Power grid weak link identification method, system and equipment

Abstract

The invention discloses a power grid weak link identification method, system and equipment, and the method comprises the steps: firstly, selecting each evaluation index of the operation state of a power grid, then determining the initial weight of each evaluation index through employing an analytic hierarchy process, and carrying out the updating of the initial weight of each evaluation index through employing a Bayesian network and a self-learning method, obtaining the weight of each evaluation index; generating a power grid blocking event sample by adopting an OPA model based on direct current power flow optimization, and performing power grid operation state evaluation by adopting the power grid blocking event sample according to each evaluation index and the corresponding weight to obtain an evaluation result; and if the evaluation result shows that the cascading failure exists in the power grid, identifying a weak link of the power grid by adopting a link analysis algorithm. The method can improve the accuracy and efficiency of weak link identification of the power grid, and can effectively predict and prevent the blocking event of the power grid.

Description

Technical Field

[0001] The present invention relates to the technical field of power system operation, and particularly to a method, system and device for identifying weak links in a power grid. Background Art

[0002] In recent years, the scale of the new power system in China has continued to expand rapidly. However, the high proportion of wind power grid connection has introduced a large amount of uncertainty, increasing the risk of cascading propagation of grid congestion faults between branches in extreme scenarios, bringing new challenges to the power system power flow stability. As a key infrastructure for power transmission and distribution, the evaluation of the operation state of the power grid and the prevention of congestion events are of great significance for preventing catastrophic power outages.

[0003] However, in terms of the evaluation of the power grid operation state, the previous evaluation index system is not perfect enough to comprehensively capture the complex and changeable operation state of the power system, resulting in deviations in the evaluation of the power grid safety state and unable to provide a reliable basis for the safe and stable operation of the power system. At the same time, there are also obvious deficiencies in the traditional method for determining index weights. Some methods simply rely on subjective experience judgment, making the weight distribution lack objectivity and unable to accurately reflect the true influence degree of each evaluation index in the actual power grid operation; while some other methods completely based on objective data are difficult to make full use of the prior knowledge and industry experience of experts and have poor adaptability in the face of complex and changeable power grid operation scenarios. This leads to the weight distribution being unable to accurately reflect the true influence degree of each evaluation index in the actual power grid operation, thus affecting the accuracy and reliability of the evaluation results.

[0004] In the field of power grid congestion event prevention, there are great limitations in the existing technologies in terms of simulation and weak link identification. On the one hand, the traditional OPA model power grid congestion event simulation method is difficult to accurately simulate the interaction of various complex factors in actual operation; on the other hand, the identification method for potential weak links in the power grid is not deep and comprehensive enough. The existing identification algorithms often can only find some superficial and obvious weak points and are difficult to dig out the potential risk points hidden in the complex power grid structure and operation state.

[0005] In summary, there are many deficiencies in the current power grid operation state evaluation and congestion event prevention technologies, and there is an urgent need for a more perfect and accurate method to ensure the safe and stable operation of the power system. Summary of the Invention

[0006] The object to be achieved by the present invention is to provide a method, system and device for identifying weak links in a power grid, and through constructing a perfect quantitative evaluation index system, an innovative weight determination method, and an advanced congestion event simulation and weak link identification algorithm, realize the accurate evaluation of the power grid safety state and the effective prevention of congestion events.

[0007] To achieve the above objectives, an embodiment of the present invention provides a method for identifying weak links in a power grid, including:

[0008] Selecting various evaluation indicators of the power grid operation status;

[0009] Using the analytic hierarchy process to determine the initial weights of the various evaluation indicators, and using the Bayesian network and self-learning method to update the initial weights of the various evaluation indicators to obtain the weights of the various evaluation indicators;

[0010] Generating power grid congestion event samples using an OPA model based on DC power flow optimization;

[0011] According to the various evaluation indicators and the corresponding weights, using the power grid congestion event samples to evaluate the power grid operation status to obtain an evaluation result;

[0012] If the evaluation result shows that there are cascading failures in the power grid, a link analysis algorithm is used to identify the weak links in the power grid.

[0013] As an improvement of the above solution, the step of using the Bayesian network and self-learning method to update the initial weights of the various evaluation indicators to obtain the weights of the various evaluation indicators includes:

[0014] Taking the initial weights of the various evaluation indicators as node attributes to construct a Bayesian network;

[0015] Calculating the weight gradient on each node in the Bayesian network;

[0016] According to the weight gradient on each node, using the post-gradient descent method to iteratively update the weights on each node until the maximum entropy principle is satisfied to obtain the weights of the various evaluation indicators; wherein, in each iteration, the iterated weights are normalized.

