An arc light high-resistance fault line selection method for active low-resistance grounding system

Abstract

The application provides an arc high-resistance fault line selection method for an active small-resistance grounding system. Zero sequence currents of first ends of each feeder after fault occurrence are collected; fundamental wave, third harmonic and fifth harmonic in the zero sequence currents are extracted by using empirical wavelet decomposition; third harmonic and fifth harmonic energy of each feeder within a quarter of a cycle after the fault is calculated, and E i The three groups of harmonic energy with the largest amplitude and corresponding feeders g, m and k are selected; based on the fundamental wave, third harmonic and fifth harmonic extracted from the feeders g, m and k, a harmonic matrix H g , H m , H k is constructed; if SSIM(H g , H m ) < SSIM(H m , H k ) and SSIM(H g , H k ) < SSIM(H m , H k ), the feeder g is the fault feeder, and the others are healthy feeders, otherwise, step 1 is returned to re-collect. The application significantly improves the sensitivity and accuracy of the line selection method.

Description

Technical Field

[0001] This invention belongs to the field of power system relay protection detection technology, specifically relating to a method for selecting arc high resistance faults in active low resistance grounding systems. Background Technology

[0002] In distribution network systems, low-resistance grounding is widely used because it can effectively limit overvoltage and quickly disconnect faulty lines. With the large-scale integration of distributed power sources, the distribution network structure has evolved from the traditional single-source radial network to a complex multi-source network, forming an active distribution network. In active low-resistance grounding systems, single-phase grounding faults are the most common type of fault. In particular, when the fault medium is tree branches, sand, or cement, the transition resistance at the grounding point exhibits nonlinear and time-varying characteristics, which can easily lead to arcing high-resistance grounding faults.

[0003] Existing line selection methods for this type of fault mainly rely on the extraction of fault electrical quantity characteristics. When an arcing high-resistance grounding fault occurs, the transition resistance is as high as hundreds or even thousands of ohms, the fault current is weak, and the waveform is accompanied by nonlinear distortion of the arc. This leads to a significant reduction in the sensitivity of traditional methods based on power frequency electrical quantities, and the reliability and accuracy are difficult to meet the needs of actual engineering.

[0004] Therefore, there is an urgent need in this field to solve the problem of fault location accuracy in active low-resistance grounding systems under complex scenarios of high resistance arcing. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a method for selecting faulty feeders in active low-resistance grounding systems during arcing. Utilizing structural similarity theory, the method selects the faulty feeder by comparing the waveform similarity between healthy and faulty lines, thereby improving the accuracy of the fault selection results.

[0006] A method for locating arcing high-resistance faults in an active low-resistance grounding system includes the following steps:

[0007] Step 1: Collect the zero-sequence current I at the beginning of each feeder after the arcing high-impedance fault occurs in the system. 0i (i=1,2,…,n), where i is the feeder number and n is the total number of feeders;

[0008] Step 2: Extract the zero-sequence current I of each feeder using Empirical Wavelet Transform (EWT). 0i The fundamental frequency, third harmonic, and fifth harmonic;

[0009] Step 3: Analyze the zero-sequence current I of each feeder. 0i Calculate the harmonic energy and E of the third and fifth harmonics within T / 4 (T is the system period) after the fault. i (i=1,2,…,n), as shown in the following formula:

[0010] ;

[0011] In the formula, The extracted third harmonic; The extracted fifth harmonic;

[0012] Step 4: Select harmonic energy and E i The three feeders with the largest amplitude are denoted as feeder g, feeder m, and feeder k, respectively.

[0013] Step 5: Based on the extracted fundamental, third, and fifth harmonic components, select the fundamental and harmonic data of feeders g, m, and k within T / 4 after the fault, and construct the harmonic matrix H respectively. g H m H k ;

[0014] Step 6: Calculate SSIM(H) based on Structural Similarity Theory (SSIM). g H m ), SSIM(H g H k ) and SSIM(H m H k );

[0015] Step 7: Based on the calculation results of Step 6, if SSIM(H) g H m ) <SSIM(H m H k And SSIM(H) g H k ) <SSIM(H m H k If the condition is met, then feeder g is a faulty feeder, and all others are healthy feeders; if not, return to step 1 to collect data again.

