Methods, systems, computer equipment, and media for locating high-resistance grounding faults in low-resistance grounding systems.

Abstract

This invention relates to the field of power distribution network fault detection technology, and particularly to a method, system, computer equipment, and medium for selecting high-resistance grounding faults in low-resistance grounding systems. The method includes: acquiring the bus zero-sequence voltage and the zero-sequence current of each feeder to obtain a set of characteristic parameters; identifying the main noise frequency band and determining the target high-frequency detection frequency band; when preset start-up conditions are met, controlling the neutral point-to-ground bypass branch to switch between at least two different impedance levels; constructing a normalized energy response sequence for each feeder to obtain the matched filter output value for each feeder; and identifying the feeder with the largest matched filter output value that meets preset significant conditions as the faulty feeder. This approach solves the technical problem of insufficient fault feature extraction accuracy in environments with strong noise interference, improving the accuracy, anti-interference capability, and adaptability to complex operating environments for selecting high-resistance grounding faults.

Description

Technical Field

[0001] This invention relates to the field of power distribution network fault detection technology, and in particular to a method, system, computer equipment, and medium for selecting high-resistance grounding faults in low-resistance grounding systems. Background Technology

[0002] Low-resistance grounding systems are widely used in medium-voltage distribution networks due to their ability to effectively limit fault current, suppress overvoltage, and provide a certain level of detection sensitivity. When a single-phase ground fault occurs, protection and fault location devices mainly rely on monitoring and analyzing the zero-sequence voltage of the bus and the zero-sequence current of each feeder to determine the fault and identify the feeder. However, when there is a large transition resistance at the ground fault point, i.e., a high-resistance ground fault occurs, the fault current amplitude decreases significantly, and the electrical characteristic signal becomes extremely weak. This poses a serious challenge to traditional fault location methods that rely on comparison of power frequency steady-state quantities, fixed threshold criteria, or analysis of simple changes before and after switching, as they suffer from insufficient sensitivity, weak anti-interference capabilities, and a sharp decline in fault location accuracy.

[0003] More importantly, the background noise characteristics of the distribution network undergo fundamental changes under the "high-voltage, high-efficiency, and high-power" operating environment. Numerous grid-connected inverters, rectifiers, active filters, and various frequency converters inject multi-source, wide-bandwidth, and time-varying power electronic switching noise into the grid during operation. These noise components are superimposed on the zero-sequence voltage and zero-sequence current signals, manifesting as complex switching frequency harmonics, sideband components, and broadband spectral interference. This type of noise not only has a wide spectral distribution but also dynamically changes with fluctuations in new energy output, network topology adjustments, power electronic equipment switching, and control strategy switching, forming a highly interfering and time-varying detection environment. In this context, traditional line selection methods based on fixed detection frequency bands or relying on single, simple disturbance excitations are easily overwhelmed by strong background noise, making it difficult to guarantee reliability in practical engineering applications. Therefore, existing high-resistance grounding fault location technology for low-resistance grounding systems generally suffers from insufficient fault feature extraction accuracy due to fixed detection frequency bands and lack of noise self-adaptation capability in environments with strong noise and time-varying interference caused by the high proportion of new energy and power electronic equipment connected, which in turn leads to insufficient reliability of fault location. Summary of the Invention

[0004] To address the aforementioned shortcomings or disadvantages, this invention provides a method, system, computer equipment, and medium for selecting high-resistance grounding faults in low-resistance grounding systems. This solution addresses the technical problem of insufficient fault feature extraction accuracy in existing technologies under strong noise interference environments.

[0005] This invention provides a method for selecting the fault location of a high-resistance grounding system in a low-resistance grounding system, comprising: Collect electrical quantity data of the low-resistance grounding system and extract a set of characteristic parameters from the collected electrical quantity data. The set of characteristic parameters includes noise intensity indicators.

[0006] Before initiating active detection, the main noise frequency band is identified from the bus zero-sequence voltage of electrical quantity data, and the target high-frequency detection frequency band is determined according to the predetermined frequency band selection rules and the spectrum after removing the main noise frequency band.

[0007] In response to the fulfillment of preset start-up conditions, within a preset detection period, the neutral point to ground bypass branch is controlled to switch between at least two different frequency selective impedance levels according to a preset coding sequence.

[0008] High-frequency components located within the target high-frequency detection band are extracted from electrical quantity data to construct the normalized energy response sequence of each feeder.

[0009] Using a matching template corresponding to a preset coding sequence, the normalized energy response sequence is subjected to matched filtering to obtain the matched filter output value of each feeder.

[0010] Among all feeders, the feeder with the largest matched filter output value and that meets the preset significant condition is identified as the faulty feeder.

[0011] According to a second aspect, the present invention provides a high-resistance grounding fault location system for a low-resistance grounding system, comprising: The data acquisition and acquisition module is used to collect electrical quantity data of the low-resistance grounding system and obtain a set of characteristic parameters from the collected electrical quantity data, including noise intensity indicators.

[0012] The frequency band determination module is used to identify the main noise frequency band from the bus zero-sequence voltage of electrical quantity data before starting active detection, and to determine the target high-frequency detection frequency band according to the predetermined frequency band selection rules and the spectrum after removing the main noise frequency band.

[0013] The disturbance injection module is used to control the neutral point-to-ground bypass branch to switch between at least two different frequency-selective impedance levels according to a preset coding sequence within a preset detection period when preset start-up conditions are met.

[0014] The response processing module is used to extract high-frequency components located in the target high-frequency detection band from electrical quantity data, in order to construct the normalized energy response sequence of each feeder.

[0015] The matched filtering module is used to perform matched filtering on the normalized energy response sequence using a matched template corresponding to the preset coding sequence, so as to obtain the matched filtering output value of each feeder.

[0016] The fault diagnosis and safety control module is used to identify the feeder with the largest matched filter output value and that meets the preset significant conditions as the faulty feeder.

[0017] According to a third aspect, the present invention provides a computer device comprising: At least one processor; and a memory communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which enables the at least one processor to perform a high-resistance grounding fault selection method for any low-resistance grounding system in the embodiments of the present invention.

[0018] According to another aspect of the present invention, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause a computer to execute a high-resistance grounding fault selection method for any low-resistance grounding system in the embodiments of the present invention.

[0019] The present invention provides a method for selecting high-resistance grounding faults in a low-resistance grounding system. This method is achieved through six core steps: data acquisition and feature acquisition, noise frequency band identification and target frequency band determination, coded disturbance signal generation, high-frequency component extraction and response sequence construction, matched filtering, and fault judgment and recovery. Specifically, the method acquires the bus zero-sequence voltage and zero-sequence current of each feeder and obtains a set of characteristic parameters including noise intensity indicators, establishing a quantitative basis for the current noise environment of the sensing system. Before initiating active detection, the main noise frequency band is identified based on power spectrum analysis combined with noise intensity indicators, and the target high-frequency detection frequency band is determined from the spectrum after removing this frequency band according to rules, achieving pre-emptive perception and active avoidance of strong interference noise. In response to meeting the activation conditions, within a preset detection period, the neutral point bypass branch containing frequency-selective impedance levels is controlled to switch levels according to the coded sequence to generate a coded disturbance signal. This design ensures that the disturbance energy is concentrated and directed to the selected target frequency band. Within a preset detection period, high-frequency components within the target frequency band are extracted from electrical quantities. A normalized energy response sequence is constructed based on the ratio of the energy changes of the high-frequency components of each feeder and the bus, thereby achieving a normalized characterization of the disturbance response and suppressing common-mode interference. Matched filtering is performed on the normalized energy response sequence using a matching template corresponding to the coding sequence to obtain the matched filter output value of each feeder. Correlation operations enhance the signal-to-noise ratio of the coded disturbance response characteristics. The feeder with the largest matched filter output value that meets the significance condition is identified as a faulty feeder, and the bypass branch is restored after the detection period ends, forming a complete closed-loop process of "sensing-avoidance-injection-extraction-decision-recovery".

[0020] In this technical solution, the present invention addresses the problem of unstable fault feature extraction caused by the fixed detection frequency band and lack of noise adaptation capability, as described in the background technology. It introduces a noise frequency band identification and target frequency band determination step based on a combination of real-time power spectrum analysis and noise indicators. Before each detection, the current dominant noise frequency band is dynamically analyzed and avoided, ensuring that the active detection signal is injected into a relatively "quiet" frequency band. This solves the fundamental defect of existing technologies, which are easily overwhelmed by strong time-varying noise due to fixed frequency band injection. To address the insufficient reliability of line selection caused by this defect, a detection mechanism combining "frequency-selective coding perturbation injection" and "normalized matched filtering extraction" is constructed. First, a frequency-selective impedance level is used to generate efficient coding perturbation within the target frequency band. Then, by constructing a normalized energy response sequence and performing matched filtering, weak fault response features highly correlated with the coding sequence are stably extracted from strong background noise, thus ensuring high signal-to-noise ratio acquisition of fault features at the signal level. Therefore, the technical solution of the present invention solves the technical problem that the existing technology has insufficient accuracy in fault feature extraction under strong noise interference environment, and improves the accuracy, anti-interference ability and adaptability to complex operating environment of high resistance grounding fault selection. Attached Figure Description

[0021] Figure 1 This is a flowchart of a high-resistance grounding fault selection method for a low-resistance grounding system according to an embodiment of the present invention; Figure 2 This diagram illustrates a complete application process of the high-resistance grounding fault location method according to another embodiment of the present invention in a specific power distribution network scenario. Figure 3 This is a schematic diagram of a typical low-resistance grounding distribution network system structure applying the high-resistance grounding fault location method of the present invention according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the evolution of active detection timing waveform and equivalent circuit parameters based on multi-level impedance encoding according to an embodiment of the present invention. Figure 5 This is a schematic diagram of the relevant output results obtained after performing matched filtering on the normalized energy response sequence of each feeder according to an embodiment of the present invention; Figure 6 This is a structural block diagram of a high-resistance grounding fault selection system of a small-resistance grounding system according to an embodiment of the present invention; Figure 7 This is a block diagram of a computer device for implementing embodiments of the present invention. Detailed Implementation

[0022] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of the present invention, including various details to aid understanding. These details should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope of the invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0023] During the development of this invention, the applicant, through extensive experiments and data analysis, discovered an intrinsic link between the spectral distribution characteristics of time-varying power electronic noise and the detectability of weak features of high-resistance grounding faults in "three-high" (high proportion of new energy, high proportion of power electronic equipment, and high proportion of power receiving) distribution networks: Although the system background noise exhibits complex characteristics of being multi-source, wide-bandwidth, and time-varying, its energy is not uniformly distributed in the spectrum, and there are always relatively "quiet" frequency band gaps; if strong noise frequency bands can be actively avoided, and the detection energy can be accurately injected and focused on these "quiet" frequency bands, the signal-to-noise ratio of the fault response signal can be significantly improved. Based on this relationship, the applicant innovatively proposed this technical solution, which utilizes real-time acquired bus zero-sequence voltage for online power spectrum analysis. By combining noise intensity indicators, it adaptively identifies and avoids the current main noise frequency band, and determines the optimal target high-frequency detection frequency band by combining predetermined frequency band selection rules, thereby realizing intelligent tracking and avoidance of the detection frequency band in the noise environment. Furthermore, by controlling the neutral point bypass branch, which includes frequency-selective impedance levels, to switch according to the coding sequence, it generates an enhanced coded disturbance signal in the target frequency band while limiting disturbances at the power frequency. Finally, by extracting high-frequency components in the target frequency band and constructing a normalized energy response sequence, and by using a matching template for matched filtering, it stably extracts fault features related to coded disturbances from a strong noise background, embodying the core concept of "first sensing and avoiding noise, and then implementing enhanced coded detection and matched filtering extraction in a clean frequency band".

[0024] Specifically, through comparative experiments, the invention team discovered that traditional line selection methods based on fixed-band injection or switching comparison suffer from technical defects such as "detection blind spots" and "susceptibility to submersion": the injected signal frequency is fixed in advance and cannot adapt to time-varying noise; when the injected frequency band falls into a strong noise region, the fault response is completely submerged, leading to line selection failure or misjudgment. However, the target frequency band determination method based on noise adaptive avoidance proposed in this invention can improve the "signal-to-noise ratio floor" of the detection signal, creating a prerequisite for weak feature detection; the frequency-selective coding perturbation injection method ensures that limited perturbation energy is efficiently concentrated in the selected "quiet" frequency band, enhancing the observability of fault features within that band; and the method combining normalized energy response sequence construction and matched filtering ensures that common-mode noise is suppressed from each feeder signal and fault responses highly correlated with known codes are extracted, achieving accurate identification and reliable judgment of faulty feeders.

[0025] Therefore, this invention provides a high-resistance grounding fault location method for a low-resistance grounding system, which can be applied to a relay protection and fault detection system (hereinafter referred to as the "system") for a low-resistance grounding distribution network. This system can run on the protection and control layer of a distribution network substation via embedded software or industrial control software. Specifically, this system can be deployed in various hardware environments, including but not limited to: dedicated fault location protection devices installed in substations, functional modules integrated into feeder protection or monitoring devices, and industrial control computers and servers deployed in the substation monitoring backend or centralized control center. This flexible deployment architecture enables the system to meet the requirements of rapid fault identification and isolation on-site, in real-time, and with high reliability, while also adapting to the needs of centralized analysis, strategy optimization, and advanced applications of multi-station information at the master station. This achieves synergy between "rapid on-site action" and "centralized intelligence at the master station," improving the efficiency and intelligence level of the entire distribution network fault handling system.

[0026] like Figure 1 As shown, the method may include: Step S110: Collect electrical quantity data of the low-resistance grounding system and obtain the characteristic parameter set from the collected electrical quantity data.

[0027] The characteristic parameter set includes a noise intensity index used to characterize the noise level of power electronic equipment. The characteristic parameter set refers to a set of quantified parameters reflecting the current operating state of the system. The noise intensity index is a parameter used to comprehensively quantify the noise intensity introduced by power electronic equipment in the zero-sequence channel. This index can be calculated using parameters such as the frequency band energy ratio, the prominence of spectral peaks, and the proportion of power electronic equipment.

[0028] Specifically, the system can synchronously acquire the instantaneous values ​​of the bus zero-sequence voltage and the zero-sequence current of each feeder through the data acquisition unit, and the processor can calculate the power spectral density and other operating statistics based on the sampled data within a preset time window, and then fuse them to generate a set of characteristic parameters.

[0029] For example, in a 10 kV distribution network, the system continuously collects electrical quantities at a sampling frequency of 20 kHz, extracts 200 ms of bus zero-sequence voltage data, calculates the power spectral density using the Welch method, and calculates the current noise intensity index by analyzing the energy concentration near the power electronic switching frequencies and harmonics in the 300 Hz to 5000 Hz frequency band.

[0030] In some embodiments, the system can perform DC-free preprocessing on the bus zero-sequence voltage baseline data collected before the active detection is started using the following formula (1): (1) Formula (1) is the formula for deDC calculation of the baseline voltage signal, which is used to eliminate the constant bias component in the signal and provide zero-mean baseline data for subsequent spectrum analysis. This represents the nth original discrete voltage sample value extracted from the zero-sequence voltage of the bus, where the subscript n is the sample point number, and the value range is... ; The total number of sample points in the baseline window, determined by the sampling frequency. Compared with baseline window length Multiplying them together yields the result, i.e. ; This indicates all data within the baseline window. One original voltage sample Calculate the arithmetic mean, which represents the DC component of the signal; This represents the nth baseline voltage sample value obtained after DC removal. For example, when the sampling frequency... Baseline window length hour, The system will calculate the average of these 2000 samples and subtract that value from each original sample to obtain a zero mean. .