[0017] As an improvement of the above solution, the step of generating power grid congestion event samples using an OPA model based on DC power flow optimization includes:

[0018] Determining the DC power flow optimization objective function and building an OPA model based on the DC power flow optimization objective function;

[0019] Collecting power grid data and setting the initial operating state of the OPA model; wherein, the power grid data includes the initial generator output, the load demand magnitude of each node, and the initial power grid topology;

[0020] Updating the loads, generators, and partial line capacities in the OPA model to simulate the power grid under different faults and operating states, and generating samples of power grid congestion events.

[0021] As an improvement of the above solution, the DC power flow optimization objective function is:

[0022]

[0023]

[0024] Where G represents the set of generator nodes, are the upper and lower limits of the power of generator i, p i represents the power of generator i, and D represents the load node set, p j represents the load of node j, p dj represents the load demand of node j, and P dj ≤p j ≤0,j∈D;c i represents the unit power generation cost of the i-th generator, w j represents the loss caused by the removal of unit load at node j; L is the line set, F l is the flow of the lth branch, F l max is the transmission capacity of the lth branch, F = (F 1 ,F 2 ,…,F m ) T ; p is the node injection power, p = (p 1 ,p 2 ,…,p n ) T ; A represents the conversion matrix from node injection power to branch power flow.

[0025] As an improvement of the above solution, the link analysis algorithm is used to identify the weak links of the power grid, including:

[0026] A link analysis algorithm is used to determine the link relationship between the power grid branch, the propagation link of the power grid fault and the vulnerable line of the power grid; wherein the power grid branch is used as a node in the link analysis algorithm, the propagation line is used as a central node in the link analysis algorithm, and the vulnerable line is used as a weight node in the link analysis algorithm;

[0027] Using the samples of the power grid congestion events, and calculating the set of power grid branches, the set of propagation links of power grid faults, and the set of vulnerable lines of the power grid according to the link relationship;

[0028] Initializing the propagation value and the vulnerability value of each power grid branch, and iteratively calculating the center value of the propagation line and the weight value of the vulnerable line according to the propagation value and the vulnerability value, until the center value and the weight value meet the preset convergence condition;

[0029] Sort each power grid branch, identify the weak links of the power grid according to the comprehensive index of weak links and the fault propagation correlation index, and obtain the weak links and the correlation of the weak links.

[0030] As an improvement of the above solution, the comprehensive index of weak links is calculated by the following formula:

[0031]

[0032] Wherein, represents the comprehensive index of weak links, represents the central value of the propagation line, α represents the weight of, represents the weight value of the vulnerable line, 1-α represents the weight of;

[0033] The fault propagation correlation index is calculated by the following formula:

[0034]

[0035] Wherein, represents the fault propagation correlation index, The larger the value of, the greater the probability that different vulnerable lines are affected by the same propagation line, and the greater the correlation between vulnerable lines; l a represents the vulnerable line, respectively represent the xth and yth lines in the vulnerable line; represents can both be affected by the same propagation line; represents the xth and yth lines in the vulnerable line the probability that both can be affected by the same propagation line, represents the probability that the yth and xth lines in the vulnerable line set can both be affected by the same propagation line.

[0036] As an improvement of the above solution, the evaluation indexes include the electrical betweenness difference index, the power flow distribution index, the voltage margin index, the load rate index and the node voltage loss index;

[0037] The electrical betweenness difference index is determined by the Gini coefficient;

[0038] The power flow distribution index is determined according to the limit transmission capacity of each branch of the power grid, the power flow distribution and the number of branches of the power grid;

[0039] The voltage margin index is determined according to the voltage amplitude of the load nodes of the power grid;

[0040] The load rate index is determined according to the transmission power of each branch of the power grid and the limit transmission capacity of each branch of the power grid;

[0041] The node voltage loss index is determined according to the node voltages after a power grid fault and the node voltages in the normal state of the power grid.

[0042] As an improvement to the above solution, the initial weights of the various evaluation indicators are determined by using the analytic hierarchy process, including:

[0043] Establish a hierarchical structure model according to the various evaluation indicators;

[0044] Construct a judgment matrix according to the hierarchical structure model and the various evaluation indicators;

[0045] Calculate the initial weights of the various evaluation indicators according to the judgment matrix;

[0046] Verify the judgment matrix according to the consistency index. If the judgment matrix does not meet the preset conditions, adjust the judgment matrix and recalculate the initial weights of the various evaluation indicators until the judgment matrix meets the preset conditions.

