Branch box grounding fault line selection method based on zero sequence current amplitude ratio

Abstract

This invention discloses a method for selecting ground fault lines in branch boxes based on zero-sequence current amplitude comparison, belonging to the field of power distribution fault detection technology. The method includes: collecting real-time zero-sequence current, inherent electrical parameters, historical normal leakage baseline, and real-time load parameters of each outgoing branch; constructing a dynamic zero-sequence current judgment threshold; and removing the normal leakage component to obtain the pure fault characteristic zero-sequence current amplitude. A comprehensive equivalent amplitude is obtained through steady-state sampling and time-series weighted calculation. A graded differentiated amplitude comparison logic is performed for four types of ground faults to initially screen candidate fault branches. A secondary verification is then performed using the bus zero-sequence current vector. After meeting the consistency requirements of deviation and continuous steady-state sampling time window, the selection result is output. Finally, the fault branch, type, and severity are determined based on the amplitude level and fluctuation characteristics, simultaneously realizing load disturbance discrimination and graded storage of fault records. This invention can accurately identify different types of ground faults, improving the reliability and practicality of branch box fault selection.

Description

Technical Field

[0001] This invention belongs to the field of power distribution fault detection technology, specifically a branch box grounding fault selection method based on zero-sequence current amplitude comparison. Background Technology

[0002] Distribution branch boxes are critical nodes in distribution networks, with numerous outgoing branches and complex lines, making them prone to single-phase grounding faults. Failure to quickly and accurately select the fault line can expand the fault area and threaten power grid safety. Existing grounding fault selection methods often rely on fixed threshold criteria, neglecting factors such as branch cable length, insulation aging, and load fluctuations, resulting in poor dynamic adaptability and a high risk of misjudgment or missed detection. Constant leakage current and load switching interference in the distribution network can mask true fault characteristics, making it difficult to identify weak faults such as high resistance, arcing, and intermittent faults. Traditional amplitude comparison methods do not classify fault types, lacking adaptability to different fault characteristics and resulting in low fault selection accuracy. Furthermore, most methods rely solely on amplitude as a criterion, lacking vector verification, making them susceptible to system unbalanced currents and exhibiting poor stability. In addition, existing technologies cannot effectively distinguish between fault disturbances and load surges, often leading to malfunctions; the fault record storage and uploading mechanism is inadequate, hindering fault tracing and remote monitoring. With the development of intelligent distribution networks, there is an urgent need for a line selection method that is highly adaptive, has good anti-interference capabilities, and can accurately identify multiple types of grounding faults, so as to improve the efficiency of fault handling and operational reliability of distribution branch boxes. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention proposes a branch box grounding fault selection method based on zero-sequence current amplitude comparison. It collects real-time zero-sequence current, inherent electrical parameters, historical normal leakage baseline, and real-time load parameters for each outgoing branch, constructs a dynamic zero-sequence current judgment threshold, and removes the normal leakage component to obtain the pure fault characteristic zero-sequence current amplitude. A comprehensive equivalent amplitude is obtained through steady-state sampling and time-series weighted calculation. A graded differentiated amplitude comparison logic is performed for four types of grounding faults to initially screen candidate fault branches. Then, the bus zero-sequence current vector is introduced for secondary verification. The selection result is output after meeting the consistency requirements of deviation and continuous steady-state sampling time window. Finally, the fault branch, type, and severity are determined based on the amplitude level and fluctuation characteristics, simultaneously realizing load disturbance discrimination and graded storage of fault records. This invention can accurately identify different types of grounding faults, improving the reliability and practicality of branch box fault selection.

[0004] To achieve the above objectives, the present invention provides the following technical solution: A branch box grounding fault location method based on zero-sequence current amplitude ratio includes: Acquire real-time zero-sequence current data, inherent electrical parameters, historical normal leakage baseline data, and real-time load parameters of all outgoing branches of the distribution branch box; Based on inherent electrical parameters and real-time load parameters, a dynamic zero-sequence current determination threshold is constructed for each outgoing branch. Using the historical normal leakage current baseline data, the normal leakage current component is extracted from the real-time zero-sequence current data to obtain the amplitude of the pure fault characteristic zero-sequence current. The amplitude of the zero-sequence current characteristic of the pure fault is sampled using a preset steady-state sampling time window. The comprehensive equivalent amplitude of each branch is obtained through time-weighted calculation. For four types of faults, namely metallic grounding, high-resistance grounding, arcing grounding, and intermittent grounding, a graded differential amplitude comparison logic is executed to initially screen out candidate fault branches. Introducing real-time zero-sequence current data from the branch box busbar, a secondary vector deviation verification is performed on the candidate fault branch. If the vector deviation is less than the preset verification threshold and the judgment results of multiple consecutive steady-state sampling time windows are consistent, a fault selection completion signal is output. Based on the fault selection completion signal, the amplitude level is determined by the comprehensive equivalent amplitude, the fluctuation characteristics are determined by the change of the comprehensive equivalent amplitude within multiple consecutive steady-state sampling time windows, and the fault branch number, ground fault type and fault severity are determined and stored.

[0005] Specifically, the acquisition of real-time zero-sequence current data, inherent electrical parameters, historical normal leakage current baseline data, and real-time load parameters of all outgoing branches of the distribution branch box includes: By deploying zero-sequence current transformers on each outgoing branch in the distribution branch box, the instantaneous zero-sequence current waveform of each branch is collected. A fast Fourier transform is performed on the instantaneous zero-sequence current waveform to extract the power frequency component amplitude, and the power frequency component amplitude is used as the real-time zero-sequence current data of the branch. The cable length, cable cross-sectional area, insulation material type, and insulation aging coefficient of each outgoing branch are retrieved from the local storage unit of the distribution branch box, which together constitute the inherent electrical parameters of the branch. Extract the average zero-sequence current amplitude under the condition of no grounding fault from the historical operation database of the distribution branch box, and use this average value as the historical normal leakage current baseline data. The active power, reactive power, and power factor angle at the current moment are collected by the load monitoring terminal installed at the end of each outgoing branch, which together constitute the real-time load parameters.

[0006] Specifically, the dynamic zero-sequence current determination threshold for each outgoing branch includes: The inherent electrical parameters are input into the insulation impedance evaluation model pre-set in the branch box controller. The insulation impedance evaluation model outputs the distributed capacitance and leakage resistance parameters of each outgoing branch under the current insulation state. The output distributed capacitance parameter, the leakage resistance parameter, and the real-time load parameter are input into the dynamic threshold generator. The dynamic threshold generator corrects the distributed capacitance parameter based on the power factor angle in the real-time load parameter to obtain the corrected distributed capacitance parameter. The corrected distributed capacitance parameter and the leakage resistance parameter are then vector-summed to obtain the dynamic zero-sequence current determination threshold. The dynamic threshold generator monitors the changes in the real-time load parameters and iterates repeatedly whenever the active power, reactive power, or power factor angle changes, updating the dynamic zero-sequence current determination threshold in real time.

[0007] Specifically, the process of generating the amplitude of the zero-sequence current characteristic of the pure fault includes: The historical normal leakage baseline data is defined as the intrinsic zero-sequence current component of each outgoing branch in a healthy state. The real-time zero-sequence current data is input into a moving average filter to smooth the real-time zero-sequence current data of each branch, and the measured zero-sequence current amplitude after removing high-frequency noise is output. The measured zero-sequence current amplitude of each branch is subtracted from the intrinsic zero-sequence current component corresponding to that branch to obtain the difference. The absolute value of the difference is then taken as the pure fault characteristic zero-sequence current amplitude of that branch.

[0008] Specifically, the process of generating the comprehensive equivalent amplitude includes: Set multiple consecutive steady-state sampling time windows, with the duration of each steady-state sampling time window being an integer multiple of the power frequency period; Within each steady-state sampling time window, the amplitude of the pure fault characteristic zero-sequence current is sampled at equal intervals to form multiple amplitude sampling points arranged in time sequence within the steady-state sampling time window; The multiple amplitude sampling points are input into a time-series weighted arithmetic unit. The time-series weighted arithmetic unit performs a weighted summation on all amplitude sampling points to obtain a weighted sum value. The weighted sum value is then divided by the total number of sampling points within the steady-state sampling time window, and the resulting quotient is used as the comprehensive equivalent amplitude of each branch within the steady-state sampling time window. The time-series weighted arithmetic unit assigns a larger weighting coefficient to sampling points that are closer to the current time.

[0009] Specifically, the screening process for the candidate faulty branch includes: Obtain the comprehensive equivalent amplitude of each branch, and obtain the pre-stored metallic grounding threshold, high-resistance grounding threshold, arc grounding threshold and background noise threshold; Compare the relationship between the comprehensive equivalent amplitude of each branch and the metallic grounding threshold, the high-resistance grounding threshold, and the arcing grounding threshold; if the comprehensive equivalent amplitude is greater than or equal to the metallic grounding threshold, it is determined to be a metallic grounding fault; if it is less than the metallic grounding threshold but greater than or equal to the high-resistance grounding threshold, it is determined to be a high-resistance grounding fault; if it is less than the high-resistance grounding threshold but greater than or equal to the arcing grounding threshold, it is determined to be an arcing grounding fault; if the comprehensive equivalent amplitude alternately crosses each threshold within multiple consecutive steady-state sampling time windows, it is determined to be an intermittent grounding fault. If the fault is determined to be a metallic grounding fault, the branch with the largest comprehensive equivalent amplitude is selected as the candidate fault branch. If the fault is determined to be a high-resistance grounding fault, branches with comprehensive equivalent amplitudes lower than the background noise threshold are first eliminated, and then the three branches with the largest comprehensive equivalent amplitudes among the remaining branches are selected as candidate fault branches. If the fault is determined to be an arc grounding fault, the mean and standard deviation of the comprehensive equivalent amplitudes of all branches are calculated, and the branches with comprehensive equivalent amplitudes greater than the mean plus twice the standard deviation are selected as candidate fault branches. If the fault is determined to be an intermittent grounding fault, the branch number with the largest comprehensive equivalent amplitude in each steady-state sampling time window is recorded, and the branch that appears most frequently is selected as the candidate fault branch.

[0010] Specifically, the introduction of real-time zero-sequence current data from the branch box busbar for secondary vector deviation verification of candidate fault branches includes: The amplitude and phase of the real-time zero-sequence current of the busbar are collected by the zero-sequence current transformer installed at the busbar of the distribution branch box, forming the measured zero-sequence current vector of the busbar. The real-time zero-sequence current amplitude and phase of the candidate fault branch are collected from the zero-sequence current transformer of the branch to form the zero-sequence current vector of the candidate fault branch. All remaining branches other than the candidate fault branches are marked as non-candidate fault branches. The intrinsic zero-sequence current component amplitude and corresponding phase of each non-candidate fault branch in the healthy state are obtained. After phase alignment, they are summed to obtain the vector sum of non-candidate fault branches. The zero-sequence current vector of the candidate fault branch is added to the vector sum of the non-candidate fault branches to obtain the theoretical bus zero-sequence current vector. Calculate the vector deviation between the measured bus zero-sequence current vector and the theoretical bus zero-sequence current vector, and use the vector deviation as the verification basis.