[0031] Next, in this embodiment, the system can also calculate the discrete Fourier transform of each sub-window signal using formula (2): (2) Formula (2) is the calculation formula for the windowed Discrete Fourier Transform (DFT), which is used to convert the time-domain sub-window signal to the frequency domain for analysis. This represents the nth baseline voltage sample value within the rth sub-window, and the sub-window signal is derived from the DC-free sequence. Obtained in segments; `n` represents the value of the window function at index `n`, such as a Hanning or Hamming window, used to reduce spectral leakage; `L` is the sub-window length, i.e., the number of sample points contained in each sub-window; `m` is the frequency domain index, with a value ranging from... j is the imaginary unit; This is the complex spectrum value of the r-th sub-window signal at frequency domain index m.

[0032] Furthermore, in this embodiment, the system can also calculate the normalization factor of the window function using formula (3): (3) Formula (3) is used to calculate the energy normalization factor of the window function to ensure the unbiasedness of the power spectral density estimation. Represents the normalization factor, whose value is the window function. The sum of squares of all sample values ​​divided by the subwindow length L; This represents the square of the window function's value at the nth sample point. For example, for a Hanning window of length L=1024, the system will calculate the sum of the squares of all 1024 sample values, then divide by 1024 to obtain the normalization factor of the window function. .

[0033] Furthermore, in this embodiment, the system can also calculate the bus zero-sequence voltage at a certain frequency using formula (4). Power spectral density estimate at: (4) Formula (4) is the formula for calculating the power spectral density using the Welch method. Indicates frequency The estimated zero-sequence voltage power spectral density of the bus is given in volts squared per hertz (V / Hertz). ); fs is the actual physical frequency (in Hz) corresponding to the frequency domain index m; fs is the sampling frequency (in Hz); K is the total number of sub-windows. Indicates the frequency of the r-th sub-window. The formula is derived by applying the periodograms of all K sub-windows at the frequency... Average the values ​​and divide by the normalization factor. and sampling frequency The final power spectral density estimate is obtained. For example, if the baseline data with a length of 2000 points is divided into K=7 sub-windows with a length of L=1024 and an overlap of 50%, the system applies formula (2) to each sub-window to calculate the spectrum, and then applies formula (4) to the 7 sub-windows at each frequency point. v at the location By averaging, the final result is obtained from 0Hz to Power spectral density curves within the frequency range .

[0034] Therefore, by combining the above formulas (1) to (4), the system can complete the entire preprocessing and spectrum analysis process from the acquisition of the original bus zero-sequence voltage signal to its accurate power spectral density estimation. This process first eliminates the constant bias by removing DC, and then uses the Welch method of windowing, segmentation, and averaging to effectively suppress random fluctuations in the spectrum estimation, obtaining a smooth, reliable power spectral density curve that accurately reflects the frequency domain distribution characteristics of noise and signal energy within the baseline window. This curve is the core input data for subsequent identification of the main noise frequency band and adaptive selection of the target high-frequency detection frequency band.

[0035] In some embodiments, the system can construct the harmonic consistency evaluation function using the following formula (5): (5) Formula (5) is the calculation formula for the harmonic consistency evaluation function, which is used to evaluate and locate the noise frequency with concentrated energy that is most likely caused by the switching of power electronic equipment and its harmonics on the power spectral density curve. This represents the evaluation function value at the fundamental frequency f; H is the preset maximum harmonic order, which is usually taken as 3 to 5. is the weighting coefficient for the h-th harmonic, which takes a value greater than zero and usually decreases as the harmonic order h increases, in order to reflect the characteristic that the fundamental energy is usually the highest and the harmonic energy gradually weakens in the actual noise spectrum. The zero-sequence voltage power spectral density of the bus calculated by formula (4) at the frequency The value at that location. For example, in one embodiment, H is set to 3, and the weighting coefficient is... The system will calculate the power spectral density values ​​of each candidate fundamental frequency f and its two preceding harmonics (i.e., f, 2f, and 3f) within a preset search range (e.g., 300 Hz to 3000 Hz), and then perform a weighted summation according to the aforementioned weights to obtain the evaluation function value of the fundamental frequency. .

[0036] Next, in this embodiment, the system can also obtain the set of characteristic frequency centers of the dominant noise source in the current operating state through formula (6): (6) Formula (6) defines a set consisting of Q dominant characteristic frequencies. The system evaluates the evaluation function values ​​across the entire search frequency band calculated by formula (5). A peak search is performed, selecting the top Q peaks (i.e., local maxima) with the highest function values. The fundamental frequency f corresponding to these peaks is determined as the characteristic frequency center of the noise source, forming a set. Q is the preset maximum number of dominant noise sources to be identified, typically ranging from 1 to 3.

[0037] Furthermore, in this embodiment, the system can also construct candidate noise bands based on the characteristic frequency center using formula (7): (7) Formula (7) is used to define the candidate noise band around the characteristic frequency and its harmonics. Represents the area around the q-th characteristic frequency center The continuous frequency range of the h-th harmonic; h is the harmonic order, with a value range of... ; This is the half-width (FWHM) of the frequency band corresponding to the h-th harmonic (in Hz). This value can be preset as a constant, or it can be set to increase appropriately as the harmonic order h increases to cover the actual broadening of the harmonic spectrum. For example, let Δ1 = 100Hz, Δ2 = 150Hz, and Δ3 = 200Hz. Assume a characteristic frequency center is identified. The system will then construct three candidate noise bands: the fundamental frequency band. Second harmonic frequency band Third harmonic frequency band .

[0038] Furthermore, in this embodiment, the system can also calculate the energy of the candidate noise frequency band and the total high-frequency energy using formulas (8) and (9), providing a basis for constructing noise intensity indices: (8) (9) Formula (8) calculates the noise frequency bands from all candidate noise bands. Total energy covered (Unit: volts squared) Its calculation method is based on sets. Each characteristic frequency center in and the frequency band corresponding to its first H harmonics For the power spectral density function, respectively Integrate within this frequency band, and then sum all the integration results. Formula (9) calculates the total high-frequency analysis interval within the preset range. Total energy within (unit: This range typically covers the main frequency band where power electronic noise is distributed, for example, it is set to 300Hz to 5000Hz.

[0039] Furthermore, in this embodiment, the system can also calculate two key intermediate indicators using formulas (10) and (11): (10) (11) Formula (10) is used to calculate the noise frequency band energy ratio index. This index is a dimensionless value between 0 and 1, reflecting the total energy of the candidate noise band. Total energy at high frequency The proportion it accounts for. It is a very small positive number introduced to prevent the denominator from being zero, for example... Formula (11) calculates the peak prominence index. This metric is used to quantify the prominence of the strongest noise peak in the power spectrum. Represents all candidate noise bands The union of; In the union Within the range, power spectral density function The maximum value; Indicates the total high-frequency range Within the range, the median of the power spectral density values ​​corresponding to all frequency points.

[0040] Finally, in this embodiment, the system calculates the final noise intensity index using formula (12): (12) Formula (12) defines the noise intensity index. The comprehensive calculation method. , , These are non-negative weighting coefficients, and at least one of them is not zero. Their values ​​can be determined based on engineering experience or simulation calibration, for example, by setting... . The normalized parameter, which characterizes the penetration rate of power electronic equipment in the system, has a value ranging from 0 to 1 and can be obtained from the Energy Management System (EMS) or monitoring system. The larger the value, the stronger the background noise with concentrated energy introduced by power electronic devices in the current system's zero-sequence channel. This provides a key quantitative basis for subsequent safety verification when starting active detection, fine selection of target high-frequency detection bands, and adaptive adjustment of coding sequence parameters (such as code length and symbol width).

[0041] Therefore, by combining the above formulas (5) to (12), the system can complete the entire analysis chain from power spectral density analysis to noise characteristic frequency identification, and then to noise intensity index quantification. This chain first locates the noise source through evaluation functions and peak search, then quantifies the noise intensity by calculating the frequency band energy and peak prominence, and finally generates a comprehensive noise intensity index. This index provides direct and objective data support for the adaptive selection of target high-frequency detection bands and the setting of safe start conditions in the core process, and is a key technical link in realizing the core inventive concept of "noise perception and avoidance" in this invention.

[0042] Step S120: Before starting active detection, the main noise frequency band is identified from the bus zero-sequence voltage of the electrical quantity data, and the target high-frequency detection frequency band is determined according to the predetermined frequency band selection rules and the spectrum after removing the main noise frequency band.

[0043] The main noise frequency band refers to the frequency band with a power spectral density significantly higher than the local background energy, which usually corresponds to the switching frequency of power electronic equipment and its harmonics. The target high-frequency detection frequency band refers to the continuous frequency band selected from the remaining spectrum after removing the main noise frequency band and used for subsequent active injection and response detection. The predetermined frequency band selection rule is configured to prioritize the continuous frequency band with the lowest average power spectral density. If there are multiple candidate frequency bands with similar average power spectral densities, the continuous frequency band with the largest interval from the center point of the noise source characteristic frequency is selected.

[0044] Specifically, the system can first estimate the set of characteristic frequencies of the noise source based on the power spectral density, and construct candidate noise bands around each characteristic frequency; by comparing the signal energy in each candidate noise band with a preset reference energy threshold, the main noise bands are identified; then, multiple consecutive candidate detection bands are generated in the remaining part of the spectrum with a fixed bandwidth, and the average power spectral density of each candidate detection band and the minimum frequency interval from its center frequency to the center of each noise band are calculated; finally, the frequency band selection rules are applied to determine the final target high-frequency detection band.

[0045] For example, system analysis revealed a significant spectral peak near 2500 Hz, and the 2400 Hz to 2600 Hz range was identified as the main noise band and eliminated. In the remaining spectrum, several candidate bands were generated with a bandwidth of 600 Hz. After calculation and comparison, the band with the lowest average power spectral density and the furthest distance from the identified noise center was determined to be the target high-frequency detection band.

[0046] Step S130: In response to the satisfaction of the preset start-up conditions, within the preset detection period, the neutral point to ground bypass branch is controlled to switch between at least two different frequency selective impedance levels according to the preset coding sequence.

[0047] The bypass branch includes at least one frequency-selective impedance level, which exhibits high impedance at the power frequency and low impedance within the target high-frequency detection band. A frequency-selective impedance level is a passive network level whose impedance value varies with frequency. In this invention, it specifically refers to a level designed to exhibit high impedance at the power frequency (50Hz) and low impedance within the selected target high-frequency detection band, thereby limiting power frequency disturbances and enhancing high-frequency detection. The encoding sequence refers to a predefined set of discrete values ​​used to control the impedance level switching according to a specific pattern to form a detection signal with good autocorrelation characteristics.

[0048] Specifically, before initiating detection, the system can verify that the current bus zero-sequence voltage amplitude is within a safe threshold range and estimate that the expected voltage fluctuation caused by the shift switch is lower than the safe limit. Once the conditions are met, within a preset detection period, based on the shift number corresponding to each code element in the coded sequence, a switching command is sent to the control module of the bypass branch at the start of the code element, driving the power electronic switch to operate and realizing the shift switch.

[0049] For example, the preset startup condition is that the bus zero-sequence voltage is between 0.5kV and 3kV, and the expected fluctuation is less than 0.1kV. The target high-frequency detection band is 900Hz to 1500Hz. The system uses a series resistor-inductor-capacitor (RLC) branch as the frequency-selective impedance level, and its resonant frequency is designed to be the center frequency of the target band, 1200Hz, where the damping resistor is 4 ohms. The inductance is approximately 1.06 millihenries (mH), and the capacitance is approximately 16.6 microfarads (µF). The impedance of this branch at 1200Hz is approximately 4. At 50Hz, the impedance is significantly higher than 100Ω. The encoding sequence uses a 13-bit Barker code to control the switching between the RLC branch and the open circuit position according to the code pattern.

[0050] Step S140: Extract the high-frequency components located in the target high-frequency detection band from the electrical quantity data to construct the normalized energy response sequence of each feeder.

[0051] Here, high-frequency components refer to the signal components of the original electrical signal that are retained only within the target high-frequency detection band after bandpass filtering. Energy change refers to the change in the signal energy of the high-frequency components relative to a certain reference value (such as the energy of the previous symbol or the baseline average energy before detection begins) within the active detection symbol interval. The normalized energy response sequence is a set of ordered values ​​composed of the energy change of each feeder within each symbol interval divided by the corresponding energy change of the bus, used to characterize the relative response strength of each feeder to injected disturbances.

[0052] Specifically, the system uses a digital bandpass filter to filter the bus zero-sequence voltage and the zero-sequence current of each feeder to obtain their respective high-frequency components. Then, the entire preset detection period is divided into multiple consecutive symbol intervals according to the symbol width of the coded sequence. For each symbol interval, the symbol energy of the bus high-frequency voltage component and the high-frequency current component of each feeder is calculated. Next, using the average energy of several symbols prior to the start of detection as a benchmark, the energy change corresponding to each symbol interval is calculated. Finally, for each feeder, its energy change in each symbol interval is divided by the energy change of the bus in the corresponding interval to obtain a ratio. Arranging these ratios in symbol order constitutes the normalized energy response sequence of that feeder.

[0053] For example, the target high-frequency detection band is 900Hz to 1500Hz, and the system uses a finite-length impulse response (FIR) bandpass filter for filtering. The symbol width is 10ms. The sum of squares of the signal within each 10ms interval is calculated as the symbol energy. Using the average energy of the first 5 symbols (50ms) as a benchmark, the energy change of each symbol relative to this benchmark is calculated. Assume that the energy change of a certain feeder in the kth symbol is... The energy change of the busbar is The normalized energy response value of the feeder at the k-th symbol is... ,in To prevent small positive numbers from being divided by zero, for example .

[0054] Step S150: Using the matching template corresponding to the preset coding sequence, perform matched filtering on the normalized energy response sequence to obtain the matched filter output value of each feeder.

[0055] The matching template refers to a reference sequence generated based on the encoded sequence. It is typically obtained by removing the mean from the original encoded sequence and is used to perform cross-correlation calculations with the normalized energy response sequence to detect whether the sequence contains features related to the encoding. The matched filter output value is the maximum correlation value obtained within a preset time delay search range after cross-correlation calculation between the normalized energy response sequence and the matching template sequence. This value reflects the degree of matching between the feeder response and the injected encoded waveform.

[0056] Specifically, the system obtains the matched filter output value in the following way: First, the preset encoded sequence is mean-reduced to generate a matching template sequence. Then, for each feeder's normalized energy response sequence, the cross-correlation value between this sequence and the matching template sequence within a preset time delay offset range is calculated. Finally, among all the calculated cross-correlation values, the value with the largest absolute value is selected as the matched filter output value for that feeder.

[0057] For example, the encoded sequence is a 13-bit Barker code. After removing the mean, a matching template is generated. For a certain feeder's normalized energy response sequence Calculate its time delay offset for Cross-correlation value at each sampling point In all Find the maximum value in, for example Then the output value of the matched filter for this feeder is 5.2.

[0058] Step S160: Among all feeders, the feeder with the largest matched filter output value and that meets the preset significant condition is identified as the faulty feeder.

