A method for power distribution network global feeder protection setting value intelligent checking and early warning

Abstract

This invention discloses a method for intelligent verification and early warning of feeder protection settings across the entire distribution network, belonging to the field of power system relay protection technology. The method includes collecting relevant distribution network data and calculating positive-sequence components, establishing a Thevenin equivalent circuit model and iteratively optimizing parameters, calculating the operating current based on Thevenin parameters, verifying the level difference of adjacent feeder settings in real time, generating a list of compliant settings and distributing it to the protection device. The relevant distribution network data includes current and voltage. The method calculates the instantaneous current increment and performs a three-level decomposition using Daubechies 4 wavelets, extracting high-frequency detail coefficients d1-d3 for fault prediction, and performing Stockwell and Hilbert transform analysis on the current data based on the fault prediction results. This invention can significantly improve the response speed and processing capability of the power grid under sudden events, ensuring the safety and stability of power grid operation.

Description

Technical Field

[0001] This invention relates to the field of power system relay protection technology, specifically to a method for intelligent verification and early warning of feeder protection settings across the entire distribution network. Background Technology

[0002] Traditional distribution network protection methods typically rely on static setting protection schemes. These schemes set protection settings based on historical experience and perform maintenance through periodic checks and manual adjustments. However, with the continuous changes in distribution network loads, the increasing intelligence of equipment, and the diversification of fault types, traditional static protection schemes have revealed some significant shortcomings. Existing distribution network protection setting verification schemes mainly rely on simple judgments based on fixed current and voltage data, lacking the ability to respond to dynamic changes. Furthermore, the interactions between feeders in the power grid are complex, and traditional methods fail to effectively handle these factors, easily leading to setting conflicts and false protection, thus affecting system stability and reliability. Therefore, how to further improve the intelligence and response speed of distribution network protection setting verification has become a key research focus. Summary of the Invention

[0003] In view of the above-mentioned problems, the present invention is proposed.

[0004] Therefore, the technical problem solved by this invention is: how to further improve the intelligence level and response speed of distribution network protection setting verification.

[0005] To address the aforementioned technical problems, this invention provides the following technical solution: a method for intelligent verification and early warning of feeder protection settings across a distribution network, comprising the following steps: collecting relevant data from the distribution network and performing positive sequence component calculations; establishing a Thevenin equivalent circuit model and iteratively optimizing the parameters; calculating the operating current based on the Thevenin equivalent circuit parameters; verifying the level difference of settings between adjacent feeders in real time; generating a list of compliant settings and sending it to the protection device; the relevant data from the distribution network includes current and voltage; calculating the instantaneous current increment and performing a three-level decomposition using Daubechies4 wavelets; extracting the first-level detail coefficient d1, the second-level detail coefficient d2, and the third-level detail coefficient d3 for fault prediction; performing Stockwell transform and Hilbert transform analysis on the current data based on the fault prediction results; calculating a mixed fault index based on the analysis results for multi-level fault early warning and generating corresponding measures; aggregating the Thevenin equivalent circuit parameters of the entire network, generating a fault heatmap and a setting conflict report; calculating the optimal setting combination for the entire network based on a game theory model; and sending the optimized settings to edge nodes via GOOSE messages.

[0006] As a preferred embodiment of the intelligent verification and early warning method for the protection settings of the entire distribution network feeder described in this invention, the step of establishing the Thevenin equivalent circuit model and performing iterative optimization of parameters includes: calculating the Thevenin equivalent voltage using positive sequence voltage data, and calculating the Thevenin equivalent impedance through changes in current and voltage; calculating the initial Thevenin equivalent voltage and initial Thevenin equivalent impedance using the three-point method; performing global optimization of the Thevenin equivalent voltage and Thevenin equivalent impedance using the particle swarm optimization algorithm to find the optimal initial solution; performing local optimization using the Gauss-Newton method based on PSO optimization, calculating the residuals and Jacobian matrix using the initial parameters obtained from PSO optimization to reflect the partial derivatives of the residuals with respect to each parameter; updating the parameters of the Thevenin equivalent voltage and Thevenin equivalent impedance using the Gauss-Newton optimization formula; setting a convergence threshold, updating the Thevenin equivalent voltage and Thevenin equivalent impedance in each iteration until the residual is less than the preset convergence threshold, and obtaining the optimized Thevenin equivalent voltage and Thevenin equivalent impedance parameters.

[0007] As a preferred embodiment of the intelligent verification and early warning method for the protection settings of the entire distribution network feeder as described in this invention, the step of calculating the operating current based on Thevenin equivalent circuit parameters, verifying the level difference of adjacent feeder settings in real time, generating a list of compliant settings and sending it to the protection device includes: calculating the operating current in the power grid based on the optimized Thevenin equivalent voltage and Thevenin equivalent impedance; calculating the setting range difference between two adjacent feeders according to the operating current of adjacent feeders in the power grid; setting a level difference threshold; if the setting range difference exceeds the level difference threshold, it indicates that the protection settings of adjacent feeders are unreasonable, leading to errors. If an operation or erroneous protection occurs, the sub-region containing adjacent feeders in the power grid is marked as a setting conflict sub-region, and an alarm message is generated to remind the operator to make adjustments. If the setting range does not exceed the threshold, the setting verification passes, and the protection setting remains unchanged. If the setting range equals the threshold, it is determined to be a critical state, and the sub-region containing adjacent feeders in the power grid is added to the setting optimization task list. Setting optimization is performed in the next setting cycle. Based on the obtained operating current and setting verification results, a complete setting list is generated. The setting list is then sent to each protection device using the GOOSE message protocol.

[0008] As a preferred embodiment of the method for intelligent verification and early warning of protection settings for the entire distribution network feeder as described in this invention, the calculation of the instantaneous current increment and the use of Daubechies 4 wavelets for three-level decomposition to extract the first-level detail coefficients d1, the second-level detail coefficients d2, and the third-level detail coefficients d3 for fault prediction includes: real-time acquisition of the current data of the current feeder and preprocessing it; calculating the difference between the current value at the current moment and the current value at the previous moment to obtain the instantaneous current increment; setting a threshold for the instantaneous current increment; and using Daubechies 4 wavelets to detect current signals with instantaneous increments exceeding the threshold for fault prediction. Hies4 wavelet transform is used for three-level decomposition. The wavelet decomposition process decomposes the signal into approximation coefficients and detail coefficients. After completing the three-level wavelet decomposition, the first-level detail coefficients d1, the second-level detail coefficients d2, and the third-level detail coefficients d3 are obtained. Energy calculation is performed on each level of detail coefficients to analyze high-frequency fluctuations in the current signal. Fourier transform is used to perform frequency domain transformation on the detail coefficients of each level. A high-frequency fluctuation threshold is set, and high-frequency fluctuations higher than the high-frequency fluctuation threshold are filtered out. The instantaneous increment of the current is combined with the energy characteristics. If the instantaneous increment of the current and the energy change of the detail coefficients both exceed the set threshold, it is determined to be a power grid fault.