[0011] Specifically, the step of outputting a fault selection completion signal if the vector deviation is less than a preset verification threshold and the judgment results of multiple consecutive steady-state sampling time windows are consistent includes: The vector deviation is obtained, and a preset verification threshold dynamically set according to the rated voltage level of the branch box is obtained; Determine whether the vector deviation is less than the preset verification threshold. If not, terminate the verification and wait for the next round of fault determination. If yes, proceed to the consistency verification of multiple consecutive steady-state sampling time windows. Within N consecutive steady-state sampling time windows, the candidate fault branch number of each steady-state sampling time window is recorded sequentially; where N is a preset integer greater than or equal to 2. Determine whether the candidate fault branch numbers of the N consecutive steady-state sampling time windows are completely consistent. If not, terminate the verification and clear all recorded steady-state sampling time window numbers; if yes, generate a fault selection completion signal.

[0012] Specifically, the determination and storage of the fault branch number, ground fault type, and fault severity includes: After receiving the fault selection completion signal, the candidate fault branch number is read and used as the fault branch number. Read the comprehensive equivalent amplitude corresponding to the fault branch number, compare the comprehensive equivalent amplitude with the preset amplitude level range, and determine the amplitude level of the branch; Read the M comprehensive equivalent amplitude values ​​of the fault branch number within M consecutive steady-state sampling time windows, calculate the difference between the comprehensive equivalent amplitude values ​​of two adjacent steady-state sampling time windows to obtain M-1 difference values, and calculate the range of the M-1 difference values. Determine the fluctuation characteristics of the branch based on the range of the range; M is a preset integer greater than or equal to 3. Input the determined ground fault type, the determined amplitude level, and the determined fluctuation characteristics into the preset fault severity determination matrix, and output the corresponding fault severity level. The fault branch number, the fault severity level, and the grounding fault type are packaged and stored in the local non-volatile memory of the branch box, and simultaneously uploaded to the remote monitoring master station through the communication interface.

[0013] Specifically, it also includes a step of detecting abrupt changes in the power factor angle in real-time load parameters to help determine the nature of the fault: While measuring the amplitude of the zero-sequence current characteristic of the pure fault, the power factor angle collected within each steady-state sampling time window is acquired simultaneously. Calculate the difference between the power factor angle in the current steady-state sampling time window and the power factor angle in the previous steady-state sampling time window, and take the absolute value as the characteristic quantity of load change. Obtain a preset angle mutation threshold and a preset amplitude fluctuation multiple threshold, compare the load mutation characteristic quantity with the angle mutation threshold, and simultaneously compare the ratio of the pure fault characteristic zero-sequence current amplitude in the current steady-state sampling time window to the pure fault characteristic zero-sequence current amplitude in the previous steady-state sampling time window with the amplitude fluctuation multiple threshold. If the load change characteristic exceeds the angle change threshold and the ratio also exceeds the amplitude fluctuation multiple threshold, it is determined that the current electrical quantity change originates from load switching, thereby suppressing the execution of the fault selection process. If the load change characteristic exceeds the angle change threshold, but the ratio does not exceed the amplitude fluctuation multiple threshold, it is determined that there is a ground fault superimposed on the load disturbance, and the fault selection process continues accordingly.

[0014] Specifically, it also includes the step of managing fault log information using a tiered storage strategy: After storing the fault record in the local non-volatile memory, obtain the preset first priority capacity value, second priority capacity value and third priority upload rate; The severity of the determined fault is mapped to a priority level, and the faults are stored in the first priority storage area, the second priority storage area, and the third priority storage area in descending order of priority level. The first priority storage area stores the K most recent fault records in a circular overwrite method, where K is a preset positive integer. Once the first priority storage area is full of K records, subsequent fault records are stored in the second priority storage area and stored using a page write method until the second priority storage area is full of the preset second priority capacity value. Once the first and second priority storage areas are full, subsequent fault records are uploaded to the remote monitoring master station in batches through the communication interface at the limited upload rate, forming the third priority storage area. After the branch box is powered off and restarted locally, the fault records in the first priority storage area and the second priority storage area are read, and the two are merged and loaded into the running memory of the branch box controller to form a fault record index table.

[0015] Compared with the prior art, the beneficial effects of the present invention are: By constructing dynamic thresholds, stripping normal leakage components, and performing time-series weighted calculations, interference from cable parameters, insulation aging, real-time load, and normal leakage is effectively eliminated, significantly improving the purity of fault feature extraction. It can accurately adapt to four typical grounding faults: metallic, high-resistance, arcing, and intermittent. By adopting graded differentiated amplitude comparison logic, it reduces the misjudgment and missed judgment rates of weak and complex faults. The accuracy and adaptability of line selection are significantly better than traditional fixed threshold methods.

[0016] By introducing secondary verification of the zero-sequence current vector of the busbar and consistency judgment of multiple consecutive steady-state sampling time windows, combined with the load change identification mechanism, the ground fault and load switching disturbance can be effectively distinguished to avoid false operation. At the same time, the fault branch, type and severity can be automatically determined, and a fault record hierarchical storage and remote upload strategy can be provided to improve fault tracing and operation and maintenance efficiency, and enhance the stability, anti-interference ability and intelligence level of ground fault selection of distribution branch box. Attached Figure Description

[0017] Fig. 1 This is a schematic diagram of the branch box grounding fault selection method based on zero-sequence current amplitude according to the present invention; Fig. 2 This is a flowchart illustrating the principle of the branch box grounding fault selection method based on zero-sequence current amplitude of the present invention. Detailed Implementation

[0018] Please see Figs. 1-2 The present invention provides an embodiment of a branch box grounding fault selection method based on zero-sequence current amplitude ratio, the method comprising S1~S4, as follows: S1: Obtain real-time zero-sequence current data, inherent electrical parameters, historical normal leakage baseline data, and real-time load parameters for all outgoing branches of the distribution branch box; This embodiment takes a typical outdoor ring network distribution branch box in a 10kV urban power distribution network as the application object. The branch box is equipped with a total of 8 outgoing branches, which correspond to different load types such as surrounding residential areas, commercial buildings, municipal street lights, and public parking lots. Each outgoing branch is laid with cross-linked polyethylene insulated power cables. The cable laying environment includes three forms: direct burial, conduit, and cable trench. During long-term operation, it is susceptible to single-phase grounding faults due to factors such as soil moisture, ground subsidence, external force damage, and insulation aging.

[0019] The acquisition of real-time zero-sequence current data, inherent electrical parameters, historical normal leakage baseline data, and real-time load parameters for all outgoing branches of the distribution branch box includes: S1.1: By deploying zero-sequence current transformers on each outgoing branch in the distribution branch box, the instantaneous zero-sequence current waveform of each branch is collected, a fast Fourier transform is performed on the instantaneous zero-sequence current waveform, the power frequency component amplitude is extracted, and the power frequency component amplitude is used as the real-time zero-sequence current data of the branch. The fast Fourier transform is a prior art in this field and is not an inventive solution of this application, so it will not be described in detail here. In this embodiment, each of the eight outgoing branches of the distribution branch box is equipped with a high-precision, wide-frequency response, and low-power zero-sequence current transformer. The transformer adopts an open-type installation structure, which can be directly mounted on the outside of the three-phase cable of the outgoing branch without disconnecting the cable conductor. It can sense the zero-sequence current signal generated during the operation of the branch in real time. Its measurement accuracy meets the 0.5-level metrological standard, and its frequency response range covers 15Hz~5000Hz. It can completely capture the zero-sequence current waveform generated by different fault types such as metallic grounding, high-resistance grounding, arcing grounding, and intermittent grounding, and can effectively avoid problems such as high-frequency component attenuation and phase shift. The zero-sequence current transformer is connected to the branch box through a shielded twisted-pair cable. The two ends of the shielding layer are reliably grounded, effectively resisting the influence of harsh electromagnetic environments such as electromagnetic interference, harmonic interference, and surge impact in the power distribution room, and ensuring the integrity and authenticity of the instantaneous zero-sequence current waveform acquisition.

[0020] Furthermore, the data acquisition port of the branch box continuously acquires the instantaneous zero-sequence current waveform of each outgoing branch at a sampling rate of 32 points per power frequency cycle, with a sampling frequency of 1600Hz, ensuring that the acquired waveform data can completely reflect the instantaneous change law of the zero-sequence current. After acquisition, the instantaneous zero-sequence current waveform data is transmitted to the branch box controller. The branch box controller performs fast Fourier transform processing on the instantaneous zero-sequence current waveform data, decomposing the fundamental component, harmonic component, and DC component in the zero-sequence current. The fundamental component is the 50Hz power frequency component, which is the core component reflecting the characteristics of grounding faults. The branch box controller automatically filters out the 3rd, 5th, and 7th harmonic components and DC offset components, extracting only the effective value amplitude of the power frequency component. This amplitude data is marked as the real-time zero-sequence current data of the corresponding branch, and a timestamp accurate to the millisecond level and branch number information are added and stored in the real-time data buffer of the controller.

[0021] S1.2: Retrieve the cable length, cable cross-sectional area, insulation material type, and insulation aging coefficient of each outgoing branch from the local storage unit of the distribution branch box, which together constitute the inherent electrical parameters of the branch. In this embodiment, the local storage unit of the distribution branch box uses an industrial-grade non-volatile flash memory chip, which does not lose data after power failure and can store the inherent electrical parameters of each outgoing branch for a long time. The parameter information is entered by on-site maintenance personnel through a handheld debugging terminal before the branch box is put into operation. For the eight outgoing branches, the inherent electrical parameters of each branch are stored and retrieved independently to avoid calculation errors caused by parameter confusion. The cable length is the actual laying length from the branch box outgoing end to the load-side distribution cabinet, accurate to 0.1m; the cable cross-sectional area is the nominal cross-sectional area of ​​the three-phase cable core wire, including 35mm². 2 50mm 2 70mm 2 120mm 2Four specifications are available; the insulation material is uniformly cross-linked polyethylene insulation, with some older branches using ethylene propylene rubber insulation. The basic characteristics of the insulation material, such as dielectric constant and dielectric loss factor, are pre-built into the controller's parameter library; the insulation aging coefficient is a dimensionless coefficient derived from a comprehensive evaluation of factors such as cable service life, operating temperature, load rate, and fault records, with a value range between 0.1 and 1.0. The closer the coefficient is to 1.0, the better the cable insulation condition; the closer the coefficient is to 0.1, the more severe the cable insulation aging and the greater the leakage current.

[0022] Furthermore, when the branch box controller initiates the fault selection process, it reads the above four parameters from the local storage unit in sequence according to the branch number, and encapsulates them into the inherent electrical parameter group of the corresponding branch. The parameter group is bound to the real-time zero-sequence current data and real-time load parameters one by one. Each branch uses exclusive inherent electrical parameters to avoid judgment deviations caused by the inability of unified parameters to adapt to different cable characteristics.

[0023] S1.3: Extract the average value of zero-sequence current amplitude under the condition of no grounding fault record in the past 72 hours from the historical operation database of the distribution branch box, and use the average value as the historical normal leakage current baseline data; In this embodiment, the branch box has a built-in historical operation database with a cyclic storage mechanism, which can save all the operation data for the most recent 30 days, including information such as zero-sequence current, load power, insulation status, and fault alarms for each branch. The data storage interval is 1 minute, which can meet the needs of historical data statistics, trend analysis, and baseline calculation. Before performing fault selection, the controller automatically filters the operation data of the past 72 hours in the historical operation database. First, it removes the data for the period with ground fault alarms, overcurrent alarms, and insulation abnormality alarms by using the fault flag bit, and only retains the zero-sequence current amplitude data under fault-free, alarm-free, and normal operating conditions.

