A method for grading protection of incoming line fault of box-type substation

Abstract

The application discloses a kind of box-type substation incoming line fault grading protection method, belong to power system operation, monitoring and protection technical field, it includes the synchronous acquisition box-type substation's incoming line three-phase current, bus three-phase voltage and all feeder outgoing line three-phase current, generates transient data set, extracts the collaborative response characteristic and voltage current space-time correlation characteristic of each feeder outgoing line three-phase current, and generates virtual opposite end logic signal, carries out the comprehensive determination of fault nature and fault point relative position, generates fault judgment result and generates grading protection instruction.Differentiated grading protection instruction is generated and issued according to the judgment of fault nature and position, so as to improve the selectivity, rapidity and reliability of protection.

Description

Technical Field

[0001] This invention relates to the field of power system operation, monitoring and protection technology, and in particular to a method for graded protection of incoming line faults in a prefabricated substation. Background Technology

[0002] As a critical node in the distribution network, the prefabricated substation relies on rapid and accurate fault diagnosis and protection actions when a fault occurs on its incoming side to ensure the safe and stable operation of the power grid. Traditional protection methods mainly rely on current and voltage transformers installed at the incoming line to collect basic electrical quantities and use fixed threshold values ​​such as current, voltage, or impedance for fault detection and tripping decisions.

[0003] Existing technical solutions typically employ simple overcurrent protection or distance protection principles. Overcurrent protection determines faults by comparing the incoming line current with a preset value, while distance protection determines the fault range by calculating the relationship between the measured impedance and the set impedance range. The implementation of these solutions is mainly based on local electrical information at the incoming line, and the protection settings are often set according to the system's maximum operating mode or typical fault conditions, and remain fixed once set.

[0004] However, existing technical solutions have significant technical shortcomings. First, they rely on a single dimension of electrical information and fail to effectively utilize the rich fault correlation information provided by the multiple outgoing circuits within the substation. This results in insufficient ability to distinguish the nature of faults and difficulty in accurately differentiating between transient and permanent faults. Second, fixed protection settings cannot adapt to changes in system operation modes and the differentiated requirements of protection action characteristics for different fault locations and types. This can easily lead to protection maloperation or failure to operate, especially when handling downstream feeder faults, posing a risk of cascading tripping and affecting power supply reliability. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a method for graded protection of incoming line faults in prefabricated substations. By synchronously acquiring multi-source transient information, fusing and analyzing the spatiotemporal correlation characteristics of feeder coordinated response and voltage and current, and mapping protection strategies based on fault diagnosis results, the method can determine the nature and location of faults and generate and issue differentiated graded protection commands accordingly, thereby improving the selectivity, speed, and reliability of protection.

[0006] The above objectives can be achieved through the following approach: A method for graded protection of incoming line faults in a prefabricated substation includes synchronously acquiring the incoming three-phase current, bus three-phase voltage, and all outgoing three-phase currents of the prefabricated substation. When the incoming three-phase current exceeds a preset current threshold or the bus three-phase voltage exceeds a preset voltage threshold, a transient data set is generated. The transient data set includes the incoming transient current vector, the bus transient voltage vector, and the outgoing transient current vector. Based on the outgoing transient current vector in the transient data set, the sudden change timing and phase relationship of the outgoing three-phase currents of each feeder are analyzed, and the transient data is extracted. The coordinated response characteristics of the three-phase currents of each feeder are analyzed, and a virtual peer logic signal representing the relative direction of the fault is generated. Based on the transient current vector of the incoming line and the transient voltage vector of the bus in the transient data set, the spatiotemporal correlation characteristics of voltage and current are extracted. The virtual peer logic signal and the spatiotemporal correlation characteristics of voltage and current are fused to make a comprehensive judgment on the nature of the fault and the relative position of the fault point, and a fault judgment result including the fault section code and the fault type identifier is generated. According to the fault judgment result, a graded protection command for the incoming circuit breaker is generated and issued.

[0007] Optionally, the generation of the transient data set includes: synchronously acquiring the incoming three-phase current, bus three-phase voltage, and all feeder outgoing three-phase current of the prefabricated substation; when the incoming three-phase current exceeds a preset current threshold or the bus three-phase voltage exceeds a preset voltage threshold, extracting the fundamental component and harmonic components of the incoming three-phase current to obtain the incoming current spectrum component; extracting the fundamental component and harmonic components of the bus three-phase voltage to obtain the bus voltage spectrum component; performing time-frequency joint analysis on the incoming current spectrum component and the bus voltage spectrum component to obtain the time-frequency joint analysis result; and generating a transient data set containing the incoming transient current vector, the bus transient voltage vector, and the outgoing transient current vector based on the time-frequency joint analysis result.

[0008] Optionally, the step of performing time-frequency joint analysis on the incoming current spectrum component and the bus voltage spectrum component to obtain the time-frequency joint analysis result includes: aligning the incoming current spectrum component and the bus voltage spectrum component at the same time point to construct a current-voltage time-frequency joint matrix; performing multi-dimensional mode decomposition on the current-voltage time-frequency joint matrix to extract the transient energy distribution vector reflecting the energy distribution of each frequency component during the transient process and the phase coupling relationship matrix reflecting the phase coupling relationship of the key frequency components; and fusing the transient energy distribution vector and the phase coupling relationship matrix to generate a time-frequency joint analysis result characterizing the joint evolution features of voltage and current spectrum during the fault transient process.

[0009] Optionally, the step of extracting the coordinated response characteristics of the three-phase currents of each feeder based on the transient current vectors in the transient data set, and generating a virtual peer logic signal representing the relative direction of the fault, includes: acquiring the transient current vectors of each feeder in the transient data set, and performing amplitude mutation detection on the transient current vectors corresponding to the three-phase currents of each feeder to extract the amplitude mutation amount of the three-phase currents of each feeder; identifying the initial mutation time of the three-phase currents of each feeder based on the amplitude mutation amount, and generating a mutation time sequence for all feeders; selecting a feeder with a preset mutation time as a reference feeder according to the mutation time sequence, and calculating the phase difference between the transient current vectors of each feeder and the transient current vectors of the reference feeder; assigning a direction logic value representing whether the direction of the current mutation is consistent with that of the reference feeder based on the phase difference and a preset phase threshold; integrating the direction logic values ​​of all feeders and associating them with the mutation time sequence to generate a virtual peer logic signal representing the coordinated response time and direction relationship of each feeder to the fault.

[0010] Optionally, the extraction of the spatiotemporal correlation features of voltage and current includes: performing differential operations on the incoming transient current vector and the bus transient voltage vector in the transient data set to calculate the instantaneous rate of change sequence of current and the instantaneous rate of change sequence of voltage; calculating the ratio of the instantaneous rate of change sequence of voltage and the instantaneous rate of change sequence of current at the same time point to generate a ratio sequence; performing statistical analysis on the ratio sequence to extract the mean, variance, and slope of the ratio change trend to obtain the spatiotemporal correlation features of voltage and current.

