A Key Transmission Section Search Method and System Based on Standard Cut and Safety Risk

Abstract

This invention relates to the field of power grid security technology and discloses a method for searching critical transmission sections based on normative tangents and security risks, comprising: Step S1: Importing historical power grid operation data from the power grid operation database; Step S2: Generating operation scenarios using the Latin hypercube sampling method; Step S3: Calculating power flow and node voltages using the Newton-Lawrence method; Step S4: Constructing a branch weight model; Step S5: Constructing a power system graph theory model based on spectral graph theory, the power system graph theory model including the number of network nodes, the number of branches, and weights; Step S6: Partitioning the power grid topology map using a spectral clustering algorithm based on normative tangents; Step S7: Screening critical and non-critical transmission sections using security risk indicators; Step S8: Searching for critical transmission sections in all operation scenarios and outputting the critical transmission section S of the power grid; This invention also relates to related systems. This invention can accurately and efficiently search for critical transmission sections of the power grid.

Description

Technical Field

[0001] This invention relates to the field of power grid security technology, specifically to a method and system for searching critical transmission sections based on standard cutting and security risks. Background Technology

[0002] Critical transmission sections are important safety features of large power grids and are key targets for power system security analysis and monitoring. Major blackouts often occur because power flow exceeds limits at critical transmission sections, leading to cascading accidents. Therefore, researching rapid search methods for critical transmission sections is a key research direction for the stable operation of power systems.

[0003] With the construction of my country's power grid, the grid structure has become increasingly complex. Coupled with the large-scale integration of intermittent new energy sources such as wind and solar power, the power flow in the power system changes frequently, and the critical transmission sections of the system also change accordingly. This poses a challenge to traditional methods of determining critical transmission sections based on manual experience. Determining critical transmission sections only for a few typical scenarios cannot adapt to the frequent changes in grid operation modes; on the other hand, pre-determining a large number of critical transmission sections increases the actual amount of analysis and calculation required for the grid, making it difficult for operators to obtain the true risk points of the system.

[0004] Currently, research on critical transmission section identification methods mainly focuses on two aspects: transient security risk and static security risk. Transient security assessment typically measures the criticality of a section based on the magnitude of its impact on system transient instability risk caused by changes in power output. However, this method involves transient simulation modeling, which is complex and time-consuming. Critical transmission section identification methods based on static security risk can be broadly categorized into two types: analysis methods based on power flow transfer relationships and methods based on grid partitioning, such as methods based on electrical betweenness, breaking coefficient, and shortest path methods. Data-driven critical section identification methods offer faster computation speeds but rely heavily on large amounts of historical data as training samples, have high computational costs, and lack flexibility.

[0005] Most of the above methods have problems such as missing or misselecting key transmission sections, and in actual power grid key transmission section search applications, they have problems such as slow calculation speed and insufficient accuracy. Summary of the Invention

[0006] To address the aforementioned problems in the existing technology, this invention provides a method and system for searching critical transmission sections based on standardization and safety risks. This method searches a large number of generated scenarios for critical transmission sections, statistically analyzing the number and probability of occurrence of each critical transmission section. This method and system can not only accurately and efficiently identify critical transmission sections currently being monitored in the power grid, but also filter out critical transmission sections that have not been discovered by operators and pose a high risk, thereby eliminating risks, improving monitoring efficiency, and significantly enhancing the accuracy of critical section identification. The specific implementation of this application is as follows:

[0007] A method for searching critical transmission sections based on standard cutting and safety risks includes:

[0008] Step S1: Import historical power grid operation data. Import historical power grid operation data from the power grid operation database. The historical power grid operation data includes generator data, bus data, transformer data, load data, and line impedance data.

[0009] Step S2: Based on historical power grid operation data, use the Latin hypercube sampling method to generate operation scenarios and simulate typical and extreme power grid operation scenarios;

[0010] Step S3: Calculate the power flow and node voltages of the power grid using the Newton-Layer method based on the operating scenario parameters;

[0011] Step S4: Construct a branch weighting model based on line impedance, node voltage, and branch power flow parameters;

[0012] Reactive power at node i Q i Voltage amplitude at node j V j The partial derivatives are:

[0013]

[0014] in, V i This represents the voltage magnitude at node i; i ij The voltage phase angle difference between node i and node j; G ij For the conductance of the branch circuit, B ij For the susceptance of the branch circuit;

[0015] Branch weights based on voltage factor w U for:

[0016]

[0017] Among them, the reactive power at node j Q j The active power of node i and node j are respectively: P i , P j ;

[0018] Branch weights based on active power flow factors w I for: w I =(| P i |+| P j |) / 2

[0019] Branch weights that take into account both voltage and active power flow factors w ij for:

[0020]

[0021] Step S5: Construct a power system graph theory model based on spectral graph theory. The power system graph theory model includes the number of network nodes, the number of branches, and the weight values.

[0022] Step S6: Partition the power grid topology map based on the spectral clustering algorithm of normalized cut;

[0023] Step S7: For the transmission sections formed between zones, use safety risk indicators to screen critical and non-critical transmission sections. or The calculation is as follows:

[0024]

[0025] in, l 1. l 2 represents two different lines at the critical transmission section S. P l1 and P l1-max They are respectively l 1. Active power flow and transmission limits of the line. P l2 For the line l The active power flow of line 2, when the line l After disconnection 1, the power flow shifts to the line. l The break distribution factor on 2 is α l1-l2 ; exp is an exponential function, max is for finding the maximum value;

[0026] Step S8: Perform a critical transmission section search for all operating scenarios, output the critical transmission section S of the power grid, and count the number of critical transmission sections S. N S and the probability of occurrence P S :

[0027]

[0028] in, N The number of running scenarios; when P S If the value exceeds the set threshold, the key power transmission section S will be monitored in real time.

[0029] Preferably, the historical power grid operation data includes generator data, bus data, transformer data, load data, and line impedance data. The generator data includes the generator's rated capacity, rated voltage, rated current, reactive power, and power factor; the transformer data includes turns ratio, resistance, and admittance; and the load data includes the load power, reactive power, and power factor of each node.

[0030] Preferably, step S2: Based on historical power grid operation data, an operation scenario is generated using the Latin hypercube sampling method to simulate typical and extreme power grid operation scenarios, including:

[0031] S21: Obtain the probability density curve of new energy output based on the historical grid operation data of the new energy power generation grid connection points. F ;

[0032] S22: Latin hypercube sampling is used, with random variables... Y 1 ,Y 2 ,Y 3 ,..., Y T For T new energy output values, Y k Let be any one of these random variables, and its cumulative probability distribution function be... Z k for: Z k = F k ( Y k );

[0033] The sampling size is N running scenarios, and the sampling method is as follows: The cumulative probability distribution curve is... Z k = F k (Y k The ordinate of ) is divided into equal parts N There are 1 interval, and the width of each interval is 1 / N Select the midpoint of each interval as Z k The sampled values ​​are then determined according to the cumulative probability distribution function. F k Find the inverse function Y k The sampled values, Z k The n'th sample value y kn' for:

[0034]

[0035] F k -1 (·)for F k The inverse function is obtained by sampling T random variables N times using the above method to form an initial sampling matrix Y of order T×N. According to the Gram-Schmidt orthogonalization method, the row and column order of the initial sampling matrix Y is adjusted using alternating forward and reverse methods, thus obtaining the transformed sampling matrix. L= [ L 1 , L 2 ,..., L T ] T , L 1 , L 2 , L T Let L be the sampling column vector matrices of the 1st, 2nd, and Tth random variables after the transformation order; find the root mean square correlation ρ of the transformed sampling matrix L. ms :

[0036]

[0037] In the formula:

[0038]

[0039] ρ kg Represents an arbitrary sampling column vector L k and L g The correlation coefficient between them cov (·) represents the covariance. var (·) represents the variance;

[0040] The Gram-Schmidt orthogonalization method is used to make the root mean square correlation ρ of the matrix columns equal. ms The transformation sampling matrix stops decreasing or reaches a set threshold of forward and reverse iterations. L That is, the Latin hypercube sampling matrix Y TN This refers to the combination of generating units that utilize new energy sources;

[0041] S23: Obtain N The output data of new energy generators under various operating scenarios, namely the Latin hypercube sampling matrix Y TN .