[0016] Preferably, in step 2, EWT adaptively segments the signal in the frequency domain and constructs a filter bank for each segmented interval, making the separated fundamental and harmonic components purer and improving the accuracy of the line selection results.

[0017] Preferably, the specific process of step 3 is as follows:

[0018] Step 3.1: Based on the fundamental and harmonic frequencies extracted in Step 2, calculate the zero-sequence current I of the first feeder within T / 4 after the fault. 01 The third and fifth harmonic energies and E1 contained in it are shown in the following formula:

[0019] ;

[0020] In the formula, The extracted third harmonic; The extracted fifth harmonic;

[0021] Step 3.2: Based on Step 3.1, the zero-sequence current I from the second feeder to the nth feeder within T / 4 after the fault can be calculated similarly. 0q The third and fifth harmonic energies and E in (q=2,3,…,n) q (q=2,3,…,n), as shown in the following formula: ;

[0022] Step 3.3: Based on steps 3.1 and 3.2, the third and fifth harmonic energies and E contained in the zero-sequence current of each of the n feeders can be obtained. i (i=1,2,…,n).

[0023] Preferably, in step 4, based on the sum of zero-sequence harmonic energies of each feeder calculated in step 3, the feeders with the highest amplitude are selected for similarity calculation and fault line selection. This utilizes the characteristic that faulty feeders and healthy feeders have amplitude and phase differences and waveform distortion during the initial transient phase of a fault, thereby reducing computational load and response time, and improving the reliability and sensitivity of line selection.

[0024] Preferably, the specific process of step 5 is as follows:

[0025] Step 5.1: Based on the extracted fundamental, third, and fifth harmonic components of each feeder, select the fundamental and harmonic data corresponding to feeder g within T / 4 after the fault, and construct a three-row, L-column harmonic matrix. , where L=f s ×T / 4 represents the number of sampling points within T / 4 after the fault, f s The system sampling rate;

[0026] Step 5.2: Similarly to step 5.1, the harmonic matrices corresponding to feeder m and feeder k can be obtained. , .

[0027] Preferably, in step 6, the structural similarity theory (SSIM) is used to solve the waveform similarity between the two signals. This method judges the similarity of signals from the overall structure, is insensitive to local noise interference or small amplitude fluctuations, and has inherent robustness. The waveform of the fault feeder current signal is distorted, and SSIM can capture the differences between waveform structures, significantly improving the accuracy of line selection for arcing high-resistance faults in active low-resistance grounding systems. The specific process includes:

[0028] Step 6.1: Calculate the harmonic matrix H obtained in Step 5 based on structural similarity theory (SSIM). g H mSSIM(H) between g H m The calculation formula is as follows:

[0029] ;

[0030] ;

[0031] ;

[0032] ;

[0033] In the formula, This indicates that the harmonic matrix H corresponds to the feed line g. g The mean of all elements in the set; This indicates that the harmonic matrix H corresponds to feeder m. m The mean of all elements in the set; This indicates that the harmonic matrix H corresponds to the feed line g. g The standard deviation of all elements within the range; This indicates that the harmonic matrix H corresponds to feeder m. m The standard deviation of all elements within the range; H represents the harmonic matrix corresponding to feed line g and feed line m. g H m The covariance; C1, C2, and C3 are constants set to prevent the denominator from being zero;

[0034] Step 6.2: Similarly to step 6.1, SSIM(H) can be calculated. g H k ), SSIM(H m H k ).

[0035] Preferably, the specific process of step 7 is as follows:

[0036] Step 7.1: Apply the SSIM(H) obtained in Step 6 to... g H m ), SSIM(H g H k ), SSIM(H m H k Compare the sizes of each pair of items;

[0037] Step 7.2: If SSIM(H) g H m ) <SSIM(H m H k And SSIM(H) g H k ) <SSIM(H m H kIf the condition is met, then feeder g is the faulty line, and the rest are healthy lines; if not, return to step 1 and repeat the measurement.

[0038] This application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described power equipment condition monitoring and diagnosis method.