[0047] To achieve the above object, an embodiment of the present invention further provides a power grid weak link identification system, including:

[0048] An index selection module, configured to select various evaluation indicators of the power grid operation status;

[0049] An index weight calculation module, configured to determine the initial weights of the various evaluation indicators by using the analytic hierarchy process, and update the initial weights of the various evaluation indicators by using the Bayesian network and self-learning method to obtain the weights of the various evaluation indicators;

[0050] A sample generation module, configured to generate power grid congestion event samples by using an OPA model based on DC power flow optimization;

[0051] An evaluation module, configured to evaluate the power grid operation status by using the power grid congestion event samples according to the various evaluation indicators and the corresponding weights to obtain an evaluation result;

[0052] A weak link identification module, configured to identify the weak links of the power grid by using a link analysis algorithm if the evaluation result shows that there are cascading failures in the power grid.

[0053] To achieve the above object, an embodiment of the present invention further provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the power grid weak link identification method described in any one of the above is implemented.

[0054] Compared with the prior art, the embodiment of the present invention provides a method, system and device for identifying weak links in a power grid. First, each evaluation index of the power grid operation state is selected, and then the initial weight of each evaluation index is determined by the hierarchical analysis method, and the initial weight of each evaluation index is updated by the Bayesian network and the self-learning method to obtain the weight of each evaluation index; then, the OPA model based on DC power flow optimization is used to generate a power grid blocking event sample, and according to each evaluation index and the corresponding weight, the power grid blocking event sample is used to evaluate the power grid operation state to obtain an evaluation result; if the evaluation result shows that there is a cascading failure in the power grid, the link analysis algorithm is used to identify the weak links of the power grid. The present invention can fully capture the operating state of the power system, thereby realizing an accurate evaluation of the safety state of the power grid. At the same time, the Bayesian network and the maximum entropy weight adaptive method are used in weight determination. This method effectively integrates subjective and objective factors, making the weight distribution more in line with the actual situation. In addition, the OPA model based on power flow optimization and the SALSA algorithm are combined to more deeply explore the potential weak links of the power grid, improve the accuracy and efficiency of weak link identification, and effectively predict and prevent power grid blocking events. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solution of the present invention, the drawings used in the implementation mode will be briefly introduced below. Obviously, the drawings described below are only some implementation modes of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0056] Figure 1 is a flow chart of a method for identifying weak links in a power grid provided by an embodiment of the present invention;

[0057] Figure 2 It is a flow chart for solving the weight of an evaluation index provided by an embodiment of the present invention;

[0058] Figure 3 A flowchart of iteratively generating power grid congestion event samples using an OPA model based on power flow optimization provided by an embodiment of the present invention;

[0059] Figure 4 It is a schematic diagram of propagation value and vulnerability value of each branch in a power system provided by an embodiment of the present invention;

[0060] Figure 5 It is a structural block diagram of a power grid weak link identification system provided by an embodiment of the present invention;

[0061] Figure 6 It is a structural block diagram of a terminal device provided by an embodiment of the present invention. Detailed implementation manners

[0062] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts belong to the scope of protection of the present invention.

[0063] Refer to Figure 1 , Figure 1 which is a flowchart of a method for identifying weak links in a power grid provided by an embodiment of the present invention. The method for identifying weak links in the power grid includes steps S1 to S5:

[0064] S1. Select each evaluation index of the power grid operation state;

[0065] Exemplarily, in specific implementation, according to the operation mechanism and fault occurrence mechanism of the transmission power grid, appropriate quantitative evaluation indexes can be selected to construct a safety state evaluation system for the transmission power grid, so that the operation state of the power grid can be preliminarily evaluated before identifying the weak links in the power grid. The preliminary evaluation can help determine the areas or links where problems currently exist in the power grid, such as large voltage fluctuations in some areas and too high load rates of some lines, making the identification of weak links more targeted, concentrating on analyzing the places where problems may really exist, avoiding blind search, and improving the accuracy and efficiency of identification.

[0066] In an alternative embodiment, the evaluation indexes include the electrical betweenness difference index, the power flow distribution index, the voltage margin index, the load rate index, and the node voltage loss index;

[0067] Specifically, the electrical betweenness difference index is determined by the Gini coefficient.