[0024] Furthermore, for each outgoing branch, the branch box controller performs an arithmetic average calculation on the filtered effective zero-sequence current data, and determines the calculated average value as the historical normal leakage current baseline data for that branch. This baseline data represents the inherent leakage current level of the branch under healthy, fault-free, and undisturbed conditions, and is a key benchmark for distinguishing normal leakage current from fault zero-sequence current. Since the cable length, cross-sectional area, and insulation condition of different branches vary, the historical normal leakage current baseline data for each branch is different. For example, the baseline data of long-distance, small-cross-sectional area, and severely aged cable branches is relatively high; the baseline data of short-distance, large-cross-sectional area, and newly commissioned cable branches is relatively low. The branch box controller stores the baseline data of each branch separately and links it with the real-time data of the corresponding branch.

[0025] S1.4: The active power, reactive power and power factor angle at the current moment are collected by the load monitoring terminal installed at the end of each outgoing branch, which together constitute the real-time load parameters.

[0026] In this embodiment, an intelligent load monitoring terminal is installed in the load-side distribution cabinet of each outgoing branch. The terminal has the functions of measuring voltage, current, power and power factor, and the measurement accuracy meets the 0.5 standard. It can collect load operation data at the end of the branch in real time and upload the data to the distribution branch box controller through power line carrier or RS485 communication. The communication delay does not exceed 50ms, ensuring that the load parameters and zero-sequence current data are collected synchronously.

[0027] Furthermore, the load monitoring terminal collects the three-phase active power and three-phase reactive power of the branch in real time, calculates the total active power and total reactive power at the current moment, and calculates the power factor angle of the branch based on the phase difference between voltage and current. The power factor angle is in rad and reflects the inductive or capacitive characteristics of the load. The branch box controller receives the load data of 8 branches in real time through the communication interface, and combines the total active power, total reactive power, and power factor angle of each branch into the real-time load parameters of that branch. The parameter data is updated every 200ms, which can track the operating status such as load switching, load fluctuation, and power change in real time. For fluctuating loads such as residential areas and commercial buildings, the load monitoring terminal can automatically smooth short-term fluctuation data to avoid distortion of real-time load parameters caused by sudden load changes. For stable loads such as municipal street lights and parking lots, it can accurately collect constant load parameters.

[0028] S2: Based on inherent electrical parameters and real-time load parameters, construct a dynamic zero-sequence current determination threshold for each outgoing branch, and use the historical normal leakage current baseline data to extract the normal leakage current component from the real-time zero-sequence current data to obtain the amplitude of the pure fault characteristic zero-sequence current. The dynamic zero-sequence current determination threshold for each outgoing branch includes: S2.1: Input the inherent electrical parameters into the insulation impedance evaluation model pre-set in the branch box controller. The insulation impedance evaluation model outputs the distributed capacitance parameters and leakage resistance parameters of each outgoing branch under the current insulation state. Furthermore, the inherent electrical parameters received by the insulation impedance assessment model include four items: cable length, cable cross-sectional area, insulation material type, and insulation aging coefficient. These four parameters are fixed parameters entered into the local storage unit before the branch box is put into operation and do not change within a single model calculation cycle. Among them, the input value of cable length ranges from 50m to 1000m. In this embodiment, the cable lengths of the eight outgoing branch lines participating in the calculation are set to 120m, 280m, 350m, 420m, 580m, 650m, 720m, and 850m respectively. The input value of cable cross-sectional area includes 30mm². 2 50mm 2 70mm 2 120mm 2 Four standard specifications are used, and in this embodiment, the eight branches correspond to 50mm in sequence. 2 50mm 2 70mm 2 70mm 2 120mm 2 120mm 2 70mm 2 50mm 2 The insulation material types are divided into two categories: cross-linked polyethylene insulation and ethylene propylene rubber insulation. In this embodiment, the first six branches are cross-linked polyethylene insulation and the last two branches are ethylene propylene rubber insulation. The insulation aging coefficient is a dimensionless parameter with a value range limited to 0.1 to 1.0. The smaller the value, the more severe the insulation aging. In this embodiment, the insulation aging coefficients of the eight branches are set to 0.95, 0.9, 0.85, 0.8, 0.75, 0.7, 0.5, and 0.6, respectively. All parameters are normalized by the controller before being input into the model, mapping parameters with different dimensions and numerical ranges to the interval between 0 and 1. The normalization rule is to divide the cable length parameter by 1000 and the cable cross-sectional area parameter by 120. Since the insulation aging coefficient itself is in the 0~1 interval, the original value is retained unchanged. At the same time, the classification parameter such as the insulation material type is coded with a digital label. The insulation material type is directly represented by the number 1 for cross-linked polyethylene insulation and the number 2 for ethylene propylene rubber insulation. After processing, a standard input vector that the model can recognize is formed.

[0029] Furthermore, the model's internal structure is divided into four sequentially connected computational units: a parameter parsing module, a dielectric property matching module, a distributed capacitance calculation module, and a leakage resistance fitting module. Each unit uses a serial data transfer method, with the output of one unit directly serving as the input of the next. There is no parallel computation or jump logic, ensuring the continuity and traceability of the computation process. The standard input vector is split field by field, extracting four independent parameters in a fixed order: cable length, cable cross-sectional area, insulation material type, and insulation aging coefficient. The splitting process strictly follows the order of the input vector, without changing the physical meaning or numerical value of the parameters. After splitting, each parameter is individually passed to its corresponding processing unit. Specifically, cable length and cable cross-sectional area are passed to the distributed capacitance calculation module, insulation material type to the dielectric property matching module, and insulation aging coefficient to both the distributed capacitance calculation module and the leakage resistance fitting module.

[0030] Furthermore, the dielectric property matching module has a pre-stored table of fixed dielectric parameters for commonly used insulation materials in power distribution cables. This table is fixed during the model training phase. The relative permittivity of cross-linked polyethylene insulation is set to 2.3, and the dielectric loss factor is set to 0.0003. The relative permittivity of ethylene propylene rubber insulation is set to 2.8, and the dielectric loss factor is set to 0.0005. Based on the numerical values ​​of the insulation material type obtained from the analysis, the module automatically matches the corresponding relative permittivity and dielectric loss factor. In this embodiment, the first six branches are matched to obtain a relative permittivity of 2.3 and a dielectric loss factor of 0.0003, while the last two branches are matched to obtain a relative permittivity of 2.8 and a dielectric loss factor of 0.0005. After matching, the relative permittivity is output to the distributed capacitance calculation module, and the dielectric loss factor is output to the leakage resistance fitting module.

[0031] Furthermore, the distributed capacitance calculation module employs a classic analytical algorithm based on cable geometric parameters and dielectric properties. The calculation process consists of three steps: First, the analyzed cable length is multiplied by the cable cross-sectional area to obtain the effective conductive cross-sectional area-length product. Second, the product is multiplied by the relative permittivity output by the dielectric property matching module, and then multiplied by a fixed scaling factor. This scaling factor is set to 0.8 multiplied by 10 according to the standard operating parameters of the 10kV distribution system. -12 The third step is to divide the above calculation result by the insulation aging coefficient to obtain the final distributed capacitance parameter; in this embodiment, the third outgoing branch is used as an example, with a cable length of 350m and a cable cross-sectional area of ​​70mm². 2The insulation material is cross-linked polyethylene insulation with an insulation aging coefficient of 0.85. The distributed capacitance parameter obtained after calculation is 0.21μF. The other 7 branches are calculated sequentially according to the same algorithm. The final output distributed capacitance parameter range is controlled between 0.15μF and 0.3μF, which is consistent with the distributed capacitance characteristics of 10kV distribution cables in actual operation. Here, μF is the capacitance unit microfarad.

[0032] Furthermore, the leakage resistance fitting module employs a linear fitting algorithm based on insulation aging characteristics and dielectric loss. During training, the fitting slope is set to 5000 and the fitting intercept to 100. First, the dielectric loss factor output by the dielectric characteristic matching module is multiplied by a fixed system rated voltage, which is set to 10kV according to the 10kV power distribution system standard, to obtain the initial leakage conductance parameter. Then, the initial leakage conductance parameter is divided by the insulation aging coefficient to obtain the corrected leakage conductance parameter. Finally, the fitting slope is divided by the corrected leakage conductance parameter, and the fitting intercept is added to obtain the final leakage resistance parameter. In this embodiment, the third outgoing branch is taken as an example, with a dielectric loss factor of 0.0003 and an insulation aging coefficient of 0.85. After calculation, the obtained leakage resistance parameter is 2300kΩ. The other seven branches are calculated using the same process, and the final output leakage resistance parameter range is controlled between 1500kΩ and 3000kΩ, which perfectly matches the insulation leakage characteristics of cables with different aging levels.

[0033] S2.2: Input the output distributed capacitance parameter, the leakage resistance parameter, and the real-time load parameter into the dynamic threshold generator. The dynamic threshold generator corrects the distributed capacitance parameter based on the power factor angle in the real-time load parameter to obtain the corrected distributed capacitance parameter. Then, it performs a vector sum calculation on the corrected distributed capacitance parameter and the leakage resistance parameter to obtain the dynamic zero-sequence current determination threshold. Furthermore, the specific steps of S2.2 include: (1) Input the distributed capacitance parameters and leakage resistance parameters output by the insulation impedance evaluation model, as well as the real-time load parameters obtained in real time, into the dynamic threshold generator. After receiving the data, the dynamic threshold generator first performs data synchronization verification to confirm that the timestamps of the distributed capacitance parameters, leakage resistance parameters and real-time load parameters are completely consistent and there is no time deviation. (2) The dynamic threshold generator first extracts the power factor angle value from the input real-time load parameters. In this embodiment, the power factor angle of each outgoing branch is set according to the stable operating state on site. The first branch is set to 0.87 rad, the second branch is set to 0.93 rad, the third branch is set to 0.95 rad, the fourth branch is set to 0.88 rad, the fifth branch is set to 0.91 rad, the sixth branch is set to 0.94 rad, the seventh branch is set to 0.90 rad, and the eighth branch is set to 0.89 rad. (3) After extraction, the generator corrects the distributed capacitance parameters. First, it calculates the cosine function value corresponding to the power factor angle. In this embodiment, the third branch is used as an example. The cosine value corresponding to the power factor angle of 0.95 rad is 0.572100. Second, the original distributed capacitance parameter output is multiplied by the cosine value to obtain the preliminary correction value. The original distributed capacitance parameter of the third branch is 0.21 μF. The preliminary correction value obtained after multiplication is 0.119942 μF. Third, the amplitude normalization of the preliminary correction value is performed. The amplitude normalization coefficient is used to compensate for the deviation between theoretical calculation and actual working conditions on site, and to offset the calculation errors caused by stray capacitance of the line, installation process, and ambient temperature and humidity. Combined with the on-site operating characteristics of 10kV distribution cable, it is set to 1.02. The preliminary correction value is multiplied by the amplitude normalization coefficient to obtain the final corrected distributed capacitance parameter. The corrected distributed capacitance parameter obtained after normalization of the third branch is 0.122341 μF. All outgoing branches are corrected in the same way. (4) Convert the corrected distributed capacitance parameters into the capacitive zero-sequence current amplitude. The system angular frequency used in the conversion is set to 314 rad / s according to the power frequency standard of 50 Hz. Multiply the corrected distributed capacitance parameters by the system angular frequency to obtain the capacitive susceptance value. Then multiply the value by the rated line voltage of the 10kV system. The line voltage value is set to 20kV to obtain the capacitive zero-sequence current amplitude. (5) Convert the leakage resistance parameter into the resistive zero-sequence current amplitude. Directly divide the rated line voltage of the 10kV system by the leakage resistance parameter to obtain the resistive zero-sequence current amplitude. In this embodiment, the leakage resistance parameter of the third branch is 2300kΩ, and the calculated resistive zero-sequence current amplitude is 4.35mA. (6) After obtaining the amplitude of capacitive zero-sequence current and resistive zero-sequence current, the operation is performed according to the vector synthesis rule. Since the capacitive current and resistive current are perpendicular to each other in phase, the vector sum calculation adopts the square root method of the square sum. The square of the capacitive zero-sequence current amplitude is added to the square of the resistive zero-sequence current amplitude, and then the square root operation is performed on the addition result. The final value is the dynamic zero-sequence current judgment threshold of the branch.