[0039] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described power equipment condition monitoring and diagnosis method.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] 1. This invention utilizes structural similarity theory for fault location, transforming traditional numerical comparisons based on electrical magnitude and phase into structural similarity analysis of waveform morphology. It fully leverages the inherent characteristic of large waveform distortion in faulty lines, effectively overcoming the weak fault characteristics, nonlinear distortion, and interference from distributed power sources during arcing high-resistance grounding by quantifying the overall structural similarity of transient zero-sequence current waveforms across lines. This method exhibits higher sensitivity to arcing high-resistance faults, strong noise immunity, and engineering adaptability, significantly improving the reliability and accuracy of arcing high-resistance fault location in active low-resistance grounding systems under complex fault scenarios. Attached Figure Description

[0042] Figure 1 This is a flowchart of the fault selection process of the present invention;

[0043] Figure 2 This is a simulation model diagram of the active low-resistance grounding system in an embodiment of the present invention;

[0044] Figure 3 This is a diagram showing the fault settings of the simulation model in an embodiment of the present invention;

[0045] Figure 4 This refers to the zero-sequence current at the beginning of each feeder when an f1 fault occurs in the embodiment. Detailed Implementation

[0046] The embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and should not be construed as limiting the scope of the invention.

[0047] First, a brief introduction to the terms used in the embodiments of this application will be given.

[0048] Active low-resistance grounding system: refers to a distribution network where the neutral point is grounded through a low resistor and one or more distributed power sources (such as photovoltaic inverters, energy storage systems, micro gas turbines, etc.) are connected to the system. Compared with traditional passive radial networks, the connection of distributed power sources in active systems changes the distribution characteristics and harmonic features of fault currents, increasing the complexity of fault location for high-resistance grounding faults.

[0049] Arc high-resistance fault: This refers to a single-phase ground fault in a distribution network where the fault medium is such as tree branches, sand, cement, or air gaps, causing the grounding point transition resistance to exhibit nonlinear and time-varying characteristics, accompanied by intermittent reignition and extinction of the arc. This type of fault has a weak current (typically below tens of amperes) and severe waveform distortion, significantly reducing the sensitivity of traditional line selection methods based on power frequency electrical quantities.

[0050] Zero-sequence current: refers to the vector sum of three-phase currents. When a single-phase ground fault occurs, the zero-sequence current of the faulted line and the non-faulted line have significant differences in amplitude, phase and waveform, which is an important criterion for fault line selection.

[0051] Empirical Wavelet Transform (EWT) is an adaptive signal decomposition method that can accurately decompose the original signal into several modal components with tightly supported spectra by adaptively segmenting the signal spectrum and constructing an empirical wavelet filter bank. In this application, EWT is used to extract the fundamental, third, and fifth harmonic components from the zero-sequence current to obtain characteristic information of fault waveform distortion.

[0052] Structural Similarity Index Measure (SSIM) is an index that measures the structural similarity between two signals or images. It quantifies the degree of similarity by comparing three dimensions: brightness (mean), contrast (variance), and structure (covariance). The SSIM value ranges from [-1, 1], with values ​​closer to 1 indicating greater structural similarity between the two signals. In this application, SSIM is applied to the similarity comparison of harmonic matrices to quantify the differences in waveform structure between faulty and healthy feeders.

[0053] Harmonic energy sum: This refers to the energy value obtained by integrating the squares (or discretely summing) of the extracted third and fifth harmonics within a specific time window after a fault. It is used to quantify the severity of harmonic distortion in a line. The feeder with the highest harmonic energy sum usually has the most severe waveform distortion and serves as a basis for initially screening candidate fault lines.

[0054] Based on the above definitions, the implementation environment of the arc-induced high-resistance fault location method for active low-resistance grounding systems based on structural similarity provided in this application embodiment will be described. Indicatively, this implementation environment includes:

[0055] Data acquisition unit: Deployed at the beginning of each feeder, including but not limited to zero-sequence current transformer, voltage transformer and high-precision sampling device. Among them, the zero-sequence current transformer is used to acquire the zero-sequence current signal at the beginning of each feeder, with a sampling frequency of not less than 10kHz, to meet the requirements for accurate extraction of fifth harmonic (250Hz) and higher frequency components.

[0056] Signal processing unit: This can be a central processing unit (CPU), digital signal processor (DSP), field-programmable gate array (FPGA), or embedded microcontroller. This unit is used to perform algorithmic operations such as empirical wavelet transform (EWT), harmonic energy calculation, harmonic matrix construction, and structural similarity (SSIM) calculation.