[0059] Among them, the preset significant conditions refer to a set of judgment conditions set to ensure the reliability of the decision. These typically include: the maximum matched filter output value must be significantly greater than the average output value of other non-faulty feeders, and it must be significantly greater than the baseline noise level under historical normal operating conditions. The system's default grounding state usually refers to the operating state where only the main grounding resistor is engaged.

[0060] Specifically, the system can find the maximum value by comparing the matched filter output values ​​of all feeders. and its corresponding feeder Subsequently, a judgment was made. Do both conditions need to be met simultaneously? (1) The ratio of the average output value of all other feeders to the average value of the other feeders is greater than a first preset threshold. ; (2) Root mean square value of historical baseline noise The ratio is greater than the second preset threshold. If both conditions are met, then the feeder is determined to be faulty. If the fault is detected, the feeder is considered faulty; otherwise, no significant fault is detected. Regardless of the determination, at the end of the preset detection cycle, the system will send a control command to switch the neutral point bypass branch to the open state or high impedance state, restoring the system's default grounding mode.

[0061] For example, suppose the matched filter output values ​​of the four feeders are respectively Then the maximum output value This corresponds to feeder 1. The average value of the remaining feeders is... .ratio If a preset threshold is used... Then condition one is satisfied. Historical baseline noise root mean square value ,ratio If a preset threshold is used... If this condition is met, then condition two is satisfied. Therefore, the system determines that feeder 1 is a faulty feeder and subsequently controls the bypass branch switch to open.

[0062] In another embodiment, such as Figure 2 This demonstration presents a complete application process example of the high-resistance grounding fault location method provided by this invention in a specific distribution network scenario. This example is based on a 10kV distribution network with 5 outgoing lines and low-resistance grounding, where the system experiences a transition resistance of 1500Ω. This method is activated when a single-phase ground fault occurs. First, step S1 performs data acquisition and parameter acquisition: After the fault occurs, the protection starting element is triggered, and the device synchronously acquires the bus zero-sequence voltage and the zero-sequence current of all five feeders at a sampling frequency of 20kHz. Based on the data from the most recent 200ms, the system calculates and acquires a set of characteristic parameters, including a noise intensity index, which reflects the background noise level caused by the current photovoltaic inverter cluster connection. Next, step S2 performs noise analysis and frequency band selection: The system selects the bus zero-sequence voltage from 100ms before the fault as baseline data and calculates its power spectral density using the Welch method. Combined with the noise intensity index, analysis reveals significant noise spectral peaks near 2250Hz and 4500Hz, and then... and The frequency bands identified as primary noise bands are eliminated. In the remaining spectrum, based on predetermined frequency band selection rules, the bands with the lowest average power spectral density and furthest from the noise center are ultimately selected. The frequency band was determined as the target high-frequency detection band for this detection. Subsequently, step S3 performs security verification and coded disturbance injection: the system verifies that the current bus zero-sequence voltage is 1.8kV, which is within the preset range. Within the threshold range, and with the estimated disturbance voltage fluctuation at 0.05kV, below the safety limit of 0.1kV, the preset startup conditions are met. The system enters a preset detection period of 130ms and, according to a preset 13-bit Barker code encoding sequence (symbol width 10ms), controls the neutral point bypass branch to switch between two impedance levels: "open circuit" and "RLC branch resonating at 1500Hz," thereby injecting a coded high-frequency disturbance current signal into the system. Then, step S4 performs response signal extraction and sequence construction: within the preset detection period, the system uses a passband of... A finite-length unit impulse response (FIR) digital filter is used to extract high-frequency components from the bus voltage and the current of each feeder. The 130ms probe data is divided into 13 symbol intervals with a symbol width of 10ms. The energy of each signal component within each interval is calculated, and the energy change of each symbol is calculated based on the average energy of the first five probe intervals. Finally, by dividing the energy change of each feeder by the corresponding change of the bus, a normalized energy response sequence containing 13 elements is constructed for each feeder. Then, step S5 performs matched filtering and feature enhancement: the system performs mean-reduction processing on the 13-bit Barker code to generate a matched template sequence. For the normalized energy response sequence of each feeder, the system calculates its time delay offset relative to the template sequence. The cross-correlation value within (unit: sampling interval) is calculated, and the absolute maximum value of the correlation values ​​under all delays is taken as the matched filter output value of the feeder. In this example, the output value of the faulty feeder (assumed to be feeder 3) is significantly higher than that of other non-faulty feeders. Finally, step S6 performs fault determination and system recovery: the system compares the matched filter output values ​​of all feeders and finds that the output value of feeder 3 is not only the largest, but its value is also 8.5 times the average output value of other feeders, which is much greater than the preset threshold of 3.0; at the same time, this value is 7.2 times the historical baseline noise level, which is greater than the preset threshold of 5.0, satisfying all preset significant conditions. Therefore, the system determines that feeder 3 is a faulty feeder and issues an alarm and trip command. After the preset detection period ends, the system controls the neutral point bypass branch to return to the open circuit state, and the system returns to the default operating mode of only being grounded by the main small resistor.

[0063] In another embodiment, such as Figure 3 This diagram illustrates a typical low-resistance grounding distribution network system structure using the high-resistance grounding fault location method of this invention. The diagram clearly depicts the complete electrical connections from the main power grid to the distribution feeders, as well as the locations of the key equipment and fault points involved in this invention. Figure 3 In the middle section, the main grid power supply on the left is connected to the distribution network via a 110kV / 21kV step-down transformer. The neutral point of the transformer is grounded through a grounding transformer, and the secondary side of this grounding transformer is connected to a neutral point grounding resistor. With a switchable compensation branch, which includes a resistor R and a reactor These components together constitute the neutral point grounding device of the system. Specifically, branch M, as indicated in the figure, is the controllable neutral point-to-ground bypass branch described in this invention. This branch includes at least one frequency-selective impedance level, whose impedance characteristics can be switched according to control commands. The right side of the system is the power distribution feeder section, containing three outgoing lines: busbar... , and Each busbar is connected to a circuit breaker, load, and distributed power sources or power electronic equipment (not shown in detail in the diagram). The diagram clearly marks six locations where single-phase ground faults may occur, namely... to .in, , and They are located on the lines of three feeders respectively. and Located in different sections of the busbar, Located near the grounding transformer, these fault points represent different scenarios where high-resistance grounding faults may occur. The fault location system of this invention (illustrated as a control and acquisition unit in the figure, not shown separately) uses a voltage transformer (PT) and a current transformer (CT) to collect the bus zero-sequence voltage and the voltage of each feeder in real time. The zero-sequence current. When the system experiences the zero-sequence current shown in the figure. When a high-impedance ground fault occurs as shown in point M, the system will execute the aforementioned complete process: First, analyze the background noise spectrum and adaptively select the target high-frequency detection band; then, after meeting safety conditions, control the switching devices in branch M to switch their input impedance values ​​according to a preset coding sequence (e.g., switching between the power frequency high impedance level and the low impedance level resonating in the target frequency band), thereby injecting a coded disturbance signal into the neutral-to-ground loop; finally, by analyzing the normalized energy response of the target frequency band component in the zero-sequence current of each feeder and performing matched filtering, accurately calculate the matched filter output value of each feeder. In this example, since the fault occurs... feeder Point, therefore The output value of the matched filter corresponding to the feeder will be significantly higher than and This allows the system to reliably identify and pinpoint faulty feeders. This schematic diagram visually illustrates the physical system structure upon which the method of this invention is based and its applicability in complex power distribution networks.

[0064] In another embodiment, to better adapt to grid frequency fluctuations or achieve wideband impedance disturbances, the system can employ a multi-level impedance coding strategy. For example... Figure 4 The diagram illustrates the evolution of the active detection timing waveform and equivalent circuit parameters based on multi-level impedance encoding. In this embodiment, the system utilizes the fast switching characteristics of switching devices (such as IGBTs or MOSFETs) in the neutral point bypass branch to switch between multiple different impedance levels according to preset encoding rules, thereby injecting a disturbance signal with specific frequency characteristics into the power grid.

[0065] Specifically, such as Figure 4 The upper part shows the encoded sequence. The discrete impedance levels to which the bypass branch should be connected during the k-th symbol period are defined. For example, the coded sequence can be set as { Each number represents a preset equivalent grounding impedance level. Next, the waveform diagram of the bypass selection switch K shows the actual timing of the physical switch's operation. It is directly controlled by the encoding sequence, and a transition occurs at each symbol boundary, switching the bypass branch to the corresponding impedance branch (e.g., contact 1 corresponds to level 1, contact 2 corresponds to level 4, etc.).

[0066] Furthermore, in Figure 4 The lower middle section shows the equivalent circuit parameters of the system at different frequency bands as the switch is activated. Near the power frequency ( The equivalent parameter of the system is expressed as equivalent resistance. , Equivalent conductance , And so on. And within the high-frequency band ( Due to the presence of inductive or capacitive components in the system (such as neutral point reactors, main transformer leakage reactance, or line-to-ground capacitance), these impedance levels exhibit an equivalent impedance modulus that is drastically different from that at power frequency. or equivalent admittance mode .

[0067] Therefore, as Figure 4 The summary at the bottom represents the total equivalent grounding admittance from the neutral point to the ground during the entire active detection process. It is a parameter that changes dynamically with time (as well as with the encoded sequence k). It is determined by the steady-state admittance of the main grounding resistance. This time-varying ground admittance is superimposed on the dynamic equivalent admittance resulting from the encoding switch. This time-varying ground admittance disrupts the original steady-state equilibrium of the system, inducing harmonic voltage and current components related to the encoding pattern in each feeder. The system then captures these electrical responses that change with the encoding sequence to accurately identify faulty feeders. This multi-level encoding method not only enriches the spectral characteristics of the disturbance signal but also effectively overcomes the problem of insufficient identification capability of single fixed impedance disturbances under certain complex operating conditions, further improving the sensitivity and reliability of fault detection.

[0068] In another embodiment, such as Figure 5 This diagram illustrates the relevant output results obtained after performing matched filtering on the normalized energy response sequences of each feeder in the method of this invention. The diagram visually presents the significant differences in response characteristics between faulty and healthy feeders, providing a clear visual basis for the final fault determination. Figure 5 The horizontal axis represents the time delay offset. The unit is the number of symbols, representing the possible relative time offset between the matching template sequence and the normalized energy response sequence. The vertical axis represents the correlation output value. This is a dimensionless numerical value that quantifies the similarity (i.e., cross-correlation value) between the normalized energy response sequence and the matching template sequence of the j-th feeder at different time delays. The figure shows multiple curves corresponding to the correlation outputs of different feeder numbers j. Delay The changing situation. As shown in the figure, the relevant output curves of all feeders. On the time delay axis All of them showed some fluctuation. Among them, one was marked as The curve with the highest correlation output value is significantly higher than other curves across the entire latency search range. This curve, at a specific latency... Reaching the global maximum value This curve corresponds to the feeder where a high-resistance ground fault occurred. In contrast, other markings are... The curves (representing healthy feeders), their related output values Overall, the level remained low. The maximum value among these healthy feeder curves was marked as... It represents the highest level of the matched filter output value among all non-faulty feeders, but is comparable to the peak value of the faulty feeder. There is a significant difference in magnitude. To illustrate with a concrete example, suppose the system has a total of 5 feeders ( ), of which the third feeder (i.e. A fault occurred. After matched filtering calculation, five curves were obtained. From Figure 5 It can be observed that the curve In time delay A sharp peak appears at [location], and its related output value [is shown]. Up to approximately 2.5. The other four intact feeders... The magnitude of the relevant output curve is very small, reaching its maximum value over the entire time delay range. It does not exceed 0.8. The characteristic that the peak output of this faulty feeder is significantly higher than that of a healthy feeder intuitively and powerfully proves that the normalized energy response sequence and matched filtering method constructed in this invention can effectively amplify fault characteristics and suppress noise interference, thereby forming a data distribution at the output end that is extremely beneficial for discrimination. Figure 5 The results shown directly serve the final decision-making step: the system only needs to identify the value with the largest matched filter output (i.e., By using the feeder, the location of the fault can be accurately pinpointed.

[0069] Therefore, according to the above implementation method, the system achieves its function through six core steps: data acquisition and feature acquisition, noise frequency band identification and target frequency band determination, coded disturbance signal generation, high-frequency component extraction and response sequence construction, matched filtering, and fault judgment and recovery. Specifically, acquiring the bus zero-sequence voltage and zero-sequence current of each feeder and obtaining a set of characteristic parameters including noise intensity indicators establishes a quantitative basis for the current noise environment of the sensing system. Before initiating active detection, the main noise frequency band is identified based on power spectrum analysis combined with noise intensity indicators, and the target high-frequency detection frequency band is determined from the spectrum after removing this frequency band according to rules, achieving pre-emptive perception and active avoidance of strong interference noise. In response to meeting the activation conditions, within a preset detection period, the neutral point bypass branch containing frequency-selective impedance levels is controlled to switch levels according to the coded sequence to generate coded disturbance signals. This design ensures that the disturbance energy is concentrated and directed to the selected target frequency band. Within a preset detection period, high-frequency components within the target frequency band are extracted from electrical quantities. A normalized energy response sequence is constructed based on the ratio of the energy changes of the high-frequency components of each feeder and the bus, thereby achieving a normalized characterization of the disturbance response and suppressing common-mode interference. Matched filtering is performed on the normalized energy response sequence using a matching template corresponding to the coding sequence to obtain the matched filter output value of each feeder. Correlation operations enhance the signal-to-noise ratio of the coded disturbance response characteristics. The feeder with the largest matched filter output value that meets the significance condition is identified as a faulty feeder, and the bypass branch is restored after the detection period ends, forming a complete closed-loop process of "sensing-avoidance-injection-extraction-decision-recovery".

[0070] Specifically, in the technical solution of this embodiment, to address the problem of unstable fault feature extraction due to fixed detection frequency band and lack of noise adaptation capability mentioned in the background technology, a noise frequency band identification and target frequency band determination step based on real-time power spectrum analysis and noise index is introduced. Before each detection, the current dominant noise frequency band is dynamically analyzed and avoided, thereby ensuring that the active detection signal is injected into a relatively "quiet" frequency band. This solves the fundamental defect of the prior art, which is easily submerged by strong time-varying noise due to fixed frequency band injection. To address the consequence of insufficient line selection reliability caused by this defect, a detection mechanism combining "frequency-selective coding perturbation injection" and "normalized matched filtering extraction" is constructed. First, a frequency-selective impedance level is used to generate efficient coding perturbation in the target frequency band. Then, by constructing a normalized energy response sequence and performing matched filtering, weak fault response features highly correlated with the coding sequence are stably extracted from strong background noise, thereby ensuring high signal-to-noise ratio acquisition of fault features at the signal level. Therefore, the technical solution of this embodiment solves the technical problem that the existing technology has insufficient accuracy in fault feature extraction under strong noise interference environment, and improves the accuracy, anti-interference ability and adaptability to complex operating environment of high resistance grounding fault selection.

[0071] In some embodiments, the step of identifying the main noise frequency band from the bus zero-sequence voltage of electrical quantity data includes: The power spectral density of the bus zero-sequence voltage is calculated based on the bus zero-sequence voltage data in the electrical quantity data.

[0072] Among them, power spectral density is a statistic used to describe how signal power is distributed in the frequency domain, and it reflects the energy intensity of the signal at different frequency components.