[0034] S2.3: The dynamic threshold generator monitors the changes in the real-time load parameters and iterates repeatedly whenever the active power, reactive power, or power factor angle changes, updating the dynamic zero-sequence current determination threshold in real time.

[0035] In this embodiment, the dynamic threshold generator monitors the changes in active power, reactive power, and power factor angle of each outgoing branch in real time with a period of 200ms. Combining the conventional load fluctuation characteristics of 10kV distribution lines and algorithm calibration requirements, it presets relative change thresholds for active power, relative change thresholds for reactive power, and absolute change thresholds for power factor angle. When any of the monitored changes exceeds the corresponding threshold, the threshold update process is immediately triggered. Specifically, when the change in active power exceeds 5%, the change in reactive power exceeds 8%, and the change in power factor angle exceeds 0.1rad, the generator automatically re-acquires the latest inherent electrical parameters, distributed capacitance parameters, leakage resistance parameters, and real-time load parameters, repeats the distributed capacitance correction and vector sum calculation, and outputs the updated dynamic zero-sequence current judgment threshold.

[0036] The process of generating the zero-sequence current amplitude of the pure fault characteristic includes: S2.4: Define the historical normal leakage baseline data as the intrinsic zero-sequence current component of each outgoing branch in a healthy state; In this embodiment, the intrinsic zero-sequence current component is the inherent zero-sequence current when the branch is fault-free and undisturbed. It is generated only by cable insulation leakage and distributed capacitance coupling and is unrelated to grounding faults. The controller directly assigns the obtained historical normal leakage baseline data to the intrinsic zero-sequence current component of the corresponding branch. The intrinsic zero-sequence current component is a constant reference value that remains unchanged in a single fault selection process and serves as the reference basis for stripping normal leakage. The intrinsic zero-sequence current components of the 8 branches are independent of each other and do not interfere with each other, ensuring the accuracy of pure fault feature extraction.

[0037] S2.5: Input the real-time zero-sequence current data into the moving average filter, smooth the real-time zero-sequence current data of each branch, and output the measured zero-sequence current amplitude after removing high-frequency noise. The moving average filter is a prior art in this field and is not an inventive solution of this application, so it will not be described in detail here. Furthermore, the filter selects real-time zero-sequence current data from five consecutive sampling points centered on the current sampling point, calculates the arithmetic mean, and replaces the current sampling point data with the average value. This process is repeated sequentially across all data points, effectively filtering out high-frequency noise, random pulses, short-term spikes, and other interferences, while retaining the true amplitude of the power frequency zero-sequence current. After filtering, the measured zero-sequence current amplitude curve is smoother, without glitches or abrupt changes, and can truly reflect the actual level of the branch zero-sequence current. The formula for calculating the arithmetic mean is existing technology in this field and is not an inventive solution of this application, so it will not be elaborated here.

[0038] S2.6: Subtract the intrinsic zero-sequence current component corresponding to each branch from the measured zero-sequence current amplitude of each branch to obtain the difference, and take the absolute value of the difference as the pure fault characteristic zero-sequence current amplitude of the branch.

[0039] In this embodiment, the measured zero-sequence current amplitude includes the intrinsic zero-sequence current component and the fault zero-sequence current component, namely, normal leakage current and fault signal. The normal leakage current component can be eliminated by subtraction, leaving only the fault-related characteristic quantity. For each branch, the controller subtracts the intrinsic zero-sequence current component from the filtered measured zero-sequence current amplitude to obtain the difference between the two. Since the fault zero-sequence current may be positive or negative, the absolute value of the difference is taken to eliminate the phase influence, resulting in a pure fault characteristic zero-sequence current amplitude that only reflects the intensity of the ground fault. This pure fault characteristic amplitude completely eliminates non-fault components such as normal leakage current, distributed capacitance inherent current, and insulation leakage current, and is not affected by the inherent characteristics and operating status of the branch, thus reflecting the severity of the ground fault in a true and intuitive way. For fault-free branches, the pure fault characteristic zero-sequence current amplitude is close to zero; for faulty branches, the amplitude will increase, and the more severe the fault, the larger the amplitude.

[0040] S3: The amplitude of the zero-sequence current of the pure fault characteristics is sampled using a preset steady-state sampling time window. The comprehensive equivalent amplitude of each branch is obtained through time-weighted calculation. For the four types of faults, namely metallic grounding, high-resistance grounding, arcing grounding, and intermittent grounding, a graded differential amplitude comparison logic is executed to initially screen out candidate fault branches. The process of generating the comprehensive equivalent amplitude includes: S3.1: Set multiple consecutive steady-state sampling time windows, the duration of each steady-state sampling time window being an integer multiple of the power frequency period; In this embodiment, the power frequency period is 20ms, and the steady-state sampling time window is set to 10 times the power frequency period, i.e., 200ms. This duration can ensure that enough steady-state fault data is collected, and also meet the real-time requirements of rapid line selection. The controller sets 5 consecutive steady-state sampling time windows to sample the amplitude of the zero-sequence current, which is a pure fault characteristic, in sequence. There are no gaps or overlaps between the steady-state sampling time windows, forming a continuous sampling sequence to fully capture the temporal change characteristics of the fault signal.

[0041] S3.2: Within each steady-state sampling time window, the amplitude of the pure fault characteristic zero-sequence current is sampled at equal intervals to form multiple amplitude sampling points arranged in time sequence within the steady-state sampling time window; In this embodiment, sampling is performed at equal intervals of 20ms within each steady-state sampling time window. Each steady-state sampling time window can collect 10 pure fault characteristic zero-sequence current amplitude sampling points. The sampling points are arranged in chronological order to form a timing sampling sequence. The sampling point data is stored in the timing buffer of the controller in real time, waiting for timing weighted calculation processing. Equal interval sampling ensures uniform data distribution and avoids feature loss due to sparse sampling.

[0042] S3.3: Input the multiple amplitude sampling points into the time-series weighted arithmetic unit. The time-series weighted arithmetic unit performs a weighted summation on all amplitude sampling points to obtain a weighted sum value. The weighted sum value is then divided by the total number of sampling points within the steady-state sampling time window. The resulting quotient is used as the comprehensive equivalent amplitude of each branch within the steady-state sampling time window. The time-series weighted arithmetic unit assigns a larger weighting coefficient to sampling points that are closer to the current time.

[0043] Furthermore, the specific steps of S3.3 include: (1) After receiving the amplitude sampling point data, the time-series weighting arithmetic unit first performs a sampling point sequence confirmation operation, sorts all sampling points according to the order of sampling time, arranges the first sampled point at the beginning of the sequence, and arranges the last sampled point, which is closest to the current time, at the end of the sequence; in this embodiment, the total number of sampling points set in each steady-state sampling time window is fixed at 10, so the time-series weighting arithmetic unit will assign corresponding weighting coefficients to the 10 amplitude sampling points arranged in sequence. The values ​​of the weighting coefficients have been set according to the linear increment rule when the algorithm is constructed. The specific allocation method is as follows: The weighting coefficient for the first sampling point is set to 0.5, the weighting coefficient for the second sampling point is set to 0.6, the weighting coefficient for the third sampling point is set to 0.7, the weighting coefficient for the fourth sampling point is set to 0.8, the weighting coefficient for the fifth sampling point is set to 0.9, the weighting coefficient for the sixth sampling point is set to 1.0, the weighting coefficient for the seventh sampling point is set to 1.1, the weighting coefficient for the eighth sampling point is set to 1.2, the weighting coefficient for the ninth sampling point is set to 1.3, and the weighting coefficient for the tenth sampling point, which is closest to the current time, is set to 1.4. (2) After the weighting coefficients are allocated, the time-series weighting arithmetic unit enters the weighted summation calculation, and multiplies the value of each amplitude sampling point with the weighting coefficient corresponding to that point in turn to obtain the weighted calculated value corresponding to each sampling point; after the weighted calculated values ​​of all individual sampling points are obtained, the arithmetic unit adds these 10 weighted calculated values ​​one by one, and determines the total value obtained by adding all of them as the weighted sum value of this calculation; (3) After obtaining the weighted sum, the time-series weighted arithmetic unit performs an average calculation operation, dividing the previously obtained weighted sum by the fixed total number of sampling points within the current steady-state sampling time window. In this embodiment, the total number of sampling points is fixed at 10, so the weighted sum is directly divided by 10 to obtain the corresponding quotient. After the arithmetic unit standardizes the quotient, it is formally determined as the comprehensive equivalent amplitude of the corresponding branch within the current steady-state sampling time window. The screening process for the candidate faulty branch includes: S3.4: Obtain the comprehensive equivalent amplitude of each branch, and obtain the pre-stored metallic grounding threshold, high-resistance grounding threshold, arc grounding threshold and background noise threshold; In this embodiment, the branch box controller pre-stores four levels of fault judgment thresholds adapted to the 10kV power distribution system: metallic grounding threshold, high-resistance grounding threshold, arcing grounding threshold, and background noise threshold. The thresholds are set comprehensively based on the system voltage level, line parameters, and operating experience, and can adapt to most power distribution branch box fault scenarios. Among them, the metallic grounding threshold corresponds to a severe grounding fault with a fault resistance close to zero, with the largest amplitude, and is set to 50mA; the high-resistance grounding threshold corresponds to a weak signal fault with a fault resistance of 1kΩ~10kΩ, with a medium amplitude, and is set to 20mA; the arcing grounding threshold corresponds to an intermittent arcing grounding fault, with a small amplitude, and is set to 5mA; the background noise threshold is used to eliminate weak signals such as system noise and measurement errors to avoid noise interference in the judgment, and is set to 1mA.