[0011] Optionally, generating a fault judgment result including a fault segment code and a fault type identifier includes: numerically encoding the virtual peer logic signal to convert it into a logic feature vector; concatenating the logic feature vector with the voltage and current spatiotemporal correlation features to generate a comprehensive feature vector; using the comprehensive feature vector to generate a segment judgment result regarding whether the fault point is located upstream or downstream of the incoming line side of the prefabricated substation, and a type judgment result regarding whether the fault is a permanent or transient fault; and synthesizing a fault judgment result including a fault segment code and a fault type identifier based on the segment judgment result and the type judgment result.

[0012] Optionally, the step of numerically encoding the virtual peer logic signal and converting it into a logic feature vector includes: parsing the virtual peer logic signal to obtain the directional logic value of each feeder and the abrupt change timing sequence; binary encoding the directional logic value corresponding to each feeder to generate a directional encoding vector; calculating the normalized time difference of each feeder relative to the reference feeder based on the abrupt change timing sequence to generate a timing feature vector; and fusing the directional encoding vector and the timing feature vector to generate a logic feature vector characterizing the cooperative relationship of current abrupt changes in each feeder.

[0013] Optionally, generating and issuing graded protection instructions for the incoming circuit breaker based on the fault judgment result includes: parsing the fault judgment result to obtain the fault section code and fault type identifier; matching the corresponding protection level and action delay parameter from a preset protection strategy mapping table based on the fault section code and fault type identifier; determining the target tripping current setting of the incoming circuit breaker based on the protection level, and generating a graded protection instruction containing the target tripping current setting and action delay parameter in combination with the action delay parameter; and issuing the graded protection instruction to the protection device of the prefabricated substation to control the incoming circuit breaker to perform the corresponding protection action.

[0014] Optionally, parsing the fault judgment result to obtain the fault segment code and fault type identifier includes: extracting the segment judgment result text description and type judgment result text description contained in the fault judgment result; performing semantic parsing on the segment judgment result text description to identify keywords indicating that the fault point is located upstream or downstream of the incoming line side of the box-type substation, and mapping them to the fault segment code; performing semantic parsing on the type judgment result text description to identify keywords indicating permanent faults or transient faults, and mapping them to the fault type identifier.

[0015] Based on the same inventive concept, the present invention also provides a graded protection system for incoming line faults in a prefabricated substation, the system comprising: The data acquisition module is used to synchronously acquire the incoming three-phase current, bus three-phase voltage, and all feeder outgoing three-phase current of the box-type substation. When the incoming three-phase current exceeds the preset current threshold or the bus three-phase voltage exceeds the preset voltage threshold, a transient data set is generated. The transient data set includes the incoming transient current vector, the bus transient voltage vector, and the outgoing transient current vector. The data processing module is used to analyze the sudden change timing and phase relationship of the three-phase current of each feeder based on the transient current vector of the outgoing line in the transient data set, extract the cooperative response characteristics of the three-phase current of each feeder, and generate a virtual opposite-end logic signal characterizing the relative direction of the fault. The feature association module is used to extract the spatiotemporal correlation features of voltage and current based on the incoming transient current vector and the bus transient voltage vector in the transient data set. The fault judgment module is used to integrate the virtual peer logic signal with the spatiotemporal correlation features of voltage and current to make a comprehensive judgment on the nature of the fault and the relative position of the fault point, and generate a fault judgment result containing the fault segment code and the fault type identifier. The instruction sending module is used to generate and issue graded protection instructions for the incoming circuit breaker based on the fault judgment result.

[0016] Compared with the prior art, the present invention has the following advantages: This invention enables in-depth analysis of the spectral evolution, phase coupling, and spatiotemporal correlation characteristics during fault transients by synchronously collecting transient electrical quantities of incoming lines, busbars, and all outgoing lines and constructing joint features. This improves the completeness and accuracy of complex fault feature extraction and lays a data foundation for subsequent judgment.

[0017] This invention integrates the coordinated response timing direction characteristics of the feeder group with the voltage and current correlation characteristics between the incoming busbars to form a comprehensive criterion. This enables the identification of the fault nature and the determination of the relative location of the fault point to corroborate and complement each other, overcoming the problems of misjudgment and omission caused by single information or one-sided features, and improving the reliability of protection decisions.

[0018] Based on the comprehensive judgment results of fault location and nature, this invention dynamically matches and issues graded protection commands containing differentiated current settings and action delays, realizing intelligent grading and coordination of protection actions. It can quickly cut off the fault in the case of severe faults, and avoid unnecessary over-level tripping in the case of minor or transient faults, thus optimizing the selectivity and speed of protection.

[0019] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating a method for graded protection of incoming line faults in a prefabricated substation according to an embodiment of the present invention.

[0022] Figure 2 This is a schematic diagram of the graded protection action characteristic curve of an embodiment of the present invention.

[0023] Figure 3 This is a schematic diagram of the structure of a box-type substation incoming line fault graded protection system according to an embodiment of the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Reference Figure 1 One embodiment of the present invention proposes a graded protection method for incoming line faults in a prefabricated substation. By synchronously collecting multi-source transient information, fusing and analyzing the spatiotemporal correlation characteristics of feeder coordinated response and voltage and current, and mapping protection strategies based on fault diagnosis results, the method can determine the nature and location of the fault and generate and issue differentiated graded protection commands accordingly, thereby improving the selectivity, speed and reliability of the protection.

[0026] The method described in this embodiment specifically includes: S1. Synchronously collect the incoming three-phase current, bus three-phase voltage, and all feeder outgoing three-phase current of the box-type substation. When the incoming three-phase current exceeds the preset current threshold or the bus three-phase voltage exceeds the preset voltage threshold, generate a transient data set. The transient data set includes the incoming transient current vector, the bus transient voltage vector, and the outgoing transient current vector. Optionally, the generation of the transient data set includes: Synchronously collect the incoming three-phase current, bus three-phase voltage, and all feeder outgoing three-phase current of the prefabricated substation; When the incoming three-phase current exceeds the preset current threshold or the bus three-phase voltage exceeds the preset voltage threshold, the fundamental component and each harmonic component in the incoming three-phase current are extracted to obtain the incoming current spectrum component. The fundamental component and harmonic components of the three-phase bus voltage are extracted to obtain the bus voltage spectrum components. The input current spectrum component and the bus voltage spectrum component are subjected to time-frequency joint analysis to obtain the time-frequency joint analysis results. Based on the time-frequency joint analysis results, a transient data set is generated, which includes the incoming transient current vector, the bus transient voltage vector, and the outgoing transient current vector.