[0057] Storage unit: Includes random access memory (RAM) and read-only memory (ROM) or solid-state drive (SSD), used to store the acquired zero-sequence current data, harmonic components after EWT decomposition, intermediate calculation results, and line selection results. The storage unit can be a distributed storage device or a centralized storage device, which is not limited here.

[0058] Output unit: includes display, alarm device or communication interface, used to output fault line selection results (fault feeder number) and trigger alarm signal when necessary, and upload the line selection results to the dispatch center or operation and maintenance platform through the communication interface.

[0059] Based on the above explanations of terms and implementation environments, the application scenarios of the embodiments of this application are described. The arc high resistance fault selection method for active low resistance grounding systems based on structural similarity provided in the embodiments of this application can be applied to scenarios including but not limited to the following:

[0060] In active power distribution network high-resistance grounding fault diagnosis scenarios, especially for low-resistance grounding systems connected to distributed power sources such as photovoltaics and wind power, this method can effectively identify faulty feeders by using zero-sequence current transformers deployed at the beginning of each feeder to collect fault current signals in real time when arcing high-resistance grounding faults caused by media such as tree branches and sand occur. It utilizes Empirical Wavelet Transform (EWT) to accurately extract the fundamental, third, and fifth harmonic components, and combines harmonic energy with waveform comparison using Screening and Structural Similarity (SSIM). Compared to traditional line selection methods that rely on power frequency amplitude or phase, this scheme maintains high accuracy even in extreme scenarios with weak fault characteristics and severe waveform distortion, avoiding cascading trips or expanded fault ranges due to incorrect line selection.

[0061] In complex distribution network scenarios with large-scale integration of distributed generation sources, this technical solution can adapt to situations where the fault current characteristics change after the integration of inverter-type distributed generation sources. Since harmonic injection from distributed generation sources can interfere with traditional fault location criteria, this solution uses EWT adaptive separation of fault harmonic components and combines it with SSIM waveform structure comparison to effectively avoid interference from distributed generation sources, ensuring the reliability of fault location results and supporting the safe and stable operation of active distribution networks.

[0062] In fault protection scenarios for core urban power supply areas and critical load power supply lines, this technical solution can be applied to substation outgoing lines or important user feeders where power supply reliability requirements are extremely high. By monitoring the zero-sequence current waveform in real time, harmonic energy calculation, candidate line screening, and SSIM comparison are quickly completed within a quarter cycle (T / 4) after the fault occurs, enabling rapid and accurate determination of the faulty line. This provides a reliable basis for line selection for protection devices, significantly shortens fault clearing time, and reduces power outage losses for users.

[0063] As an illustration, the arc high resistance fault selection method for active low resistance grounding systems based on structural similarity provided in this application embodiment can also be applied to other scenarios that require high sensitivity high resistance grounding fault identification. This is only an example and does not limit the specific application scenario.

[0064] In one embodiment, the present invention provides a method for selecting the fault location in an active low-resistance grounding system during arcing high-resistance faults, such as... Figure 1 As shown, it includes the following steps:

[0065] Step 1: Collect the zero-sequence current I at the beginning of each feeder after the arcing high-impedance fault occurs in the system. 0i (i=1,2,…,n), where i is the feeder number and n is the total number of feeders;

[0066] Step 2: Extract the zero-sequence current I of each feeder using Empirical Wavelet Transform (EWT). 0i The fundamental frequency, third harmonic, and fifth harmonic;

[0067] Specifically, the process of extracting the fundamental and higher harmonics using empirical wavelet transform (EWT) in step 2 above includes: First, performing Fourier transform on the zero-sequence current of each feeder to obtain the spectrum; adaptively dividing the Fourier spectrum interval according to the spectral maximum; then constructing an empirical wavelet filter bank in each segmented interval; and accurately separating the fundamental and harmonic components through filtering.

[0068] The operating steps are as follows: First, collect the zero-sequence current I at the beginning of each feeder. 0i (Original data), and then the EWT algorithm is used to decompose the original signal:

[0069] (1) Perform a Fourier transform on the zero-sequence current signal to obtain the spectrum function;

[0070] (2) Detect the local maxima of the spectrum, sort them from largest to smallest amplitude, take the first M maxima (M=5, i.e. the frequency band corresponding to the fundamental wave and the 2nd to 5th harmonics), take the midpoint of adjacent maxima as the frequency band boundary, and adaptively divide the spectrum into M continuous intervals;

[0071] (3) In each frequency band, according to the Meyer wavelet construction method, construct the empirical wavelet scaling function and the empirical wavelet function respectively to form a filter bank;

[0072] (4) Perform convolution operation between the original zero-sequence current signal and each filter to extract the fundamental, third harmonic and fifth harmonic components respectively.