[0009] As a preferred embodiment of the intelligent verification and early warning method for the protection settings of the entire distribution network feeder as described in this invention, the following steps are included: performing Stockwell and Hilbert transform analysis on the current data based on the fault judgment results, calculating the mixed fault index based on the analysis results, performing multi-level fault early warning, and generating corresponding measures. This includes: using wavelet transform to denoise the current signal determined to be a power grid fault, and using Stockwell transform to capture the short-time frequency features in the current signal to obtain the frequency components corresponding to each moment; finding the extreme points of the signal, constructing upper and lower envelopes using cubic Hermite interpolation, calculating the average value of the upper and lower envelopes, removing the envelope calculated from the average value from the original signal to obtain the first IMF component, removing the extracted first IMF component from the original signal to obtain the remaining signal, and continuing to perform IMF analysis on the remaining signal. Extracting signals until all IMF components are decomposed; applying Hilbert transform to each IMF component to obtain its instantaneous frequency; calculating the signal's resolution in the frequency domain by analyzing its spectral and time-domain bandwidth; calculating the mixed fault index of the power grid based on the amplitude spectrum obtained from Stockwell transform, the instantaneous frequency obtained from Hilbert transform, and the frequency resolution; setting high-risk and medium-risk thresholds, with the high-risk threshold being greater than the medium-risk threshold; if the mixed fault index is greater than or equal to the high-risk threshold, it indicates a serious potential fault in the power grid, triggering a high-risk alarm, automatically isolating faulty equipment, and adjusting protection settings; if the mixed fault index is less than the high-risk threshold but greater than or equal to the medium-risk threshold, it indicates a medium potential fault in the power grid, prompting operators to monitor the power grid status; if the mixed fault index is less than the medium-risk threshold, it indicates normal power grid operation, requiring continuous monitoring of the power grid.

[0010] As a preferred embodiment of the method for intelligent verification and early warning of setting values ​​for feeder protection across the entire distribution network as described in this invention, the following steps are taken: The aggregation of the entire network's Thevenin equivalent circuit parameters to generate a fault heatmap and setting conflict report refers to collecting data from the Thevenin equivalent circuit parameters and mixed fault index records of all feeders, storing all aggregated data on a cloud platform, generating a fault heatmap using data visualization technology, classifying areas into low, medium, and high risk levels according to the mixed fault index, and marking them with varying shades of the same color; based on the aggregated data, checking for setting conflicts in the power grid and automatically generating a setting conflict report.

[0011] As a preferred embodiment of the intelligent verification and early warning method for feeder protection settings across the entire distribution network as described in this invention, the following steps are taken: Calculating the optimal setpoint combination for the entire network based on a game theory model, and distributing the optimized setpoints to edge nodes via GOOSE messages, refers to dividing the distribution network into multiple sub-regions based on its overall structure, modeling the feeders and protection devices within each sub-region; optimizing the protection setpoints for each sub-region using a non-cooperative game theory model, where each protection device is considered a participant in the game; and selecting the most suitable protection setpoint based on the operating state and objective function of the sub-region, by solving the Nash equation of the game. The equilibrium point yields the optimal combination; during the game, the gradient descent method is used to optimize the protection settings of each sub-region. By iteratively updating the setting parameters of each sub-region, the comprehensive objective of minimizing the cost and risk of each sub-region is achieved. After multiple rounds of iterative calculations, the optimal protection setting combination is finally obtained; a list of optimal setting configurations for the entire network is generated based on the optimal protection setting combination calculated by the game theory model, and the optimal setting configuration list is distributed to each protection device in the distribution network through the GOOSE message protocol; the positive sequence component calculation refers to analyzing the three-phase current using the positive sequence component method after collecting current and voltage data, and calculating the positive sequence components.

[0012] The beneficial effects of this invention are as follows: This invention adopts a technical solution that combines dynamic optimization and intelligent learning. By introducing positive-sequence and zero-sequence component calculations for Thevenin equivalent circuit parameters, iterative optimization, wavelet transform, and Stockwell transform, it can collect current and voltage data in real time and calculate accurate operating currents. At the same time, it calculates the optimal setpoint combination through a game theory model, realizing automated optimization and dynamic verification of distribution network protection settings. This invention optimizes the shortcomings of existing technologies in dealing with power grid load fluctuations, fault warnings, and setpoint conflicts, and can significantly improve the power grid's response speed and processing capabilities under emergencies, ensuring the safety and stability of power grid operation. Attached Figure Description

[0013] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 The above is a flowchart of a method for intelligent verification and early warning of feeder protection settings across the entire distribution network, provided as an embodiment of the present invention.

[0015] Figure 2 This is a schematic flowchart illustrating the fault detection process of a method for intelligent verification and early warning of feeder protection settings across the entire distribution network, provided as an embodiment of the present invention.

[0016] Figure 3 This is a schematic diagram of the game theory-based optimization setting issuance process for a method of intelligent verification and early warning of feeder protection settings across the entire distribution network, provided as an embodiment of the present invention. Detailed Implementation

[0017] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0018] Example 1, referring to Figures 1-3 This is the first embodiment of the present invention, which provides a method for intelligent verification and early warning of feeder protection settings across the entire distribution network, comprising:

[0019] S1: Collect relevant data of the distribution network and perform positive sequence component calculation, establish the Thevenin equivalent circuit model and perform iterative optimization of parameters, calculate the operating current based on the Thevenin parameters, verify the level difference of adjacent feeder settings in real time, generate a list of qualified values ​​and send it to the protection device.

[0020] Specifically, collecting relevant data from the power distribution network and performing positive-sequence component calculations includes:

[0021] Start various monitoring devices in the distribution network, such as current transformers (CT), voltage transformers (PT), and temperature sensors. These devices collect real-time operating data of the power grid, including current and voltage. This data is the basis for subsequent power grid protection setting verification and fault early warning.

[0022] After collecting current and voltage data, the positive sequence component method is used to analyze the three-phase current and calculate the positive sequence component.

[0023] For example, suppose the instantaneous value of the collected three-phase current is A, A, A; among them. , , These are the phasor values ​​of the currents in phases A, B, and C, respectively. Representing the phase angle, the positive sequence component... .

[0024] in, ; j is the imaginary unit. Let be the rotation factor, then we can calculate... , It is the complex rotation factor in the symmetric component method.