[0044] S3.5: Compare the relationship between the comprehensive equivalent amplitude of each branch and the metallic grounding threshold, the high-resistance grounding threshold, and the arcing grounding threshold; if the comprehensive equivalent amplitude is greater than or equal to the metallic grounding threshold, it is determined to be a metallic grounding fault; if it is less than the metallic grounding threshold but greater than or equal to the high-resistance grounding threshold, it is determined to be a high-resistance grounding fault; if it is less than the high-resistance grounding threshold but greater than or equal to the arcing grounding threshold, it is determined to be an arcing grounding fault; if the comprehensive equivalent amplitude alternately crosses each threshold within multiple consecutive steady-state sampling time windows, it is determined to be an intermittent grounding fault. In this embodiment, the controller performs hierarchical judgment on the comprehensive equivalent amplitude of each branch according to the threshold from high to low, so as to accurately identify the fault type: If the comprehensive equivalent amplitude of any branch reaches the metallic grounding threshold, it is determined to be a metallic grounding, with extremely small fault resistance, large zero-sequence current amplitude, and obvious fault characteristics. If the amplitude is between the metallic grounding threshold and the high-resistance grounding threshold, it is determined to be a high-resistance grounding, with a large fault resistance and a small zero-sequence current amplitude, which belongs to a weak fault scenario. If the amplitude is between the high-resistance grounding threshold and the arc grounding threshold, it is determined to be an arc grounding fault, which exists in the form of intermittent electric arcs with small amplitude fluctuations. If the amplitude repeatedly exceeds or falls below different thresholds within multiple consecutive steady-state sampling time windows, and there is no stable state, it is determined to be intermittent grounding, with an unstable fault state and intermittent connection.

[0045] Furthermore, through hierarchical determination, four common grounding fault types can be accurately distinguished. In this embodiment, the comprehensive equivalent amplitude of the third outgoing branch continuously exceeds the high-resistance grounding threshold but is lower than the metallic grounding threshold, so it is determined that the branch has a high-resistance grounding fault; the amplitudes of the remaining branches are all lower than the arc grounding threshold, and are determined to be fault-free.

[0046] S3.6: If the fault is determined to be a metallic grounding fault, the branch with the largest comprehensive equivalent amplitude is selected as the candidate fault branch; if the fault is determined to be a high-resistance grounding fault, branches with a comprehensive equivalent amplitude lower than the background noise threshold are first eliminated, and then the three branches with the largest comprehensive equivalent amplitude among the remaining branches are selected as candidate fault branches; if the fault is determined to be an arc grounding fault, the mean and standard deviation of the comprehensive equivalent amplitude of all branches are calculated, and the branches with a comprehensive equivalent amplitude greater than the mean plus twice the standard deviation are selected as candidate fault branches; if the fault is determined to be an intermittent grounding fault, the branch number with the largest comprehensive equivalent amplitude in each steady-state sampling time window is recorded in multiple consecutive steady-state sampling time windows, and the branch that appears most frequently is selected as the candidate fault branch.

[0047] In this embodiment, a differentiated amplitude screening rule is adopted for different fault types to improve the accuracy of candidate branches: Metallic grounding fault: The fault characteristics are extremely strong. Only the single branch with the largest amplitude needs to be selected to pinpoint the faulty branch. High-impedance grounding fault: The fault signal is weak and easily affected by noise. First, eliminate the noisy branches, and then select the three branches with the highest amplitude as candidates to narrow down the investigation scope. Arc grounding fault: The fault amplitude is small and the distribution is scattered. Statistical methods are used to screen branches with abnormal amplitude to avoid incorrect selection. Intermittent grounding fault: The fault state is unstable. By statistically analyzing the frequency through a multi-steady-state sampling time window, the branch with the most frequent occurrence and the largest amplitude is selected to improve reliability.

[0048] Furthermore, in this embodiment, a high-resistance grounding fault is identified. The controller first eliminates 5 branches out of 8 branches whose comprehensive equivalent amplitude is lower than the background noise threshold. Among the remaining 3 branches, the 3rd branch has the largest comprehensive equivalent amplitude, followed by the 5th and 7th branches. Therefore, these three branches are identified as candidate fault branches, completing the initial screening.

[0049] S4: Introduce real-time zero-sequence current data of the branch box bus to perform secondary vector deviation verification on the candidate fault branch. If the vector deviation is less than the preset verification threshold and the judgment results of multiple consecutive steady-state sampling time windows are consistent, output the fault selection completion signal. The introduction of real-time zero-sequence current data from the branch box busbar for secondary vector deviation verification of candidate fault branches includes: S4.1: Collect the amplitude and phase of the real-time zero-sequence current of the busbar from the zero-sequence current transformer installed at the busbar of the distribution branch box to form the measured zero-sequence current vector of the busbar; Furthermore, the specific steps of S4.1 include: (1) The controller sends a synchronous acquisition command to the zero-sequence current transformer installed on the bus side. The zero-sequence current transformer is a high-precision switch-type zero-sequence current transformer adapted to the 10kV power distribution system. Its measurement accuracy class is 0.5, and its frequency response range covers 15Hz to 5000Hz. It can fully respond to the power frequency zero-sequence current and the transient and harmonic components generated by the fault. The acquisition circuit adopts shielded twisted pair connection, and the shielding layer is reliably grounded at both ends, which can effectively resist the electromagnetic interference on site and ensure that the signal is not distorted. When the controller sends the acquisition command, it will mark the system timestamp at the current moment. The timestamp accuracy reaches the ms level to ensure that the bus zero-sequence current data and the zero-sequence current data of each outgoing branch are completely aligned in time. (2) Perform continuous acquisition of the real-time zero-sequence current waveform of the bus. The acquisition frequency is fixed at 1600Hz according to the power frequency characteristics of the system and the fault analysis requirements. 32 sampling points are completed in each power frequency cycle. This sampling density can completely restore the waveform characteristics of the zero-sequence current and ensure the accuracy of amplitude and phase calculation. The acquisition process lasts for a complete power frequency cycle. The power frequency cycle is fixed at 20ms according to the State Grid standard. 32 instantaneous current sampling values ​​are continuously acquired within this time. All sampling values ​​are stored in the temporary waveform buffer of the controller in the order of sampling, without losing any instantaneous data. (3) The controller performs a fast Fourier transform operation on the 32 instantaneous current sampling values ​​in the buffer area. The fast Fourier transform decomposes the power frequency component in the zero-sequence current, filters out the DC offset and the third, fifth, and seventh harmonic components, and retains only the effective value of the 50Hz power frequency component. The effective value is determined as the amplitude of the real-time zero-sequence current of the bus. The fast Fourier transform is the prior art in this field and is not an inventive solution of this application. It will not be described in detail here. (4) By comparing the zero-crossing time difference between the zero-sequence current power frequency component and the voltage reference phase, the phase difference between the two is calculated. In this embodiment, the real-time zero-sequence current phase of the bus is uniformly referenced with the system rated voltage phase, and the phase value collected is stable between 2.0 rad and 2.5 rad. (5) The zero-sequence current amplitude and phase obtained by the acquisition and calculation at the same time are combined and bound together to form a complete vector data structure in a fixed order of amplitude first and phase later. This vector data can simultaneously represent the magnitude and direction of the zero-sequence current. The controller stores the constructed measured bus zero-sequence current vector into a dedicated vector buffer area and adds the timestamp information of the acquisition time to ensure that it strictly corresponds to the zero-sequence current vector of each branch in time.

[0050] S4.2: Collect the real-time zero-sequence current amplitude and phase of the candidate fault branch from the zero-sequence current transformer of the branch to form the zero-sequence current vector of the candidate fault branch. Furthermore, the specific steps of S4.2 include: (1) The controller locates the zero-sequence current transformer installed on the corresponding branch according to the previously selected candidate fault branch number. In this embodiment, the candidate fault branches are the 3rd branch, the 5th branch and the 7th branch. The controller starts independent data acquisition for each candidate fault branch in order of branch number from small to large. Only one candidate fault branch is acquired at the same time to avoid signal interference or timing disorder caused by the simultaneous transmission of multiple data. The zero-sequence current transformer configured for each candidate fault branch is a 0.5-level high-precision wideband response model with a frequency response range covering 15Hz to 5000Hz. (2) When the controller starts the acquisition of each candidate fault branch, it generates a timestamp at the ms level that is exactly the same as the bus zero-sequence current acquisition. The time deviation does not exceed 1ms, eliminating the vector calculation error caused by timing asynchrony from the source. Then the controller performs real-time zero-sequence current waveform acquisition according to fixed parameters. The sampling frequency is set to 1600Hz. 32 instantaneous current sampling values ​​are uniformly acquired in each power frequency cycle. The power frequency cycle is fixed at 20ms according to the national standard. The acquisition process lasts for a complete power frequency cycle. The 32 instantaneous sampling values ​​are stored in the temporary waveform buffer area allocated separately by the controller for the candidate branch in chronological order, without missing any instantaneous data points. (3) After completing waveform acquisition, the controller performs fast Fourier transform on 32 instantaneous sampled values, decomposes the 50Hz power frequency component in the instantaneous waveform through fast Fourier transform, automatically removes the DC offset component and the third, fifth, and seventh harmonic components, and retains only the effective value of the power frequency component, and determines the effective value as the amplitude of the real-time zero-sequence current of the candidate fault branch. (4) The controller calculates the difference between the zero-sequence current of the candidate branch and the reference phase by comparing the time difference between the zero-crossing time of the power frequency component of the zero-sequence current and the zero-crossing time of the rated voltage phasor. (5) After the amplitude and phase are calculated, the zero-sequence current amplitude and phase of the branch calculated at the same time are bound and combined according to the same data structure as the bus zero-sequence current vector. The amplitude value is recorded first, and the phase value is recorded later to form a complete zero-sequence current vector that belongs only to the candidate fault branch. The controller stores the generated candidate fault branch zero-sequence current vector into a dedicated vector operation buffer area and adds a timestamp consistent with the bus acquisition time to ensure that the vector data of each candidate branch can be directly compared and calculated with the bus vector data. (6) For scenarios with multiple candidate fault branches, the controller repeatedly executes (1)-(5) to generate the zero-sequence current vector of all candidate branches in sequence, thereby generating accurate and reliable zero-sequence current vector of candidate fault branches.

[0051] S4.3: Mark all remaining branches other than the candidate fault branches as non-candidate fault branches, obtain the intrinsic zero-sequence current component amplitude and corresponding phase of each non-candidate fault branch in the healthy state, and sum them after phase alignment to obtain the vector sum of non-candidate fault branches; Furthermore, the specific steps of S4.3 include: (1) The controller performs the traversal and marking operation of the remaining branches other than the candidate fault branches. Based on the configuration of 8 outgoing branches in this embodiment, the 3rd branch, the 5th branch, and the 7th branch have been identified as candidate fault branches. The controller compares the branches one by one in the order from the 1st to the 8th branch according to their numbers, and marks all the remaining branches whose numbers do not match the candidate fault branches as non-candidate fault branches. In this embodiment, the marked non-candidate fault branches are the 1st branch, the 2nd branch, the 4th branch, the 6th branch, and the 8th branch, a total of 5 branches. (2) The controller reads the intrinsic zero-sequence current component amplitude and phase of each non-candidate fault branch in the order of the non-candidate branch list. In this embodiment, the intrinsic zero-sequence current component amplitudes of the five non-candidate fault branches are set as follows during algorithm construction: 0.3mA for the first branch, 0.4mA for the second branch, 0.5mA for the fourth branch, 0.4mA for the sixth branch, and 0.3mA for the eighth branch. The corresponding phases are uniformly set to 2.2rad during algorithm construction. The intrinsic phases of all non-candidate branches are consistent. (3) After retrieving the intrinsic zero-sequence current amplitude and phase of all non-candidate fault branches, the controller enters the phase alignment operation. The core purpose of phase alignment is to ensure that all intrinsic zero-sequence current components participating in the summation are vectored under the same phase reference, so as to avoid calculation errors caused by phase differences. Since the intrinsic zero-sequence current phase of all non-candidate fault branches has been uniformly set to 2.2 rad during algorithm construction, there is no need to perform phase offset correction in this embodiment. It is directly assumed that the intrinsic zero-sequence current components of all non-candidate branches are in a fully aligned phase state. If there is an inconsistency in the intrinsic phase of different branches, the controller will use the unified system rated voltage phasor as the reference to adjust the intrinsic phase of each branch to the reference phase one by one, so as to ensure that all vectors participate in the summation in the same direction. (4) The controller, in the order of the non-candidate branch list, sequentially accumulates the 0.3mA of the first branch, the 0.4mA of the second branch, the 0.5mA of the fourth branch, the 0.4mA of the sixth branch, and the 0.3mA of the eighth branch. First, the amplitudes of the first and second branches are added together to get 0.7mA, then the amplitude of the fourth branch is added together to get 1.2mA, then the amplitude of the sixth branch is added together to get 1.6mA, and finally the amplitude of the eighth branch is added together to get 1.9mA. Because all components have the same phase, the vector phase after summing is still 2.2rad. Finally, the total amplitude obtained by summing is combined with the unified phase to form a complete vector sum of non-candidate fault branches. (5) Store the calculated non-candidate fault branch vectors into the vector operation cache area.