[0042] Preferably, step S5: constructing a power system graph theory model based on spectral graph theory, wherein the power system graph theory model includes the number of network nodes, the number of branches, and weight values, specifically including:

[0043] Construct a graph G=(V,E,W), where V represents the set of nodes, E represents the set of branches, and W is the weight matrix. W ij for:

[0044]

[0045] In the formula, w ij The branch weights between node i and node j;

[0046] The connection relationships between nodes are represented by a degree matrix. D= [ d 1 , d 2 , d 3 ,..., d n ] indicates that D is a diagonal matrix, and the diagonal elements It is represented as the sum of the elements in the i-th row of the weight matrix, where n is the number of power grid nodes;

[0047] The Laplace matrix of graph G constructed from the weight matrix W and the degree matrix D is: L=DW.

[0048] Preferably, step S6 involves partitioning the power grid topology map using a spectral clustering algorithm based on normalized cuts. Based on the normalized cut criterion, the graph G is divided into partition A and partition B. The partitioning indicator vector divides the system into two partitions. When the cross-section between the partitions is a critical transmission section, it is considered the optimal partition.

[0049] The objective function for obtaining the optimal power grid partitioning based on the standard cutting method is as follows. f :

[0050]

[0051] In the formula, V A set of nodes; x i , x j These are the split indicator vector elements in partition A and partition B, respectively. When the node is in partition A, the split indicator vector element is... x i When the value is 1, the split indicator vector element is in partition B. x j =-1; w ij The branch weights between node i and node j; w iz Let be the branch weight between node i and node z; w jz Let be the branch weight between node j and node z;

[0052] The objective function for finding the optimal partition of the power grid is the normalized tangent, which will be 2 n The -1 partitioning is simplified to n-1 partitioning; according to the solution principle of canonical cut, the eigenvalues ​​of the Laplacian matrix L of graph G and the Fiedler eigenvectors are calculated to obtain n-1 different partitioning indicator vectors. X 1, X 2,..., X n-1 Substitute the segmentation indicator vectors into the objective function respectively. f The partition corresponding to the segmentation indicator vector that minimizes the objective function value is the optimal partition of the power grid.

[0053] Preferably, step S3: calculating power flow and node voltage using the Newton-Lager method based on the operating scenario parameters, includes: first, initializing the voltage amplitude and phase angle of the nodes as the starting point of the iteration; calculating the active power and reactive power deviation of the nodes and determining whether a set threshold has been reached; then calculating the Jacobian matrix to obtain the voltage correction amount; finally, determining whether the target value or the number of iterations has been reached, and using the node voltage and branch power flow obtained from the iteration as the final power flow calculation result;

[0054] Let the number of nodes in the power grid be n, and the node power balance equation be:

[0055]

[0056] in, P i , Q i These are the active power and reactive power of node i, respectively; Vi , V j These are the voltage amplitudes at nodes i and j, respectively. Y ij For branch admittance, when the branch conductance is G ij The susceptance of the branch circuit is B ij The active power of node i can be obtained by decomposing the formula. P i and reactive power Q i They are respectively:

[0057]

[0058] In the formula: i ij The voltage phase angle difference between node i and node j. V i This represents the voltage magnitude at node i. V j This represents the voltage magnitude at node j; the formula for calculating power imbalance is:

[0059]

[0060] P is 、Q is Let Δ be the given active power and given reactive power of node i, respectively. P i Δ Q i These are the active power deviation and reactive power deviation of the node, respectively;

[0061] The node types include PQ load nodes, PV generator nodes, and Slack balancing nodes. When the number of PQ load nodes is m and the number of Slack balancing nodes is 1, the Jacobian matrix J is calculated using a matrix block format as follows:

[0062]

[0063] Where H is an (n-1)×(n-1) order square matrix, its elements , representing the power deviation between node i and node j. i j Let be the voltage phase angle at node j; R is an (n-1)×m matrix with elements of . O is an m×(n-1) matrix whose elements C is an m × m matrix whose elements .

[0064] This application also provides a key transmission section search system based on standard cutting and safety risks, including: a power grid historical operation data import module, which imports power grid historical operation data from the power grid operation database, wherein the power grid historical operation data includes generator data, bus data, transformer data, load data, and line impedance data;

[0065] The operation scenario generation module generates operation scenarios based on historical power grid operation data and uses the Latin hypercube sampling method to simulate typical and extreme power grid operation scenarios.

[0066] The power flow and node voltage calculation module uses the Newton-Lambert method to calculate power flow and node voltage based on operating scenario parameters.

[0067] A branch weighting model module is constructed based on line impedance, node voltage, and branch power flow parameters to build a branch weighting model.

[0068] Reactive power at node i Q i Voltage amplitude at node j V j The partial derivatives are:

[0069]

[0070] in, V i This represents the voltage magnitude at node i; i ij The voltage phase angle difference between node i and node j; G ij For the conductance of the branch circuit, B ij For the susceptance of the branch circuit;

[0071] Branch weights based on voltage factor w U for:

[0072]

[0073] Among them, the reactive power at node j Q j The active power of node i and node j are respectively: P i , P j ;

[0074] Branch weights based on active power flow factors w I for: w I =(| Pi |+| P j |) / 2

[0075] Branch weights that take into account both voltage and active power flow factors w ij for:

[0076]

[0077] A module for constructing a power system graph theory model is provided, which constructs a power system graph theory model based on spectral graph theory. The power system graph theory model includes the number of network nodes, the number of branches, and the weight values.

[0078] The power grid topology partitioning module partitions the power grid topology map based on the spectral clustering algorithm of normal cut.

[0079] The safety risk index calculation module uses safety risk indicators to screen critical and non-critical transmission sections for transmission sections formed between zones. The safety risk index η is calculated as follows:

[0080]

[0081] in, l 1. l 2 represents two different lines at the critical transmission section S. P l1 and P l1-max They are respectively l 1. Active power flow and transmission limits of the line. P l2 For the line l The active power flow of line 2, when the line l After disconnection 1, the power flow shifts to the line. l The break distribution factor on 2 is α l1-l2 ; exp is an exponential function, max is for finding the maximum value;

[0082] The critical section output module searches for critical transmission sections in all operating scenarios, outputs the critical transmission sections S of the power grid, and counts the number of critical transmission sections S. N S and the probability of occurrence P S :

[0083]

[0084] in, N The number of running scenarios; when P S If the value exceeds the set threshold, the key power transmission section S will be monitored in real time.