[0027] Optionally, the step of performing time-frequency joint analysis on the incoming current spectrum component and the bus voltage spectrum component to obtain the time-frequency joint analysis result includes: Align the input current spectrum component with the bus voltage spectrum component at the same time point to construct a current-voltage time-frequency joint matrix. Multidimensional mode decomposition is performed on the current-voltage time-frequency joint matrix to extract the transient energy distribution vector reflecting the energy distribution of each frequency component during the transient process and the phase coupling matrix reflecting the phase coupling relationship of the key frequency components. By fusing the transient energy distribution vector with the phase coupling matrix, a time-frequency joint analysis result characterizing the joint evolution of voltage and current spectra during the fault transient process is generated.

[0028] In one embodiment of the present invention, step S1 includes the following steps: Specifically, sensor arrays installed within the prefabricated substation simultaneously measure the three-phase current at the incoming line, the three-phase voltage at the busbar, and the three-phase current at each feeder's outgoing line, forming a set of raw electrical quantity data that is strictly aligned in time. The sampling frequency of these sensors is set to 10 kHz, a setting based on the analysis of measured data from 200 sets of industrial sensors, confirming their ability to reliably capture high-frequency components of transient processes in the power system. The preset current threshold is 1.2 times the rated incoming current, and the preset voltage threshold is 0.85 times or 1.15 times the rated bus voltage. When the monitoring system detects that the measured incoming current value of any phase continuously exceeds the preset current threshold for 3 milliseconds, or the measured bus voltage value of any phase continuously exceeds the preset voltage threshold for 3 milliseconds, the system determines that the power grid has entered a transient process and immediately triggers subsequent processing procedures.

[0029] The system extracts waveform data of the incoming three-phase current and the bus three-phase voltage from the measurement data window spanning 5 cycles before and 10 cycles after the trigger moment. For each phase current and voltage waveform, a Fast Fourier Transform is applied to separate the fundamental component at 50 Hz, and the harmonic components with frequencies that are integer multiples of the fundamental frequency up to the 13th harmonic (650 Hz). For the incoming phase A current, its... The complex form of the subharmonic component is expressed as: ,in The value range is an integer from 1 to 13, and the upper bound is determined based on the distribution range of major harmonic components in engineering practice. The voltage of phase A of the busbar... The complex form of the second harmonic component is expressed as follows: , The value range is also from 1 to 13. Therefore, for each time sampling point... Each of them can obtain a set of spectral components containing the fundamental wave and harmonics, which together constitute the spectral components of the incoming current and the bus voltage.

[0030] To establish the joint relationship between current and voltage in the time and frequency dimensions, the same time point is used as the reference. All current harmonic components and voltage harmonic components are arranged. For a three-phase system, a time-frequency joint matrix is ​​constructed. Its dimensions are The mathematical description of this matrix is: , in, Representative at Time of the first Phase current The amplitude of a subharmonic wave is measured in amperes. Representative at Time of the first Phase voltage The amplitude of the first harmonic is measured in volts. Each row of the matrix corresponds to one phase, and each column corresponds to the amplitude of the 1st to 13th harmonics of the current and the amplitude of the 1st to 13th harmonics of the voltage, thus encapsulating the spectral information of the three-phase current and voltage at that moment within a single matrix.

[0031] Multidimensional mode decomposition is performed on a time-frequency joint matrix sequence consisting of multiple consecutive time points. This process first calculates the covariance matrix of the matrix sequence in the time dimension, and then performs eigenvalue decomposition on the covariance matrix. (Before extraction) The eigenvectors corresponding to the largest eigenvalues The values ​​are determined by the cumulative variance contribution rate exceeding 95%. These eigenvectors constitute the dominant transient mode. Transient energy distribution vector. It is A column vector of dimension, whose ? element By the eigenvalues After normalization, the result is as follows: This is used to characterize the proportion of each transient mode in the total energy. Phase coupling matrix. It is A matrix whose elements By calculating the first The and the first The cosine value of the phase difference between the components of each eigenvector in a specific frequency band, such as the 2nd to 6th harmonic frequency band, is obtained. This reflects the phase correlation strength between key frequency components of different modes. The calculation formula is as follows: ,in and This represents the average phase angle of the corresponding feature vector in the target frequency band.

[0032] Transient energy distribution vector Phase coupling matrix The upper triangular elements, excluding the diagonal, are expanded row-wise and connected to form a one-dimensional comprehensive eigenvector. This comprehensive eigenvector is the result of time-frequency joint analysis, which quantifies the energy distribution of voltage and current spectral components and their phase coupling relationships during fault transient processes.

[0033] Finally, based on the above time-frequency joint analysis results, the system generates a transient data set. This set contains three core parts: the incoming line transient current vector, which is extracted and reconstructed from the dimension specifically corresponding to the current amplitude characteristics in the time-frequency joint analysis results; the bus transient voltage vector, which is extracted and reconstructed from the dimension specifically corresponding to the voltage amplitude characteristics in the time-frequency joint analysis results; and the outgoing line transient current vector, which is generated in a similar way to the incoming line transient current vector, except that the input data is the spectral components of the three-phase currents of each feeder outgoing line, and is obtained through the same time-frequency joint analysis and feature extraction process.

[0034] For example, suppose a prefabricated substation has an incoming line rated current of 630 amps and a bus rated line voltage of 10 kV. The preset current threshold is set to 756 amps, and the preset voltage threshold has a lower limit of 8.5 kV and an upper limit of 11.5 kV. When the sensor detects that the incoming line C-phase current reaches 800 amps within 4 milliseconds, exceeding the preset current threshold of 756 amps, the system immediately initiates the transient data recording and processing procedure.

[0035] The system extracted data for a total of 15 cycles, or a 300-millisecond time window, before and after the event. After performing a Fast Fourier Transform on the incoming C-phase current, the fundamental 50Hz component amplitude was found to be 810 amperes. Significant 3rd and 5th harmonics were also detected, with the 3rd harmonic amplitude being 8% of the fundamental frequency, or 64.8 amperes. The same analysis of the bus C-phase voltage yielded a fundamental voltage amplitude of 5.6 kV, while the 3rd harmonic voltage amplitude was 5% of the fundamental frequency, or 280 V. At the fault initiation time... The constructed time-frequency joint matrix contains the fundamental and harmonic amplitude information of the aforementioned current and voltage.

[0036] For containing Multidimensional mode decomposition was performed on the time-frequency joint matrix sequence of 100 consecutive sampling points. The first three eigenvalues ​​of the covariance matrix were calculated as follows: , , The cumulative contribution rate reached 97.5%, therefore The transient energy distribution vector is calculated as follows: This indicates that the first dominant mode carries 85.6% of the transient energy. Further calculations yield the phase coupling matrix... , , This indicates that there is a strong anti-phase coupling relationship between Mode 1 and Mode 3.