[0073] Specifically, the system can estimate the power spectral density using the Welch periodogram method on a discrete sampling sequence of bus zero-sequence voltage with a preset time length. This method segments the data, applies a window, calculates the Discrete Fourier Transform (DFT) of each segment, and then averages the periodograms of each segment to obtain a smooth power spectral density estimate. For example, the system acquires 200ms of bus zero-sequence voltage data at a sampling frequency of 20kHz, divides it into eight 50ms segments with 50% overlap, applies a Hanning window to each segment, calculates the DFT of each segment and squares the modulus, and finally averages the results of the eight segments to obtain a power spectral density estimation curve in the frequency range from 0Hz to 10kHz. .

[0074] Based on the power spectral density, the set of characteristic frequencies of the dominant noise sources in the current system is determined.

[0075] The characteristic frequency set refers to the set of frequency values ​​identified from the power spectral density curve that correspond to the main power electronic noise sources (such as converter switching frequencies) and their significant harmonic components in the current system.

[0076] Specifically, the system can perform peak search on the power spectral density curve within a preset high-frequency band (e.g., 300Hz to 5000Hz). This is achieved by constructing a harmonic consistency evaluation function. (h is the harmonic order,) (as weights), and find the ones that make The frequency points that achieve local maxima constitute a preliminary candidate set of characteristic frequencies. Further thresholding is used to retain frequencies whose peak prominence (the ratio of the power spectral density at that frequency point to the local background median) is greater than a preset value, ultimately forming the final set of characteristic frequencies. For example, in the range of 300Hz to 5000Hz, the system calculates... (consider It was found that it has significant peaks at 2500Hz and 5000Hz, and the peak prominence at these two frequencies is greater than 5. Therefore, it is... It is determined to be the set of characteristic frequencies.

[0077] Based on the set of characteristic frequencies, construct one or more candidate noise bands.

[0078] Among them, the candidate noise band refers to a continuous band formed by extending a certain width to both sides of each frequency point in the characteristic frequency set as the center, and is used to evaluate the concentration of noise energy near that frequency point.

[0079] Specifically, the system can target each frequency in the feature frequency set. Define a symmetrical frequency band. As a candidate noise frequency band, among which, The preset half-width of the frequency band can be set according to the typical spectral diffusion characteristics of the noise source or empirical values. Different values ​​can also be set for different orders of harmonics. For example, for characteristic frequencies of 2500Hz and 5000Hz, set... The system then constructs two candidate noise frequency bands: and .

[0080] Traverse one or more candidate noise frequency bands, calculate the signal energy in each candidate noise frequency band, and compare the signal energy with a preset reference energy threshold.

[0081] In the frequency domain, the signal energy can be obtained by integrating the power spectral density over a specified frequency band. The reference energy threshold is a benchmark used to determine whether the noise in the frequency band is "significant". It can be the median of the signal energy in several reference frequency bands of the same width near the candidate noise band, or it can be the total energy in the entire analysis frequency band (such as the total high-frequency range).

[0082] Specifically, the system analyzes each candidate noise frequency band. Calculate its band energy Meanwhile, in the frequency band Several non-overlapping reference frequency bands of equal width are selected on both sides, and the median of the energy within these reference frequency bands is calculated as the local background energy. .Will and The ratio, and With high frequency total energy The ratios of these values ​​are respectively compared with preset thresholds. and Comparisons are made. For example, for candidate noise bands. Calculate its band energy .exist and The energy was calculated within two reference frequency bands, and the median was taken as the average. Assuming Preset threshold If so, then the comparison condition is satisfied. Simultaneously calculate... The proportion of the total high-frequency energy (300~5000Hz) is 0.15, with a preset threshold. The comparison condition is also satisfied.

[0083] Based on the comparison results, from one or more candidate noise frequency bands, the candidate noise frequency band that meets the preset energy comparison conditions is determined as the main noise frequency band.

[0084] Among them, the preset energy comparison condition is a rule used to determine whether a candidate noise frequency band constitutes "major" interference, which is usually a logical combination of the above ratio comparison conditions (such as "AND" relationship).

[0085] Specifically, the system examines each candidate noise band. If it carries internal energy Simultaneously satisfy: (1) , and (2) (in To prevent division by zero (small positive numbers), the candidate noise band is... The frequency bands identified as primary noise bands are the set of all frequency bands determined as primary noise bands, representing the interference bands that need to be avoided. For example, candidate noise bands. Since all the above conditions are met, it is therefore identified as the main noise frequency band. Although it meets condition (1), its energy proportion is low, and it does not meet condition (2) (assuming) ),therefore This was not identified as a major noise frequency band. Ultimately, the system identified the following as the major noise frequency band: For safety reasons, guard bands are usually added on both sides of this frequency band, for example, extending it by 50 Hz on each side, forming the final frequency band to be avoided. .

[0086] In some embodiments, the system can calculate the candidate noise band using the following formula (13). Internal energy : (13) Formula (13) is the frequency band energy integral formula, which is used to quantify the signal energy intensity in each candidate noise frequency band. Represents the center of the q-th characteristic frequency Candidate noise band corresponding to the h-th harmonic Total energy within (unit: ); Integration interval The frequency range of the candidate noise band (unit: Hz); The power spectral density function of the zero-sequence voltage of the bus (unit: The result is calculated using the aforementioned formula (4). For example, for a specific candidate noise band... The system calculates the area under the power spectral density curve in this interval using numerical integration methods (such as the trapezoidal rule). If the calculation result is... This indicates that the signal energy contained in that frequency band is .

[0087] Next, in this embodiment, the system can also identify the candidate noise frequency band as the main noise frequency band by using the discrimination conditions defined by formulas (14) and (15): (14) or (15) Formulas (14) and (15) together constitute the dual-threshold discrimination logic for the main noise frequency band. Formula (14) is the local signal-to-noise ratio discrimination formula, which is used to determine whether the energy of the candidate frequency band is significantly higher than its local background noise level. Indicates candidate noise band Local background energy (unit: The calculation method is as follows: Several reference frequency bands with the same width are symmetrically selected on both sides of the candidate frequency band, and the in-band energy of these reference frequency bands is calculated respectively. Then, the median of these energies is taken as the median value. ; It is a very small positive number that prevents the denominator from being zero, for example... ; It is a preset local signal-to-noise ratio threshold, which is a constant greater than 1, for example, set to... Formula (15) is the global energy proportion discriminant, used to determine whether the candidate frequency band energy occupies a significant proportion in the entire high-frequency energy. Indicates the preset high-frequency analysis total interval Total energy within (unit: ), which is calculated using formula (9); It is a preset global energy percentage threshold, which is a decimal between 0 and 1. For example, setting it to... As long as the candidate noise band If either formula (14) or formula (15) is satisfied, it is determined to be the main noise frequency band.

[0088] Furthermore, in this embodiment, the system can also define the final set of frequency bands to be avoided using formula (16). : (16) Formula (16) is the union operation formula of the set of frequency bands to be avoided. Its purpose is to add an extra guard band on both sides of each identified main noise frequency band to form the final frequency band range that needs to be completely removed from the spectrum, so as to ensure that the target high-frequency detection frequency band selected later is far away from the sideband leakage area of ​​noise energy. Represents the set of all frequency bands that need to be avoided; symbol This indicates that the union operation is performed on the indices q (center index of characteristic frequency) and h (harmonic order) corresponding to all identified major noise frequency bands; The h-th harmonic frequency (unit: Hz) is the center of the q-th characteristic frequency. The half-width of the frequency band corresponding to the h-th harmonic is defined in formula (7) (unit: Hz). The preset protection bandwidth (unit: Hz), for example, set The meaning of this formula is: for each identified main noise frequency band (whose center is...) The original half width is Expand it to both the left and right sides. The width creates a new, wider avoidance zone. Then, take the union of all such intervals to obtain the final result. For example, suppose a major noise frequency band is identified with a center of 2500Hz, the original half-width Δ1 = 100Hz, and the guard bandwidth... The corresponding avoidance zone is .

[0089] Therefore, through the combination of the above formulas (13) to (16), the system can complete the entire logical chain from calculating the energy of candidate noise frequency bands, to accurately identifying the main noise frequency bands based on dual thresholds (local signal-to-noise ratio and global energy ratio), and then to constructing the final avoidance frequency band set including the guard band. This process first quantifies the energy of each potential noise frequency band, then filters out the frequency bands with truly concentrated energy that may cause strong interference to the detection through strict threshold conditions, and finally ensures a safe distance by extending the guard band, thereby facilitating subsequent steps to remove the "remaining part of the spectrum" (i.e., from the total analysis frequency band). The robust and reliable selection of the optimal target high-frequency detection band in the latter part provides accurate input and assurance.

[0090] Therefore, according to the above implementation method, the system can dynamically and accurately identify the main interference noise frequency bands generated by power electronic devices in the current power distribution network operating environment before each active detection is initiated. This provides accurate input for the subsequent adaptive selection of "quiet" target detection frequency bands, thereby avoiding the possible submersion of active detection signals by strong background noise from the source.

[0091] In some embodiments, the step of determining the target high-frequency detection band based on a predetermined frequency band selection rule and the spectrum after removing the main noise frequency band includes: Multiple consecutive candidate frequency bands with preset bandwidths are generated from the spectrum after removing the main noise frequency bands.

[0092] The remaining spectrum refers to the usable spectrum range remaining after removing all identified primary noise bands and their adjacent guard bands from the preset high-frequency analysis range. The preset bandwidth refers to the fixed frequency width pre-set for each generated candidate frequency band. Candidate frequency bands refer to a series of continuous frequency ranges generated from the remaining spectrum range according to the preset bandwidth and sliding step size for participation in the final target frequency band selection.

[0093] Specifically, the system first extends the identified main noise frequency bands to both sides by a guard band width (e.g., 50Hz), merging them to obtain a set of frequency bands that need to be completely avoided. Then, these avoided frequency bands are removed from a preset total high-frequency analysis range (e.g., 300Hz to 5000Hz), yielding the remaining spectrum. Finally, using a fixed preset bandwidth (e.g., 600Hz) as the window width, the system slides within the remaining spectrum with a certain step size (e.g., the same as or smaller than the window width) to generate a series of continuous and non-overlapping (or partially overlapping) candidate frequency bands. For example, the total high-frequency analysis range is... The identified and expanded frequency bands to be avoided are The remaining part of the spectrum is... and With a preset bandwidth of 600Hz, candidate frequency bands can be generated starting from 300Hz: .

[0094] In some embodiments, the system can calculate the set of remaining available frequency bands after removing the main interference frequency bands from the preset high-frequency analysis total interval using the following formula (17): (17) Formula (17) is a set difference operation used to obtain a “clean” spectral range of selectable target detection bands. This indicates the preset high-frequency analysis range, which is a continuous frequency range, for example... ; This represents the union of the main noise frequency bands and their guard bands that need to be completely avoided, calculated by formula (16). It is a set of one or more discontinuous frequency bands; the symbol "\" represents the set difference operation, i.e., from the set Remove all elements belonging to the set. Element; This is the set of remaining frequency bands obtained after performing the difference operation. It consists of all continuous or discontinuous frequency sub-intervals that are not occupied by the main noise.

[0095] Next, in this embodiment, the system can also construct a series of consecutive candidate detection bands within the remaining frequency band set using formula (18): (18) Formula (18) defines the k-th candidate frequency band. The construction method. It is a continuous closed interval, representing a potential detection frequency band to be selected; Indicates the starting frequency of the candidate frequency band (unit: Hz); This represents a preset fixed bandwidth (unit: Hz), for example, 600Hz; the meaning of this formula is that, in order to... Starting from this point, extending towards higher frequencies. The width of the bandwidth forms a candidate frequency band. All constructed candidate frequency bands... It must be completely contained within the remaining frequency band set. conditions, that is The system starts by sliding the initial frequency in steps of a certain size. , can Multiple consecutive candidate frequency bands are generated internally.

[0096] Furthermore, in this embodiment, the system can also calculate the average power spectral density of each candidate frequency band using formula (19): (19) Formula (19) is the formula for calculating the average power spectral density, used to quantize candidate frequency bands. The average background noise energy level within. Indicates candidate frequency band Average power spectral density (unit: ); Indicates candidate frequency band The width, its value is equal to ; Represents the power spectral density function In frequency band Integrate the values ​​above to obtain the total energy within the frequency band (unit: Divide the total energy by the bandwidth. This yields the average energy per unit frequency within that frequency band, i.e., the average power spectral density. For example, for a... A candidate frequency band of Hz, if its total integral energy is Then its average power spectral density .

[0097] Furthermore, in this embodiment, the system can also calculate the minimum geometric distance from the center frequency of each candidate frequency band to the center point of the characteristic frequency of the nearest noise source using formula (20): (20) Formula (20) is used to calculate the minimum frequency interval. (Unit: Hz) This indicator quantifies the "safe distance" between candidate frequency bands and known strong noise sources. Indicates candidate frequency band The center frequency of is , and its value is ; This represents the set of characteristic frequencies determined by formula (6). The specific value of the center frequency of the r-th noise band constructed by the harmonics is: (where h is the harmonic order,) (characteristic frequency); symbol " "" indicates the center frequency of all possible noise bands Perform a traversal and calculate the center frequency of the candidate frequency bands. With each The absolute value of the difference between them is taken, and then the minimum value among all these absolute values ​​is selected as the minimum. The larger this value, the farther the candidate frequency band is from all known strong noise sources, and the lower the risk of being affected by their sideband interference or frequency aliasing.

[0098] Furthermore, in this embodiment, the system can initially select the optimal candidate with the smallest average power spectral density from all candidate frequency bands using formula (21): ;(twenty one) Among them, in formula (21) Operators represent the methods to find the objective function. The independent variable that achieves the minimum value This step aims to select from all candidate frequency bands. Find the average power spectral density in the middle. The frequency band with the smallest value is denoted as . This was initially selected as the optimal detection frequency band. This reflects the primary principle of selecting the "quietest" frequency band.

[0099] Meanwhile, in this embodiment, the system defines a criterion for similar power spectral densities using formula (22) to trigger a secondary selection mechanism: ;(twenty two) Formula (22) defines the mathematical conditions for similar power spectral densities. and Representing any two different candidate frequency bands and The average power spectral density value; For the preset density similarity threshold (unit: It is a small positive number, for example... This condition means that if the absolute value of the difference between the average power spectral density values ​​of two candidate frequency bands is less than or equal to a threshold... If they are considered to be "similar" in terms of background noise level, then they are considered to be "similar" in terms of background noise level. When there are multiple (more than one) candidate frequency bands that meet the condition of being "similar" to each other, it means that multiple frequency bands are difficult to distinguish in terms of "quietness" and need to enter the next round of selection based on "safe distance".

[0100] Finally, in this embodiment, when there are multiple candidate frequency bands with similar power spectral densities, the system ultimately determines the target high-frequency detection frequency band using formula (23): ;(twenty three) Among them, in formula (23) Operators represent the methods to find the objective function. The independent variable that achieves its maximum value This step means that, among all candidate frequency bands that satisfy the condition of similar power spectral density (Equation (22)), the one with the largest minimum frequency spacing is selected. The frequency band furthest from the noise source will be used as the final high-frequency detection frequency band for the target. This rule ensures that the frequency band with the highest anti-interference potential is selected from multiple low-noise candidate frequency bands.