[0085] Preferably, the historical power grid operation data includes generator data, bus data, transformer data, load data, and line impedance data. The generator data includes the generator's rated capacity, rated voltage, rated current, reactive power, and power factor; the transformer data includes turns ratio, resistance, and admittance; and the load data includes the load power, reactive power, and power factor of each node.

[0086] Preferably, the module for constructing the power system graph theory model: constructs a power system graph theory model based on spectral graph theory. The power system graph theory model includes the number of network nodes, the number of branches, and weights; specifically, it includes: establishing a graph G=(V,E,W), where V represents the set of nodes, E represents the set of branches, and its weight matrix is ​​W, with matrix elements... W ij for:

[0087]

[0088] In the formula, w ij The branch weights between node i and node j;

[0089] The connection relationships between nodes are represented by a degree matrix. D= [ d 1 , d 2 , d 3 ,..., d n ] indicates that D is a diagonal matrix, and the diagonal elements It is represented as the sum of the elements in the i-th row of the weight matrix, where n is the number of power grid nodes;

[0090] The Laplace matrix of graph G constructed from the weight matrix W and the degree matrix D is: L=DW.

[0091] Preferably, the power grid topology map partitioning module partitions the power grid topology map using a spectral clustering algorithm based on normalized cuts. Based on the normalized cut criterion, the graph G is divided into partition A and partition B. The partitioning indicator vector divides the system into two partitions. When the cross-section between the partitions is a critical transmission section, it is the optimal partition.

[0092] The objective function for obtaining the optimal power grid partitioning based on the standard cutting method is as follows. f :

[0093]

[0094] In the formula, V A set of nodes; x i ,x j These are the split indicator vector elements in partition A and partition B, respectively. When the node is in partition A, the split indicator vector element is... x i When the value is 1, the split indicator vector element is in partition B. x j =-1; w ij The branch weights between node i and node j; w iz Let be the branch weight between node i and node z; w jz Let be the branch weight between node j and node z;

[0095] The objective function for finding the optimal partition of the power grid is the normalized tangent, which will be 2 n The -1 partitioning is simplified to n-1 partitioning; according to the solution principle of canonical cut, the eigenvalues ​​of the Laplacian matrix L of graph G and the Fiedler eigenvectors are calculated to obtain n-1 different partitioning indicator vectors. X 1, X 2,..., X n-1 Substitute the segmentation indicator vectors into the objective function respectively. f The partition corresponding to the segmentation indicator vector that minimizes the objective function value is the optimal partition of the power grid.

[0096] This invention provides a method and system for searching critical transmission sections based on standard cutting and safety risks, and the beneficial technical effects it can achieve are as follows:

[0097] 1. This invention generates a large number of operating scenarios based on historical operating data using Latin hypercube sampling, simulating typical and extreme operating scenarios of the power grid. Based on the operating scenario parameters, the Newton-Lager method is used to calculate power flow and node voltage. A branch weight model is constructed based on line impedance, node voltage, and branch power flow parameters. A power system graph theory model is built based on spectral graph theory, including the number of network nodes, branches, and weights. A spectral clustering algorithm based on normalized cuts is used to partition the power grid topology under specific scenarios. For the transmission sections formed between partitions, safety risk indicators are used to screen critical and non-critical transmission sections. All generated scenarios' critical transmission sections are output, and the number and probability of each critical transmission section are statistically analyzed. These sequential steps not only accurately and efficiently identify the critical transmission sections currently being monitored in the power grid but also screen out critical transmission sections that have not been discovered by operators and pose a high risk, thereby helping operation and dispatch personnel eliminate risks and improve monitoring efficiency.

[0098] 2. In the spectral clustering algorithm based on normalized cuts, the transmission section corresponding to the partition with the smallest objective function value may not be a critical transmission section, but may be a transmission section corresponding to the second or third smallest objective function value. Therefore, this invention uses safety risk indicators to determine critical and non-critical transmission sections. For transmission sections formed between partitions, safety risk indicators are used to screen critical and non-critical transmission sections. The critical section output module outputs all critical transmission sections of the generated scenarios, and counts the number and probability of occurrence of critical transmission sections. When the probability of occurrence is greater than a set threshold, the critical transmission section is monitored in real time, which greatly improves the accuracy of critical section judgment.

[0099] 3. This invention uses a spectral clustering algorithm based on canonical cuts to partition the power grid topology map in specific scenarios. The canonical cuts will partition the 2... n -1 partitioning is simplified to n-1 partitioning, resulting in n-1 different partitioning indicator vectors. Based on the solution principle of normalized cut, the eigenvalues ​​of the Laplacian matrix L of graph G and the Fiedler eigenvectors are calculated, which greatly improves the calculation accuracy.

[0100] 4. In the process of calculating branch weights, this invention combines voltage factors and active power flow factors for comprehensive judgment, so as to achieve accurate calculation of branch weights. Attached Figure Description

[0101] 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 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.

[0102] Figure 1 This is a schematic diagram of the steps of the key transmission section search method based on standard cutting and safety risks of the present invention;

[0103] Figure 2 This is a schematic diagram of the rapid search process for key power transmission sections in this invention;

[0104] Figure 3 This is a schematic diagram of the node system partitioning of the present invention. Detailed Implementation

[0105] 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, and 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.

[0106] Example 1:

[0107] In view of the aforementioned problems mentioned in the prior art, and in order to solve these technical problems, such as Figure 1 The method provided here is a fast search method for critical transmission sections based on standard cutting and safety risks, including:

[0108] In some embodiments, key transmission sections are searched in real-time online or typical scenarios of the power grid. The following analysis uses data from the peak summer load operation scenario (Xia Da scenario) of the power grid.

[0109] The first step is to import the network topology data of the power grid, including generator parameters, load parameters, and line impedance parameters. The second step is to generate system operation data for the Xia Da scenario, including thermal power output, renewable energy output, and load demand. The third step is to calculate the power flow and node voltages under the Xia Da scenario. The fourth step is to construct a branch weight model based on the node voltage and power flow data and assign weights to the branches. The fifth step is to abstract the power system of the Xia Da scenario into a graph model consisting of 24 nodes and 34 branches. The sixth step is to partition the power grid based on the system graph model using a spectral clustering algorithm with standard cuts. The seventh step is to obtain the transmission line cut sets between the partitions, i.e., transmission sections, based on the power grid partitions, calculate the safety risk index of the transmission sections, and select sections with a safety risk index greater than 1 as critical transmission sections.

[0110] Step S1: Import historical power grid operation data. Import historical power grid operation data from the power grid operation database. This historical power grid operation data includes generator data, bus data, transformer data, load data, and line impedance data. First, import the basic data, which includes:

[0111] 1. Network topology data, including the connection relationships between nodes and branches of the power system, as well as parameters such as the admittance and resistance of the branches.

[0112] 2. Load data, including load information at each node, such as load power, reactive power, and power factor.

[0113] 3. Generator data: including the generator's rated capacity, voltage and current characteristics, reactive power and power factor, etc.

[0114] 4. Transformer data, such as turns ratio, resistance, and admittance.

[0115] 5. Historical operating data of the power system.