[0052] S4.4: Perform vector addition on the zero-sequence current vector of the candidate fault branch and the vector sum of the non-candidate fault branches to obtain the theoretical bus zero-sequence current vector; Furthermore, the specific steps of S4.4 include: (1) The controller retrieves the zero-sequence current vector of the candidate fault branch and the vector sum of the non-candidate fault branch from the dedicated vector operation buffer in sequence. In this embodiment, there are 3 candidate fault branches. Therefore, the controller will perform vector synthesis calculations for the 3rd branch, the 5th branch and the 7th branch in sequence to ensure that each candidate branch can obtain a set of corresponding calculation bus zero-sequence current vectors. When reading data, the controller will first confirm that the timestamps of the two sets of vectors are completely consistent to ensure that the vector data involved in the calculation belong to the same time. (2) Decompose each vector into horizontal and vertical components, sum the components in the same direction, and finally recombine the synthesized components into a complete vector. (3) Extract the amplitude and phase values ​​of the complete vector. Taking the third candidate fault branch in this embodiment as an example, its zero-sequence current vector amplitude is 18.25mA and the phase is 2.2rad. The controller first calculates the cosine and sine values ​​corresponding to the phase. The cosine value corresponding to 2.2rad is -0.5048 and the sine value is 0.8634. Then, the amplitude of 18.25mA is multiplied by the cosine and sine values ​​respectively to obtain the horizontal component as -9.2032mA and the vertical component as 15.7616mA. (4) Extract the magnitude and phase values ​​of the vector sum. In this embodiment, the magnitude of the vector sum of the non-candidate fault branch is 1.9mA and the phase is 2.2rad. The controller multiplies 1.9mA by the cosine value corresponding to 2.2rad, which is -0.5048, to obtain the horizontal component as -0.9591mA. Then, it multiplies 1.9mA by the sine value corresponding to 2.2rad, which is 0.8634, to obtain the vertical component as 1.6433mA. (5) After completing the component decomposition of the two vectors, perform algebraic addition on the horizontal and vertical components respectively: First, add the horizontal component of the candidate fault branch to the horizontal component of the vector sum of the non-candidate fault branches, i.e., -9.2032 + (-0.9591) = -10.1623mA, and obtain the total horizontal component after synthesis as -10.1623mA. Then add the vertical component of the candidate fault branch to the vertical component of the vector sum of the non-candidate fault branches, i.e., 15.7616 + 1.6433 = 17.4049mA; (6) After the component synthesis is completed, the controller enters the calculation stage of the amplitude and phase of the zero-sequence current vector of the bus. First, the total amplitude is calculated, and then the total phase is calculated. When calculating the amplitude, the values ​​of the total horizontal component and the total vertical component are squared respectively, and then the two squared results are added together. Finally, the square root of the sum is performed to obtain the final vector amplitude. Taking the data of this embodiment as an example, the square of the total horizontal component -10.1623mA is 103.2723, the square of the total vertical component 17.4049mA is 302.9305, the two are added together to get 406.2028, and the square root is taken to obtain the amplitude of the zero-sequence current vector of the bus of 20.15mA. The result is kept to two decimal places. (7) When calculating the phase, the total vertical component is divided by the total horizontal component to obtain the corresponding tangent value. Then, the corresponding phase angle is calculated by arctangent operation. During the operation, the controller will automatically determine the quadrant of the vector based on the positive and negative signs of the total horizontal and vertical components to ensure that the phase value is accurate. In this embodiment, the total vertical component is positive and the total horizontal component is negative. The vector is located in the second quadrant, and the phase value obtained after arctangent operation is 2.2 rad. (8) The controller combines the calculated amplitude and phase according to a fixed structure to form a complete theoretical bus zero-sequence current vector, and stores it in a dedicated buffer for vector verification, which is used to compare the deviation with the measured bus zero-sequence current vector.

[0053] S4.5: Calculate the vector deviation between the measured bus zero-sequence current vector and the theoretical bus zero-sequence current vector, and use the vector deviation as the verification basis.

[0054] In this embodiment, the vector deviation includes amplitude deviation and phase deviation. The controller calculates the deviation between the measured bus zero-sequence current vector and the calculated bus zero-sequence current vector of each group through vector difference operation. The smaller the deviation value, the higher the probability that the candidate branch is a faulty branch. Among them, the vector deviation corresponding to the third candidate branch is the smallest, which is much smaller than that of the other two candidate branches. It is preliminarily verified that the third branch is a real faulty branch.

[0055] If the vector deviation is less than a preset verification threshold and the judgment results of multiple consecutive steady-state sampling time windows are consistent, then a fault selection completion signal is output, including: S4.6: Obtain the vector deviation and obtain the preset verification threshold dynamically set according to the rated voltage level of the branch box; Furthermore, to avoid false verification caused by transformer measurement errors, normal three-phase imbalance of the line, and environmental electromagnetic interference, the verification threshold is calibrated based on a large amount of field operation data of 10kV distribution network, equipment accuracy indicators and ground fault discrimination requirements. Since 1mA is a commonly used empirical value for zero-sequence current verification of 10kV distribution network branch boxes, in this embodiment, the preset verification threshold is set to 1mA for the 10kV rated voltage level. S4.7: Determine whether the vector deviation is less than the preset verification threshold. If not, terminate the verification and wait for the next round of fault determination. If yes, proceed to the consistency verification of multiple consecutive steady-state sampling time windows. S4.8: Within N consecutive steady-state sampling time windows, the candidate fault branch number of each steady-state sampling time window is recorded sequentially; where N is a preset integer greater than or equal to 2; In this embodiment, N=3 is preset, that is, three consecutive steady-state sampling time windows. The controller records the candidate fault branch number determined in each steady-state sampling time window in sequence. In all three steady-state sampling time windows, the third branch is determined to be a candidate fault branch.

[0056] S4.9: Determine whether the candidate fault branch numbers of the N consecutive steady-state sampling time windows are completely consistent. If not, terminate the verification and clear all recorded steady-state sampling time window numbers; if yes, generate a fault selection completion signal.

[0057] In this embodiment, the candidate branch number is the 3rd branch in three consecutive steady-state sampling time windows. The judgment results are completely consistent. The controller immediately generates a fault selection completion signal, indicating that the fault branch is accurately locked and the selection process enters the final judgment stage.

[0058] S5: Based on the fault selection completion signal, determine the amplitude level using the comprehensive equivalent amplitude, determine the fluctuation characteristics using the change of the comprehensive equivalent amplitude within multiple consecutive steady-state sampling time windows, and determine and store the fault branch number, grounding fault type, and fault severity.

[0059] The determination and storage of the fault branch number, ground fault type, and fault severity includes: S5.1: After receiving the fault selection completion signal, read the candidate fault branch number and use the candidate fault branch number as the fault branch number. In this embodiment, after the controller receives the fault selection completion signal, it automatically reads the locked candidate fault branch number as 3 and determines the 3rd outgoing branch as the fault branch. S5.2: Read the comprehensive equivalent amplitude corresponding to the fault branch number, compare the comprehensive equivalent amplitude with the preset amplitude level range, and determine the amplitude level of the branch; Furthermore, the amplitude level range is a pre-defined continuous numerical range for quantifying the severity of a fault. All range boundaries are fixedly configured during the algorithm construction phase based on measured data of grounding faults in the 10kV distribution system. It is divided into four levels, corresponding to minor faults, general faults, severe faults, and extremely severe faults, respectively. There is no overlap, discontinuity, or ambiguity between the ranges, covering all possible fault amplitude ranges. The specific numerical settings are as follows: the amplitude range corresponding to the minor fault level is greater than or equal to 0mA and less than 5mA; the amplitude range corresponding to the general fault level is greater than or equal to 5mA and less than 20mA; the amplitude range corresponding to the severe fault level is greater than or equal to 20mA and less than 40mA; and the amplitude range corresponding to the extremely severe fault level is greater than or equal to 40mA.

[0060] Furthermore, the specific steps in S5.2 include: (1) After receiving the fault selection completion signal, the controller reads the unique number of the finally determined fault branch from the fault result storage area after verification. This number has been verified twice by amplitude comparison and vector deviation in the early stage and has unique determination. In this embodiment, the fault branch number finally locked is the 3rd branch. After the controller reads this number, it immediately establishes a dedicated data reading direction. (2) The controller retrieves the comprehensive equivalent amplitude of the third branch at the current fault time based on the locked branch number. In this embodiment, the specific value of the amplitude is 18.25mA. (3) The controller reads four complete amplitude level ranges from the local fixed parameter storage area; (4) Compare 18.25mA with the amplitude level range. Since the amplitude level range is between 5 and 20, it falls completely within the range corresponding to the general fault level. (5) After the determination is completed, the controller will temporarily store the determined amplitude level locally and mark it as the amplitude level corresponding to the third branch of the fault branch, i.e. the general fault level. The temporary data is bound to the fault branch number and the comprehensive equivalent amplitude for storage.

[0061] S5.3: Read the M comprehensive equivalent amplitude values ​​of the fault branch number within M consecutive steady-state sampling time windows, calculate the difference between the comprehensive equivalent amplitude values ​​of two adjacent steady-state sampling time windows to obtain M-1 difference values, and calculate the range of the M-1 difference values. Determine the fluctuation characteristics of the branch based on the range of the range; M is a preset integer greater than or equal to 3; the range is defined as the difference between the maximum and minimum values ​​in a set of values, used to reflect the fluctuation amplitude of the data. In this embodiment, M=5 is set, and the comprehensive equivalent amplitude of five consecutive steady-state sampling time windows is read. The adjacent differences are calculated and the range is obtained. The small range indicates that the fault state is stable and there are no violent fluctuations. Therefore, the fluctuation characteristics are determined to be stable.

[0062] Furthermore, this embodiment predefines three fixed range judgment intervals, corresponding to three characteristics: stable fluctuation, slight fluctuation, and severe fluctuation. Specifically, the intervals are set as follows: a range greater than or equal to 0 mA and less than 0.5 mA corresponds to the stable fluctuation characteristic, indicating that the fault signal is stable without significant fluctuations; a range greater than or equal to 0.5 mA and less than 2 mA corresponds to the slight fluctuation characteristic, indicating that the fault signal has small changes; and a range greater than or equal to 2 mA corresponds to the severe fluctuation characteristic, indicating that the fault signal fluctuates significantly, mostly intermittent or arcing ground.