[0073] Wherein, the sampling frequency f s Satisfying the Nyquist sampling theorem, by setting it to above 10kHz, ensures accurate extraction of the highest 5th harmonic (250Hz).

[0074] Step 3: Analyze the zero-sequence current I of each feeder. 0i Calculate the harmonic energy and E of the third and fifth harmonics within T / 4 after the fault. i (i=1,2,…,n), as shown in the following formula:

[0075] ;

[0076] In the formula, The extracted third harmonic; The extracted fifth harmonic; The sampling time interval (which is expressed as the time derivative, i.e., the time step of discrete sampling); For time variables (which represent continuous sampling moments on the time axis after the fault occurs; unit: seconds (s) or milliseconds (ms); and Indicates time At this specific moment, the third and fifth harmonic current values ​​are extracted using Empirical Wavelet Transform (EWT); T is the system period.

[0077] Specifically, the process of step 3 above is as follows:

[0078] Step 3.1: Based on the fundamental and harmonic frequencies extracted in Step 2, calculate the zero-sequence current I of the first feeder within T / 4 after the fault. 01 The third and fifth harmonic energies and E1 contained in it are shown in the following formula:

[0079] ;

[0080] In the formula, The extracted third harmonic; The extracted fifth harmonic;

[0081] Step 3.2: Based on Step 3.1, the zero-sequence current I from the second feeder to the nth feeder within T / 4 after the fault can be calculated similarly. 0q The third and fifth harmonic energies and E in (q=2,3,…,n) q (q=2,3,…,n), as shown in the following formula: ;

[0082] Step 3.3: Based on steps 3.1 and 3.2, the third and fifth harmonic energies and E contained in the zero-sequence current of each of the n feeders can be obtained. i (i=1,2,…,n);

[0083] For discrete sampled signals, the sum of harmonic energies can also be approximated by integration using a summation method:

[0084] ;

[0085] in, This represents the number of sampling points within the T / 4 time window after the fault. f s The system sampling rate is used; the fault initiation time is determined by detecting zero-sequence voltage change or phase voltage change.

[0086] Step 4: Select harmonic energy and E i The three feeders with the largest amplitude are denoted as feeder g, feeder m, and feeder k, respectively.

[0087] Specifically, in step 4 above, the harmonic energy and E are selected. i The three feeders with the largest amplitude are selected, and subsequent structural similarity calculations and line selection operations are only performed on these three feeders; by excluding feeders with weak harmonic energy, the amount of data processing is effectively reduced, and the overall line selection speed is improved.

[0088] Step 5: Based on the extracted fundamental, third, and fifth harmonic components, select the fundamental and harmonic data of feeders g, m, and k within T / 4 after the fault, and construct the harmonic matrix H respectively. g H m H k ;

[0089] Specifically, the process of step 5 above includes:

[0090] Step 5.1: Based on the extracted fundamental, third, and fifth harmonic components of each feeder, select the fundamental and harmonic data corresponding to feeder g within T / 4 after the fault, and construct a three-row, L-column harmonic matrix. , where L=fs ×T / 4 represents the number of sampling points within T / 4 after the fault, f s The system sampling rate;

[0091] Step 5.2: Similarly to step 5.1, the three-row, L-column harmonic matrix corresponding to feeder m and feeder k can be obtained. , ;

[0092] Step 6: Calculate SSIM(H) based on Structural Similarity Theory (SSIM). g H m ), SSIM(H g H k ) and SSIM(H m H k );

[0093] Specifically, the process of step 6 above includes:

[0094] Step 6.1: Based on structural similarity theory, calculate the harmonic matrix H obtained in step 5. g H m SSIM(H) between g H m The calculation formula is as follows:

[0095] ;

[0096] ;

[0097] ;

[0098] ;