[0025] In power distribution networks, current transformers (CTs), voltage transformers (PTs), and temperature sensors collectively constitute a highly efficient real-time data acquisition system. These devices enable the accurate and real-time acquisition of key operating parameters of the power grid (such as current, voltage, load, and temperature). This data provides crucial foundational information for the power grid's protection system, ensuring precise monitoring and assessment during normal operation. After acquiring current and voltage data, the positive sequence component method is used to analyze the three-phase currents and calculate the positive sequence components. Positive sequence component calculation accurately distinguishes between normal and fault states in the power grid. When a fault occurs in the power grid, the negative sequence and zero sequence components exhibit abnormal changes, while the positive sequence component represents the current or voltage waveform under normal conditions. By calculating the positive sequence components, the potential for faults in the power grid can be accurately identified, providing data support for subsequent protection setting adjustments.

[0026] Furthermore, establishing the Thevenin equivalent circuit model and performing iterative parameter optimization includes:

[0027] Calculate the Thevenin equivalent voltage using positive-sequence voltage data. :

[0028] ;

[0029] In the formula, It is a positive sequence voltage. It is the phase angle of the voltage (i.e., the phase difference between the voltage and the current).

[0030] j is the imaginary unit;

[0031] Thevenin equivalent impedance is calculated by varying the current and voltage. :

[0032] ;

[0033] In the formula, It is a resistor. Is it a sensory or capacitive antibody?

[0034] The initial Thevenin equivalent voltage was calculated using the three-point method. and the initial Thevenin equivalent impedance :

[0035] ;

[0036] In the formula, and These are the positive sequence voltage and current at time k, respectively. and These are the positive sequence voltage and current at time k-1, respectively.

[0037] For example, measurement data from three consecutive time points are selected: hour, kV, A; hour, kV, A; hour, kV, A, where Indicates the first The positive sequence voltage value at time 1. Indicates the first The positive sequence current value at time t is determined by the formula. Calculate, substitute Data obtained Ω (a negative value indicates that the current increases when the voltage decreases). kV.

[0038] The particle swarm optimization algorithm (PSO) is used to globally optimize the Thevenin equivalent voltage and the Thevenin equivalent impedance parameters to find the optimal initial solution, including:

[0039] In the particle swarm optimization algorithm, each particle represents a set of Thevenin equivalent voltage and Thevenin equivalent impedance parameters. The particle swarm is initialized and the number of particles is set. Each particle has randomly initialized Thevenin equivalent voltage and Thevenin equivalent impedance parameters. The fitness function is used to measure the quality of the current particle parameters. The particle updates its position and velocity according to the fitness value to obtain the initially optimized Thevenin equivalent voltage and Thevenin equivalent impedance parameters.

[0040] For example, setting the number of particles Each particle represents a set of parameters to be optimized. Its initial position is randomly distributed in the parameter space: Thevenin equivalent voltage kV (determined based on system rated voltage), Thevenin equivalent impedance parameters Ω (determined based on the range of line parameters). The particle velocity is initialized to 10% of the dimensional parameter search range, i.e., the upper limit of voltage velocity is 1.5kV / cycle, and the upper limit of impedance velocity is 0.19Ω / cycle.

[0041] Based on the particle swarm optimization algorithm, the Gauss-Newton method is used for local optimization. By adjusting the Thevenin equivalent voltage and Thevenin equivalent impedance parameters, the model error is further minimized, including:

[0042] Calculate the residuals using the initial parameters obtained from PSO optimization:

[0043] ;

[0044] In the formula, It is a residual. This is the actual measured voltage. It is the estimated Thevenin equivalent voltage. These are the estimated Thevenin equivalent impedance parameters. It is positive sequence current;

[0045] Calculate the Jacobian matrix J, which reflects the partial derivatives of the residuals with respect to each parameter:

[0046] ;

[0047] In the formula, The parameters to be optimized are... It is a resistor. It is inductive reactance, and T represents the transpose operation. Represents the error norm. Representing partial derivatives;

[0048] Update the parameters of Thevenin equivalent voltage and impedance using the Gauss-Newton optimization formula:

[0049] ;

[0050] In the formula, It is the parameter update amount. It is the transpose of the Jacobian matrix. It is the pseudo-inverse of the Jacobian matrix.

[0051] The convergence threshold is set by the maximum number of iterations. The Thevenin equivalent voltage and Thevenin equivalent impedance are updated in each iteration until the residual ϵ ​​is less than the preset convergence threshold, and the optimized Thevenin equivalent voltage and Thevenin equivalent impedance parameters are obtained.

[0052] In one implementation, Thevenin equivalent parameter identification, applied to 10kV distribution network fault identification to assist fault location and system stability analysis, firstly involves continuously acquiring three sets of complex measurements of positive-sequence voltage and positive-sequence current during normal system operation using a synchronous phasor measurement unit (PMU) installed at the distribution network bus. The positive-sequence voltages are respectively... , , The positive sequence currents are respectively , , Next, parameter initialization is performed, setting the initial value of the Thevenin equivalent voltage to... The initial value of the equivalent resistance is set to The initial value of the equivalent reactance is set to (corresponding equivalent impedance) At the same time, set the convergence threshold. (Convergence is determined if the residual is less than this value), Maximum number of iterations Then, the steps of "calculate residuals → check convergence → update parameters" are repeated. During the first iteration, the residuals are calculated. The (L2 norm) is approximately 5.23, which is greater than the threshold of 0.5. Therefore, the parameter is adjusted based on the residual's "sensitivity" to the parameter. , , In subsequent iterations, the residual gradually decreases. When the 8th iteration is performed, the calculated residual is approximately 0.48, which is less than the threshold of 0.5, satisfying the convergence condition and stopping the iteration. The final optimized Thevenin equivalent parameters are obtained.