[0037] The combined time-frequency analysis results generated from the above information, such as a 15-dimensional eigenvector containing energy distribution and phase coupling relationships, are used to construct the final transient data set. The incoming line transient current vector in this set can be characterized as follows: The numerical sequence of the bus transient voltage vector can be characterized as follows: The numerical sequence is used, while the transient current vector of the outgoing line is formed into a similar structure based on the actual measurement and processing results of each feeder. These vectors together serve as the basis for subsequent steps to analyze fault characteristics.

[0038] S2. Based on the transient current vector of the outgoing line in the transient data set, analyze the sudden change timing and phase relationship of the three-phase current of each feeder, extract the cooperative response characteristics of the three-phase current of each feeder, and generate a virtual opposite-end logic signal characterizing the relative direction of the fault. Optionally, the step of extracting the coordinated response characteristics of the three-phase currents of each feeder based on the transient current vectors in the transient data set, and generating a virtual opposite-end logic signal characterizing the relative direction of the fault, includes: Obtain the outgoing transient current vector from the transient data set, and perform amplitude change detection on the outgoing transient current vector corresponding to the three-phase current of each feeder, and extract the amplitude change amount of the three-phase current of each feeder. Based on the amplitude mutation amount, the initial mutation time of the three-phase current of each feeder is identified, and a mutation time sequence of all feeders is generated. Based on the mutation time sequence, a feeder at a preset mutation time is selected as a reference feeder, and the phase difference between the outgoing transient current vector of each feeder and the outgoing transient current vector of the reference feeder is calculated. Based on the phase difference and the preset phase threshold, each feeder is assigned a directional logic value that characterizes whether the direction of the current change is consistent with the reference feeder. By integrating the directional logic values ​​of all feeders and associating them with the mutation timing sequence, a virtual peer logic signal is generated that characterizes the timing and directional relationship of the coordinated response of each feeder to a fault.

[0039] In one embodiment of the present invention, step S2 includes the following steps: Specifically, the transient current vectors of each feeder are obtained from the transient data set generated in step S1. The transient current vector represents the current waveform data sequence formed by synchronous sampling and preprocessing of the three-phase current at the feeder's outgoing position within the transient process recording time window. Its sampling frequency is consistent with step S1, which is 10 kHz. Assuming the prefabricated substation has N feeders, the transient current vector of the i-th feeder is denoted as... It contains the values ​​of all sampling points of the three-phase current from 5 cycles before transient triggering to 10 cycles after triggering.

[0040] Amplitude mutation detection is performed on the transient current vector of each feeder. For the three-phase current of feeder i, the instantaneous amplitude of each phase current at each sampling point t is first calculated. This amplitude is obtained by taking the absolute value of the sampled value of the phase current, denoted as . Where p represents the phase, taking values ​​of A, B, or C. Next, the amplitude abrupt change is calculated. It is defined as the absolute value of the difference between the current amplitude and the amplitude one cycle ago, and the calculation formula is: , in, This represents the number of sampling points corresponding to one power frequency cycle, which is 200 points at a 10 kHz sampling rate. This calculation method can highlight the rapid changes in current during transient processes. The sequence of the largest amplitude abrupt changes in the three-phase current of each feeder is extracted. This serves as the representative mutation sequence for the feeder.

[0041] The initial abrupt change in the three-phase current of each feeder is identified based on the magnitude of the abrupt change. A threshold for this abrupt change is set. This threshold is derived from statistical analysis of load fluctuation data measured by 200 sets of industrial sensors, and is set to 15% of the rated current amplitude of the corresponding feeder. When the representative abrupt change sequence of a feeder exceeds a certain threshold for more than three consecutive sampling points... When a sudden change in current occurs in the feeder, the time of the first sampling point exceeding the threshold is recorded as the initial change time of the feeder. The initial abrupt change times of all N feeders constitute a time-series abrupt change. .

[0042] Reference feeders are selected based on the mutation time series. The selection rule is to choose the feeder with the earliest initial mutation time as the reference feeder, denoted by its index r, which satisfies the following condition: This setting is based on the fact that transient processes caused by faults usually first appear on the feeder with the closest electrical distance.

[0043] Calculate the phase difference between the outgoing transient current vector of each feeder and the outgoing transient current vector of the reference feeder. First, for each feeder i, extract the A-phase current data of the first complete power frequency cycle after the transient onset from its outgoing transient current vector, and apply Fast Fourier Transform to obtain the phase angle of its fundamental component. The fundamental phase angle of the reference feed line r is denoted as... The phase difference between feed line i and the reference feed line is... Calculated using the following formula: , in, For modulo operations, this process constrains the phase difference range to within to The units are degrees. The phase difference reflects the relative directional relationship of each feeder current during its initial abrupt change.

[0044] Each feeder is assigned a directional logic value based on the phase difference and a preset phase threshold. Phase threshold Set as This value is set based on the directional protection principle commonly used in power system fault analysis. Directional logic value. The assignment rule is: if ,but This indicates that the direction of the current change in the feeder is basically consistent with that of the reference feeder; otherwise... This indicates a misalignment in direction. This logic value indicates whether the direction of the current abrupt change is consistent with the reference feeder.

[0045] The directional logic values ​​of all feeders are integrated and correlated with abrupt timing sequences to generate a virtual peer logic signal. This signal is a structured data set used to comprehensively characterize the coordinated timing and directional relationship of each feeder's response to a fault. Specifically, the generation method involves combining the directional logic values ​​of each feeder... Its normalized initial mutation time Combination. Normalized time. The calculation formula is: , in, It is the latest moment in the mutation time sequence. Ultimately, the virtual peer logic signal... Represented as a sequence of binary pairs corresponding to all feeders: , This signal serves as one of the inputs for the comprehensive determination in step S4.

[0046] For example, suppose a prefabricated substation has four feeders, numbered 1 to 4. The transient current vectors of their outgoing lines are obtained from the transient data set. After amplitude mutation detection, a sequence of representative mutations for each feeder is calculated, and a mutation threshold of 15% (15 amperes) of the rated current amplitude of 100 amperes is used. The representative mutation amount for feeder 1 continuously exceeds 15 amperes starting from time index 2050; for feeder 2, it's at 2052; for feeder 3, at 2065; and for feeder 4, at 2051. Therefore, the identified initial mutation times are as follows: , , , , forming a mutation time sequence .

[0047] Feeder 1, corresponding to the earliest abrupt change time of 2050, is selected as the reference feeder r. The fundamental phase angle of the A-phase current in the first cycle after the transient start of each feeder is extracted, assuming it is obtained through calculation. , , , The phase difference relative to the reference feed line is then calculated as follows: , , (After modulo operation, the value is -180°) .