[0116] Then, historical power grid operation data is imported: Generator data: including the rated capacity (e.g., 500MW), output voltage (e.g., 22kV), output current (e.g., 13636A), reactive power (e.g., 150MVar), and power factor (e.g., 0.8) for each generator. Bus data: covering the voltage level (e.g., 110kV), current voltage value (e.g., 108kV), and bus connection configuration for each bus. Transformer data: including the transformer turns ratio (e.g., 220 / 110kV), resistance (e.g., 0.5Ω), and admittance (e.g., 0.01Siemens). Load data: recording the load power (e.g., 100MW), reactive power (e.g., 30MVar), and power factor (e.g., 0.9) for each node. Line impedance data: including the impedance value for each transmission line, consisting of resistance (e.g., 0.05Ω / km) and reactance (e.g., 0.1Ω / km).

[0117] Data import process: First, the system imports generator data, bus data, transformer data, load data, and line impedance data from the power grid operation database. This data is extracted through predefined data interfaces and can be real-time data or historical data at a specific point in time. After import, the system performs preliminary data verification to ensure data integrity and accuracy.

[0118] Step S2: Based on historical power grid operation data, use the Latin hypercube sampling method to generate operation scenarios and simulate typical and extreme power grid operation scenarios. In the power system, due to the obvious randomness and volatility of new energy output, coupled with the difficulty in accurately predicting load, there are a large number of uncertain scenarios.

[0119] This invention, based on historical operating data of renewable energy sources and node loads, employs scenario generation technology to generate a large number of typical and extreme operating scenarios, providing search scenarios for key transmission sections. Latin hypercube sampling: Samples are generated from a probability distribution using Latin hypercube sampling technology. This method can sample uniformly across the entire sample space, ensuring the representativeness of the scenarios. For example, for wind power, we might generate a series of combinations of different wind speeds and directions; for solar power, we might generate combinations of different irradiance levels; for load, we might generate different load levels. Scenario generation: Based on the sampling results, a series of typical and extreme operating scenarios are created. Typical scenarios might include normal renewable energy output and load levels, while extreme scenarios might include very high or very low renewable energy output and / or load levels. For example, an extreme scenario might be a sudden drop in wind and solar power generation during peak load periods. Application scenarios: These generated scenarios will be used to simulate the operating state of the power grid under different conditions, such as power flow calculations and node voltage analysis. By analyzing these scenarios, we can better understand the behavior of the power grid in the face of fluctuations in renewable energy output and load changes, thereby providing support for optimizing grid operation and enhancing its stability.

[0120] Step S3: Calculate the power flow and node voltages of the power grid using the Newton-Layer method based on the operating scenario parameters; calculate the node voltages and network power flow of the power grid based on the operating scenario of the power grid.

[0121] The baseline capacity is selected as 100MW. The parameters to be imported into the power grid include bus base voltage, line parameters, transformer parameters, generator output parameters, and load parameters. Except for the bus voltage, all other data are expressed in per-unit values. Node types are set. The node types in this application include PQ load nodes (i.e., PQ nodes), PV generator nodes (i.e., PV nodes), and Slack nodes (i.e., Slack nodes). The number of PQ load nodes is m, the number of Slack nodes is 1, and the number of PV generator nodes is the number of grid nodes nm-1, where n is the number of grid nodes. Initialization: A specific operating scenario is selected, which includes specific generator output, load demand, and possible line states. The voltage amplitude and phase angle of each node are initialized. Typically, the voltage amplitude can be set to 1.0 pu (per-unit), and the phase angle can be initialized to 0.

[0122] It's important to explain that the Newton-Raphson method is used to solve the nonlinear power flow equations of a power system. These equations describe the active and reactive power balance at each node in the system. Power balance equations are written for each node, and the Jacobian matrix, which contains partial derivatives with respect to voltage magnitude and phase angle, is calculated. These equations are solved using an iterative method. Each iteration calculates the power imbalance based on the current voltage and phase angle values ​​and updates the voltage and phase angle until a preset convergence criterion is met. The iterative process involves calculating the power imbalance at each node (i.e., the difference between actual and expected power). Then, the Jacobian matrix is ​​calculated, and a system of linear equations is solved to obtain corrections for voltage and phase angle. The node's voltage and phase angle are updated, and it is determined whether the convergence condition is met (e.g., the power imbalance at all nodes is less than a certain threshold). Convergence and output: When the iteration converges, the current voltage and phase angle values ​​represent the desired power flow solution. The output shows the final voltage magnitude and phase angle of each node, as well as the power flow on the transmission line. This example demonstrates the role of step S3 in calculating power flow and node voltages under specific power grid operating scenarios. The Newton-Raphson method is a commonly used and effective approach for solving power flow problems in power systems; it can handle the nonlinear characteristics of the system and provide fast and accurate solutions.

[0123] Step S4: Construct a branch weighting model based on line impedance, node voltage, and branch power flow parameters;

[0124] Reactive power at node i Q i Voltage amplitude at node j V j The partial derivatives are:

[0125]

[0126] in, V i This represents the voltage magnitude at node i; i ij The voltage phase angle difference between node i and node j; G ij For the conductance of the branch circuit, B ij For the susceptance of the branch circuit;

[0127] Branch weights based on voltage factor w U for:

[0128]

[0129] Among them, the reactive power at node j Qj The active power of node i and node j are respectively: P i , P j ;

[0130] Branch weights based on active power flow factors w I for: w I =(| P i |+| P j |) / 2

[0131] Branch weights that take into account both voltage and active power flow factors w ij for:

[0132]

[0133] Step S5: Construct a power system graph theory model based on spectral graph theory. The power system graph theory model includes the number of network nodes, the number of branches, and weights. Definition of network nodes and branches: In this power system, each generator, load center, or any other electrical equipment is considered a node. Each transmission line connecting two nodes is considered a branch. For example, in some embodiments, the power system has 30 nodes, including 9 PV generator nodes, 20 PQ load nodes, and 1 Slack balancing node, and 40 transmission lines connecting these nodes. Determining weights: For each branch, we need to determine a weight. This weight can be based on multiple factors, such as the line impedance, capacity, or the electrical force flowing through the line. For example, we can use the line capacity or reverse impedance (1 / impedance) as the weight.

[0134] Construction of the graph theory model: Construct a graph G, where each node and each branch corresponds to a node and a branch of the power system, respectively. Construct a weight matrix W for graph G, whose elements... W ij The weight matrix represents the branch weights between nodes i and j. Simultaneously, a degree matrix D is constructed, which is a diagonal matrix where the elements on the diagonal are the sum of the weights of all connected branches of the corresponding node. Application of spectral theory: Based on the weight matrix W and degree matrix D of graph G, the Laplace matrix L of graph G is calculated, where L = D - W. The Laplace matrix L can be used for further spectral analysis, such as identifying critical nodes and critical paths in a power system, or for subsequent system partitioning and cross-sectional analysis.

[0135] Step S6: Partition the power grid topology map based on the spectral clustering algorithm of normalized cut;

[0136] In some embodiments, there is a power system comprising multiple generators, load centers, and transmission lines. The objective of this application is to partition this power grid using a spectral clustering algorithm to optimize its operation and improve its stability.

[0137] Creating the Power Grid Topology: First, create a topology graph based on the actual layout of the power system. In this graph, each generator, load center, or electrical device is considered a node, and each transmission line connecting two nodes is considered an edge. For example, our power grid might contain 50 nodes and 70 edges. Constructing the Weight and Degree Matrices: Assign weights to each edge based on the actual parameters of the power grid (such as line impedance, capacity, etc.), constructing a weight matrix. Simultaneously, construct a degree matrix, where the degree of each node is the sum of the weights of all its connected edges.