[0053] Accurate calculation of Thevenin equivalent voltage and impedance effectively simplifies power grid analysis, providing a foundation for subsequent protection setting verification and fault early warning. By applying dynamic optimization algorithms to the model, voltage and impedance parameters can be adjusted in real time during actual power grid operation, resulting in more accurate power grid protection settings. Using Particle Swarm Optimization (PSO) to globally optimize the Thevenin equivalent voltage and impedance parameters quickly finds the global optimum in the parameter space. PSO avoids the pitfalls of getting stuck in local optima, a problem common in traditional methods, by simulating a swarm intelligence search process. This algorithm efficiently handles multi-dimensional parameter optimization problems in power grids, ensuring accurate optimization of protection settings in large-scale power grid environments, thereby improving the flexibility and response speed of power grid protection strategies. Based on the initial parameters obtained from PSO optimization, the Gauss-Newton method is used for local optimization, bringing the Thevenin equivalent voltage and impedance parameters to their optimal state. The Gauss-Newton method improves the accuracy and stability of the power grid model by continuously adjusting parameters to minimize errors. Compared to single optimization methods, the combination of PSO and Gauss-Newton enables the model to achieve excellent results in both global and local optimization, avoiding the limitations of relying solely on a single algorithm. This invention employs techniques such as residual calculation and Jacobian matrix calculation to minimize errors and improve model accuracy by continuously adjusting the Thevenin equivalent voltage and impedance parameters. This step effectively reduces the risk of setting inaccuracies caused by dynamic changes in the power grid, improving the real-time performance and adaptability of power grid protection settings. Compared to traditional static methods, error minimization is particularly important in variable power grid environments, ensuring that the power grid maintains efficient and accurate protection performance even under load fluctuations and equipment failures. This invention provides accurate optimized configurations for feeder protection settings across the entire distribution network by calculating the optimal Thevenin equivalent voltage and impedance parameters. The optimization method based on network-wide data comprehensively considers the power grid status of each feeder and region, avoiding false or missed protection due to setting conflicts.

[0054] Furthermore, based on the Thevenin parameters, the operating current is calculated, the difference in settings between adjacent feeders is checked in real time, and a list of compliant values ​​is generated and sent to the protection device, including:

[0055] Calculate the operating current Y in the power grid based on the optimized Thevenin equivalent voltage and Thevenin equivalent impedance:

[0056] ;

[0057] In the formula, It is the optimized Thevenin equivalent voltage. It is the optimized Thevenin equivalent impedance;

[0058] For example, if the optimization yields kV (Thevenin equivalent voltage). Ω ( (where 0.8Ω is the resistive component and 0.6Ω is the reactive component). modulus Ω, phase angle Therefore Ω, operating current A indicates that the operating current amplitude is 11200A and the phase angle lag is 36.87°.

[0059] The operating current is the standard value that triggers the protection device to operate. It reflects the minimum current required by the power grid when a fault or abnormality occurs. The calculated operating current will provide basic data for subsequent setting verification. This current value will be used to help determine whether the current protection settings of the power grid are applicable and whether the parameters of the protection device need to be adjusted.

[0060] Calculate the setting range difference between two adjacent feeders based on their operating currents in the power grid. :

[0061] ;

[0062] In the formula, and These are the operating currents of two adjacent feeders.

[0063] A fixed range threshold is set based on historical data and statistical analysis. If the set value range threshold is exceeded, it indicates that the protection settings of adjacent feeders are unreasonable, which may lead to malfunctions or false protection. The area where adjacent feeders are located in the power grid is marked as a set value conflict area, and an alarm message is generated to remind the operator to make adjustments. If the set value range... If the set value range threshold is not exceeded, the set value verification passes, and the protection set value remains unchanged; if the set value range threshold is exceeded, the protection set value is verified. If the value is equal to the set value range threshold, it is determined to be a critical state. The area where the adjacent feeder is located in the power grid is added to the set value optimization task list, and set value optimization is performed in the next set value tuning cycle.

[0064] For example, based on historical operational data statistics, a fixed value difference threshold is set. ,in The rated current of the feeder is 0.3, which is an empirical coefficient (representing a 30% margin). If the operating currents of two adjacent feeders F1 and F2 are respectively... A and A, then the constant range A. If the rated current A, then the fixed value difference threshold A.

[0065] because A A. This is determined to be a setting conflict, and the protection settings need to be adjusted to ensure selectivity.

[0066] Based on the obtained operating current and setting verification results, a complete setting list is generated, listing the operating current, setting value and related protection parameters of each feeder.

[0067] The setpoint list is sent to each protection device using Generic Object Oriented SubstationEvent (GOOSE message protocol).

[0068] By using optimized Thevenin equivalent voltage and impedance, the operating current in the power grid is calculated to provide data support for subsequent setting verification. The operating current reflects the standard operating value of the protection device in the event of a fault or power grid anomaly, and is one of the key parameters in the protection system. The calculated operating current allows determination of the applicability of the current protection settings in the power grid. By calculating the operating current of adjacent feeders and comparing their setting differences, effective verification of power grid protection settings can be achieved. It can automatically identify and correct protection setting conflicts in the power grid that may lead to maloperation, thereby reducing power grid protection failures caused by improper setting settings. By automatically generating a setting list and accurately distributing it to each protection device, the protection settings of the entire network can be adjusted quickly and accurately, ensuring that protection measures in all aspects of the power grid are synchronized, thus significantly improving the fault response speed of the power grid and enhancing overall safety and stability. The compliant setting list refers to the list of protection device setting parameters that meet the power grid protection requirements, generated after a complete verification and optimization process.

[0069] S2. Calculate the instantaneous current increment and use the Daubechies 4th order compactly supported orthogonal wavelet (Daubechies4 wavelet) for 3-level decomposition. Extract the first level detail coefficient d1, the second level detail coefficient d2 and the third level detail coefficient d3 for fault prediction. Based on the fault prediction results, perform Stockwell transform and Hilbert transform analysis on the current data. Based on the analysis results, calculate the hybrid fault index to perform multi-level fault early warning and generate corresponding measures.

[0070] Specifically, the instantaneous current increment is calculated and decomposed into three levels using Daubechies4 wavelets. The first-level detail coefficients d1, the second-level detail coefficients d2, and the third-level detail coefficients d3 are extracted for fault prediction, including:

[0071] After real-time acquisition and preprocessing of the current data of the current feeder, the difference between the current value at the current moment and the current value at the previous moment is calculated to obtain the instantaneous current increment. The instantaneous current increment threshold is set according to the current load conditions of the power grid and the system operating status. For current signals whose instantaneous increment exceeds the instantaneous current increment threshold, Daubechies4 wavelet transform is used for three-level decomposition.

[0072] For example, under normal load conditions, a threshold for instantaneous current increment is set. ,in A is the rated current of the feeder, and 0.2 is the sensitivity coefficient (a 20% mutation). A.

[0073] When the current difference between two consecutive sampling points is detected At time A, where The sampling interval (e.g., 0.02s) is determined by... A A triggers wavelet decomposition analysis.

[0074] Wavelet decomposition decomposes the signal into approximate coefficients (low-frequency information) and detail coefficients (high-frequency information). After completing three levels of wavelet decomposition, the first level detail coefficients d1, the second level detail coefficients d2, and the third level detail coefficients d3 are obtained. Energy calculation is performed on each level detail coefficient to analyze the high-frequency fluctuations in the current signal.