[0048] Based on phase threshold Perform a judgment to obtain the direction logic value. , , , Calculating the normalized time, the latest time is 2065, therefore... , , , The final generated virtual peer logic signal is: This signal indicates that the current abrupt change direction of feeders 1, 2, and 4 is consistent with that of the reference feeder and the response is earlier, while the direction of feeder 3 is opposite and the response is the latest, accurately characterizing the coordinated response features of each feeder.

[0049] S3. Based on the incoming transient current vector and the bus transient voltage vector in the transient data set, the spatiotemporal correlation features of voltage and current are extracted. Optionally, the extracted spatiotemporal correlation features of voltage and current include: Differentiate the incoming transient current vector and the bus transient voltage vector in the transient data set to calculate the instantaneous rate of change sequence of current and the instantaneous rate of change sequence of voltage; Calculate the ratio of the instantaneous rate of change of voltage sequence to the instantaneous rate of change of current sequence at the same time point, and generate a ratio sequence; Statistical analysis is performed on the ratio sequence to extract the ratio mean, ratio variance, and ratio change trend slope, thereby obtaining the spatiotemporal correlation characteristics of voltage and current.

[0050] In one embodiment of the present invention, step S3 includes the following steps: Specifically, the incoming line transient current vector and the bus transient voltage vector are obtained from the transient data set generated in step S1. The incoming line transient current vector represents the characteristic data sequence formed by the three-phase current at the incoming line position of the prefabricated substation after processing in step S1 within the transient process recording time window. The bus transient voltage vector represents the characteristic data sequence formed by the three-phase voltage of the bus after processing in step S1 within the same time window. These two vectors are strictly synchronized in time, and their data length corresponds to the transient recording time window, for example, from 5 cycles before the transient trigger to 10 cycles after the trigger, a total of 300 milliseconds, containing 3000 synchronized data points at a 10 kHz sampling rate.

[0051] The transient current vector of the incoming line is differentiated to calculate the instantaneous rate of change of the current. Since the vector already contains the amplitude information of the fundamental and harmonic currents, to reflect the overall rate of change, the equivalent comprehensive amplitude sequence of the incoming line current must first be reconstructed. The equivalent amplitude is obtained at each sampling point. The formula for calculating this is: (The formula is missing from the provided text.) , in, Representative at Incoming current at time 1 The complex representation of a second harmonic component, whose magnitude is the amplitude of that frequency component. For the equivalent amplitude sequence... The instantaneous rate of change of the current was calculated using the central difference method, resulting in a sequence of instantaneous rate of change of the current. : , in, The sampling time interval is 0.1 milliseconds. The unit is amperes per second. The central difference method is verified based on the analysis of 200 sets of measured transient data from industrial sensors. Compared with forward or backward difference methods, it can more effectively suppress noise and accurately reflect the changing trend.

[0052] A similar differentiation operation is performed on the bus transient voltage vector to calculate the instantaneous rate of change of voltage. The equivalent composite magnitude sequence of the bus voltage is then reconstructed. : , in, Representative at Bus voltage at time 1 Complex representation of the second harmonic component. The same central difference method is applied to obtain the voltage instantaneous rate of change sequence. : , The unit is volts per second.

[0053] Calculate the ratio of the instantaneous rate of change of voltage sequence to the instantaneous rate of change of current sequence at the same time point, and generate a ratio sequence. This ratio physically has the dimension of impedance, reflecting the instantaneous change characteristics of the system's equivalent impedance during transient processes. Its calculation formula is: , in, The unit is ohms. To ensure the stability of the calculation, when the denominator... The absolute value is less than a local minimum. (For example The ratio at that moment (amperes per second). The value is assigned to zero.

[0054] For the complete ratio sequence Statistical analysis was conducted to extract three core statistical features to constitute the spatiotemporal correlation characteristics of voltage and current. These three features are the ratio, mean, and... Ratio variance and the slope of the ratio change trend .

[0055] Ratio Mean The average level of the system's dynamic impedance during the transient process is characterized and calculated as follows: , in, is the total length of the ratio sequence.

[0056] Ratio variance It characterizes the degree of fluctuation of dynamic impedance around its mean, reflects transient stability, and is calculated as follows: , slope of ratio change trend Used to characterize the direction and rate of change of dynamic impedance during the main transient phase. The selection is from the transient triggering time. After that A subsequence of ratio data within milliseconds (e.g., 50 milliseconds), and the numerical values ​​of that subsequence. With time index Perform a least-squares linear fit. The slope of the fitted line is... Its calculation formula can be obtained through the following formula: , in, Given the length of the selected subsequence, all summation operations are performed within this subsequence. Slope The unit is ohms per millisecond. A positive value indicates that the impedance is increasing, and a negative value indicates that it is decreasing.

[0057] Finally, the spatiotemporal correlation characteristics of voltage and current A three-dimensional eigenvector representation consisting of these three statistics: This feature vector comprehensively describes the spatiotemporal correlation characteristics of voltage and current change rates during fault transients, serving as another key input for comprehensive determination in step S4.

[0058] For example, suppose that the incoming transient current vector and the bus transient voltage vector obtained from the transient data set are reconstructed to obtain an equivalent amplitude sequence. and The central difference method is applied to calculate the result at a certain sampling point. (During a transient process), we obtain ampere, ampere, Milliseconds. Then the instantaneous rate of change of current. kiloamperes per second. Similarly, the instantaneous rate of change of voltage at that point is calculated. kilovolts per second.

[0059] Therefore, the dynamic impedance ratio at that moment can be calculated. Ohms. Calculate this ratio for all sampling points within the complete 300-millisecond window, forming a ratio sequence.

[0060] Perform statistical analysis on this sequence. Assume that the mean of the ratios is calculated. Ohm. Calculate the variance of the ratio, assuming... Ohm². A linear fit was performed on 500 data points from 0 to 50 milliseconds after the transient trigger, and the trend slope was calculated using the least squares method. Ohms per millisecond.

[0061] The final extracted spatiotemporal correlation features of voltage and current are The smaller mean in this feature With negative slope It can be correlated with specific types of faults, validating the effectiveness of the technical process from rate of change sequence to statistical feature extraction.