[0138] Calculate the Laplace matrix and eigenvectors: Using spectral graph theory, calculate the Laplace matrix of the power grid topology, defined as the difference between the degree matrix and the weight matrix. Calculate the eigenvalues ​​and eigenvectors of the Laplace matrix. Eigenvectors represent the distribution of nodes along a specific dimension. Apply spectral clustering to partition the power grid: Using the information in the eigenvectors, apply a spectral clustering algorithm to partition the power grid. This typically involves selecting an appropriate number of eigenvectors and then grouping the nodes using standard clustering techniques such as K-means. For example, we might decide to divide the power grid into three regions to optimize operating efficiency or enhance the system's resilience. Result of power grid partitioning: Ultimately, the power grid is divided into multiple regions, each containing a specific set of nodes. These regional divisions can be based on various factors, such as the distribution of power load, generation capacity, or other operating parameters. The results of partitioning help to better manage and control the power grid, such as limiting the spread of problems in the event of a fault or optimizing the distribution and use of electricity.

[0139] Step S7: For the transmission sections formed between zones, use safety risk indicators to screen critical and non-critical transmission sections. or The calculation is as follows:

[0140]

[0141] in, l 1. l 2 represents two different lines at the critical transmission section S. P l1 and P l1-max They are respectively l 1. Active power flow and transmission limits of the line. P l2 For the line l The active power flow of line 2, when the linel After disconnection 1, the power flow shifts to the line. l The break distribution factor on 2 is α l1-l2 ; exp is an exponential function, max is for finding the maximum value;

[0142] Step S8: Perform a critical transmission section search for all operating scenarios, output the critical transmission section S of the power grid, and count the number of critical transmission sections S. N S and the probability of occurrence P S :

[0143]

[0144] in, N The number of running scenarios; when P S If the value exceeds the set threshold, the key power transmission section S will be monitored in real time.

[0145] In some embodiments, the historical power grid operation data includes generator data, bus data, transformer data, load data, and line impedance data. The generator data includes the generator's rated capacity, rated voltage, rated current, reactive power, and power factor. The transformer data includes turns ratio, resistance, and admittance. The load data includes the load power, reactive power, and power factor of each node.

[0146] In some embodiments, step S2: generating operating scenarios based on historical power grid operating data using the Latin hypercube sampling method, simulating typical and extreme power grid operating scenarios, including:

[0147] S21: Obtain the probability density curve of new energy output based on the historical grid operation data of the new energy power generation grid connection points. F ;

[0148] S22: Latin hypercube sampling is used, with random variables... Y 1 ,Y 2 ,Y 3 ,..., Y T For T new energy output values, Y k Let be any one of these random variables, and its cumulative probability distribution function be... Z k for: Z k = F k ( Y k);

[0149] The sampling size is N running scenarios, and the sampling method is as follows: The cumulative probability distribution curve is... Z k = F k ( Y k The ordinate of ) is divided into equal parts N There are 1 interval, and the width of each interval is 1 / N Select the midpoint of each interval as Z k The sampled values ​​are then determined according to the cumulative probability distribution function. F k Find the inverse function Y k The sampled values, Z k The n'th sample value y kn' for:

[0150]

[0151] in, F k -1 (·)for F k The inverse function is obtained by sampling T random variables N times using the above method to form an initial sampling matrix Y of order T×N. According to the Gram-Schmidt orthogonalization method, the row and column order of the initial sampling matrix Y is adjusted using alternating forward and reverse methods, thus obtaining the transformed sampling matrix. L= [ L 1 , L 2 ,..., L T ] T , L 1 , L 2 , L T Let L be the sampling column vector matrices of the 1st, 2nd, and Tth random variables after the transformation order; find the root mean square correlation ρ of the transformed sampling matrix L. ms :

[0152]

[0153] In the formula:

[0154]

[0155] ρ kg Represents an arbitrary sampling column vectorL k and L g The correlation coefficient between them cov (·) represents the covariance. var (·) represents the variance;

[0156] The Gram-Schmidt orthogonalization method is used to make the root mean square correlation ρ of the matrix columns equal. ms The transformation sampling matrix stops decreasing or reaches a set threshold of forward and reverse iterations. L That is, the Latin hypercube sampling matrix Y TN This refers to the combination of generating units that utilize new energy sources;

[0157] S23: Obtain N The output data of new energy generators under various operating scenarios, namely the Latin hypercube sampling matrix Y TN .

[0158] In some embodiments, step S5: constructing a power system graph theory model based on spectral graph theory, wherein the power system graph theory model includes the number of network nodes, the number of branches, and weight values, specifically including:

[0159] Construct a graph G=(V,E,W), where V represents the set of nodes, E represents the set of branches, and W is the weight matrix. W ij for:

[0160]

[0161] In the formula, w ij The branch weights between node i and node j;

[0162] The connection relationships between nodes are represented by a degree matrix. D= [ d 1 , d 2 , d 3 ,..., d n ] indicates that D is a diagonal matrix, and the diagonal elements It is represented as the sum of the elements in the i-th row of the weight matrix, where n is the number of power grid nodes;

[0163] The Laplace matrix of graph G constructed from the weight matrix W and the degree matrix D is: L=DW.

[0164] Further, in step S6: the power grid topology map is partitioned based on the spectral clustering algorithm of the normalized cut. Based on the normalized cut criterion, the map G is divided into partition A and partition B. The partitioning indicator vector divides the system into two partitions. When the cross section between the partitions is a critical transmission section, it is the optimal partition.

[0165] The objective function for obtaining the optimal power grid partitioning based on the standard cutting method is as follows. f :

[0166]

[0167] In the formula, V A set of nodes; x i , x j These are the split indicator vector elements in partition A and partition B, respectively. When the node is in partition A, the split indicator vector element is... x i When the value is 1, the split indicator vector element is in partition B. x j =-1; w ij The branch weights between node i and node j; w iz Let be the branch weight between node i and node z; w jz Let be the branch weight between node j and node z;

[0168] The objective function for finding the optimal partition of the power grid is the normalized tangent, which will be 2 n The -1 partitioning is simplified to n-1 partitioning; according to the solution principle of canonical cut, the eigenvalues ​​of the Laplacian matrix L of graph G and the Fiedler eigenvectors are calculated to obtain n-1 different partitioning indicator vectors. X 1, X 2,..., X n-1 Substitute the segmentation indicator vectors into the objective function respectively. f The partition corresponding to the segmentation indicator vector that minimizes the objective function value is the optimal partition of the power grid.

[0169] In some embodiments, step S3: calculating power flow and node voltage using the Newton-Lager method based on operating scenario parameters includes: first, initializing the voltage amplitude and phase angle of the node as the starting point of the iteration; calculating the active power and reactive power deviation of the node and determining whether a set threshold has been reached; then calculating the Jacobian matrix to obtain the voltage correction amount; finally, determining whether the target value or the number of iterations has been reached, and using the node voltage and branch power flow obtained from the iteration as the final power flow calculation result.