[0075] ;

[0076] In the formula, It is the energy of the i-th layer. It is the nth sampling point of the detail coefficient of the i-th layer, and N is the number of sampling points of the signal;

[0077] For example, for the first level detail coefficient d1, assume the number of sampling points... The values ​​of the first 5 points are (Unit: per unit), according to the formula, the energy contributed by the first 5 points is:

[0078]

[0079] The total energy is obtained after calculating all 100 points. .

[0080] Fourier transform is used to perform frequency domain transformation on the detail coefficients of each layer. A high-frequency fluctuation threshold is set by support vector machine to filter out high-frequency fluctuations that exceed the high-frequency fluctuation threshold. The instantaneous increment of current is combined with energy characteristics. If the instantaneous increment of current and the energy change of detail coefficients both exceed the set threshold, it is determined to be a power grid fault.

[0081] For example, a high-frequency fluctuation threshold can be determined by training a support vector machine. (per unit value). When diagnosing a fault, two conditions must be met simultaneously: (1) The instantaneous current increment exceeds the corresponding threshold. As in the previous example (2) The high-frequency components of the detail coefficients exceed the high-frequency fluctuation threshold. And the energy changes are significant. For example, for After performing a Fourier transform on the layer, the amplitude of the 1250Hz frequency component was detected to be 0.15 (per unit), which exceeds the high-frequency fluctuation threshold. At the same time, the energy of this layer changes from its normal state. Sudden increase to The energy increase reached 204%. Since the instantaneous current increment (60A>40A) and the energy change of the detail coefficient (204% increase) both exceeded their respective set thresholds, it was determined to be a power grid fault.

[0082] The three-level decomposition using Daubechies4 wavelet transform involves the input signal x[n] first passing through a set of low-pass filters (corresponding to the scaling function of the db4 wavelet) and a set of high-pass filters (corresponding to the wavelet function of the db4 wavelet). The result after low-pass filtering is downsampled to obtain the first-level approximation coefficients A1, and the result after high-pass filtering is downsampled to obtain the first-level detail coefficients d1.

[0083] Using the obtained first-level approximation coefficients A1 as the new input signal, the same low-pass and high-pass filter banks are applied again, and the results are downsampled to obtain the second-level approximation coefficients A2 and the second-level detail coefficients d2, respectively.

[0084] Using the second-layer approximation coefficient A2 as input, low-pass and high-pass filter banks are applied and downsampled to obtain the third-layer approximation coefficient A3 and the third-layer detail coefficient d3.

[0085] Ultimately, the original signal is decomposed into three levels: the first level detail coefficient d1, the second level detail coefficient d2, and the third level detail coefficient d3, as well as the third level approximation coefficient A3 at the coarsest scale.

[0086] By acquiring current signals in real time and calculating instantaneous increments, this invention can effectively capture the initial stages of load fluctuations or sudden faults in the power grid. Instantaneous current increments are precursors to power grid faults, rapidly reflecting sudden changes or abnormal fluctuations in current before a fault occurs. Using wavelet transform for signal decomposition, particularly the db4 wavelet, can separate high-frequency noise and fault characteristics in the current signal, thus more accurately identifying faults in the power grid. During wavelet decomposition, by extracting all detail coefficients, this invention can capture high-frequency fluctuations in the current signal. Wavelet transform has excellent time-frequency localization characteristics, accurately extracting rapid fluctuations in the current signal over short periods, which is crucial for predicting power grid faults. By comprehensively considering the energy changes of instantaneous increments, detail coefficients, and frequency domain characteristics, this invention can promptly detect potential faults and provide early warnings under conditions of power grid load fluctuations or imbalances. Compared to traditional static setting protection methods, this fault determination method has higher sensitivity and real-time response capabilities, effectively avoiding false protection and missed protection phenomena.

[0087] Furthermore, based on the fault diagnosis results, Stockwell and Hilbert transform analyses are performed on the current data. Based on the analysis results, a hybrid fault index is calculated to provide multi-level fault early warning and generate corresponding measures, including:

[0088] Wavelet transform is used to denoise the current signal identified as a power grid fault, and Stockwell transform is used to capture the short-time frequency features in the current signal to obtain the frequency components corresponding to each time moment.

[0089] The extreme points of the signal are found, and upper and lower envelopes are constructed using cubic Hermite interpolation (a polynomial interpolation method that interpolates not only the function value but also its derivative). The average value of the upper and lower envelopes is calculated, and the envelope calculated from the average value is removed from the original signal to obtain the first IMF component. The extracted first IMF component is then removed from the original signal to obtain the remaining signal. IMF component extraction is continued on the remaining signal until all IMF components are decomposed. Here, IMF represents an oscillation mode with a single frequency scale in the signal, and each IMF component represents a local feature of the signal at different time scales.

[0090] For example, suppose the sequence of extreme points detected within a 0.1-second time window is as follows:

[0091] ;

[0092] in, Let x represent time (in seconds), and x∈[1,2,3,4]. This indicates the current amplitude at that moment (per unit value, based on the rated current).

[0093] The upper envelope is obtained by three Hermite interpolations. and lower envelope Then the envelope average value For example, at time... At point s, if the interpolation calculation yields... , ,but .

[0094] Apply the Hilbert transform to each IMF component to obtain the instantaneous frequency of each IMF;

[0095] By analyzing the signal's spectral width and time domain width The resolution of the signal in the frequency domain is calculated. ;

[0096] ;

[0097] The hybrid fault index H of the power grid is calculated based on the amplitude spectrum obtained from the Stockwell transform, the instantaneous frequency obtained from the Hilbert transform, and the frequency resolution.

[0098] ;

[0099] In the formula, M is the number of signal data points. It is the amplitude spectrum of the signal. It is the time variable of the signal. It is the frequency variable of the signal. It is the instantaneous frequency, and i is the index;

[0100] Based on historical data, statistical analysis is performed to set high-risk and medium-risk thresholds, with the high-risk threshold being greater than the medium-risk threshold. If the mixed fault index is greater than or equal to the high-risk threshold, it indicates a serious potential fault in the power grid, triggering a high-risk alarm, automatically isolating faulty equipment, and adjusting protection settings. If the mixed fault index is less than the high-risk threshold but greater than or equal to the medium-risk threshold, it indicates a medium potential fault in the power grid, prompting operators to monitor the power grid status. If the mixed fault index is less than the medium-risk threshold, it indicates that the power grid is operating normally, and continuous monitoring of the power grid is conducted.

[0101] For example, a high-risk threshold can be set by statistical analysis of historical fault data. medium threshold Among them, the mixed failure index The value range is [0,1], where 0 represents normal and 1 represents the most severe fault. When calculated... At that time, among them This represents the time-frequency amplitude spectrum (values ​​range from 0 to 1). Frequency resolution (rad / s) The instantaneous frequency (Hz) is due to This triggered a high-risk alert.