[0063] Furthermore, the specific steps in S5.3 include: (1) The controller reads the comprehensive equivalent amplitude generated sequentially by the branch during the current fault occurrence period for the locked fault branch number. In this embodiment, the fault branch is the third branch, and the comprehensive equivalent amplitudes corresponding to the five consecutive steady-state sampling time windows are 17.82mA, 18.15mA, 18.25mA, 18.30mA, and 18.40mA, respectively. (2) After reading the five comprehensive equivalent amplitudes, subtract the comprehensive equivalent amplitude of the previous steady-state sampling time window from the comprehensive equivalent amplitude of the next steady-state sampling time window to obtain the difference between the two adjacent values. Since there are a total of five amplitude data and four adjacent combinations, four difference results can be obtained in the end: 0.33mA, 0.10mA, 0.05mA, and 0.10mA. Store these four differences into the temporary operation buffer in the order of calculation. (3) The controller first iterates through the four differences to find the maximum value. The maximum value this time is 0.33mA. Then it continues to iterate to find the minimum value among the four differences. The minimum value this time is 0.05mA. After determining the maximum and minimum values, the maximum value is subtracted from the minimum value to obtain the range value, which is 0.28mA. (4) The controller compares the calculated range of 0.28 with the preset range judgment interval one by one, and determines that the value is greater than or equal to 0mA and less than 0.5mA, which falls completely into the range corresponding to the stable fluctuation. Therefore, the fluctuation characteristics of the fault branch are directly determined to be stable. (5) After the judgment is completed, the controller binds and stores the fluctuation characteristic results with information such as fault branch number, comprehensive equivalent amplitude, amplitude level, etc.

[0064] S5.4: Input the determined ground fault type, the determined amplitude level, and the determined fluctuation characteristics into the preset fault severity determination matrix, and output the corresponding fault severity level; Furthermore, the fault severity judgment matrix is ​​a fixed three-dimensional judgment matrix pre-compiled based on the measured data of grounding faults in the 10kV distribution system and the operation and maintenance standards. It covers four fault types: metallic grounding, high-resistance grounding, arcing grounding, and intermittent grounding; four amplitude levels: minor fault, general fault, severe fault, and extremely severe fault; and three fluctuation characteristics: stable, slight fluctuation, and severe fluctuation. Each combination has a unique corresponding fault severity level. The fault severity level is divided into four levels, from low to high: minor fault, medium fault, severe fault, and extremely severe fault.

[0065] Furthermore, the judgment rules for the fault severity judgment matrix are set as follows: When the ground fault type is metallic grounding, regardless of the amplitude level and fluctuation characteristics, it is directly judged as an extremely severe fault; when the ground fault type is high-resistance grounding, if the amplitude level is a minor fault and the fluctuation characteristics are stable, it is judged as a minor fault; if the amplitude level is a minor fault and the fluctuation characteristics are slight or severe fluctuations, it is judged as a medium-level fault; if the amplitude level is a general fault and the fluctuation characteristics are stable, it is judged as a medium-level fault; if the amplitude level is a general fault and the fluctuation characteristics are slight or severe fluctuations, it is judged as a severe fault; if the amplitude level is a severe fault or an extremely severe fault, regardless of the fluctuation characteristics, it is judged as an extremely severe fault; when the ground fault type is arc grounding, if the amplitude level is a minor fault or a general fault, it is judged as a medium-level fault; if the amplitude level is a severe fault or an extremely severe fault, it is judged as a severe fault; when the ground fault type is intermittent grounding, if the fluctuation characteristics are stable or slight fluctuations, it is judged as a severe fault; if the fluctuation characteristics are severe fluctuations, it is judged as an extremely severe fault.

[0066] In this embodiment, the fault type is high-resistance grounding, the amplitude level is general fault, and the fluctuation characteristic is stable, which fully conforms to the fixed combination of high-resistance grounding, general fault, and stable type in the matrix. According to the preset rules, the intermediate fault level should be output.

[0067] S5.5: Pack the fault branch number, the fault severity level, and the grounding fault type into the local non-volatile memory of the branch box, and simultaneously upload them to the remote monitoring master station through the communication interface.

[0068] The method further includes a step of detecting abrupt changes in the power factor angle in real-time load parameters to assist in determining the nature of the fault: A1: Simultaneously with the zero-sequence current amplitude of the pure fault characteristic, the power factor angle collected within each steady-state sampling time window is acquired. A2: Calculate the difference between the power factor angle in the current steady-state sampling time window and the power factor angle in the previous steady-state sampling time window, and take the absolute value as the characteristic quantity of load change. The larger the value, the more severe the load switching and fluctuation. A3: Obtain the preset angle change threshold and the preset amplitude fluctuation multiple threshold. In order to effectively identify the electrical quantity fluctuation caused by load switching and avoid misjudging load disturbance as ground fault, two sets of judgment thresholds are set in combination with the actual measured conditions of the distribution network. In this embodiment, the angle change threshold is set to 0.1 rad and the amplitude fluctuation multiple threshold is set to 1.5 times. The load change characteristic quantity is compared with the angle change threshold. At the same time, the ratio of the pure fault characteristic zero-sequence current amplitude of the current steady-state sampling time window to the pure fault characteristic zero-sequence current amplitude of the previous steady-state sampling time window is compared with the amplitude fluctuation multiple threshold. A4: If the load change characteristic exceeds the angle change threshold and the ratio also exceeds the amplitude fluctuation multiple threshold, it is determined that the current electrical quantity change originates from load switching, thereby suppressing the execution of the fault selection process; A5: If the load change characteristic exceeds the angle change threshold, but the ratio does not exceed the amplitude fluctuation multiple threshold, it is determined that there is a ground fault superimposed on the load disturbance, and the fault selection process continues accordingly.

[0069] The method also includes the step of managing fault log information using a tiered storage strategy: B1: After storing the fault record in the local non-volatile memory, obtain the preset first priority capacity value, second priority capacity value and third priority upload rate; Furthermore, the controller sequentially acquires three key parameters—first priority capacity value, second priority capacity value, and third priority upload rate—in a fixed reading order. These three parameters are all system-level fixed configurations, precisely set during the algorithm construction phase based on industrial-grade storage chip specifications, power distribution site communication stability, and fault record management standards. The first priority capacity value represents the maximum number of fault records that can be stored in the highest priority area of ​​local storage. It is primarily used to store critical fault records requiring rapid retrieval, such as extremely severe or serious faults. During construction, it is explicitly set to 100 records, ensuring sufficient storage of high-priority fault records without consuming excessive storage resources, thus conforming to the hardware requirements of embedded controllers. The configuration specifications for the first priority area are as follows: The second priority capacity value represents the maximum number of fault records that can be stored in the second priority area of ​​local storage. It is used to store general fault records such as intermediate faults. It is explicitly set to 500 records during the construction process, which can meet the long-term local storage needs of a moderate number of fault records. It forms a reasonable hierarchical match with the first priority capacity, and realizes the classification and storage of fault records of different importance. The third priority upload rate represents the speed limit for uploading fault records to the remote monitoring master station after the local storage is full. It is explicitly set to 1 record per second during the construction process. It can avoid network congestion or packet loss caused by the centralized upload of a large amount of data while ensuring stable communication. It is compatible with the wireless and wired communication transmission capabilities commonly found in power distribution sites.

[0070] B2: Map the severity of the determined fault to a priority level, and store them in the first priority storage area, the second priority storage area, and the third priority storage area in descending order of priority level. The first priority storage area stores the K most recent fault records in a circular overwrite method, where K is a preset positive integer. B3: When the first priority storage area is full of K records, the subsequent fault records are stored in the second priority storage area and stored using the page write method until the second priority storage area is full of the preset second priority capacity value. B4: When the first priority storage area and the second priority storage area are both full, the subsequent fault records are uploaded to the remote monitoring master station in batches through the communication interface at the limited upload rate to form the third priority storage area. B5: After the branch box is powered off and restarted locally, read the fault records in the first priority storage area and the second priority storage area, merge the two and load them into the running memory of the branch box controller to form a fault record index table.

[0071] Furthermore, the specific steps of B5 include: (1) The controller reads the valid fault records in the first priority storage area one by one. The reading operation starts from the starting physical address of the storage area and is executed in ascending order of address. Data verification is performed once for each fault record read to ensure that the record is complete and undamaged. During the reading process, blank storage units are automatically skipped and only the written valid fault records are read until all 100 storage capacity of the first priority storage area is traversed. In this embodiment, a total of 87 valid fault records have been stored in the first priority storage area. The controller reads all 87 fault records and temporarily stores them in the temporary cache area opened in the running memory. (2) After completing the reading of the first priority storage area, the controller immediately locates the physical storage block corresponding to the second priority storage area. The physical storage block is used to store fault records of medium priority. The controller reads the valid fault records one by one from the starting address according to the same reading rules as the first priority storage area, automatically skipping blank units, and completing the traversal reading of all 500 storage capacity. In this embodiment, a total of 312 valid fault records have been saved in the second priority storage area. The controller reads these 312 fault records into another temporary cache area of ​​the running memory and stores them separately from the first priority fault records to avoid data confusion during the reading process. (3) After the data in the two storage areas is read, the 87 records in the first priority storage area and the 312 records in the second priority storage area are mixed and arranged according to the time of the fault occurrence from early to late. The smaller the timestamp value, the earlier the fault occurred and the earlier the arrangement position. During the merging process, the controller performs a uniqueness check on all fault records to ensure that there are no duplicate records. If duplicate entries are found, the record with the latest timestamp is automatically retained and the duplicate entries are deleted to ensure that the final merging result is clean and unique. After the merging is completed, the running memory contains a total of 399 consecutively arranged and time-ordered fault records. The field format of all records is completely uniform and the arrangement order is clear and standardized. (4) The controller assigns consecutive index numbers to the 399 merged fault records in sequence. The index numbers start from 1 and end at 399. Each record corresponds to a unique index number. The controller binds the memory start address, timestamp, branch number, severity level of each fault record with the index number to form a complete index item. Then, all index items are combined in sequence to generate the final fault record index table.

[0072] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments under the guidance of the present invention without departing from the spirit and scope of the present invention. All of these variations are within the protection scope of the present invention.

Claims

1. A method for selecting branch box grounding faults based on zero-sequence current amplitude ratio, characterized in that, include: Acquire real-time zero-sequence current data, inherent electrical parameters, historical normal leakage baseline data, and real-time load parameters of all outgoing branches of the distribution branch box; Based on inherent electrical parameters and real-time load parameters, a dynamic zero-sequence current determination threshold is constructed for each outgoing branch. Using the historical normal leakage current baseline data, the normal leakage current component is extracted from the real-time zero-sequence current data to obtain the amplitude of the pure fault characteristic zero-sequence current. The amplitude of the zero-sequence current characteristic of the pure fault is sampled using a preset steady-state sampling time window. The comprehensive equivalent amplitude of each branch is obtained through time-weighted calculation. For four types of faults, namely metallic grounding, high-resistance grounding, arcing grounding, and intermittent grounding, a graded differential amplitude comparison logic is executed to initially screen out candidate fault branches. Introducing real-time zero-sequence current data from the branch box busbar, a secondary vector deviation verification is performed on the candidate fault branch. If the vector deviation is less than the preset verification threshold and the judgment results of multiple consecutive steady-state sampling time windows are consistent, a fault selection completion signal is output. Based on the fault selection completion signal, the amplitude level is determined by the comprehensive equivalent amplitude, the fluctuation characteristics are determined by the change of the comprehensive equivalent amplitude within multiple consecutive steady-state sampling time windows, and the fault branch number, ground fault type and fault severity are determined and stored.