[0099] In the formula, This indicates that the harmonic matrix H corresponds to the feed line g. g The mean of all elements in the set; This indicates that the harmonic matrix H corresponds to feeder m. m The mean of all elements in the set; This indicates that the harmonic matrix H corresponds to the feed line g. g The standard deviation of all elements within the range; This indicates that the harmonic matrix H corresponds to feeder m. m The standard deviation of all elements within the range; H represents the harmonic matrix corresponding to feed line g and feed line m. g H m The covariance; C1, C2, and C3 are constants set to prevent the denominator from being zero (constants C1, C2, and C3 are set according to the dynamic range L of the elements in the harmonic matrix: C1 = (K1 × L) 2 C2 = (K2 × L) 2C3 = C2 / 2; where K1 = 0.01, K2 = 0.03, and L is the maximum absolute value of all elements in the harmonic matrix H. (When the denominator is zero or close to zero, the constant plays a role in stabilizing the numerical calculation).

[0100] Step 6.2: Similarly to step 6.1, SSIM(H) can be calculated. g H k ), SSIM(H m H k );

[0101] Furthermore, in step 6, when calculating the correlation between two signals, the Structural Similarity Theory (SSIM) can determine the similarity of signals from the overall waveform structure. It is not sensitive to local noise interference or small amplitude fluctuations and has natural robustness. Moreover, the waveform of the fault feeder current signal is distorted, and SSIM can capture the differences between waveform structures, which significantly improves the line selection accuracy of arc high resistance grounding faults in active low resistance grounding systems.

[0102] Step 7: Based on the calculation results of Step 6, if SSIM(H) g H m ) <SSIM(H m H k And SSIM(H) g H k )< SSIM(H m H k If the condition is met, then feeder g is a faulty feeder, and all others are healthy feeders; if not, return to step 1 to collect data again.

[0103] In one embodiment, the difference from the embodiments described above is:

[0104] Step 7: Based on the calculation results of Step 6, if SSIM(H) g H m ) <SSIM(H m H k And SSIM(H) g H k ) <SSIM(H m H k If feed line g is faulty, then feed line g is the faulty feed line, and all others are healthy feed lines; if the judgment condition is not met, then perform the following operations:

[0105] (1) If the cumulative number of re-collections does not exceed 3, return to step 1 and re-collect data for the next time window;

[0106] (2) If the cumulative number of re-collection exceeds 3 times and the condition is still not met, it is determined that the line selection has failed, an alarm signal is output, and the backup protection strategy (such as delayed tripping or manual confirmation) is activated.

[0107] As shown above, the arc fault location method for active low-resistance grounding systems based on structural similarity fully utilizes the waveform structural differences of the transient zero-sequence current in the initial stage of the fault. This method transforms the traditional comparison of electrical magnitude and phase values ​​into a structural similarity analysis of waveform morphology. By quantifying the overall structural similarity of transient waveforms between lines, it overcomes the interference caused by weak fault characteristics and nonlinear distortion during arc fault high-resistance grounding. Selecting feeders with higher harmonic energy for location selection significantly reduces the computational load and improves accuracy. Furthermore, this method uses the fundamental wave and third and fifth harmonic components to construct a harmonic matrix for structural similarity calculation, exhibiting high sensitivity, strong noise immunity, and engineering adaptability, thus enhancing the reliability and accuracy of location selection.

[0108] The following provides a further explanation of the principles behind the above steps:

[0109] 1. Preliminary screening based on harmonic energy.

[0110] From a physical perspective, faults in power systems disrupt the sinusoidal nature of current waveforms, generating abundant harmonic components. The more severe the fault and the more intense the distortion, the higher the harmonic amplitude typically is. Therefore, the magnitude of harmonic energy can quantify the degree of line distortion, with the line having the highest energy being most likely to be faulty. From an engineering perspective, directly performing pairwise SSIM comparisons on all five sets of data requires calculating ten combinations, resulting in a large computational load. However, by first identifying the three most suspicious sets through energy sorting, subsequent comparisons only require comparing three combinations, improving computational efficiency and reducing judgment time. Therefore, using energy indicators for initial selection, followed by more detailed judgment of high-harmonic-content lines using SSIM shape, achieves a complementary relationship between abnormal energy amplitude and waveform structure similarity. The combination of these two approaches allows for more reliable and accurate location of faulty lines.