[0062] S4. Integrate the virtual peer logic signal with the spatiotemporal correlation features of voltage and current to make a comprehensive judgment on the nature of the fault and the relative position of the fault point, and generate a fault judgment result containing the fault segment code and the fault type identifier. Optionally, generating a fault judgment result that includes a fault segment code and a fault type identifier includes: Analyze the virtual peer logic signal to obtain the directional logic value of each feeder and the abrupt timing sequence; The direction logic value corresponding to each feeder is binary encoded to generate a direction encoding vector; Based on the mutation time series, the normalized time difference of each feeder relative to the reference feeder is calculated to generate a time series feature vector. The direction encoding vector is fused with the timing feature vector to generate a logical feature vector characterizing the cooperative relationship of current abrupt changes in each feeder. The logical feature vector is concatenated with the voltage and current spatiotemporal correlation features to generate a comprehensive feature vector; Using the comprehensive feature vector, the system generates a section determination result regarding whether the fault point is located upstream or downstream of the incoming line side of the prefabricated substation, and a type determination result regarding whether the fault is permanent or transient. Based on the segment determination result and the type determination result, a fault judgment result containing the fault segment code and the fault type identifier is synthesized.

[0063] In one embodiment of the present invention, step S4 includes the following steps: Specifically, the virtual peer logic signal generated in step S2 and the spatiotemporal correlation features of voltage and current generated in step S3 are obtained. The virtual peer logic signal is a data sequence containing N tuples, and its general form is as follows: ,in It is the direction logic value of the i-th feeder, which can be 1 or 0; This corresponds to the normalized initial abrupt change time. The spatiotemporal correlation feature of voltage and current is a three-dimensional vector. It includes the ratio mean, ratio variance, and the slope of the ratio change trend.

[0064] First, analyze the virtual peer logic signals. Extract the direction logic values ​​of all feeders from each tuple to form a set. Simultaneously, the normalized initial mutation times of all feeders are extracted to construct a mutation time series sequence. Reference feeder It is always 0.

[0065] Next, the direction logic value corresponding to each feeder is binary encoded. This process directly converts the logic value... As a binary code, it generates an N-dimensional direction encoding vector. : , This vector intuitively represents the consistency between the direction of current change in each feeder and the reference feeder.

[0066] Based on the abrupt change time series, the normalized time difference of each feeder relative to the reference feeder is calculated to generate a time series feature vector. (Time series feature vector) The elements are directly derived from the mutation time sequence. constitute: , because This vector characterizes the degree of delay of each feeder current change relative to the earliest changing feeder.

[0067] The direction encoding vector and the timing feature vector are fused to generate a logical feature vector representing the cooperative relationship of current abrupt changes in each feeder. The fusion method uses vector concatenation, that is: , Among them, symbols This represents a vector concatenation operation. Therefore, the logical feature vector... The dimension is The first N elements are directional logic values, and the last N elements are normalized times.

[0068] logical feature vector Spatiotemporal correlation feature vectors of voltage and current The features are concatenated to generate a comprehensive feature vector. : , Comprehensive feature vector The dimension is It integrates all information from feeder coordinated response and incoming-bus transient correlation.

[0069] Fault determination results are generated using comprehensive feature vectors. This process relies on a pre-generated decision rule base, which is trained on 200 sets of industrial sensor measured data and corresponding simulation data covering different fault types and locations. The rule base contains two decision functions. The first function... Output a binary value :like Then the fault point is determined to be upstream of the incoming line side of the prefabricated substation; if Then it is determined to be downstream. The second function. Output a binary value :like Then the fault is determined to be a permanent fault; if If the fault is transient, it can be determined as a transient fault. The specific implementation of these two decision functions can be a logic tree based on threshold comparisons set for each dimension of the comprehensive feature vector, or a set of classification model parameters trained using historical data.

[0070] Based on the segment determination results With type determination result The final fault diagnosis result is then synthesized. This result is represented by a structured data object, which consists of two components: the fault segment code and the fault type identifier. Fault segment code... The generation rule is This means directly using the segment determination result. Fault type identifier. The generation rule is Ultimately, the fault diagnosis result... Represented as: This result will be passed to step S5 to generate a protection instruction.

[0071] For example, assume a prefabricated substation has four feeders. The set of directional logic values ​​is obtained by parsing the virtual peer logic signals. Mutation time sequence The spatiotemporal correlation characteristics of voltage and current are as follows: .

[0072] Generate directional encoding vector Generate time-series feature vectors. The fusion yields the logical feature vector. .

[0073] and After concatenation, an 11-dimensional comprehensive feature vector is generated. .

[0074] Input this vector into a preset decision rule base. The rule base for segment determination might be: if the normalized time corresponding to an element in the logical feature vector with a directional logical value of 0 is greater than the threshold 0.8, then it is determined to be a downstream fault. In this example, corresponding Meanwhile, the dynamic impedance slope This satisfies another association rule. Therefore, the decision function... Output This is determined to be a downstream fault. The rule for type determination might be: if the dynamic impedance variance... And the ratio mean If so, it is determined to be a permanent fault. In this example... and Decision function Output It was determined to be a permanent fault.

[0075] The final synthesized fault diagnosis result is In other words, a fault segment code of 0 represents a downstream fault, and a fault type identifier of 1 represents a permanent fault. This result verifies the effectiveness of the entire technical process from feature fusion to the application of decision rules.

[0076] S5. Based on the fault judgment result, generate and issue graded protection instructions for the incoming circuit breaker.

[0077] Optionally, generating and issuing graded protection instructions for the incoming circuit breaker based on the fault judgment result includes: Extract the section determination result text description and type determination result text description contained in the fault determination result; Semantic parsing is performed on the text description of the section determination result to identify keywords indicating that the fault point is located upstream or downstream of the incoming line side of the box-type substation, and these keywords are mapped to fault section codes. Semantic parsing is performed on the text description of the type determination result to identify keywords representing permanent or transient faults and map them as fault type identifiers; Based on the fault segment code and fault type identifier, the corresponding protection level and action delay parameters are matched from the preset protection strategy mapping table; Based on the protection level, the target tripping current setting of the incoming circuit breaker is determined, and combined with the action delay parameter, a graded protection command containing the target tripping current setting and the action delay parameter is generated. The hierarchical protection command is sent to the protection device of the prefabricated substation to control the incoming circuit breaker to perform the corresponding protection action.

[0078] like Figure 2 As shown, in one embodiment of the present invention, step S5 includes the following steps: Specifically, the system receives and reads the fault judgment result generated in step S4. This fault judgment result is a data structure containing two defined fields: fault segment code and fault type identifier. The fault segment code is an integer with a value of 0 or 1, and the fault type identifier is also an integer with a value of 0 or 1.

[0079] The textual description information contained in the fault judgment result is extracted. In this embodiment, the fault segment code and the fault type identifier are themselves discrete values, and their semantics have been determined through pre-definition. A fault segment code of 0 indicates a "downstream fault," meaning the fault point is located downstream of the incoming line side of the prefabricated substation; a code of 1 indicates an "upstream fault." A fault type identifier of 0 indicates a "transient fault"; an identifier of 1 indicates a "permanent fault." This mapping relationship is stored in the system memory.