[0170] Let the number of nodes in the power grid be n, and the node power balance equation be:

[0171]

[0172] in, P i , Q i These are the active power and reactive power of node i, respectively; V i , V j These are the voltage amplitudes at nodes i and j, respectively. Y ij For branch admittance, when the branch conductance is G ij The susceptance of the branch circuit is B ij The active power of node i can be obtained by decomposing the formula. P i and reactive power Q i They are respectively:

[0173]

[0174] In the formula: i ij The voltage phase angle difference between node i and node j. V i This represents the voltage magnitude at node i. V j This represents the voltage magnitude at node j; the formula for calculating power imbalance is:

[0175]

[0176] in, P is 、Q is Let Δ be the given active power and given reactive power of node i, respectively. P i Δ Q i These are the active power deviation and reactive power deviation of the node, respectively;

[0177] The node types include PQ load nodes, PV generator nodes, and Slack balancing nodes. When the number of PQ load nodes is m and the number of Slack balancing nodes is 1, the Jacobian matrix J is calculated using a matrix block format as follows:

[0178]

[0179] Where H is an (n-1)×(n-1) order square matrix, its elements , representing the power deviation between node i and node j. i j Let be the voltage phase angle at node j; R is an (n-1)×m matrix with elements of . O is an m×(n-1) matrix whose elements C is an m × m matrix whose elements .

[0180] Example 2:

[0181] This application also provides a rapid search system for critical transmission sections based on standard cutting and safety risks, including:

[0182] The power grid historical operation data import module imports historical power grid operation data from the power grid operation database. This historical operation data includes generator data, bus data, transformer data, load data, and line impedance data. In some embodiments, the system includes: a central processing unit (CPU): a powerful central processing unit for performing computationally intensive tasks such as data processing, power flow calculation, and spectral clustering algorithms; storage devices: large-capacity hard disk drives (HDDs) or solid-state drives (SSDs) for storing large amounts of power grid operation data, including historical and real-time data; fast random access memory (RAM) to support high-speed data access and processing; data acquisition devices: sensors and data acquisition devices for collecting data from various parts of the power system, such as smart meters and power grid monitoring devices; communication devices: network interface cards (NICs) and related communication hardware to ensure high-speed and stable data exchange with external data sources (such as power grid control centers, weather stations, etc.); and servers and computing clusters: servers or computing clusters for processing large-scale data and running complex algorithms. These servers may be equipped with high-performance CPUs and GPUs to accelerate the computation process. Power and Cooling Systems: A reliable power system ensures continuous system operation, and an efficient cooling system prevents hardware damage from overheating. User Interface Devices: Including monitors, keyboards, and mice for system operation and monitoring. Backup and Disaster Recovery Equipment: Such as uninterruptible power supply systems (UPS) and data backup solutions, ensuring data security and business continuity in the event of power failures or other emergencies.

[0183] The operation scenario generation module generates operation scenarios based on historical power grid operation data and uses the Latin hypercube sampling method to simulate typical and extreme power grid operation scenarios.

[0184] The power flow and node voltage calculation module uses the Newton-Lambert method to calculate power flow and node voltage based on operating scenario parameters.

[0185] A branch weighting model module is constructed based on line impedance, node voltage, and branch power flow parameters to build a branch weighting model.

[0186] Reactive power at node i Q i Voltage amplitude at node j V j The partial derivatives are:

[0187]

[0188] in, V i This represents the voltage magnitude at node i; i ij The voltage phase angle difference between node i and node j; G ij For the conductance of the branch circuit, B ij For the susceptance of the branch circuit;

[0189] Branch weights based on voltage factor w U for:

[0190]

[0191] Among them, the reactive power at node j Q j The active power of node i and node j are respectively: P i , P j ;

[0192] Branch weights based on active power flow factors w I for: w I =(| P i |+| P j |) / 2

[0193] Branch weights that take into account both voltage and active power flow factors w ij for:

[0194]

[0195] A module for constructing a power system graph theory model is provided, which constructs a power system graph theory model based on spectral graph theory. The power system graph theory model includes the number of network nodes, the number of branches, and the weight values.

[0196] The power grid topology partitioning module partitions the power grid topology map based on the spectral clustering algorithm of normal cut.

[0197] The safety risk index calculation module uses safety risk indicators to screen critical and non-critical transmission sections for transmission sections formed between zones. The safety risk index η is calculated as follows:

[0198]

[0199] in, l 1. l 2 represents two different lines at the critical transmission section S. P l1 and P l1-max They are respectively l 1. Active power flow and transmission limits of the line. P l2 For the line l The active power flow of line 2, when the line l After disconnection 1, the power flow shifts to the line. l The break distribution factor on 2 is α l1-l2 ; exp is an exponential function, max is for finding the maximum value;

[0200] Critical section output module, step S8: Search for critical transmission sections in all operating scenarios, output the critical transmission sections S of the power grid, and count the number of critical transmission sections S. N S and the probability of occurrence P S :

[0201]

[0202] in, N The number of running scenarios; when P S If the value exceeds the set threshold, the key power transmission section S will be monitored in real time.

[0203] In some embodiments, historical power grid operation data includes generator data, bus data, transformer data, load data, and line impedance data. Generator data includes the generator's rated capacity, generator voltage and current, generator reactive power, and power factor. Transformer data includes turns ratio, resistance, and admittance. Load data includes the load power, reactive power, and power factor of each node.

[0204] In some embodiments, a power system graph theory model construction module is used: Based on spectral graph theory, a power system graph theory model is constructed, which includes the number of network nodes, the number of branches, and weights; specifically, this includes: establishing a graph G=(V,E,W), where V represents the set of nodes, E represents the set of branches, and its weight matrix is ​​W, with matrix elements... W ij for:

[0205]

[0206] In the formula, w ij The branch weights between node i and node j;

[0207] The connection relationships between nodes are represented by a degree matrix. D= [ d 1 , d 2 , d 3 ,..., d n ] indicates that D is a diagonal matrix, and the diagonal elements It is represented as the sum of the elements in the i-th row of the weight matrix, where n is the number of power grid nodes;

[0208] The Laplace matrix of graph G constructed from the weight matrix W and the degree matrix D is: L=DW.

[0209] In some embodiments, the power grid topology partitioning module partitions the power grid topology map using a spectral clustering algorithm based on canonical tangents. Based on the canonical tangent criterion, the graph G is divided into partition A and partition B. The partitioning indicator vector divides the system into two partitions. When the cross-section between the partitions is a critical transmission section, it is the optimal partition.

[0210] The objective function for obtaining the optimal power grid partitioning based on the standard cutting method is as follows. f :

[0211]

[0212] In the formula, V A set of nodes; x i , x j These are the split indicator vector elements in partition A and partition B, respectively. When the node is in partition A, the split indicator vector element is... x i When the value is 1, the split indicator vector element is in partition B. x j =-1; w ijThe branch weights between node i and node j; w iz Let be the branch weight between node i and node z; w jz Let be the branch weight between node j and node z;

[0213] The objective function for finding the optimal partition of the power grid is the normalized tangent, which will be 2 n The -1 partitioning is simplified to n-1 partitioning; according to the solution principle of canonical cut, the eigenvalues ​​of the Laplacian matrix L of graph G and the Fiedler eigenvectors are calculated to obtain n-1 different partitioning indicator vectors. X 1, X 2,..., X n-1 Substitute the segmentation indicator vectors into the objective function respectively. f The partition corresponding to the segmentation indicator vector that minimizes the objective function value is the optimal partition of the power grid.