[0102] Wavelet transform in power grid fault detection helps improve fault diagnosis accuracy through denoising. Multi-scale analysis effectively removes high-frequency noise from the signal while preserving its main features. Stockwell transform, by combining time and frequency domain analysis, can accurately capture short-time frequency characteristics in the power grid, especially in the initial stages of a fault. This advantage makes it more suitable than traditional Fourier transform for instantaneous fault detection in the power grid. By acquiring the changes in current signals at different times and frequencies, the ST transform can identify fault signals in the power grid in real time and provide data support for subsequent fault warning and protection setting adjustments. The upper and lower envelopes constructed using cubic Hermite interpolation can accurately extract the intrinsic mode functions (IMFs) of the signal. These IMF components can effectively separate local frequency components in the current signal when a fault occurs, helping to identify short-time, high-frequency abnormal changes in the power grid. Applying Hilbert transform to each IMF component yields the instantaneous frequency, which is crucial for further analysis of power grid faults. Instantaneous frequencies reflect the dynamic changes of the signal, especially when frequency changes drastically, enabling timely detection of faults or anomalies in the power grid. By calculating the instantaneous frequency, not only can the occurrence time of a fault be determined, but also its type (such as short circuit or open circuit) can be revealed. Based on the above transformation results (amplitude spectrum, instantaneous frequency, frequency resolution), the calculated hybrid fault index plays an important role in power grid fault early warning. The hybrid fault index can quantify the potential faults in the power grid and determine the severity of the fault based on set high-risk and medium-risk thresholds. The introduction of the hybrid fault index not only improves the intelligence of fault early warning but also enhances the automation level of power grid fault handling.

[0103] S3. Aggregate the Thevenin parameters of the entire network, generate a fault heat map and a setpoint conflict report, calculate the optimal setpoint combination of the entire network based on the game theory model, and send the optimized setpoints to the edge nodes through GOOSE messages;

[0104] Specifically, aggregating the Thevenin parameters of the entire network to generate fault heatmaps and setpoint conflict reports involves collecting data from the Thevenin equivalent circuit parameters and fault index records of all feeders. This data comes from monitoring equipment in various areas of the distribution network, including current, voltage, load, and equipment health status. All aggregated data is stored on a cloud platform, and fault heatmaps are generated using data visualization technology. Based on the mixed fault index, areas are classified into low, medium, and high risk levels and marked with different shades of the same color. For example, low-risk areas are green, medium-risk areas are yellow, and high-risk areas are red.

[0105] Based on aggregated Thevenin equivalent voltage, Thevenin equivalent impedance parameters, and hybrid fault index data, the system checks for setting conflicts in the power grid, calculates the setting difference between adjacent feeders, and compares it with the set safety standard. If the difference is greater than the safety standard, it is considered a setting conflict. After detecting a setting conflict area, a setting conflict report is automatically generated. The report details all conflict areas, including the conflict area, conflict type, conflict impact, and optimization suggestions.

[0106] For example, suppose the safety standard for the setting difference between adjacent feeders is set as follows: ,in Rated current, lower limit Ensure sufficient selectivity (avoid simultaneous actions), upper limit To avoid excessively large level differences that could lead to excessively long backup protection delays. If If A, then the safety range is [30A, 100A]. If the calculated level difference is... If the setting exceeds the upper limit of the safety range, it is considered a setting conflict. The conflict area refers to the specific distribution network area or feeder location, the conflict type refers to over-protection, false protection, etc., the conflict impact refers to the possible failure or malfunction of power grid equipment, and the optimization suggestions include setting adjustment suggestions to resolve the conflict, such as modifying the setting of the protection device, adjusting the step difference, etc.

[0107] Send the setting conflict report along with optimization suggestions to the power grid operation and maintenance personnel. The operation and maintenance personnel can adjust the power grid protection device according to the suggestions in the report to ensure the rationality and effectiveness of the power grid protection settings.

[0108] After adjustment, the new settings will be sent to each protection device via the GOOSE message protocol, triggering the reconfiguration of the protection devices. The adjusted settings will be verified again through the fault heat map to ensure the effectiveness of the adjustment measures and to monitor whether the setting conflicts have been resolved.

[0109] By aggregating Thevenin equivalent circuit parameters (such as voltage and impedance) and hybrid fault index data from the entire network, key data from various regions of the power grid can be centrally managed. The generation and visualization of fault heatmaps allow for a graphical representation of the power grid's operational status, intuitively presenting the fault risks in different areas. By calculating the setting differences between adjacent feeders and comparing them with preset safety standards, potential setting conflicts in the power grid can be detected in real time. These conflicts may lead to malfunctions or missed trips in protection devices, affecting grid stability. The generated setting conflict report details the conflict area, type, and impact, and provides targeted optimization suggestions. These optimization suggestions guide maintenance personnel in making targeted setting adjustments. For example, the report may recommend adjusting the protection settings of certain feeders or narrowing the differences between protection settings to ensure the rationality of power grid protection settings. These optimization measures can significantly reduce the occurrence of malfunctions and missed trips, improving the protection performance and stability of the power grid. The adjusted settings are then verified again using fault heatmaps to ensure the effectiveness of the adjustments. This feedback mechanism not only ensures that the power grid operates more stably after the settings are adjusted, but also monitors the adjustment effect in real time, thereby optimizing the power grid operation and maintenance strategy.

[0110] Furthermore, based on the game theory model, the optimal setpoint combination for the entire network is calculated, and the optimized setpoint is sent to the edge nodes via GOOSE messages. Based on the overall structure of the distribution network, the power grid is divided into multiple sub-regions, and the feeders and protection devices in each sub-region are modeled.

[0111] The protection settings of each sub-region are optimized using a non-cooperative game model. Each protection device in the sub-region is regarded as a participant in the game. Each protection device in the sub-region selects the most suitable protection settings according to the operating status of its own region and the objective function. By solving the Nash equilibrium point of the game, it is ensured that the setting selection of each sub-region is the optimal combination based on the overall benefits, so as to avoid protection failure caused by the incoordination of protection settings between different sub-regions in the power grid.

[0112] For example, suppose the distribution network is divided into 3 sub-regions, and the cost function of each sub-region is: ,in, For the first The overall performance index of the sub-region Let A be the protection setting for the i-th sub-region. The second-order weighting coefficient (A) - ²), The first-order weighting coefficient (A) -1 Setting parameters: A - ², A -1; A - ², A -1 ; A - ², A -1 By solving The Nash equilibrium point is obtained: A, A, A, , , This represents the optimal protection setpoint for each subregion obtained by solving for the Nash equilibrium. The superscript * is a standard symbol in mathematical optimization, used for distinction: Let be the protection setting (decision variable) for the i-th sub-region. This is the optimal protection setpoint (Nash equilibrium solution) for the i-th sub-region.