[0080] The fault segment code is parsed to identify the relative location of the fault point it represents. The parsing process is a direct numerical mapping: if the read fault segment code... The fault point is located downstream of the incoming line of the prefabricated substation; if If it is, then it is marked as upstream.

[0081] The fault type identifier is parsed to identify the nature of the fault it represents. The parsing process is also a numerical mapping: if the read fault type identifier... If the system internally marks the fault as transient, then the fault is considered transient. If it is, then it is marked as a permanent fault.

[0082] Based on the fault location and nature obtained from the analysis, the corresponding protection level and action delay parameters are matched from the preset protection strategy mapping table. The protection strategy mapping table is a two-dimensional lookup table stored in the system's non-volatile memory. Its row index is determined by the fault segment code and fault type identifier, and the column output is the protection level code. and motion delay parameters This mapping table is optimized based on fault recording data from 200 sets of industrial sensors and corresponding system stability simulation results. An exemplary mapping relationship is shown below: When In the event of a permanent downstream fault, the mapped output is the protection level. Action delay milliseconds; when That is, in the event of an upstream transient failure, the mapped output is , millisecond.

[0083] Protection level based on matching Determine the target tripping current setting of the incoming circuit breaker. The target tripping current setting is the current threshold value at which the circuit breaker is instructed to perform a tripping operation. Its calculation formula is as follows: , in, This is the rated current of the incoming line of the prefabricated substation, in amperes. This value is a known system parameter. It is the protection level The corresponding current coefficient, whose value is also preset in the protection strategy mapping table, is related to the protection level. Related. For example, protection level. Possible corresponding coefficients Protection level Corresponding coefficient The coefficients were set based on the principle of ensuring reliable operation during faults while avoiding cascading tripping during load fluctuations or lower-level faults. The results were determined through statistical analysis of the aforementioned 200 sets of measured and simulated data.

[0084] Combined with the target tripping current setting With motion delay parameters This generates the final hierarchical protection instruction. This instruction is a formatted digital command that contains at least two core data fields: the tripping current setting. and motion delay Once the instruction is generated, it is sent to the protection device or intelligent controller connected to the incoming circuit breaker via the standard communication bus inside the prefabricated substation, such as Modbus RTU or IEC 61850 GOOSE message.

[0085] Upon receiving the graded protection command, the protection device immediately uses the instructions in the command. Update the setting register of its instantaneous overcurrent protection element, and Set the appropriate delay timer parameters. Afterward, the protection device continuously monitors the real-time current of the incoming line; when the current value exceeds... And the duration reached At that time, the drive actuator causes the incoming circuit breaker to trip, thereby completing the graded protection action.

[0086] For example, suppose the fault determination result received from step S4 is: fault segment code. Fault type identifier The system parses the code and identifies it as a downstream permanent fault.

[0087] Query the preset protection policy mapping table. Based on the combined condition of a downstream permanent fault, the mapping table returns the protection level. Action delay parameters Milliseconds. Meanwhile, the mapping table indicates the protection level. Corresponding current coefficient .

[0088] The rated current of the incoming line of this prefabricated substation is known. Ampere. Calculate the target tripping current setting using the formula: The generated graded protection command then contains two key data points: a tripping current setting of 756 amperes and an action delay of 300 milliseconds. This command is sent to the incoming circuit breaker protection device via the communication network. After the device updates the setting, if it detects that the incoming current has continuously exceeded 756 amperes for 300 milliseconds, it immediately issues a tripping command.

[0089] Based on the same inventive concept, such as Figure 3As shown, the present invention also provides a graded protection system for incoming line faults in a prefabricated substation, the system comprising: The data acquisition module is used to synchronously acquire the incoming three-phase current, bus three-phase voltage, and all feeder outgoing three-phase current of the box-type substation. When the incoming three-phase current exceeds the preset current threshold or the bus three-phase voltage exceeds the preset voltage threshold, a transient data set is generated. The transient data set includes the incoming transient current vector, the bus transient voltage vector, and the outgoing transient current vector. The data processing module is used to analyze the sudden change timing and phase relationship of the three-phase current of each feeder based on the transient current vector of the outgoing line in the transient data set, extract the cooperative response characteristics of the three-phase current of each feeder, and generate a virtual opposite-end logic signal characterizing the relative direction of the fault. The feature association module is used to extract the spatiotemporal correlation features of voltage and current based on the incoming transient current vector and the bus transient voltage vector in the transient data set. The fault judgment module is used to integrate the virtual peer logic signal with the spatiotemporal correlation features of voltage and current to make a comprehensive judgment on the nature of the fault and the relative position of the fault point, and generate a fault judgment result containing the fault segment code and the fault type identifier. The instruction sending module is used to generate and issue graded protection instructions for the incoming circuit breaker based on the fault judgment result.

[0090] It should be noted that the electrical connections between the various units described above do not necessarily represent direct or indirect connections. Any indirect connection method can be applied to the embodiments of the present invention as long as it achieves the purpose of the present invention. The above descriptions are merely exemplary embodiments of the present invention and should not be construed as limiting the scope of the present invention.

[0091] All equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or conventional techniques in the art not described herein.

Claims

1. A method for graded protection of incoming line faults in a prefabricated substation, characterized in that, The method includes: The system synchronously collects the incoming three-phase current, bus three-phase voltage, and all feeder outgoing three-phase current of the box-type substation. When the incoming three-phase current exceeds a preset current threshold or the bus three-phase voltage exceeds a preset voltage threshold, a transient data set is generated. The transient data set includes the incoming transient current vector, the bus transient voltage vector, and the outgoing transient current vector. Based on the transient current vectors of the outgoing lines in the transient data set, the sudden change timing and phase relationship of the three-phase currents of each feeder are analyzed, the coordinated response characteristics of the three-phase currents of each feeder are extracted, and a virtual opposite-end logic signal representing the relative direction of the fault is generated. Based on the incoming transient current vector and the bus transient voltage vector in the transient data set, the spatiotemporal correlation features of voltage and current are extracted. By integrating the virtual peer logic signal with the spatiotemporal correlation features of voltage and current, a comprehensive judgment is made on the nature of the fault and the relative position of the fault point, generating a fault judgment result that includes fault segment code and fault type identifier; Based on the fault diagnosis results, hierarchical protection instructions for the incoming circuit breaker are generated and issued.

2. The method for graded protection of incoming line faults in a prefabricated substation according to claim 1, characterized in that, The generated transient data set includes: Synchronously collect the incoming three-phase current, bus three-phase voltage, and all feeder outgoing three-phase current of the prefabricated substation; When the incoming three-phase current exceeds the preset current threshold or the bus three-phase voltage exceeds the preset voltage threshold, the fundamental component and each harmonic component in the incoming three-phase current are extracted to obtain the incoming current spectrum component. The fundamental component and harmonic components of the three-phase bus voltage are extracted to obtain the bus voltage spectrum components. The input current spectrum component and the bus voltage spectrum component are subjected to time-frequency joint analysis to obtain the time-frequency joint analysis results. Based on the time-frequency joint analysis results, a transient data set is generated, which includes the incoming transient current vector, the bus transient voltage vector, and the outgoing transient current vector.