[0214] The critical transmission section output module can output the identification results and safety risk values ​​of critical transmission sections. On the one hand, when the system is used for online real-time searching of critical transmission sections, it can output the critical transmission sections and their risk values ​​for exceeding limits in online scenarios. On the other hand, by generating a massive number of simulated scenarios based on historical data scenarios for searching, it can output the critical transmission sections under all power grid scenarios and output the probability of occurrence of critical transmission sections through probabilistic statistical methods.

[0215] In some embodiments, such as Figure 3 As shown in the Xia Da scenario, the system is divided into 5 partitions, forming 6 transmission sections. Table 1 shows the search results for critical transmission sections. Transmission sections 1 and 2 have a safety risk index greater than 1, therefore they are considered critical transmission sections. The safety risk index of other transmission sections is less than 1, therefore they are considered non-critical transmission sections.

[0216] Table 1 Search Results for Key Transmission Sections

[0217]

[0218] Search for critical transmission sections under multiple scenarios. A search for critical transmission sections was performed on N operating scenarios generated based on historical power grid operating data. The results were compared with the existing monitored critical transmission sections of the power grid, and the probability of occurrence of each critical transmission section was calculated, as shown in Table 2.

[0219] Table 2 Comparison of key transmission sections monitored by the power grid and key transmission sections searched by this method

[0220]

[0221] This application has been validated in engineering applications for identifying critical transmission sections in actual power grids. It can not only accurately and efficiently identify critical transmission sections currently under operation and monitoring, but also screen out critical transmission sections that have not been discovered by operators and pose a high risk, thereby helping operation and dispatch personnel to eliminate risks and improve monitoring efficiency.

[0222] The above provides a detailed description of the method and system for searching critical transmission sections based on standard cutting and safety risks. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas and methods of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A method for searching critical transmission sections based on standard cutting and safety risks, characterized in that, include: Step S1: Import historical power grid operation data. Import historical power grid operation data from the power grid operation database. The historical power grid operation data includes generator data, bus data, transformer data, load data, and line impedance data. Step S2: Based on historical power grid operation data, use the Latin hypercube sampling method to generate operation scenarios and simulate typical and extreme power grid operation scenarios; Step S3: Calculate the power flow and node voltages of the power grid using the Newton-Layer method based on the operating scenario parameters; Step S4: Construct a branch weighting model based on line impedance, node voltage, and branch power flow parameters; The reactive power Q at node i i With respect to the voltage amplitude V at node j j The partial derivatives are: Among them, V i θ represents the voltage magnitude at node i; ij G represents the voltage phase angle difference between node i and node j. ij For the conductance of the branch, B ij For the susceptance of the branch circuit; Branch weights w based on voltage factor U for: Among them, the reactive power Q at node j j The active power of node i and node j are respectively P i P j ; Branch weights w based on active power flow factors I For: w I =(|P i |+|P j |) / 2 Branch weight w that comprehensively considers voltage factors and active power flow factors ij for: Step S5: Construct a power system graph theory model based on spectral graph theory. The power system graph theory model includes the number of network nodes, the number of branches, and the weight values. Step S6: Partition the power grid topology map based on the spectral clustering algorithm of normalized cut; Step S7: For the transmission sections formed between zones, key and non-key transmission sections are screened using safety risk indicators. The safety risk indicator η is calculated as follows: Among them, l1 and l2 are two different lines of the critical transmission section S, P l1 and P l1-max These represent the active power flow and transmission limit of line L1, respectively, P l2 For the active power flow of line l2, when line l1 is disconnected, the power flow transfer to line l2 is affected by the interruption distribution factor α. l1-l2 ; exp is an exponential function, max is for finding the maximum value; Step S8: Perform a critical transmission section search for all operating scenarios, output the critical transmission section S of the power grid, and count the number N of critical transmission sections S. S and the probability of occurrence P S : Where N is the number of running scenarios; when P S If the value exceeds the set threshold, the key power transmission section S will be monitored in real time.

2. The critical transmission section search method based on standard cutting and safety risks as described in claim 1, characterized in that, The historical power grid operation data includes generator data, bus data, transformer data, load data, and line impedance data. Among them, generator data includes the generator's rated capacity, rated voltage, rated current, reactive power, and power factor; transformer data includes turns ratio, resistance, and admittance; and load data includes the load power, reactive power, and power factor of each node.

3. The critical transmission section search method based on standard cutting and safety risks as described in claim 1, characterized in that, Step S2: Based on historical power grid operation data, an operation scenario is generated using the Latin hypercube sampling method to simulate typical and extreme power grid operation scenarios, including: S21: Obtain the probability density curve F of new energy output based on the historical power grid operation data of the new energy power generation grid connection point; S22: Latin hypercube sampling is used. Let random variables be Y1, Y2, Y3, ..., Y... T For T new energy power outputs, Y k Let Z be any one of these random variables, and its cumulative probability distribution curve be Z. k Z k =F k (Y k ); The sampling size is N running scenarios, and the sampling method is as follows: The cumulative probability distribution curve Z... k =F k (Y k The ordinate of the graph is divided into N equal intervals, each with a width of 1 / N. The midpoint of each interval is selected as Z. k The sampled values ​​are then determined according to the cumulative probability distribution function F. k Find the inverse function of Y k The sampled value, Z k The n'th sample value y kn' for: F k -1 (·) is F k The inverse function of the initial sampling matrix Y is obtained by sampling T random variables N times to form an initial sampling matrix Y of order T×N. According to the Gram-Schmidt orthogonalization method, the row and column order of the initial sampling matrix Y is adjusted using alternating forward and reverse methods, thus obtaining the transformed sampling matrix L = [L1, L2, ..., L...]. T ] T L1, L2, L T Let L be the sampling column vector matrices of the 1st, 2nd, and Tth random variables after the transformation order; find the root mean square correlation ρ of the transformed sampling matrix L. ms : In the formula: ρ kg L represents any sampled column vector k With L g The correlation coefficient between them, where cov(·) is the covariance and var(·) is the variance; The Gram-Schmidt orthogonalization method is used to make the root mean square correlation ρ of the matrix columns equal. ms The sampling matrix L stops decreasing or reaches a set threshold number of forward and reverse iterations. At this point, the transformed sampling matrix L is the Latin hypercube sampling matrix Y. TN This refers to the combination of generating units that utilize new energy sources; S23: Obtain the output data of new energy generators under N operating scenarios, i.e., the Latin hypercube sampling matrix Y. TN .

4. The critical transmission section search method based on standard cutting and safety risks as described in claim 2, characterized in that, Step S5: Constructing a power system graph theory model based on spectral graph theory. The power system graph theory model includes the number of network nodes, the number of branches, and weight values, specifically including: Construct a graph G = (V, E, W), where V represents the set of nodes, E represents the set of branches, and W is the weight matrix. ij for: The connection relationships between nodes are represented by the degree matrix D = [d1, d2, d3, ..., d n ] indicates that D is a diagonal matrix, and the diagonal elements It is represented as the sum of the elements in the i-th row of the weight matrix, where n is the number of power grid nodes; The Laplace matrix of graph G constructed from the weight matrix W and the degree matrix D is: L = DW.