[0101] Therefore, by combining the above formulas (17) to (23), the system can execute a complete, intelligent, and dual-layer optimized adaptive selection process for the target high-frequency detection band. This process first determines the available "remaining spectrum," and then constructs a series of candidate bands of equal width within this range. The average background noise level (BNR) is calculated for each candidate band. ) and minimum distance from known noise sources ( The system first selects the optimal frequency band based on the principle of "lowest noise". When multiple "optimal" candidates with extremely similar noise levels appear, a second round of decision-making based on "maximum safety interval" is initiated to ensure that the selected frequency band simultaneously meets the dual requirements of "low background noise" and "high interference isolation". This set of rigorous mathematical rules is the key guarantee for realizing the core inventive concept of "noise adaptive avoidance" of this invention, and lays the optimal frequency domain foundation for subsequent safe and efficient coding perturbation injection and weak fault feature extraction.

[0102] Traverse multiple candidate frequency bands, calculate the average power spectral density of each candidate frequency band, and calculate the minimum frequency interval between the center frequency of the candidate frequency band and the center frequencies of each noise frequency band determined by the characteristic frequency set.

[0103] The average power spectral density is an indicator that measures the average energy level of background noise within a frequency band. It is obtained by integrating the power spectral density value within the band and then dividing by the bandwidth. The minimum frequency interval is used to quantify the "distance" between a candidate frequency band and the known dominant noise frequency band. It is defined as the minimum geometric distance from the center frequency of the candidate frequency band to the center frequencies of all noise frequency bands defined by the set of characteristic frequencies (and their harmonics).

[0104] Specifically, for each candidate frequency band Its average power spectral density The calculation formula is: ,in This represents the previously calculated power spectral density. Its center frequency is... It can be simply taken as Assume that the center frequencies of the M noise bands determined by the characteristic frequency set are... Then the minimum frequency interval For example, for candidate frequency bands Its center frequency Assume the set of characteristic frequencies is The corresponding noise band center frequencies are 2500Hz and 5000Hz (assuming only the fundamental frequency is considered). Simultaneously, the average power spectral density within this frequency band is calculated. .

[0105] Based on the frequency band selection rules, the target high-frequency detection frequency band is determined from multiple candidate frequency bands based on the average power spectral density and minimum frequency spacing calculated for each candidate frequency band.

[0106] The predetermined frequency band selection rule is a decision logic based on multi-objective optimization. Its core principle is to prioritize the frequency band with the "quietest" background noise (lowest average power spectral density); if there are multiple candidate frequency bands with similar background noise levels, the frequency band that is farthest from the known noise source (largest minimum frequency interval) is selected to minimize the impact of noise sidebands or leakage.

[0107] Specifically, the system first finds the average power spectral density in all candidate frequency bands. The smallest frequency band is denoted as Then, check if there are other candidate frequency bands. Its average power spectral density and The absolute value of the difference is less than a preset density similarity threshold. (For example, Set as 10% of that). If such a percentage does not exist. Then select directly. As a high-frequency detection band for the target. If one or more such bands exist... Then these frequency bands (including These groups form a "proximity candidate group". Finally, from this "proximity candidate group", the candidate with the largest minimum frequency interval is selected. The candidate frequency band is selected as the final target high-frequency detection band. For example, the average power spectral density of all candidate frequency bands is calculated as follows: (Unit: any relative unit). Among them... of Minimum. Assume a threshold for similar densities. The inspection revealed that... The difference between 0.03 and 0.02 is greater than 0.005. The difference between 0.06 and 0.02 is greater than 0.005, and the differences for other values ​​are even larger. Therefore, no other candidate frequency bands are found. Their densities are similar. Therefore, we directly choose... Assuming a high-frequency detection band for the target. Let's consider another scenario... of , of And the difference 0.001 < 0.005, then and Enter the "similar candidate group". Then compare their minimum frequency intervals, assuming... , Then choose the one with the larger interval. As a target high-frequency detection band.

[0108] Therefore, according to the above implementation method, the system can intelligently and automatically select a comprehensively optimal detection frequency band from the available spectrum resources based on real-time noise spectrum analysis. This frequency band not only has the lowest background noise level, but also, when multiple low-noise candidates exist, it can preferentially select a frequency band far away from the main noise source. This provides a "clean" channel with the highest potential signal-to-noise ratio and the least impact from known interference for subsequent active coding perturbation injection and response detection, fundamentally improving the reliability and accuracy of weak fault feature detection.

[0109] In some embodiments, the frequency band selection rule is configured to: From multiple candidate frequency bands, the candidate frequency band with the lowest average power spectral density is determined.

[0110] Among them, the candidate frequency band with the lowest average power spectral density refers to the frequency band with the smallest average power spectral density value among all generated candidate frequency bands.

[0111] Specifically, the system can traverse all candidate frequency bands and read the pre-calculated average power spectral density value for each candidate frequency band. The algorithm compares these values ​​to find the minimum value and records the candidate frequency band number corresponding to that minimum value. For example, suppose there are 4 candidate frequency bands. to The pre-calculated average power spectral density values ​​are as follows: Unit / Hertz (unit: volts squared per Hertz) ), By comparison, the system determines To be the minimum value, therefore the candidate frequency band It was identified as the candidate frequency band with the lowest average power spectral density.

[0112] Determine whether any candidate frequency band satisfies the density similarity threshold condition. The density similarity threshold condition means that the difference between the average power spectral density of the candidate frequency band and the average power spectral density of the candidate frequency band with the lowest average power spectral density is less than a preset density similarity threshold.

[0113] The density similarity threshold is a preset tolerance value used to determine whether the background noise levels of two candidate frequency bands are "similar". This threshold defines the range of differences in the average power spectral density value that are acceptable, within which multiple frequency bands are considered to have similar low noise levels, requiring the initiation of a second round of decision-making (based on frequency spacing).

[0114] Specifically, the system determines the candidate frequency band with the lowest average power spectral density. and their corresponding density values After that, it will iterate through all the exceptions. Every other candidate frequency band For each Calculate its average power spectral density and The absolute value of the difference Then, take each one Close to the preset density threshold Compare. As long as at least one exists... satisfy If the condition is met, it is determined that the entity "exists"; otherwise, it is determined that the entity "does not exist". For example, continuing from the previous example, it has been determined that... For the lowest density frequency band, Assume a preset density similarity threshold. Calculate other frequency bands and Density difference: greater than ; greater than ; greater than Because of all All are not less than Therefore, the system judges that there are no other candidate frequency bands that meet the conditions.

[0115] If it does not exist, the candidate frequency band with the lowest average power spectral density will be determined as the target high-frequency detection frequency band.

[0116] This branch deals with situations where the background noise has a significant advantage in the lowest frequency band, making it unmatched by competitors.

[0117] Specifically, when the system determines that there are no other candidate frequency bands with similar densities, the decision-making logic ends, and the system directly selects the candidate frequency band with the lowest average power spectral density previously determined. This was directly selected as the target high-frequency detection band for this detection. For example, based on the above judgment, since there is no density and In other similar frequency bands, the system will (For example The high-frequency detection band was directly identified as the target for this round of detection.

[0118] If it exists, then from all candidate frequency bands that meet the density similarity threshold condition, select the candidate frequency band with the maximum and minimum frequency interval to the center frequency of each noise frequency band as the target high-frequency detection frequency band.

[0119] Among them, candidate frequency bands that meet the density similarity threshold constitute a "similar candidate group", which includes the candidate frequency band with the lowest average power spectral density. And all that is satisfied Other candidate frequency bands. The maximum and minimum frequency spacing refers to the pre-calculated minimum frequency spacing relative to the center frequency of each noise band within this "similar candidate group". The candidate frequency band with the largest value.

[0120] Specifically, when the system determines that other candidate frequency bands with similar density "exist," it first considers all candidates that meet the criteria... Candidate frequency bands (including) The candidate frequency bands are placed into a temporary set called the "similarity candidate group". Then, the system iterates through each candidate frequency band in this "similarity candidate group" and reads the pre-calculated minimum frequency interval for each frequency band. Finally, the system compares these. Choose the value with the largest value. The corresponding candidate frequency band is determined, and this frequency band is identified as the final target high-frequency detection frequency band. For example, consider another scenario: there are candidate frequency bands. Their average power spectral densities are respectively .set up .but This is the lowest density frequency band. Calculate the difference: ,therefore and Similar densities; ,therefore Not similar. Therefore, the "similar candidate group" includes... and Assuming that it has been calculated in advance... minimum frequency interval , minimum frequency interval .because The system selects the frequency with the maximum and minimum frequency spacing. The ultimate target is the high-frequency detection band.

[0121] Therefore, according to the above implementation method, the system can perform a two-stage intelligent optimization decision. This decision mechanism first seeks the "quiet" frequency band with the weakest background noise; when multiple excellent candidates with comparable noise levels appear, it further introduces the "safe distance" principle, selecting the frequency band farthest from known strong noise sources to maximize the avoidance of potential noise leakage and sideband interference risks. This rule ensures that the selected target high-frequency detection band achieves comprehensive optimization in both "low background noise" and "high interference isolation," creating optimal initial signal-to-noise ratio conditions for subsequent active detection.

[0122] In some embodiments, the step of controlling the neutral point-to-ground bypass branch to switch between at least two different frequency-selective impedance levels according to a preset coding sequence includes: Before initiating active detection, an initiation condition judgment is performed. The initiation condition judgment includes: verifying whether the amplitude of the current bus zero-sequence voltage is within the preset voltage threshold range, and estimating whether the expected fluctuation value of the bus zero-sequence voltage caused by impedance level switching is lower than the preset safety limit based on the characteristic parameter set.

[0123] The voltage threshold range is a preset voltage amplitude interval to ensure that active detection only initiates when the system is under a minor ground fault or in a normal asymmetrical operating state, avoiding unnecessary or potentially unsafe operations when the voltage is too high (e.g., metallic grounding) or too low. The safety limit is a preset upper limit for voltage fluctuation amplitude to ensure that actively injected disturbances do not cause excessive power frequency voltage fluctuations, thereby ensuring the safety and stability of system operation. Estimating the expected fluctuation value of the bus zero-sequence voltage requires calculation based on the current system characteristic parameter set (especially the equivalent zero-sequence network parameters) and the impedance level parameters to be switched.

[0124] Specifically, the system monitors the effective value or peak value of the zero-sequence voltage of the bus in real time. The startup condition judgment logic first checks... Does it meet the requirements? ,in and These are the preset low voltage threshold and high voltage threshold, respectively. Next, the system calculates the system's zero-sequence equivalent admittance based on the currently estimated system value. Main grounding resistance And the impedance values ​​of the planned frequency-selective impedance levels at the power frequency. Calculate the expected fluctuation amplitude of the zero-sequence voltage of the bus when switching from the current gear to the target gear. Only if both conditions are met Within the threshold range and Less than the safety limit Only then is the activation condition deemed met. For example, the preset voltage threshold range is 0.5kV to 3kV, and the safety limit... The current bus zero-sequence voltage is 0.15kV. The voltage is 1.2kV, satisfying the first condition. The system estimates the current zero-sequence equivalent admittance and calculates the switching from the open-circuit position to the frequency-selective impedance position (impedance at power frequency is...). When ), expected fluctuations The voltage is approximately 0.08kV, which is less than the safety limit of 0.15kV, so the second condition is also met. Therefore, the start-up condition is met.

[0125] If the result of the start condition judgment is that all conditions are met, the control flow enters the preset detection cycle.

[0126] The preset detection period refers to a fixed or variable time interval from the start of injecting the coded perturbation to the completion of data acquisition and processing. Entering the preset detection period means that the system will initiate the transmission of the coded sequence and the switching process of the impedance level according to a predetermined procedure.

[0127] Specifically, when the start condition judgment logic outputs a "satisfied" signal, the system's central processing unit or dedicated control logic generates a cycle start instruction. This instruction clears the timer, initializes the encoded sequence pointer, configures the data acquisition buffer, and sets the system status flag to "actively probing," thus formally entering the preset probing cycle. For example, the total length of the preset probing cycle is 130ms, corresponding to a 13-bit encoded sequence with a symbol width of 10ms. When the start condition is met, the system state machine jumps to the "preset probing cycle" state and begins execution from the first symbol of the encoded sequence.

[0128] Within the preset detection period, according to the preset coding sequence, at the start time corresponding to each symbol, a switching command indicating the target impedance level is sent to the control module of the neutral point-to-ground bypass branch.

[0129] The preset encoding sequence is a set of predefined discrete value sequences (such as binary sequences) used to control the impedance level switching order. Each value corresponds to a specific impedance level. A symbol is a basic unit in the encoded sequence, and its duration is called the symbol width. The start time of a symbol is determined by the symbol width and the sequence index. The control module is the hardware circuit (such as a programmable gate array FPGA or microcontroller) and its firmware responsible for receiving instructions and driving the switching devices to perform actions. The switching instruction is a digital or pulse signal containing the target impedance level number or code.

[0130] Specifically, the system has an internal real-time clock or timer. The timer starts after the preset detection period begins. The timer operates according to the preset symbol width. (e.g., 10ms), the system at each time point At any given time, the value c[k] of the k-th symbol in the preset encoding sequence is queried. Based on the value of c[k], the system sends a switching command containing the target gear information to the control module of the neutral-to-ground bypass branch via a communication interface (such as a serial peripheral interface SPI or a digital output port). For example, the preset encoding sequence is a 13-bit Barker code, with a symbol width of... .exist At each time, the system sends the corresponding The switching command. Assuming... This corresponds to the "frequency-selective impedance setting". This corresponds to "engaging in high resistance mode (or opening the circuit)".

[0131] In response to each received switching command, the control module drives the switching devices in the neutral-to-ground bypass branch to switch the current input impedance level of the neutral-to-ground bypass branch to the target impedance level indicated by the switching command.

[0132] Switching devices refer to power electronic components used to connect or disconnect circuit paths, such as insulated-gate bipolar transistors (IGBTs), metal-oxide-semiconductor field-effect transistors (MOSFETs), or electromechanical relays. Driving refers to the control module outputting sufficient voltage or current signals to change the switching device from one conducting state to another.

[0133] Specifically, after receiving a switching command from the main processor, the control module's internal logic circuits or microprograms parse the target impedance level. Subsequently, the control module's output drive circuit generates corresponding control signals (such as high / low levels or pulse-width modulation (PWM) waves) and applies them to the corresponding switching devices (such as the control terminals of solid-state switches or relays). The change in the state of the switching devices causes a change in the electrical connection of the bypass branch, thereby switching the currently engaged impedance level to the new level specified by the command. For example, the bypass branch may contain two branches: one is the aforementioned frequency-selective impedance branch (level 1), and the other is in a high-impedance state or open circuit (level 2), controlled by two anti-parallel thyristors or a relay. When the control module receives the command "switch to level 1," it triggers the switching device connected to the frequency-selective impedance branch to conduct, while simultaneously ensuring that the switching device connected to the other branch is turned off, thus completing the switching.

[0134] According to the symbol width of the preset encoding sequence, the steps of sending switching commands and driving switching devices are repeatedly executed until all symbols in the preset encoding sequence have been traversed. Thus, by switching from the current input impedance level to the target impedance level, a voltage or current disturbance signal corresponding to the preset encoding sequence is generated in the neutral point to ground loop.

[0135] In this context, traversal refers to processing each symbol in the encoded sequence sequentially. Generating voltage or current disturbance signals corresponding to the encoded sequence means that, due to the periodic variation of the neutral point-to-ground impedance according to the encoding rules, corresponding zero-sequence voltage and current changes will be excited in the system, and the envelope or characteristics of these changes are strongly correlated with the encoded sequence used.