[0113] During the game, the gradient descent method is used to optimize the protection setpoints of each sub-region. By iteratively updating the setpoint parameters of each sub-region, the comprehensive objective of minimizing the cost and risk of each sub-region is achieved. After multiple rounds of iterative calculation, the optimal combination of protection setpoints is finally obtained.

[0114] Based on the optimal protection setting combination calculated by the game theory model, a list of optimal setting configurations for the entire network is generated. This list contains the optimized settings for each feeder and protection device, as well as the corresponding load and equipment parameters. The optimal setting configuration list is then distributed to each protection device in the distribution network via the GOOSE message protocol.

[0115] While existing distribution network protection technologies achieve basic grid protection through static setting adjustments, they lack flexibility and adaptability in response to varying grid operating environments. This invention, by introducing a game theory model and combining a gradient descent optimization algorithm with the GOOSE message protocol, successfully addresses the problems of lack of coordination, slow response, and setting conflicts inherent in traditional technologies. By dynamically optimizing protection settings and achieving network-wide coordination, the grid can respond more efficiently and intelligently to load fluctuations and fault warnings, ensuring stable grid operation, reducing the risk of accidents caused by improper protection settings, and improving the economy and reliability of the distribution network.

[0116] This embodiment also provides a computer device applicable to the method of intelligent verification and early warning of feeder protection settings across the entire distribution network, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the method of intelligent verification and early warning of feeder protection settings across the entire distribution network as proposed in the above embodiment.

[0117] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0118] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the method for intelligent verification and early warning of feeder protection settings across the entire distribution network as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0119] Example 2 is the second embodiment of the present invention. This embodiment is different from the previous embodiments. In order to verify the beneficial effects of the present invention, the present invention is compared with the prior art. The comparison results are shown in Table 1.

[0120] Existing technologies mainly refer to traditional static setting protection and simple setting verification methods for distribution networks. Specifically, these methods rely on historical experience to set fixed protection settings, lacking the ability to adjust to dynamic changes in the power grid (such as load fluctuations, complex interactions after equipment upgrades, and diverse faults). Furthermore, they rely on fixed current and voltage data for simple judgments, without incorporating time-frequency analysis techniques such as wavelet transform and Stockwell transform, failing to accurately capture the dynamic fault characteristics of the power grid and exhibiting insufficient responsiveness to dynamic changes. They also fail to effectively handle the complex interactions between feeders in the power grid, lacking global collaborative optimization and rapid response mechanisms such as "game theory model calculation of the optimal setting combination for the entire network" and "efficient distribution of optimized settings via GOOSE messages," easily leading to setting conflicts, false protection / missed protection issues, and impacting power grid stability and reliability.

[0121] Table 1: Comparison of Experiments between the Invention and Existing Technologies

[0122]

[0123] In summary, this invention employs a technical solution combining dynamic optimization and intelligent learning. By introducing techniques such as positive-sequence and zero-sequence component calculation, Iterative optimization of Thevenin equivalent circuit parameters, wavelet transform, and Stockwell transform, it can collect current and voltage data in real time and calculate accurate operating currents. Simultaneously, by calculating the optimal setpoint combination through a game theory model, it achieves automated optimization and dynamic verification of distribution network protection settings. This invention optimizes the shortcomings of existing technologies in dealing with power grid load fluctuations, fault warnings, and setpoint conflicts, and can significantly improve the power grid's response speed and processing capabilities under emergencies, ensuring the safety and stability of power grid operation.

[0124] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0125] Example 3 is the third embodiment of the present invention. This embodiment provides a system for intelligent verification and early warning of feeder protection settings across the entire distribution network, comprising:

[0126] The operational status modeling module is used to collect relevant data of the distribution network and perform positive sequence component calculation, establish the Thevenin equivalent circuit model and perform iterative optimization of parameters, calculate the operating current based on the Thevenin parameters, verify the level difference of adjacent feeder settings in real time, generate a list of qualified values ​​and send it to the protection device; the relevant data of the distribution network includes current and voltage;

[0127] The fault feature analysis module is used to calculate the instantaneous current increment and perform a 3-level decomposition using Daubechies4 wavelet to extract high-frequency detail coefficients d1-d3 for fault prediction. Based on the fault prediction results, the module performs Stockwell transform and Hilbert transform analysis on the current data. Based on the analysis results, the module calculates the hybrid fault index to perform multi-level fault early warning and generates corresponding measures.

[0128] The global collaborative decision-making module is used to aggregate the Thevenin parameters of the entire network, generate fault heatmaps and setpoint conflict reports, calculate the optimal setpoint combination of the entire network based on game theory models, and send the optimized setpoints to edge nodes through GOOSE messages.

[0129] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for intelligent verification and early warning of feeder protection settings across the entire distribution network, characterized in that: include, Collect relevant data from the distribution network and perform positive sequence component calculations, establish a Thevenin equivalent circuit model and perform iterative optimization of parameters, calculate the operating current based on the Thevenin equivalent circuit parameters, verify the level difference of adjacent feeder settings in real time, generate a list of qualified values ​​and send it to the protection device; the relevant data from the distribution network includes current and voltage; The instantaneous current increment is calculated and decomposed into three levels using Daubechies4 wavelets. The first level detail coefficient d1, the second level detail coefficient d2, and the third level detail coefficient d3 are extracted for fault prediction. Based on the fault prediction results, Stockwell transform and Hilbert transform are performed on the current data. Based on the analysis results, a hybrid fault index is calculated to perform multi-level fault early warning and generate corresponding measures. Aggregate the Thevenin equivalent circuit parameters of the entire network, generate a fault heat map and a setpoint conflict report, calculate the optimal setpoint combination of the entire network based on the game theory model, and send the optimized setpoints to the edge nodes through GOOSE messages; The process of establishing the Thevenin equivalent circuit model and iteratively optimizing the parameters includes: Using positive sequence voltage data, the Thevenin equivalent voltage is calculated, and the Thevenin equivalent impedance is calculated by the changes in current and voltage. The initial Thevenin equivalent voltage and initial Thevenin equivalent impedance are calculated using the three-point method; The particle swarm optimization algorithm is used to perform global optimization of Thevenin equivalent voltage and Thevenin equivalent impedance to find the optimal initial solution. Based on the optimization of the particle swarm optimization algorithm, the Gauss-Newton iterative method is used for local optimization. The initial parameters obtained by the particle swarm optimization algorithm are used to calculate the residuals and Jacobian matrix, which reflect the partial derivatives of the residuals with respect to each parameter. The parameters of Thevenin equivalent voltage and Thevenin equivalent impedance are updated using the Gauss-Newton iterative optimization formula; Set a convergence threshold, update the Thevenin equivalent voltage and Thevenin equivalent impedance in each iteration until the residual is less than the preset convergence threshold, and obtain the optimized Thevenin equivalent voltage and Thevenin equivalent impedance parameters. The process of calculating the operating current based on Thevenin equivalent circuit parameters, verifying the level difference between adjacent feeder settings in real time, generating a list of compliant values, and sending it to the protection device includes: The operating current in the power grid is calculated based on the optimized Thevenin equivalent voltage and Thevenin equivalent impedance; the setting range between two adjacent feeders is calculated based on the operating current of adjacent feeders in the power grid. If the set value range exceeds the set value threshold, it indicates that the protection set value of the adjacent feeder is not set properly, resulting in malfunction or false protection. The sub-region where the adjacent feeder is located in the power grid is marked as the set value conflict sub-region, and an alarm message is generated to remind the operator to make adjustments. If the set value range does not exceed the range threshold, the set value verification is passed, and the protection set value setting remains unchanged; If the set value range is equal to the grade difference threshold, it is determined to be a critical state. The sub-region where the adjacent feeder in the power grid is located is added to the set value optimization task list, and set value optimization is performed in the next set value tuning cycle. Based on the obtained operating current and setting verification results, a complete set value list is generated; The GOOSE message protocol is used to send the setting list to each protection device.