5. The critical transmission section search method based on standard cutting and safety risks as described in claim 4, characterized in that, Step S6: The power grid topology map is partitioned based on the spectral clustering algorithm of normalized cut. Based on the normalized cut criterion, the graph G is divided into partition A and partition B. The partitioning indicator vector divides the system into two partitions. When the cross section between the partitions is a critical transmission section, it is the optimal partition. The objective function f for obtaining the optimal power grid partitioning based on the standard cutting method is: In the formula, V is the set of nodes; x i x j These are the split indicator vector elements in partition A and partition B, respectively. When the node is in partition A, the split indicator vector element x... i When the value is 1, the split indicator vector element x is in partition B. j -1; w iz w represents the branch weight between node i and node z. jz Let be the branch weight between node j and node z; The objective function for finding the optimal partition of the power grid is the normalized tangent, which will be 2 n The -1 partitioning is simplified to n-1 partitioning; according to the solution principle of canonical cut, the eigenvalues ​​of the Laplacian matrix L of graph G and the Fiedler eigenvectors are calculated, resulting in n-1 different partitioning indicator vectors X1, X2, ..., X... n-1 Substitute the segmentation indicator vectors into the objective function f, and the segmentation indicator vector that minimizes the objective function value corresponds to the optimal power grid segment.

6. The critical transmission section search method based on standard cutting and safety risks as described in claim 1, characterized in that, Step S3: Based on the operating scenario parameters, the Newton-Lager method is used to calculate the power flow and node voltage, including: first, initializing the voltage amplitude and phase angle of the node as the starting point of the iteration; calculating the active power and reactive power deviation of the node and determining whether it has reached the set threshold; then calculating the Jacobian matrix to obtain the voltage correction amount; finally, determining whether the target value or the number of iterations has been reached, and using the node voltage and branch power flow obtained from the iteration as the final power flow calculation result. Let the number of nodes in the power grid be n, and the node power balance equation be: Among them, P i Q i These represent the active power and reactive power of node i, respectively; V i V j Y represents the voltage magnitudes at nodes i and j, respectively; ij For branch admittance, when the branch conductance is G ij The susceptance of the branch is B. ij The active power P at node i can be obtained by decomposing the formula. i and reactive power Q i They are respectively: In the formula: θ ij The voltage phase angle difference between node i and node j is given by the formula for calculating power imbalance: Among them, P is Q is Let ΔP be the given active power and given reactive power of node i, respectively. i ΔQ i These are the active power deviation and reactive power deviation of node i, respectively; The node types include PQ load nodes, PV generator nodes, and Slack balancing nodes. When the number of PQ load nodes is m and the number of Slack balancing nodes is 1, the Jacobian matrix J is calculated using a matrix block format as follows: Where H is an (n-1)×(n-1) order square matrix, its elements θ represents the power deviation between node i and node j. j Let be the voltage phase angle at node j; R is an (n-1)×m matrix with elements of . O is an m×(n-1) matrix whose elements C is an m×m matrix whose elements 7. A critical transmission section search system based on standard cutting and safety risks, characterized in that, include: The power grid historical operation data import module imports historical power grid operation data from the power grid operation database. The historical power grid operation data includes generator data, bus data, transformer data, load data, and line impedance data. The operation scenario generation module generates operation scenarios based on historical power grid operation data and uses the Latin hypercube sampling method to simulate typical and extreme power grid operation scenarios. The power flow and node voltage calculation module uses the Newton-Lambert method to calculate power flow and node voltage based on operating scenario parameters. A branch weighting model module is constructed based on line impedance, node voltage, and branch power flow parameters to build a branch weighting model. The reactive power Q at node i i With respect to the voltage amplitude V at node j j The partial derivatives are: Among them, V i θ represents the voltage magnitude at node i; ij G represents the voltage phase angle difference between node i and node j. ij For the conductance of the branch, B ij For the susceptance of the branch circuit; Branch weights w based on voltage factor U for: Among them, the reactive power Q at node j j The active power of node i and node j are respectively P i P j ; Branch weights w based on active power flow factors I For: w I =(|P i |+|P j |) / 2 Branch weight w that comprehensively considers voltage factors and active power flow factors ij for: A module for constructing a power system graph theory model is provided, which constructs a power system graph theory model based on spectral graph theory. The power system graph theory model includes the number of network nodes, the number of branches, and the weight values. The power grid topology partitioning module partitions the power grid topology map based on the spectral clustering algorithm of normal cut. The safety risk index calculation module uses safety risk indicators to screen critical and non-critical transmission sections for transmission sections formed between zones. The safety risk index η is calculated as follows: Among them, l1 and l2 are two different lines of the critical transmission section S, P l1 and P l1-max These represent the active power flow and transmission limit of line L1, respectively, P l2 For the active power flow of line l2, when line l1 is disconnected, the power flow transfer to line l2 is affected by the interruption distribution factor α. l1-l2 ; exp is an exponential function, max is for finding the maximum value; The critical section output module searches for critical transmission sections in all operating scenarios, outputs the critical transmission sections S of the power grid, and counts the number N of critical transmission sections S. S and the probability of occurrence P S : Where N is the number of running scenarios; when P S If the value exceeds the set threshold, the key power transmission section S will be monitored in real time.

8. The critical transmission section search system based on standard cutting and safety risks as described in claim 7, characterized in that, The historical power grid operation data includes generator data, bus data, transformer data, load data, and line impedance data. Among them, generator data includes the generator's rated capacity, rated voltage, rated current, reactive power, and power factor; transformer data includes turns ratio, resistance, and admittance; and load data includes the load power, reactive power, and power factor of each node.

9. The critical transmission section search system based on standard cutting and safety risks as described in claim 7, characterized in that, The module for constructing a power system graph theory model: Based on spectral graph theory, a power system graph theory model is constructed. This model includes the number of network nodes, the number of branches, and their weights. Specifically, it includes: establishing a graph G = (V, E, W), where V represents the set of nodes, E represents the set of branches, and its weight matrix is ​​W, with matrix elements W... ij for: The connection relationships between nodes are represented by the degree matrix D = [d1, d2, d3, ..., d n ] indicates that D is a diagonal matrix, and the diagonal elements It is represented as the sum of the elements in the i-th row of the weight matrix, where n is the number of power grid nodes; The Laplace matrix of graph G constructed from the weight matrix W and the degree matrix D is: L = DW.

10. The critical transmission section search system based on standard cutting and safety risks as described in claim 7, characterized in that, The power grid topology partitioning module partitions the power grid topology map using a spectral clustering algorithm based on normalized cut. Based on the normalized cut criterion, the graph G is divided into partition A and partition B. The partitioning indicator vector divides the system into two partitions. When the cross-section between the partitions is a critical transmission section, it is the optimal partition. The objective function f for obtaining the optimal power grid partitioning based on the standard cutting method is: In the formula, V is the set of nodes; x i x j These are the split indicator vector elements in partition A and partition B, respectively. When the node is in partition A, the split indicator vector element x... i When the value is 1, the split indicator vector element x is in partition B. j -1; w iz w represents the branch weight between node i and node z. jz Let be the branch weight between node j and node z; The objective function for finding the optimal partition of the power grid is the normalized tangent, which will be 2 n The -1 partitioning is simplified to n-1 partitioning; according to the solution principle of canonical cut, the eigenvalues ​​of the Laplacian matrix L of graph G and the Fiedler eigenvectors are calculated, resulting in n-1 different partitioning indicator vectors X1, X2, ..., X... n-1 Substitute the segmentation indicator vectors into the objective function f, and the segmentation indicator vector that minimizes the objective function value corresponds to the optimal power grid segment.

Images