2. The branch box grounding fault selection method based on zero-sequence current amplitude as described in claim 1, characterized in that, The acquisition of real-time zero-sequence current data, inherent electrical parameters, historical normal leakage baseline data, and real-time load parameters for all outgoing branches of the distribution branch box includes: By deploying zero-sequence current transformers on each outgoing branch in the distribution branch box, the instantaneous zero-sequence current waveform of each branch is collected. A fast Fourier transform is performed on the instantaneous zero-sequence current waveform to extract the power frequency component amplitude, and the power frequency component amplitude is used as the real-time zero-sequence current data of the branch. The cable length, cable cross-sectional area, insulation material type, and insulation aging coefficient of each outgoing branch are retrieved from the local storage unit of the distribution branch box, which together constitute the inherent electrical parameters of the branch. Extract the average zero-sequence current amplitude under the condition of no grounding fault from the historical operation database of the distribution branch box, and use this average value as the historical normal leakage current baseline data. The active power, reactive power, and power factor angle at the current moment are collected by the load monitoring terminal installed at the end of each outgoing branch, which together constitute the real-time load parameters.

3. The branch box grounding fault selection method based on zero-sequence current amplitude as described in claim 2, characterized in that, The dynamic zero-sequence current determination threshold for each outgoing branch includes: The inherent electrical parameters are input into the insulation impedance evaluation model pre-set in the branch box controller. The insulation impedance evaluation model outputs the distributed capacitance and leakage resistance parameters of each outgoing branch under the current insulation state. The output distributed capacitance parameter, the leakage resistance parameter, and the real-time load parameter are input into the dynamic threshold generator. The dynamic threshold generator corrects the distributed capacitance parameter based on the power factor angle in the real-time load parameter to obtain the corrected distributed capacitance parameter. The corrected distributed capacitance parameter and the leakage resistance parameter are then vector-summed to obtain the dynamic zero-sequence current determination threshold. The dynamic threshold generator monitors the changes in the real-time load parameters and iterates repeatedly whenever the active power, reactive power, or power factor angle changes, updating the dynamic zero-sequence current determination threshold in real time.

4. The branch box grounding fault selection method based on zero-sequence current amplitude as described in claim 3, characterized in that, The process of generating the zero-sequence current amplitude of the pure fault characteristic includes: The historical normal leakage baseline data is defined as the intrinsic zero-sequence current component of each outgoing branch in a healthy state. The real-time zero-sequence current data is input into a moving average filter to smooth the real-time zero-sequence current data of each branch, and the measured zero-sequence current amplitude after removing high-frequency noise is output. The measured zero-sequence current amplitude of each branch is subtracted from the intrinsic zero-sequence current component corresponding to that branch to obtain the difference. The absolute value of the difference is then taken as the pure fault characteristic zero-sequence current amplitude of that branch.

5. The branch box grounding fault selection method based on zero-sequence current amplitude as described in claim 4, characterized in that, The process of generating the comprehensive equivalent amplitude includes: Set multiple consecutive steady-state sampling time windows, with the duration of each steady-state sampling time window being an integer multiple of the power frequency period; Within each steady-state sampling time window, the amplitude of the pure fault characteristic zero-sequence current is sampled at equal intervals to form multiple amplitude sampling points arranged in time sequence within the steady-state sampling time window; The multiple amplitude sampling points are input into a time-series weighted arithmetic unit. The time-series weighted arithmetic unit performs a weighted summation on all amplitude sampling points to obtain a weighted sum value. The weighted sum value is then divided by the total number of sampling points within the steady-state sampling time window, and the resulting quotient is used as the comprehensive equivalent amplitude of each branch within the steady-state sampling time window. The time-series weighted arithmetic unit assigns a larger weighting coefficient to sampling points that are closer to the current time.

6. The branch box grounding fault selection method based on zero-sequence current amplitude as described in claim 5, characterized in that, The screening process for the candidate faulty branch includes: Obtain the comprehensive equivalent amplitude of each branch, and obtain the pre-stored metallic grounding threshold, high-resistance grounding threshold, arc grounding threshold and background noise threshold; Compare the relationship between the comprehensive equivalent amplitude of each branch and the metallic grounding threshold, the high-resistance grounding threshold, and the arcing grounding threshold; if the comprehensive equivalent amplitude is greater than or equal to the metallic grounding threshold, it is determined to be a metallic grounding fault; if it is less than the metallic grounding threshold but greater than or equal to the high-resistance grounding threshold, it is determined to be a high-resistance grounding fault; if it is less than the high-resistance grounding threshold but greater than or equal to the arcing grounding threshold, it is determined to be an arcing grounding fault; if the comprehensive equivalent amplitude alternately crosses each threshold within multiple consecutive steady-state sampling time windows, it is determined to be an intermittent grounding fault. If the fault is determined to be a metallic grounding fault, the branch with the largest comprehensive equivalent amplitude is selected as the candidate fault branch. If the fault is determined to be a high-resistance grounding fault, branches with comprehensive equivalent amplitudes lower than the background noise threshold are first eliminated, and then the three branches with the largest comprehensive equivalent amplitudes among the remaining branches are selected as candidate fault branches. If the fault is determined to be an arc grounding fault, the mean and standard deviation of the comprehensive equivalent amplitudes of all branches are calculated, and the branches with comprehensive equivalent amplitudes greater than the mean plus twice the standard deviation are selected as candidate fault branches. If the fault is determined to be an intermittent grounding fault, the branch number with the largest comprehensive equivalent amplitude in each steady-state sampling time window is recorded, and the branch that appears most frequently is selected as the candidate fault branch.

7. The branch box grounding fault selection method based on zero-sequence current amplitude as described in claim 6, characterized in that, The introduction of real-time zero-sequence current data from the branch box busbar for secondary vector deviation verification of candidate fault branches includes: The amplitude and phase of the real-time zero-sequence current of the busbar are collected by the zero-sequence current transformer installed at the busbar of the distribution branch box, forming the measured zero-sequence current vector of the busbar. The real-time zero-sequence current amplitude and phase of the candidate fault branch are collected from the zero-sequence current transformer of the branch to form the zero-sequence current vector of the candidate fault branch. All remaining branches other than the candidate fault branches are marked as non-candidate fault branches. The intrinsic zero-sequence current component amplitude and corresponding phase of each non-candidate fault branch in the healthy state are obtained. After phase alignment, they are summed to obtain the vector sum of non-candidate fault branches. The zero-sequence current vector of the candidate fault branch is added to the vector sum of the non-candidate fault branches to obtain the theoretical bus zero-sequence current vector. Calculate the vector deviation between the measured bus zero-sequence current vector and the theoretical bus zero-sequence current vector, and use the vector deviation as the verification basis.

8. The branch box grounding fault selection method based on zero-sequence current amplitude as described in claim 7, characterized in that, If the vector deviation is less than a preset verification threshold and the judgment results of multiple consecutive steady-state sampling time windows are consistent, then a fault selection completion signal is output, including: The vector deviation is obtained, and a preset verification threshold dynamically set according to the rated voltage level of the branch box is obtained; Determine whether the vector deviation is less than the preset verification threshold. If not, terminate the verification and wait for the next round of fault determination. If yes, proceed to the consistency verification of multiple consecutive steady-state sampling time windows. Within N consecutive steady-state sampling time windows, the candidate fault branch number of each steady-state sampling time window is recorded sequentially; where N is a preset integer greater than or equal to 2. Determine whether the candidate fault branch numbers of the N consecutive steady-state sampling time windows are completely consistent. If not, terminate the verification and clear all recorded steady-state sampling time window numbers; if yes, generate a fault selection completion signal.

9. The branch box grounding fault selection method based on zero-sequence current amplitude as described in claim 8, characterized in that, The determination and storage of the fault branch number, ground fault type, and fault severity includes: After receiving the fault selection completion signal, the candidate fault branch number is read and used as the fault branch number. Read the comprehensive equivalent amplitude corresponding to the fault branch number, compare the comprehensive equivalent amplitude with the preset amplitude level range, and determine the amplitude level of the branch; Read the M comprehensive equivalent amplitude values ​​of the fault branch number within M consecutive steady-state sampling time windows, calculate the difference between the comprehensive equivalent amplitude values ​​of two adjacent steady-state sampling time windows to obtain M-1 difference values, and calculate the range of the M-1 difference values. Determine the fluctuation characteristics of the branch based on the range of the range; M is a preset integer greater than or equal to 3. Input the determined ground fault type, the determined amplitude level, and the determined fluctuation characteristics into the preset fault severity determination matrix, and output the corresponding fault severity level. The fault branch number, the fault severity level, and the grounding fault type are packaged and stored in the local non-volatile memory of the branch box, and simultaneously uploaded to the remote monitoring master station through the communication interface.

10. The branch box grounding fault selection method based on zero-sequence current amplitude as described in claim 9, characterized in that, It also includes a step of detecting abrupt changes in the power factor angle in real-time load parameters to help determine the nature of the fault: While calculating the amplitude of the zero-sequence current characteristic of the pure fault, the power factor angle collected within each steady-state sampling time window is simultaneously acquired; Calculate the difference between the power factor angle in the current steady-state sampling time window and the power factor angle in the previous steady-state sampling time window, and take the absolute value as the characteristic quantity of load change. Obtain a preset angle mutation threshold and a preset amplitude fluctuation multiple threshold, compare the load mutation characteristic quantity with the angle mutation threshold, and simultaneously compare the ratio of the pure fault characteristic zero-sequence current amplitude in the current steady-state sampling time window to the pure fault characteristic zero-sequence current amplitude in the previous steady-state sampling time window with the amplitude fluctuation multiple threshold. If the load change characteristic exceeds the angle change threshold and the ratio also exceeds the amplitude fluctuation multiple threshold, it is determined that the current electrical quantity change originates from load switching, thereby suppressing the execution of the fault selection process. If the load change characteristic exceeds the angle change threshold, but the ratio does not exceed the amplitude fluctuation multiple threshold, it is determined that there is a ground fault superimposed on the load disturbance, and the fault selection process continues accordingly.

11. The branch box grounding fault selection method based on zero-sequence current amplitude as described in claim 10, characterized in that, It also includes the step of using a tiered storage strategy to manage fault log information: After storing the fault record in the local non-volatile memory, obtain the preset first priority capacity value, second priority capacity value and third priority upload rate; The severity of the determined fault is mapped to a priority level, and the faults are stored in the first priority storage area, the second priority storage area, and the third priority storage area in descending order of priority level. The first priority storage area stores the K most recent fault records in a circular overwrite method, where K is a preset positive integer. Once the first priority storage area is full of K records, subsequent fault records are stored in the second priority storage area and stored using a page write method until the second priority storage area is full of the preset second priority capacity value. Once the first and second priority storage areas are full, subsequent fault records are uploaded to the remote monitoring master station in batches through the communication interface at the limited upload rate, forming the third priority storage area. After the branch box is powered off and restarted locally, the fault records in the first priority storage area and the second priority storage area are read, and the two are merged and loaded into the running memory of the branch box controller to form a fault record index table.

Images