[0111] 2. Structural Similarity Line Selection Principle

[0112] The SSIM index comprehensively compares the characteristics in three aspects: brightness (i.e., the mean shift of harmonic amplitudes), contrast (i.e., the fluctuation amplitude of amplitudes), and structure (i.e., the correlation between harmonic waveforms). When the change trends, relative amplitude sizes, and fluctuation characteristics of the fault current waveforms of two lines are more consistent within the same time period, their SSIM values are closer to 1; conversely, if the waveform trends are opposite or the fluctuation characteristics are significantly different, the SSIM value will decrease or even become negative. In addition, the fundamental wave determines the amplitude difference of the fault waveforms to a certain extent, and the third and fifth harmonics determine the distortion degree of the fault waveforms. Therefore, the fundamental wave, the third and fifth harmonics are selected for comparison, which also reduces the calculation amount of SSIM.

[0113] The logic of fault line selection is based on the assumption that the fault waveforms of the fault feeder should be significantly different from those of the healthy feeders. Among the top three groups of feeders with high harmonic content screened by harmonic energy ranking, if one of them is the fault feeder, its distorted waveform must be essentially different from the harmonic modes of the other two healthy feeders. Therefore, the program performs pairwise SSIM comparisons on these three groups of lines, meaning that it has the largest structural difference from the fault waveforms of the other two lines and is thus determined to be the fault line. This process is essentially using SSIM to quantify the waveform structural differences to achieve fault identification.

[0114] Example: Build an active low-resistance grounded distribution network model as shown in Figure 2 which includes five feeders and three inverter-type distributed power sources. The feeders include overhead lines, cables, and overhead-cable hybrid lines. The specific network parameters are shown in Table 1:

[0115] Table 1 Cable and overhead line parameters

[0116]

[0117] As shown in Figure 3 taking the occurrence of fault f1 as an example, that is, an arc flash high-resistance fault occurs on feeder l1, the zero-sequence current waveforms at the heads of each feeder as shown in Figure 4 can be collected. After extracting the harmonics in the zero-sequence current of each feeder through empirical wavelet transform and calculating the sum of the third and fifth harmonic energies of each feeder, the feeders with the top three amplitudes are l1, l4, and l5; then form their harmonic matrices and calculate the SSIM values between each pair, and the calculation order is SSIM(l1, l4), SSIM(l1, l5), and SSIM(l4, l5), and the calculation results are [0.2371, 0.0746, 0.6125]. From this, it can be seen that SSIM(l1, l4) < SSIM(l4, l5) and SSIM(l1, l5) < SSIM(l4, l5). Therefore, it is determined that feeder l1 has a fault and is the fault feeder, and the remaining feeders are all healthy feeders. The determination result is correct.

[0118] Here, to verify the feasibility of the detection method proposed in this invention, as Figure 3 shown, more fault scenarios are further set up for verification, and the specific simulation results are shown in Table 2 below:

[0119] Table 2 Line selection results under different fault location scenarios

[0120]

[0121] As can be seen from the above table, the SSIM values between the faulty feeder and the sound lines are always smaller than the SSIM values between the sound lines. For example, when a fault occurs at f2, there are SSIM(l2, l1) = 0.1470; SSIM(l2, l4) = 0.1665; SSIM(l1, l4) = 0.9210, and there is SSIM(l2, l1) < SSIM(l1, l4) and SSIM(l2, l4) < SSIM(l1, l4). Therefore, line 2 is determined as the faulty feeder. Similarly, in other different faulty line scenarios, this method can accurately select the faulty line.

[0122] When the distance of the fault location from the bus changes, simulation verification is carried out again. At this time, set l 11 = 6 km, l 21 = 5 km, l 31 = 8 km, l 41 = 3 km and l 51 = 6 km for testing, and the detection results shown in Table 3 below can be obtained.

[0123] Table 3 Line selection results when changing the fault distance scenario

[0124]

[0125] As can be obtained from Table 3 above, when the position of the fault point from the bus is changed, the calculated SSIM values still satisfy that the SSIM values between the faulty feeder and the sound feeders are always smaller than the SSIM values between the sound lines. Therefore, this line selection method can still accurately and effectively select the faulty feeder.

[0126] Through the above method, an arc-fault high-resistance fault line selection method for an active small-resistance grounding system in this invention, through the strategies of EWT adaptive decomposition, harmonic energy screening, and structural similarity index comparison, not only overcomes the problem of insufficient sensitivity of traditional methods under arc-fault high-resistance faults, but also takes into account the calculation efficiency and anti-interference ability, and significantly improves the accuracy of fault line selection.