[0136] Specifically, this is a cyclical process. After sending the switching command for the k-th symbol and driving the switching, the system waits for one symbol width. The time, and then At time k, a switching command for the (k+1)th symbol is sent, driving the switch to the new gear. This process is repeated until k equals the length of the encoded sequence. Subtract 1, which means the whole thing is done. The switching of each code element. Each impedance level switch changes the neutral-to-ground parallel admittance, thereby introducing an impedance related to the coded sequence into the bus zero-sequence voltage and feeder zero-sequence current. The perturbation component matches the waveform characteristics. For example, for a Barker code sequence with a length of 13 and a symbol width of 10ms, the system will execute a total of 13 "send command-drive switch" loops, lasting 130ms. During these 130ms, the neutral point-to-ground impedance follows the pattern of the Barker code. Switching between high impedance and low impedance (frequency-selective impedance) ultimately generates a specific high-frequency disturbance signal in the system that lasts for 130ms and whose waveform is consistent with the autocorrelation characteristics of a 13-bit Barker code.

[0137] In some embodiments, the system can calculate the neutral point-to-ground equivalent grounding admittance using the following formula (24): ;(twenty four) Formula (24) is the formula for calculating the equivalent grounding admittance, which is used to characterize the admittance characteristics of the entire grounding system when the neutral point to ground bypass branch is at a specific impedance level. This represents the system admittance when the p-th bypass impedance setting is engaged; it is a complex number that varies with frequency ω. The main grounding resistance is a real resistance that does not change with frequency, and its unit is 1000 kJ / m². ; Let be the impedance value of the p-th bypass impedance setting. It is a complex function, and its specific form depends on the circuit topology used for that setting. This formula means that when the p-th setting is engaged, the total system admittance is equal to the admittance of the main grounding resistance. Admittance of bypass branch impedance The sum of parallel connections. If the p-th gear indicates that the bypass is disconnected, then ,thereby At this time, the total absorbance The control module uses a preset encoding sequence. Switching between different impedance levels dynamically changes the impedance. This generates a disturbance signal.

[0138] Next, in this embodiment, the system can also define the expression for the frequency-selective impedance as the key level using formula (25): (25) Formula (25) is the impedance expression of the series damping resistor-inductor-capacitor (RLC) branch, which is configured as a frequency selective impedance level to achieve the core requirement of "high impedance at power frequency and low impedance in the high frequency detection band of the target". This is the complex impedance for this range. A damping resistor, used to limit the current at the resonant point, with units of . ; Inductance is measured in Henry (H). Capacitance is expressed in farads (F); j is the imaginary unit; ω is the angular frequency. , where f is the frequency (in Hz). The resonant frequency of this series RLC branch. Determined by formula (26): (26) When angular frequency When, the inductive resistance term in formula (25) With resistance item If the magnitudes are equal but the signs are opposite, they cancel each other out. At this point, the branch impedances... It exhibits low impedance characteristics. At power frequency... At that point, because the frequency is far from the resonant point The inductive reactance and capacitive reactance cannot cancel each other out, resulting in a very high amplitude of the total impedance of the branch, thus meeting the high impedance requirement at power frequency.

[0139] Furthermore, in this embodiment, the system can also use formula (27) to determine the resonant frequency. Set to match the target's high-frequency detection band: (27) in, The center frequency of the high-frequency detection band of the target can be calculated using formula (28): or (28) In formula (28), and These are the start and end frequencies (in Hz) of the target's high-frequency detection band, respectively. First form The second form is the arithmetic mean. The geometric mean is used. To finely design the parameters of the RLC branch, the system introduces the concept of quality factor Q and correlates it with the target bandwidth B through formula (29): (29) in, Q represents the bandwidth (in Hz) of the high-frequency detection band of the target. The Q value reflects the frequency selectivity of the branch at the resonant frequency; the higher the Q value, the stronger the frequency selectivity and the narrower the bandwidth.

[0140] Then, in this embodiment, given a damping resistance Given the design premise of quality factor Q, the system can calculate the required inductance using formulas (30) and (31). and capacitor The specific value is determined to complete the tuning of the frequency-selective impedance branch: (30) (31) Formulas (30) and (31) provide closed-form solutions for the RLC branch parameters. For example, assuming the determined target high-frequency detection band is... ,but If a damping resistor is set... To limit the current, the following formula (30) can be used to calculate: Subsequently, according to formula (31), the following can be calculated: This design of the series RLC branch will... The surrounding area presents approximately It has low impedance, but exhibits high impedance of up to several hundred ohms at 50Hz power frequency.

[0141] Therefore, by combining the above formulas (24) to (31), the system can complete the entire theoretical link from system equivalent admittance modeling to frequency-selective impedance parameter design. This link first establishes a mathematical model for changing the system admittance by switching the bypass branch position (Formula 24). Then, to achieve the core detection requirements, the series RLC branch is selected as the implementation method of frequency-selective impedance, and its impedance expression (Formula 25) and the parameter relationship that determines its resonance characteristics (Formula 26) are given. The system further aligns the resonant frequency with the center of the adaptively selected target high-frequency detection band (Formulas 27-28) and introduces the quality factor (Formula 29) to describe the bandwidth characteristics, and finally derives the design formula (Formulas 30-31) for directly calculating the required inductance and capacitance values ​​based on the target band and engineering constraints (such as damping resistor values). This series of rigorous mathematical derivations provides a direct and reliable theoretical basis and engineering design method for physically constructing a disturbance injection branch that meets the requirements of "high impedance at power frequency and low impedance at target band".

[0142] Therefore, according to the above implementation method, the system can achieve a controlled, safe, and highly recognizable active detection signal injection. Dual safety checks before startup ensure operational safety; by strictly following the timing and logic control impedance switching according to the encoded sequence, a pre-designed detection signal with good correlation characteristics is generated, laying a solid foundation for subsequent signal processing techniques such as matched filtering to extract weak fault response features from strong noise.

[0143] In some embodiments, high-frequency components located within the target high-frequency detection band are extracted from electrical quantity data to construct a normalized energy response sequence for each feeder, including: The bus zero-sequence voltage and each feeder zero-sequence current in the electrical quantity data are bandpass filtered to extract the high-frequency voltage component of the bus and the high-frequency current component of each feeder located in the target high-frequency detection frequency band.

[0144] Bandpass filtering is a signal processing operation that allows signal components within a specific frequency range (i.e., the passband) to pass through while significantly attenuating frequency components outside that range. In this step, the purpose of this processing is to separate the signal components located only within the target high-frequency detection band (e.g., 900Hz to 1500Hz) from the original zero-sequence voltage and current signals, which contain multiple frequency components.

[0145] Specifically, the system can filter the sampled discrete-time sequence signal using digital filters, such as finite-length unit impulse response (FIR) filters or infinite-length unit impulse response (IIR) filters. The passband frequency range of the filter is set to perfectly match the target's high-frequency detection band. For the bus zero-sequence voltage signal... After filtering, the high-frequency voltage component of the bus is obtained. ; for the zero-sequence current signal of the i-th feeder The high-frequency current component of the feeder is obtained after filtering. For example, the target high-frequency detection band is The system sampling frequency is 20kHz. A FIR bandpass filter with a passband of 900Hz to 1500Hz is designed for the system. The acquired bus zero-sequence voltage data stream is passed through this filter, and the output is the high-frequency voltage component of the bus. The zero-sequence current data stream of each feeder is passed through the same filter, and the output is the high-frequency current component of each feeder.

[0146] In some embodiments, the system can verify the first necessary condition before active detection is initiated, namely the bus zero-sequence voltage amplitude safety condition, by using the following formula (32): (32) Formula (32) is a voltage amplitude window discriminant, which is used to ensure that active detection is only started when the system is in a suitable operating range, and to avoid unnecessary or potentially harmful disturbance operations in the event of a serious fault or extreme asymmetry. This indicates the current measured zero-sequence voltage amplitude of the bus (unit: V or kV). This is a positive constant, representing the preset low voltage threshold. For the preset high voltage threshold, one greater than The constant. The meaning of this inequality is that only when the measured voltage amplitude is... At the same time greater than or equal to and less than or equal to Only then is the first startup condition deemed met. This condition excludes conditions of extremely low voltage (potentially without fault) or excessively high voltage (such as near metallic grounding). Specifically, the system calculates the root mean square or peak value of the zero-sequence bus voltage in real time as... and with preset thresholds stored in the system and The system compares the inequalities. Only when the inequality is true does the system consider the first starting condition satisfied and proceed to verify the next condition. For example, for a 10kV distribution network, it can be set... If the current measurement yields... Then the condition is satisfied (because) ).like If the voltage is 4.0kV or higher, this condition will not be met, and the active detection process will not be initiated.

[0147] Next, in this embodiment, the system can also estimate the second necessary condition before active detection is initiated, namely the safety condition for expected voltage fluctuations caused by the disturbance operation, using formula (33): (33) Formula (33) is the calculation formula for the expected fluctuation value of the zero-sequence voltage of the bus, which is used to quantify the neutral point to ground bypass branch from the tap position. Shift to gear At that time, the bus voltage fluctuation amplitude may be caused under power frequency. This represents the estimated expected voltage fluctuation amplitude (unit: V or kV). This represents the current zero-sequence voltage amplitude at the bus. and These represent the gears calculated according to formula (24). and gear The system's equivalent ground admittance at the power frequency angular frequency when put into operation The complex values ​​under; This represents the equivalent admittance (unit: Siemens S) of the current zero-sequence network of the distribution network. It is a complex number whose value can be estimated from the system characteristic parameter set based on the network topology and parameters. The physical meaning of this formula is the relative voltage fluctuation (…). This is approximately equal to the ratio of the change in grounding admittance to the total system admittance (the sum of the network equivalent admittance and the grounding admittance before the switch). Only if this estimate is... Less than the preset safety limit Only then is the second activation condition deemed to be met.

[0148] Specifically, the system obtains the equivalent admittance of the current zero-order network by solving the network equations or mapping parameters based on the real-time acquired feature parameter set. Then, for the upcoming coded switching action (such as switching from an open circuit to a frequency-selective impedance range), the system calculates the ground admittance before and after the switching based on the pre-stored range parameters. and Finally, substituting these parameters into formula (33), the expected fluctuation amplitude is calculated. And determine whether it is less than For example, setting security limits. The current zero-sequence equivalent admittance is estimated. Siemens. Setting gears. To "pave the way", then Siemens (assuming main grounding resistance) (Set gear) This is a frequency-selective impedance setting, exhibiting high impedance at the power frequency of 50Hz. ,but Siemens. Substituting into formula (33), we get... This value is significantly greater than 0.15kV, therefore the second start condition is not met, and the system will terminate this detection. This example illustrates that formula (33) can effectively identify and prevent dangerous operations that may lead to excessive power frequency voltage disturbances.

[0149] Furthermore, in this embodiment, in order to improve the adaptability of the detection method under different operating environments, the system can also adaptively determine the length of the encoded sequence (also known as the code length) used in this detection using formula (34): (34) Formula (34) is an adaptive code length calculation formula, which is used to dynamically adjust the length of the coding sequence (i.e. the total number of code elements) according to the current noise environment and system strength. The calculated adaptive code length (unit: code); This is a limiting function, its purpose is to limit the input value x when it is less than the lower limit. When, output When x is greater than the upper limit When, output Otherwise, output x. This represents the floor function. The base code length (unit: number), for example, 13; and This is the adjustment coefficient, a positive constant. SCR is the noise intensity index calculated by formula (12); SCR is the short circuit ratio of the system, which is a real number that characterizes the strength of the system (the ratio of the output of new energy to the short circuit capacity of the system). Its value can be obtained through real-time monitoring data or system configuration parameters. It is a very small positive number that prevents the denominator from being zero; and These are the minimum and maximum allowed values ​​for the code length, respectively. This formula indicates that the code length... It will change with the noise intensity The increase is due to the increase in the intensity of the signal and the decrease in the system strength SCR. A longer coding sequence means a longer signal energy accumulation time, which helps to extract more stable correlation features from strong noise.

[0150] Furthermore, in this embodiment, the system can also adaptively determine the symbol width used for this detection using formula (35): (35) Formula (35) is the adaptive symbol width calculation formula, which is used to dynamically adjust the duration of each symbol according to the current noise environment and system strength. The calculated adaptive symbol width (unit: s); The reference symbol width (unit: s), for example, 0.01 seconds (i.e., 10 ms); and This is the adjustment coefficient, a positive constant. and These represent the minimum and maximum allowed values ​​for the symbol width, respectively. The meanings of other symbols are the same as in formula (34). This formula indicates that the symbol width... Similarly, it will change with noise intensity. The increase is due to the increase in energy density and the decrease in system strength SCR. A larger symbol width means a longer energy injection and response settling time in each state (gear), which helps to improve the signal-to-noise ratio of the response signal within a single symbol, thereby improving the overall robustness of detection.

[0151] Therefore, through the combination of the above formulas (32) to (35), the system can achieve complete dual security verification before active detection, as well as dynamic adaptive optimization of key detection parameters (code length and symbol width). Formulas (32) and (33) together construct a safety start gate based on voltage amplitude and disturbance estimation, ensuring that active detection is only carried out under suitable voltage conditions and controllable disturbance expectations, thus guaranteeing the operational safety of the power system from the source. Formulas (34) and (35) constitute a parameter adaptive regulator based on noise intensity and system state, enabling the detection method to intelligently adjust the duration and complexity of the encoded signal according to the severity of the real-time operating environment (high noise, weak system), thereby maintaining a high success rate and high robustness in detecting weak fault features in the diverse "three high" distribution network scenarios. The synergistic effect of these four formulas fully reflects the core design idea of ​​this invention, which organically combines operational safety and detection performance.

[0152] In some embodiments, the system can perform bandpass filtering on the acquired discrete signal using the following formulas (36) and (37) to extract the signal components located within the target high-frequency detection band: (36) (37) Among them, formulas (36) and (37) are the discrete convolution calculation formulas of the high-frequency voltage component of the bus and the high-frequency current component of the feeder, respectively, which are used to separate the effective components in the target frequency band from the original time domain signal. This represents the high-frequency component of the zero-sequence voltage of the bus at the nth discrete moment. This represents the high-frequency component of the zero-sequence current at the nth discrete moment of the extracted i-th feeder; the symbol " " represents the discrete linear convolution operation; It is a high-frequency detection band for the target The impulse response sequence of a matched discrete-time filter (such as a finite-length unit impulse response (FIR filter); and These are the discrete-time sequences of the original acquired bus zero-sequence voltage and the zero-sequence current of the i-th feeder. This step, by performing convolution operations in the time domain, is equivalent to performing bandpass filtering in the frequency domain, thereby accurately obtaining the voltage and current signal components within the target detection frequency band.

[0153] Next, in this embodiment, the system can also divide the entire active detection period into multiple consecutive symbol intervals according to the symbol structure of the encoded sequence using formula (38): (38) Wherein, formula (38) is the code element interval The index definition is used to divide consecutive time-domain sampling points into data blocks corresponding to each symbol of the encoded sequence. This represents the k-th symbol interval, which contains the indices of all sampling points corresponding to the k-th coded symbol in time; It is the number of sampling points contained in each symbol; it is an integer determined by the system sampling frequency. (Unit: Hz) and preset symbol width The product (unit: seconds) is calculated as follows: The variable k is the code element index, with a value ranging from 0 to... ,in This represents the total number of symbols in the encoded sequence. For example, when the sampling frequency... Symbol width At that time, the number of sampling points contained in each symbol Then the 0th code segment Corresponding sampling index 0 to 199, the first symbol interval The corresponding indices are 200 to 399, and so on.