3. The method for graded protection of incoming line faults in a prefabricated substation according to claim 2, characterized in that, The step of performing time-frequency joint analysis on the incoming current spectrum component and the bus voltage spectrum component to obtain the time-frequency joint analysis results includes: Align the input current spectrum component with the bus voltage spectrum component at the same time point to construct a current-voltage time-frequency joint matrix. Multidimensional mode decomposition is performed on the current-voltage time-frequency joint matrix to extract the transient energy distribution vector reflecting the energy distribution of each frequency component during the transient process and the phase coupling matrix reflecting the phase coupling relationship of the key frequency components. By fusing the transient energy distribution vector with the phase coupling matrix, a time-frequency joint analysis result characterizing the joint evolution of voltage and current spectra during the fault transient process is generated.

4. The method for graded protection of incoming line faults in a prefabricated substation according to claim 2, characterized in that, The step of extracting the coordinated response characteristics of the three-phase currents of each feeder based on the transient current vectors in the transient data set, and generating a virtual opposite-end logic signal characterizing the relative direction of the fault, includes: Obtain the outgoing transient current vector from the transient data set, and perform amplitude change detection on the outgoing transient current vector corresponding to the three-phase current of each feeder, and extract the amplitude change amount of the three-phase current of each feeder. Based on the amplitude mutation amount, the initial mutation time of the three-phase current of each feeder is identified, and a mutation time sequence of all feeders is generated. Based on the mutation time sequence, a feeder at a preset mutation time is selected as a reference feeder, and the phase difference between the outgoing transient current vector of each feeder and the outgoing transient current vector of the reference feeder is calculated. Based on the phase difference and the preset phase threshold, each feeder is assigned a directional logic value that characterizes whether the direction of the current change is consistent with the reference feeder. By integrating the directional logic values ​​of all feeders and associating them with the mutation timing sequence, a virtual peer logic signal is generated that characterizes the timing and directional relationship of the coordinated response of each feeder to a fault.

5. The method for graded protection of incoming line faults in a prefabricated substation according to claim 4, characterized in that, The extracted spatiotemporal correlation features of voltage and current include: Differentiate the incoming transient current vector and the bus transient voltage vector in the transient data set to calculate the instantaneous rate of change sequence of current and the instantaneous rate of change sequence of voltage; Calculate the ratio of the instantaneous rate of change of voltage sequence to the instantaneous rate of change of current sequence at the same time point, and generate a ratio sequence; Statistical analysis is performed on the ratio sequence to extract the ratio mean, ratio variance, and ratio change trend slope, thereby obtaining the spatiotemporal correlation characteristics of voltage and current.

6. The method for graded protection of incoming line faults in a prefabricated substation according to claim 5, characterized in that, The generation of fault judgment results, which include fault segment codes and fault type identifiers, includes: The virtual peer logic signal is numerically encoded and converted into a logic feature vector; The logical feature vector is concatenated with the voltage and current spatiotemporal correlation features to generate a comprehensive feature vector; Using the comprehensive feature vector, the system generates a section determination result regarding whether the fault point is located upstream or downstream of the incoming line side of the prefabricated substation, and a type determination result regarding whether the fault is permanent or transient. Based on the segment determination result and the type determination result, a fault judgment result containing the fault segment code and the fault type identifier is synthesized.

7. The method for graded protection of incoming line faults in a prefabricated substation according to claim 6, characterized in that, The step of numerically encoding the virtual peer logic signal and converting it into a logic feature vector includes: Analyze the virtual peer logic signal to obtain the directional logic value of each feeder and the abrupt timing sequence; The direction logic value corresponding to each feeder is binary encoded to generate a direction encoding vector; Based on the mutation time series, the normalized time difference of each feeder relative to the reference feeder is calculated to generate a time series feature vector. The direction encoding vector is fused with the timing feature vector to generate a logical feature vector characterizing the cooperative relationship of current abrupt changes in each feeder.

8. A method for graded protection of incoming line faults in a prefabricated substation according to claim 6, characterized in that, The step of generating and issuing graded protection instructions for the incoming circuit breaker based on the fault judgment result includes: The fault determination result is analyzed to obtain the fault segment code and fault type identifier; Based on the fault segment code and fault type identifier, the corresponding protection level and action delay parameters are matched from the preset protection strategy mapping table; Based on the protection level, the target tripping current setting of the incoming circuit breaker is determined, and combined with the action delay parameter, a graded protection command containing the target tripping current setting and the action delay parameter is generated. The hierarchical protection command is sent to the protection device of the prefabricated substation to control the incoming circuit breaker to perform the corresponding protection action.

9. A method for graded protection of incoming line faults in a prefabricated substation according to claim 8, characterized in that, The process of parsing the fault judgment result to obtain the fault segment code and fault type identifier includes: Extract the section determination result text description and type determination result text description contained in the fault determination result; Semantic parsing is performed on the text description of the section determination result to identify keywords indicating that the fault point is located upstream or downstream of the incoming line side of the box-type substation, and these keywords are mapped to fault section codes. Semantic parsing is performed on the text description of the type determination result to identify keywords representing permanent or transient faults and map them as fault type identifiers.

10. A graded protection system for incoming line faults in a prefabricated substation, characterized in that, The system includes: The data acquisition module is used to synchronously acquire the incoming three-phase current, bus three-phase voltage, and all feeder outgoing three-phase current of the box-type substation. When the incoming three-phase current exceeds the preset current threshold or the bus three-phase voltage exceeds the preset voltage threshold, a transient data set is generated. The transient data set includes the incoming transient current vector, the bus transient voltage vector, and the outgoing transient current vector. The data processing module is used to analyze the sudden change timing and phase relationship of the three-phase current of each feeder based on the transient current vector of the outgoing line in the transient data set, extract the cooperative response characteristics of the three-phase current of each feeder, and generate a virtual opposite-end logic signal characterizing the relative direction of the fault. The feature association module is used to extract the spatiotemporal correlation features of voltage and current based on the incoming transient current vector and the bus transient voltage vector in the transient data set. The fault judgment module is used to integrate the virtual peer logic signal with the spatiotemporal correlation features of voltage and current to make a comprehensive judgment on the nature of the fault and the relative position of the fault point, and generate a fault judgment result containing the fault segment code and the fault type identifier. The instruction sending module is used to generate and issue graded protection instructions for the incoming circuit breaker based on the fault judgment result.

Images