[0127] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of a power equipment status monitoring and diagnosis method as previously described are implemented.

[0128] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0129] The embodiments of the present invention are given for the purposes of illustration and description. Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for selecting the fault location in an active low-resistance grounding system for arcing high-resistance faults, characterized in that, Includes the following steps: Step 1: Collect the zero-sequence current I at the beginning of each feeder after the arcing high-impedance fault occurs in the system. 0i (i=1,2,…,n), where i is the feeder number and n is the total number of feeders; Step 2: Extract the zero-sequence current I of each feeder using empirical wavelet transform. 0i The fundamental frequency, third harmonic, and fifth harmonic; Step 3: Analyze the zero-sequence current I of each feeder. 0i Calculate the harmonic energy and E of the third and fifth harmonics within a time window of T / 4 after the fault. i (i=1,2,…,n); Step 4: Select harmonic energy and E i The three feeders with the largest amplitude are denoted as feeder g, feeder m, and feeder k, respectively. Step 5: Based on the extracted fundamental, third, and fifth harmonic components, select the fundamental and harmonic data of feeders g, m, and k within T / 4 after the fault, and construct the harmonic matrix H respectively. g H m H k ; Step 6: Based on structural similarity theory, calculate SSIM(H g H m ), SSIM(H g H k ) and SSIM(H m H k ); Step 7: Based on the calculation results of Step 6, if SSIM(H) g H m ) <SSIM(H m H k And SSIM(H) g H k ) <SSIM(H m H k If the condition is met, then feeder g is a faulty feeder, and all others are healthy feeders; if not, return to step 1 to collect data again.

2. The method for selecting the fault location in an active low-resistance grounding system for arcing high resistance as described in claim 1, characterized in that: In step 3, harmonic energy and E i (i=1,2,…,n), as shown in the following formula: ； In the formula, The extracted third harmonic; The extracted fifth harmonic; The sampling time interval; is a time variable; T is the system period.

3. The method for selecting the fault location in an active low-resistance grounding system for arcing high resistance as described in claim 1, characterized in that: In step 4, the harmonic energy and E are selected. i The three feeders with the largest amplitudes are selected, and subsequent structural similarity calculations and line selection operations are only performed on these three feeders.

4. The method for selecting the fault location in an active low-resistance grounding system for arcing high resistance as described in claim 1, characterized in that: The specific process of step 5 includes: Step 5.1: Based on the extracted fundamental, third, and fifth harmonic components of each feeder, select the fundamental and harmonic data corresponding to feeder g within T / 4 after the fault, and construct a three-row, L-column harmonic matrix. , where L=f s ×T / 4 represents the number of sampling points within T / 4 after the fault, f s The system sampling rate; Step 5.2: Similarly to step 5.1, the three-row, L-column harmonic matrix corresponding to feeder m and feeder k can be obtained. , .

5. The method for selecting the fault location in an active low-resistance grounding system for arcing high resistance as described in claim 1, characterized in that: The specific process of step 6 includes: Step 6.1: Based on structural similarity theory, calculate the harmonic matrix H obtained in step 5. g H m SSIM(H) between g H m The calculation formula is as follows: ； ； ； ； In the formula, This indicates that the harmonic matrix H corresponds to the feed line g. g The mean of all elements in the set; This indicates that the harmonic matrix H corresponds to feeder m. m The mean of all elements in the set; This indicates that the harmonic matrix H corresponds to the feed line g. g The standard deviation of all elements within the range; This indicates that the harmonic matrix H corresponds to feeder m. m The standard deviation of all elements within the range; H represents the harmonic matrix corresponding to feed line g and feed line m. g H m The covariance; C1, C2, and C3 are constants set to prevent the denominator from being zero; Step 6.2: Similarly to step 6.1, SSIM(H) can be calculated. g H k ), SSIM(H m H k ).

6. The method for selecting the fault location in an active low-resistance grounding system for arcing high resistance as described in claim 1, characterized in that: The specific process of extracting the fundamental and higher harmonics using empirical wavelet transform in step 2 includes: First, performing Fourier transform on the zero-sequence current of each feeder to obtain the spectrum; adaptively dividing the Fourier spectrum interval according to the spectral maxima; then constructing an empirical wavelet filter bank in each segmented interval; and separating the fundamental and harmonic components through filtering.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.

Images