2. The method for intelligent verification and early warning of feeder protection settings across the entire distribution network as described in claim 1, characterized in that: The instantaneous current increment is calculated and decomposed into three levels using Daubechies4 wavelets. The first-level detail coefficients d1, the second-level detail coefficients d2, and the third-level detail coefficients d3 are extracted for fault prediction, including: After real-time acquisition and preprocessing of the current data of the current feeder, the difference between the current value at the current moment and the current value at the previous moment is calculated to obtain the instantaneous current increment. A threshold for the instantaneous current increment is set, and for current signals whose instantaneous increment exceeds the threshold, Daubechies4 wavelet transform is used for three-level decomposition. The wavelet decomposition process decomposes the signal into approximation coefficients and detail coefficients. After completing the three-level wavelet decomposition, the first level detail coefficient d1, the second level detail coefficient d2, and the third level detail coefficient d3 are obtained. Energy calculation is performed on each level detail coefficient to analyze the high-frequency fluctuations in the current signal. Fourier transform is used to perform frequency domain transformation on the detail coefficients of each layer. A high-frequency fluctuation threshold is set, and high-frequency fluctuations that exceed the high-frequency fluctuation threshold are filtered out. The instantaneous increment of the current is combined with the energy characteristics. If the instantaneous increment of the current and the energy change of the detail coefficients both exceed the set threshold, it is determined to be a power grid fault.

3. The method for intelligent verification and early warning of feeder protection settings across the entire distribution network as described in claim 2, characterized in that: The process involves performing Stockwell and Hilbert transform analyses on the current data based on the fault assessment results, calculating a hybrid fault index based on the analysis results, providing multi-level fault early warning, and generating corresponding measures, including: Wavelet transform is used to denoise the current signal identified as a power grid fault, and Stockwell transform is used to capture the short-time frequency features in the current signal to obtain the frequency components corresponding to each time moment. Find the extreme points of the signal and construct the upper and lower envelopes using cubic Hermite interpolation. Calculate the average value of the upper and lower envelopes and remove the envelope calculated from the average value from the original signal to obtain the first IMF component. Remove the extracted first IMF component from the original signal to obtain the remaining signal. Continue to extract IMF components from the remaining signal until all IMF components are decomposed. Apply Hilbert transform to each IMF component to obtain the instantaneous frequency of each IMF. The resolution of the signal in the frequency domain is calculated by analyzing its spectral width and time domain width. The hybrid fault index of the power grid is calculated based on the amplitude spectrum obtained by Stockwell transform, the instantaneous frequency obtained by Hilbert transform, and the frequency resolution. Set high-risk and medium-risk thresholds, with the high-risk threshold being greater than the medium-risk threshold. If the mixed fault index is greater than or equal to the high-risk threshold, it indicates a serious potential fault in the power grid, triggering a high-risk alarm, automatically isolating faulty equipment, and adjusting protection settings. If the mixed fault index is less than the high-risk threshold but greater than or equal to the medium-risk threshold, it indicates a medium potential fault in the power grid, prompting the operator to monitor the power grid status. If the mixed fault index is less than the medium-risk threshold, it indicates that the power grid is operating normally, and continuous monitoring of the power grid is performed.

4. The method for intelligent verification and early warning of feeder protection settings across the entire distribution network as described in claim 3, characterized in that: The process of aggregating the entire network's Thevenin equivalent circuit parameters to generate fault heatmaps and setpoint conflict reports involves collecting data from the Thevenin equivalent circuit parameters and mixed fault index records of all feeders, storing all aggregated data on a cloud platform, using data visualization technology to generate fault heatmaps, classifying areas into low, medium, and high risk levels based on the mixed fault index, and marking them with varying shades of the same color. Based on aggregated data, the system checks for setting conflicts in the power grid and automatically generates setting conflict reports.

5. The method for intelligent verification and early warning of feeder protection settings across the entire distribution network as described in claim 4, characterized in that: The calculation of the optimal setpoint combination for the entire network based on the game theory model, and the distribution of the optimized setpoint to the edge nodes through GOOSE messages, refers to the division of the power grid into multiple sub-regions based on the overall structure of the distribution network, and the modeling of the feeders and protection devices in each sub-region. The protection setpoints for each sub-region are optimized using a non-cooperative game model. Each protection device in a sub-region is considered a participant in the game. The protection device selects the protection setpoints based on the operating state of the sub-region and the objective function. The optimal combination is obtained by solving the Nash equilibrium point of the game. During the game, the gradient descent method is used to optimize the protection setpoints of each sub-region. By iteratively updating the setpoint parameters of each sub-region, the comprehensive objective of minimizing the cost and risk of each sub-region is achieved. After multiple rounds of iterative calculation, the optimal combination of protection setpoints is finally obtained. Based on the optimal protection setting combination calculated by the game theory model, a list of optimal setting configurations for the entire network is generated, and the list of optimal setting configurations is distributed to each protection device in the distribution network through the GOOSE message protocol. The calculation of positive sequence components refers to analyzing the three-phase current using the positive sequence component method after acquiring current and voltage data, and calculating the positive sequence components.

Images