[0154] Furthermore, in this embodiment, the system can also calculate the signal energy in each symbol interval using formulas (39) and (40): (39) (40) Formula (39) is used to calculate the high-frequency voltage energy of the bus for the kth symbol. Formula (40) is used to calculate the high-frequency current energy of the i-th feeder for the k-th symbol. The formula. The unit is or Specifically, it depends on the normalization; The unit is ampere square (A / m²). ) or ampere squared second ( The calculation process is as follows: traverse the code element intervals. Each sampling point within the range is indexed n, and high-frequency signals are taken. (or The square of the amplitude of the signal is calculated, and the squares of all sampling points within that interval are summed. This step quantifies the total energy carried by the signal components within the target frequency band during each symbol duration.

[0155] Furthermore, in this embodiment, the system can also calculate the baseline energy average value before the start of active detection using formulas (41) and (42) as a steady-state reference benchmark for subsequent comparisons: (41) (42) Among them, formulas (41) and (42) are the average energy values ​​of the bus baseline, respectively. and feeder baseline energy mean The calculation formula. The baseline symbol count used to calculate the average is a preset positive integer, for example, 5; the summation variable k ranges from... -1 indicates the time before the probe starts (defined as k=0). A series of consecutive code elements; and For this The energy values ​​corresponding to each baseline symbol are pre-calculated using formulas (39) and (40). By calculating the arithmetic mean of the energies of these baseline symbols, the system obtains a stable benchmark characterizing the inherent energy level of the signal within the target frequency band when not affected by active disturbances.

[0156] Furthermore, in this embodiment, the system can also calculate the specific change in signal energy of each symbol relative to the baseline energy mean during active detection using formulas (43) and (44): (43) (44) Formula (43) defines the change in the high-frequency voltage energy of the bus for the kth symbol. Formula (44) defines the change in high-frequency current energy of the i-th feeder for the k-th symbol. . and These represent the energy of the k-th symbol within the preset detection period; and The baseline energy mean is calculated using formulas (41) and (42). This calculation process is simple and direct: subtract the steady-state baseline mean from the energy value corresponding to each symbol within the preset detection period, and the resulting difference is... and This quantifies the energy response increment caused by the actively injected coding perturbation. This change reflects the alteration of the system's zero-order network characteristics due to the perturbation injection, serving as the direct data basis for subsequent construction of normalized response sequences and fault feature identification. In another implementation, the energy change can also be calculated using the difference between adjacent symbols, i.e. Its purpose is to highlight the instantaneous response changes during the detection period.

[0157] Therefore, by combining the above formulas (36) to (44), the system can complete the entire signal processing link from high-frequency signal extraction to energy quantization and then to energy change calculation. This link first uses digital filter technology to accurately separate the effective components located in the target high-frequency detection band from the mixed original signal (formulas 36-37); then, according to the temporal structure of the coding sequence, the preset detection period is divided into code intervals that can be analyzed independently (formula 38); then, in each code interval, the time domain signal is converted into an energy value that can characterize the strength of the response through square and integral operations (formulas 39-40); and a steady-state energy benchmark is established using the signal before the detection begins (formulas 41-42); finally, through simple difference calculation, a quantitative index that can clearly reflect the strength of the energy response induced by active disturbance—energy change (formulas 43-44)—is obtained. This series of rigorous mathematical operations systematically converts the electrical disturbance at the physical level into a numerical feature that is highly representative and resistant to interference and can be directly used by subsequent matched filtering and fault decision algorithms. This is the key technical link for achieving accurate target selection in this invention.

[0158] In some embodiments, the system can construct the normalized energy response sequence of each feeder relative to the bus using the following formula (45): (45) Wherein, formula (45) is the normalized energy response sequence. The definition of this formula is used to calculate the normalized energy response value of the i-th feeder in the k-th symbol interval. This formula calculates the change in high-frequency current energy of the feeder for each symbol. Changes in high-frequency voltage energy at the busbar The ratio is normalized to eliminate common-mode interference caused by fluctuations in injected signal strength and changes in system common noise, thereby highlighting the response differences between fault-coupled paths and non-faulty paths. It is a dimensionless ratio; For a very small positive number that prevents the denominator from being zero (e.g. This ensures computational stability.

[0159] Next, in this embodiment, the system can also construct a matching template sequence corresponding to the preset encoding sequence using formula (46): (46) Wherein, formula (46) is the mean-free matching template sequence. The core of the calculation formula is to generate a reference sequence with a mean of zero by performing zero-mean processing on the original coding sequence, so as to enhance the detection sensitivity of subsequent cross-correlation operations for known coding features. This represents the matching template value for index k; This represents the original value of the preset encoded sequence at the k-th symbol (e.g. or ); The total length of the encoded sequence (i.e., the number of code elements); That is, the entire encoded sequence The arithmetic mean.

[0160] Furthermore, in this embodiment, the system can also calculate the cross-correlation result of each feeder under different time delay offsets τ using formula (47): (47) Formula (47) is the time delay cross-correlation calculation formula, used to quantify the normalized energy response sequence of the i-th feeder. Matching template sequence The degree of similarity under different relative time delays τ. This represents the cross-correlation value of the i-th feeder under time delay τ; the summation variable k ranges from 0 to... Covers the entire encoded sequence length; index expression This indicates that when τ is positive, the template sequence g is relative to the response sequence. Slide forward (time advance) τ positions, ensuring during calculation. The index value falls within the valid sequence range (e.g., by padding with zeros or truncating boundary data).

[0161] Furthermore, in this embodiment, the system can also determine the final matched filter output value for each feeder using formula (48): (48) Formula (48) is the maximum cross-correlation amplitude extraction formula, which is used to find the value with the largest absolute value from the cross-correlation results under all time delay offsets τ as the final characteristic quantity of the feeder. This represents the output value of the matched filter for the i-th feeder; The maximum allowable time delay alignment deviation is a positive integer, which defines the search range limit of time delay τ in formula (47); the operator max means taking the maximum value.

[0162] Therefore, by combining the above formulas (45) to (48), the system can complete the transformation from raw energy change data to high signal-to-noise ratio eigenvalues ​​that can be used for accurate decision-making. The complete feature extraction process is described. First, normalization (Equation 45) suppresses common-mode fluctuations, making the response differences caused by fault coupling more significant. Next, a mean-free matching template is constructed (Equation 46) to establish an optimal reference benchmark for feature matching. Then, by calculating the cross-correlation sequence with time delay search (Equation 47), the maximum correlation is captured even with small synchronization errors. Finally, by extracting the maximum correlation amplitude under all possible time delays (Equation 48), a robust and comparable quantitative indicator is obtained for each feeder. This indicator directly reflects the strength of the feeder's response to coding disturbances, and is crucial for the next step of faulty feeder determination (such as determining...). It provides a direct and reliable basis for decision-making.

[0163] In some embodiments, the system can preliminarily determine the feeder with the largest matched filter output value as a suspected faulty feeder using the following formula (49): (49) Formula (49) is the maximum value index determination formula, used to determine the sequence of matched filter output values ​​from all feeders. Find the feeder number corresponding to the maximum value. This indicates the calculated suspected faulty feeder number; Indicates the desire to make The independent variable i that achieves the maximum value is the feeder index; Let be the matched filter output value of the i-th feeder calculated by formula (48). This formula performs a simple comparison and index lookup operation, initially identifying the feeder with the maximum output value as the most likely fault location.

[0164] Next, in this embodiment, the system can also perform a first significance check on the suspected faulty feeder using formula (50): (50) Formula (50) is the relative significance criterion, used to determine whether the response intensity of a suspected faulty feeder is significantly higher than the average response level of all other healthy feeders. The fraction on the left side of the inequality has a numerator... It is a suspected faulty feeder. The matched filter output value; the denominator consists of two parts, Indicates except for suspected faulty feeders Apart from that, all the rest Matched filter output value of a healthy feeder The arithmetic mean of the fractions, where ε is a very small positive number introduced to prevent the denominator from being zero (e.g., ...). ); The preset first threshold is a real number greater than 1 (e.g., 3.0). The physical meaning of this inequality is that the response value of a suspected faulty feeder must be at least equal to the average response value of other feeders. The first significance check is passed only if the inequality is true.

[0165] Furthermore, in this embodiment, the system can also perform a second significance check on the suspected faulty feeder using formula (51): (51) Formula (51) is the absolute significance criterion, used to determine whether the response intensity of a suspected faulty feeder is significantly higher than the inherent background noise level of the system. The fraction on the left side of the inequality, with its numerator still being... ; denominator The root mean square value of the historical baseline noise is obtained by calculating the root mean square of the matched filter output value sequence obtained by collecting data from historical non-detection periods (i.e. normal operation periods without active disturbance injection) according to the same matched filter processing procedure. The preset second threshold is a real number greater than 1 (e.g., 5.0). The physical meaning of this inequality is that the response value of a suspected faulty feeder must be at least equal to the historical baseline noise level. The second significance check is passed only if the inequality is true.

[0166] Furthermore, in this embodiment, to ensure the safety of the active detection process, the system can also calculate and monitor the actual fluctuation amplitude of the bus zero-sequence voltage in real time using formula (52): (52) Formula (52) is the formula for calculating the actual fluctuation of the zero-sequence voltage of the bus. This represents the instantaneous fluctuation amplitude of the zero-sequence voltage of the bus relative to its reference value at any time t during the active detection process (unit: V or kV). Represents the real-time bus zero-sequence voltage measurement value at time t; The reference voltage at time t can be obtained by recording the bus zero-sequence voltage waveform for a period of time before the detection starts, or by estimating the expected voltage under undisturbed conditions in real time during the detection period using a filtering or prediction algorithm. This formula quantifies the actual voltage fluctuation caused by impedance switching disturbances in real time by calculating the absolute value of the difference between the instantaneous voltage and the reference voltage. The system will continuously monitor... Once its value exceeds the preset safety action threshold If the current active detection operation is stopped immediately, the neutral point bypass branch will be restored to the default ground state to prevent unsafe impacts on the system.

[0167] Therefore, by combining the above formulas (49) to (52), the system can construct a complete fault determination and safety protection logic that includes preliminary screening, double significance verification, and full-process safety monitoring. First, the suspected faulty feeder with the strongest response is quickly located using formula (49). Then, formulas (50) and (51) are applied sequentially to perform strict "relative significance" and "absolute significance" verification on it. Only when both conditions are met simultaneously can the feeder be finally determined. This ensures that the fault feeder is a true fault line, thus effectively avoiding misjudgment of a single maximum value due to noise interference or measurement error, and improving the reliability and accuracy of line selection. At the same time, the real-time voltage fluctuation monitoring mechanism provided by formula (52) forms a closed loop with the safety prediction before detection starts (formula 33), constituting a double safety defense line throughout the detection process, ensuring the robustness of the method of the present invention and the safety of system operation in practical applications.

[0168] The preset detection period is divided into multiple consecutive symbol intervals according to the symbol width of the preset coding sequence.

[0169] Here, a symbol interval refers to a time segment whose length is equal to the symbol width and corresponds to a symbol in the preset coding sequence. The entire preset detection period is uniformly divided into... Such a continuous segment, in which The length of the encoded sequence.

[0170] Specifically, the system uses the symbol width of the preset encoding sequence. (Unit: seconds) and sampling frequency (Unit: Hz) Calculate the number of sampling points contained in each symbol interval. Then, the sampling points within the entire preset detection period are indexed, starting from 0, and so on. Each point is used to divide the space into intervals. The k-th code element interval is... The corresponding sampling point index range is For example, the symbol width of the preset encoding sequence. sampling frequency Then the number of sampling points contained in each symbol interval. For a coded sequence of length 13, the entire preset detection period is divided into 13 consecutive symbol intervals. The 0th interval corresponds to sampling points 0 to 199, the 1st interval corresponds to sampling points 200 to 399, and so on.

[0171] For each symbol interval, calculate the symbol energy of the high-frequency voltage component of the bus located in that interval, and calculate the symbol energy of the high-frequency current component corresponding to each feeder.

[0172] The symbol energy refers to the sum of the squares of the amplitudes of all sampling points of a signal within a certain symbol interval, and is used to quantify the energy of the signal within that time period.

[0173] Specifically, for the k-th symbol interval, the system extracts the high-frequency voltage component of the bus within that interval. Calculate the symbol energy of all sampling points. The summation range n covers all sampling point indices of the k-th symbol interval. Similarly, for the i-th feeder, its high-frequency current component within that interval is extracted. Calculate the symbol energy of all sampling points. For example, for the k=2th symbol interval (corresponding to sampling points 400 to 599), the system calculates the sum of squares of the amplitudes of the 200 bus high-frequency voltage component sampling points within this interval, obtaining... Similarly, calculate the energy of the first feeder in this interval. .

[0174] Based on the symbol energy of the high-frequency voltage component of the bus and the symbol energy of the high-frequency current component of each feeder, the energy changes of the zero-sequence voltage of the bus and the zero-sequence current of each feeder in each symbol interval relative to a preset reference are calculated.

[0175] The energy change refers to the difference between the signal energy of the current symbol interval and a preset reference energy value, used to characterize the energy increment caused by active perturbation. The preset reference can be the average signal energy of several symbol intervals before the start of detection (called the baseline energy), or it can be the energy value of the previous symbol interval.

[0176] Specifically, the system first determines a preset benchmark. In one implementation, the preset benchmark is the time before the detection begins. The average signal energy within each symbol interval (called the baseline interval). Calculate the bus baseline energy. Where m is the index of the baseline interval (e.g., Similarly, calculate the baseline energy for each feeder. Therefore, in the k-th symbol interval, the change in the bus energy is... The energy change of the i-th feeder For example, the baseline interval is set to 5 symbols before the start of the probe ( The average symbol energy of the high-frequency voltage components of the bus within these five intervals is calculated as... For the 0th symbol interval within the preset detection period, its energy change... Assuming (Unit: Volts squared second) ), ,but .

[0177] For each feeder and each symbol interval, the normalized energy response value of the feeder in the corresponding symbol interval is obtained by dividing the energy change of the feeder by the sum of the energy change of the bus and a small positive number to prevent division by zero.

[0178] The normalized energy response value is a dimensionless ratio, representing the proportion of the energy response of a feeder to an active disturbance relative to the overall energy response of the bus within a specific symbol interval. A small positive number (often denoted as ) is introduced to prevent division by zero. This is to avoid changes in the busbar energy. The problem of computational overflow occurs when the value is zero or close to zero.

[0179] Specifically, for the normalized energy response value of the i-th feeder in the k-th symbol interval Calculate according to the following formula: .in, It is the energy change of the feeder in the k-th interval. It is the energy change of the busbar in the k-th interval. It is a very small positive number, for example For example, for the first feeder (i=1) in the second symbol interval (k=2), assuming the calculation yields... , , Then the normalized energy response value .

[0180] According to the symbol interval order, all normalized energy response values ​​corresponding to each feeder are combined to form a normalized energy response sequence.

[0181] The normalized energy response sequence is a one-dimensional ordered array consisting of the normalized energy response values ​​of a feeder across all symbol intervals (typically corresponding to the entire coding sequence length) arranged in chronological order. This sequence characterizes the dynamic response pattern of the feeder relative to the bus energy variation throughout the entire active detection period.

[0182] Specifically, for a total of The preset detection period for each symbol interval, and the normalized energy response sequence of the i-th feeder. It is a length of vector: The system constructs such a sequence independently for each feeder. For example, for a feeder of length 13 ( The encoded sequence of the system, which is the normalized energy response sequence constructed for the first feeder. Contains 13 elements: Each of them (k=0 to 12) are all normalized energy response values ​​calculated using the method described above.

