteristic mode of feeder l; The fault characteristic mode x l The accumulated product of the time t is the charge Q l , when the maximum charge magnitude is greater than the sum of the remaining charge magnitudes, the feeder corresponding to the maximum charge magnitude is determined to be a faulty feeder, otherwise, it is determined to be a busbar fault; Step 4: Calculate the current change rate V l , expressed as: In the formula, t represents time, x l represents the fault characteristic mode of feeder l; The fault characteristic mode x l The current change rate V is obtained by accumulating the derivative with respect to time t. l , when the maximum amplitude of the current change rate is greater than the sum of the amplitudes of the other current change rates, the feeder corresponding to the maximum amplitude of the current change rate is determined to be a faulty feeder, otherwise, it is determined to be a busbar fault; Step 5: When the determination results of steps 2 to 4 are consistent, the final fault feeder is directly obtained; Otherwise, the final fault feeder is determined by voting based on the principle of majority rule; If the faulty feeder still cannot be determined, return to step 1 to step 5.

2. The method for detecting high-resistance arc faults in a distribution network according to claim 1, characterized in that: The specific process of step 1 is as follows: Step 1.1: Number each feeder l in the low-resistance grounding system, expressed as: l＝1,2,…,n Where n is the total number of feeders in the low-resistance grounding system, n>3; Step 1.2: Collect the zero-sequence current ZSC of each feeder in the data window l , where the data window length is from 0.04s before the fault to 0.05s after the fault; Step 1.3: Use the multivariate variational mode decomposition algorithm to calculate the zero-sequence current ZSC of each feeder. l Decompose, perform Fourier transform on the decomposed mode, obtain the corresponding frequency, and select the mode in the frequency band of 45Hz to 65Hz as the fault characteristic mode x l .

3. The method for detecting high-resistance arc faults in a distribution network according to claim 1, characterized in that: The specific process of step 2 is: Step 2.1, define =1,2,…,n, and have the same value range as l, use ρ l represents the fault characteristic mode x of feeder l l The fault characteristic mode x of feeder h h The cosine similarity of the waveform is calculated by calculating the waveform similarity matrix D, which is expressed as: Step 2.2: Calculate the waveform similarity coefficient ρ of each feeder l , expressed as: Step 2.3, set a threshold β of the waveform similarity coefficient. When the calculated waveform similarity coefficient is less than β, the feeder corresponding to the waveform similarity coefficient with the smallest value is determined to be a faulty feeder. Otherwise, it is determined to be a busbar fault.

4. The method for detecting high-resistance arc faults in a distribution network according to claim 3, characterized in that: The fault characteristic mode x of feeder line l in step 2.1 l The fault characteristic mode x of feeder h h The cosine similarity ρ l It is expressed as: Where U is x l and x h The sampling point.

5. The method for detecting high-resistance arc faults in a distribution network according to claim 3, characterized in that: In step 2.3, the threshold β is set to 0.8.

Images