[0183] Therefore, according to the above implementation method, the system can transform the original voltage and current time-domain signals into a series of normalized sequences characterizing the relative response strength of each feeder. This construction process effectively suppresses common-mode interference caused by factors such as grid common fluctuations and unstable amplitude of injected signals, and highlights the difference in energy response between faulty and non-faulty feeders under disturbances, providing high-quality, interference-resistant input data for subsequent accurate fault feature identification using matched filtering technology.

[0184] In some embodiments, a matched filter is applied to the normalized energy response sequence using a matching template corresponding to a preset coding sequence to obtain the matched filter output value of each feeder, including: Based on the preset encoding sequence, a matching template sequence that has undergone mean removal processing is generated.

[0185] Mean reduction is a signal preprocessing operation that subtracts the arithmetic mean of the original sequence, resulting in a mean of zero. This eliminates the influence of DC components or fixed bias, allowing subsequent matched filtering to focus on detecting sequence variation patterns and thus improving the sensitivity to coded correlation features. The matched template sequence, obtained after mean reduction, is a reference sequence used for cross-correlation with the normalized energy response sequence.

[0186] Specifically, the system first reads a preset encoded sequence c, which is a sequence containing... A discrete sequence of symbols, for example Then, the system calculates the arithmetic mean of the sequence. Where k ranges from 0 to Finally, by subtracting this average from each element of the original sequence, a matching template sequence g is generated, i.e. For example, the default encoding sequence is a simple 5-bit sequence: The mean of this sequence After mean removal, the resulting matching template sequence .

[0187] For each feeder, the matched filtering result of the normalized energy response sequence and the matched template sequence of the feeder is calculated, and the matched filtering result under different time delays is calculated traversed within the preset time delay offset range.

[0188] In discrete signal processing, the matched filtering result is typically obtained by calculating the cross-correlation function between the normalized energy response sequence and the matched template sequence. This result quantifies the similarity between the two sequences under different relative time delays. The preset time delay offset range is a finite interval centered at zero, containing both positive and negative integers, for example... This is used to accommodate any small time alignment deviations that may exist between the normalized energy response sequence and the matching template sequence. The traversal calculation refers to, within this time delay offset range, sequentially sliding the matching template sequence relative to the normalized energy response sequence for each possible time delay value, and calculating the cross-correlation value once for each.

[0189] Specifically, for the i-th feeder, its normalized energy response sequence is: The matching template sequence is Let the preset time delay offset range be... For each integer delay offset within this range The system calculates the matched filter result (i.e., the cross-correlation value) under this time delay. .when When positive, it indicates that the template sequence g is relative to the response sequence. Slide forward (time advance) The calculation is performed at each location, and boundary conditions may need to be handled during the calculation. A common calculation method is: The range of the summation variable k needs to be ensured. The index is within the valid range (0 to 1). For example, the normalized energy response sequence of the i-th feeder. Length is 5: The matching template sequence g is as described in the example above. Preset time delay offset range. The system needs to calculate. .

[0190] calculate : No offset.

[0191] .

[0192] calculate : Template g relative to Slide forward 1 position. At this point, the effective range of k must satisfy k-1 within... That is, k in . .

[0193] calculate : Template g relative to Slide one position to the right. At this point, the effective range of k must satisfy k+1 within... That is, k in . .

[0194] The maximum value among all the matched filter results calculated for the feeder under all time delays is determined as the matched filter output value of the feeder.

[0195] The matched filter output value is a scalar representing the highest similarity that the response sequence of the feeder and the matched template sequence can achieve under all possible small time alignment deviations. This maximum value is the core feature for determining whether the feeder is a faulty feeder.

[0196] Specifically, after completing all tasks within the preset time delay offset range... Matched filtering results of values After calculation, the system compares all of these. The numerical value. Usually, the comparison is... absolute value This is because matched filtering focuses on the strength of the correlation rather than its sign. The system finds the maximum value among these absolute values, denoted as . .this This is determined to be the matched filter output value for that feeder. For example, continuing from the previous example, the calculated value is... After taking the absolute value, they are respectively The maximum value is 1.84. Therefore, the system determines the matched filter output value of this feeder to be 1.84.

[0197] Therefore, according to the above implementation method, the system can calculate a quantitative index—the matched filter output value—for each feeder, characterizing the degree to which its response matches the known injected coded waveform. This processing effectively enhances the signal components related to the coded features and suppresses unrelated noise through cross-correlation operations; it overcomes minor synchronization errors that may exist in the actual system through traversal search within a preset time delay range. Ultimately, the matched filter output value of the faulty feeder will be significantly higher than that of the non-faulty feeder because it has the strongest response to coded disturbances, providing a clear and reliable criterion for accurate subsequent determination of the faulty feeder.

[0198] Figure 6 This is a structural block diagram of a high-resistance grounding fault location system in a low-resistance grounding system according to an embodiment of the present invention.

[0199] like Figure 6 As shown, the high-resistance grounding fault location system of this low-resistance grounding system includes: The data acquisition and acquisition module 210 is used to acquire electrical quantity data of the low-resistance grounding system and obtain a set of characteristic parameters from the acquired electrical quantity data. The set of characteristic parameters includes noise intensity indicators.

[0200] The frequency band determination module 220 is used to identify the main noise frequency band from the bus zero-sequence voltage of electrical quantity data before starting active detection, and to determine the target high-frequency detection frequency band according to the predetermined frequency band selection rules and the spectrum after removing the main noise frequency band.

[0201] The disturbance injection module 230 is used to control the neutral point to ground bypass branch to switch between at least two different frequency selective impedance levels within a preset detection period according to a preset coding sequence when the preset start-up conditions are met.

[0202] The response processing module 240 is used to extract high-frequency components located in the target high-frequency detection band from electrical quantity data, in order to construct the normalized energy response sequence of each feeder.

[0203] The matched filtering module 250 is used to perform matched filtering on the normalized energy response sequence using a matched template corresponding to the preset coding sequence, so as to obtain the matched filtering output value of each feeder.

[0204] The fault determination and safety control module 260 is used to determine the feeder with the largest matched filter output value and that meets the preset significant conditions as the faulty feeder.

[0205] According to embodiments of the present invention, the above-described method of the present invention can be applied to a computer device and a readable storage medium.

[0206] Figure 7 A schematic block diagram of a computer device 600 that can be used to implement embodiments of the present invention is shown. The computer device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The computer device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0207] like Figure 7 As shown, the computer device 600 includes a computing unit 601, which can perform various appropriate actions and processes based on a computer program stored in a read-only memory (ROM) 602 or a computer program loaded from a storage unit 608 into a random access memory (RAM) 603. The RAM 603 may also store various programs and data required for the operation of the computer device 600. The computing unit 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0208] Multiple components in computer device 600 are connected to I / O interface 605, including: input unit 606, such as keyboard, mouse, etc.; output unit 607, such as various types of monitors, speakers, etc.; storage unit 608, such as disk, optical disk, etc.; and communication unit 609, such as network card, modem, wireless transceiver, etc. Communication unit 609 allows computer device 600 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0209] The computing unit 601 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 performs the various methods and processes described above, such as a method for selecting a high-resistance ground fault in a low-resistance grounding system. For example, in some embodiments, a method for selecting a high-resistance ground fault in a low-resistance grounding system can be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program can be loaded and / or installed on the computer device 600 via ROM 602 and / or communication unit 609. When the computer program is loaded into RAM 603 and executed by the computing unit 601, one or more steps of the method for selecting a high-resistance ground fault in a low-resistance grounding system described above can be performed. Alternatively, in other embodiments, the computing unit 601 may be configured by any other suitable means (e.g., by means of firmware) to perform a high-resistance ground fault location method for a low-resistance grounding system.

[0210] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0211] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0212] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0213] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0214] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0215] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0216] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.

[0217] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for selecting the fault location of a high-resistance grounding system in a low-resistance grounding system, characterized in that, include: Collect electrical quantity data of a low-resistance grounding system, and obtain a set of characteristic parameters from the collected electrical quantity data, the set of characteristic parameters including noise intensity index; Before initiating active detection, the main noise frequency band is identified from the bus zero-sequence voltage of the electrical quantity data, and the target high-frequency detection frequency band is determined according to the predetermined frequency band selection rules and the spectrum after removing the main noise frequency band. In response to the fulfillment of preset start-up conditions, within a preset detection period, the neutral point to ground bypass branch is controlled to switch between at least two different frequency selective impedance levels according to a preset coding sequence. High-frequency components located within the target high-frequency detection band are extracted from the electrical quantity data to construct the normalized energy response sequence of each feeder; Using a matching template corresponding to the preset coding sequence, the normalized energy response sequence is subjected to matched filtering to obtain the matched filtering output value of each feeder. Among all feeders, the feeder with the largest matched filter output value and that meets the preset significant condition is identified as the faulty feeder.

2. The method according to claim 1, characterized in that, The step of identifying the main noise frequency band from the bus zero-sequence voltage of the electrical quantity data includes: Based on the bus zero-sequence voltage data in the electrical quantity data, calculate the power spectral density of the bus zero-sequence voltage; Based on the power spectral density, determine the set of characteristic frequencies of the dominant noise sources in the current system; Based on the set of characteristic frequencies, construct one or more candidate noise frequency bands; Traverse the one or more candidate noise frequency bands, calculate the signal energy in each candidate noise frequency band, and compare the signal energy with a preset reference energy threshold; Based on the comparison results, from the one or more candidate noise frequency bands, the candidate noise frequency band that meets the preset energy comparison conditions is determined as the main noise frequency band.

3. The method according to claim 2, characterized in that, The step of determining the target high-frequency detection band based on predetermined frequency band selection rules and the spectrum after removing the main noise frequency band includes: In the spectrum after removing the main noise frequency band, multiple continuous candidate frequency bands with preset bandwidths are generated; Traverse the multiple candidate frequency bands, and for each candidate frequency band, calculate the average power spectral density of the candidate frequency band, and calculate the minimum frequency interval between the center frequency of the candidate frequency band and the center frequencies of each noise frequency band determined by the set of characteristic frequencies. According to the frequency band selection rules, the target high-frequency detection frequency band is determined from the plurality of candidate frequency bands based on the average power spectral density and the minimum frequency interval calculated for each candidate frequency band.

4. The method according to claim 3, characterized in that, The frequency band selection rule is configured to be used for: From the plurality of candidate frequency bands, determine the candidate frequency band with the lowest average power spectral density; Determine whether any candidate frequency band satisfies the density similarity threshold condition. The density similarity threshold condition means that the difference between the average power spectral density of the candidate frequency band and the average power spectral density of the candidate frequency band with the lowest average power spectral density is less than a preset density similarity threshold. If it does not exist, the candidate frequency band with the lowest average power spectral density is determined as the target high-frequency detection frequency band; If it exists, then from all candidate frequency bands that satisfy the density similarity threshold condition, select the candidate frequency band that has the largest minimum frequency interval with the center frequency of each noise frequency band as the target high-frequency detection frequency band.

5. The method according to claim 1, characterized in that, The step of controlling the neutral point-to-ground bypass branch to switch between at least two different frequency-selective impedance levels according to a preset coding sequence includes: Before initiating active detection, an initiation condition judgment is performed. The initiation condition judgment includes: verifying whether the amplitude of the current bus zero-sequence voltage is within the preset voltage threshold range, and estimating whether the expected fluctuation value of the bus zero-sequence voltage caused by impedance level switching is lower than the preset safety limit based on the feature parameter set. If the result of the start condition judgment is that all conditions are met, the control process enters the preset detection cycle; Within the preset detection period, according to the preset encoding sequence, at the start time corresponding to each symbol, a switching command indicating the target impedance level is sent to the control module of the neutral point to ground bypass branch. In response to each received switching command, the control module drives the switching device in the neutral point-to-ground bypass branch to switch the current input impedance level of the neutral point-to-ground bypass branch to the target impedance level indicated by the switching command. According to the symbol width timing of the preset encoding sequence, the steps of sending switching instructions and driving switching device actions are repeatedly executed until all symbols in the preset encoding sequence have been traversed. Thus, through the switching from the current input impedance level to the target impedance level, a voltage or current disturbance signal corresponding to the preset encoding sequence is generated in the neutral point to ground loop.

6. The method according to claim 1, characterized in that, The step of extracting high-frequency components located within the target high-frequency detection band from the electrical quantity data to construct the normalized energy response sequence for each feeder includes: The bus zero-sequence voltage and each feeder zero-sequence current in the electrical quantity data are respectively subjected to bandpass filtering to extract the high-frequency voltage component of the bus and the high-frequency current component of each feeder located in the target high-frequency detection frequency band. The preset detection period is divided into multiple consecutive symbol intervals according to the symbol width of the preset coding sequence; For each of the symbol intervals, the symbol energy of the high-frequency voltage component of the bus located in that interval is calculated, and the symbol energy of the high-frequency current component corresponding to each feeder is calculated. Based on the symbol energy of the high-frequency voltage component of the bus and the symbol energy of the high-frequency current component of each feeder, the energy change of the zero-sequence voltage of the bus and the zero-sequence current of each feeder in each symbol interval relative to a preset reference is calculated. For each feeder and each symbol interval, the energy change of the feeder is divided by the sum of the energy change of the bus and a small positive number to prevent division by zero, to obtain the normalized energy response value of the feeder in the corresponding symbol interval. The normalized energy response values ​​corresponding to each feeder are combined in the order of symbol intervals to form the normalized energy response sequence.

7. The method according to claim 1, characterized in that, The step of performing matched filtering on the normalized energy response sequence using a matching template corresponding to the preset coding sequence to obtain the matched filter output value of each feeder includes: Based on the preset encoding sequence, a matching template sequence that has undergone mean removal processing is generated; For each feeder, the normalized energy response sequence of the feeder and the matching template sequence are calculated, and the matching filtering results under different time delays are calculated traversally within the preset time delay offset range. The maximum value among all the matched filter results calculated for the feeder under all time delays is determined as the matched filter output value for the feeder.

8. A high-resistance grounding fault location system for a low-resistance grounding system, characterized in that, include: The data acquisition and acquisition module is used to collect electrical quantity data of the low-resistance grounding system and obtain a set of characteristic parameters from the collected electrical quantity data, the set of characteristic parameters including noise intensity index; The frequency band determination module is used to identify the main noise frequency band from the bus zero-sequence voltage of the electrical quantity data before starting active detection, and to determine the target high-frequency detection frequency band according to the predetermined frequency band selection rules and the spectrum after removing the main noise frequency band. The disturbance injection module is used to control the neutral point to ground bypass branch to switch between at least two different frequency selective impedance levels according to a preset coding sequence within a preset detection period when the preset start-up conditions are met. The response processing module is used to extract high-frequency components located in the target high-frequency detection band from the electrical quantity data, in order to construct the normalized energy response sequence of each feeder. The matched filtering module is used to perform matched filtering on the normalized energy response sequence using a matching template corresponding to the preset coding sequence, so as to obtain the matched filtering output value of each feeder. The fault diagnosis and safety control module is used to identify the feeder with the largest matched filter output value and that meets the preset significant conditions as the faulty feeder.

9. A computer device, characterized in that, include: At least one processor; and a memory that is communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-7.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, in, Computer instructions are used to cause a computer to perform the method according to any one of claims 1-7.

Images