[0068] It should be noted that in a power system, a generator is a device that generates electric energy, a generator node is the connection point of the generator in the power grid, the electric energy generated by the generator is injected into the power grid through the generator node, and the active power output of the generator is realized through the generator node to interact with the power grid. A load is a device or system that consumes electric energy, a load node is the point where the load is connected to the power grid, and the load node obtains electric energy from the power grid to meet its own power consumption needs, that is, consumes the active power transmitted from the generator through the power grid. That is, the generator node is the injection end of the active power, and the load node is the receiving end of the active power. They are connected to each other through the operation branches of the power grid (such as transmission lines, etc.) to form a complete electric energy transmission path, so that the power flows from the generator node to the load node to realize the transmission and distribution of electric energy.

[0069] Exemplarily, based on Kirchhoff's law, in the embodiments of the present invention, the electrical betweenness of the operating branch i-j in the power grid is defined as:

[0070]

[0071] In the formula, G is the set of generator nodes, D is the set of load nodes, represents the active power generated on branch i-j when injecting unit active power from the generator-load node pair p-q; considering that power can only flow from the generator node to the load node, in the embodiments of the present invention, the active power P = 1 is injected from the generator node p, and the active power P = -1 is injected from the load node q; W pq = min(S p , S q ), where W pq is the coefficient of electrical betweenness, representing the maximum available transmission power between nodes p and q, S p represents the rated power generation capacity of the generator node p, and S q represents the maximum load of the load node q.

[0072] It should be noted that in the embodiments of the present invention, the Gini coefficient is introduced to measure the uniformity of the branch electrical betweenness, and further evaluate the vulnerability of the power grid network structure, so as to evaluate the security state of the power grid. Among them, the larger the Gini coefficient of the branch electrical betweenness, the more vulnerable the power grid network structure is.

[0073] It should be noted that generally, the Gini coefficient is a common indicator used to measure the fairness of the income distribution of residents in a country or region, and the Gini coefficient is more commonly calculated based on the Lorenz curve. The Lorenz curve is in a plane rectangular coordinate system, with the vertical axis representing the cumulative percentage of income and the horizontal axis representing the cumulative percentage of the population. The population of a country is sorted by income from low to high, and then the income ratio corresponding to a certain proportion of the population is calculated, and these points are connected to obtain the Lorenz curve.

[0074] In the embodiments of the present invention, the electrical betweenness of N branches in the transmission grid can be arranged in ascending order, denoted as B 1 , B 2 , …, B i , …, B N , with the cumulative node percentage as the abscissa, that is, the ratio of the subscript i of each electrical betweenness B to N with the cumulative electrical betweenness percentage as the ordinate, that is, the ratio of the cumulative electrical betweenness to the sum of the electrical betweenness

[0075] Then the calculation formula of the Gini coefficient G' in the embodiments of the present invention is:

[0076]

[0077] Among them, the larger G is, the greater the difference in electrical betweenness in each branch, indicating that the network structure of the power grid is more uneven, that is, more fragile; on the contrary, it indicates that the network structure of the transmission grid is more uneven.

[0078] Specifically, the power flow distribution index is determined according to the limit transmission capacity of each branch of the power grid, the power flow distribution and the number of branches of the power grid.

[0079] It should be noted that the power flow distribution index refers to the ratio of the sum of the differences between the limit transmission capacity and the power flow distribution of each branch in the power grid system to the number of branches, and it can reflect the static stability of the power system.

[0080] In the embodiment of the present invention, the power flow distribution index K s is defined as:

[0081]

[0082] In the formula, F i is the power flow distribution of branch i. Since this index reflects the static stability of the power grid system, only the active power flow is considered; F imax is the limit transmission capacity of branch i; N is the total number of branches. The larger K s , the greater the difference between the limit transmission capacity of the system branch and the power flow distribution, and the more stable and safe the power grid system is.

[0083] Specifically, the voltage margin index is determined according to the voltage amplitude of the load nodes of the power grid.

[0084] It should be noted that the voltage margin index refers to the average value of the ratio of the difference between the maximum allowable voltage amplitude and the current voltage amplitude of all load nodes to the current voltage amplitude, and it can reflect the ability of the load nodes of the power system to withstand voltage disturbances and the reactive power reserve ability.

[0085] In the embodiment of the present invention, the voltage margin index K V is defined as:

[0086]

[0087] In the formula, U j is the voltage amplitude of load node j; U jmax is the maximum allowable voltage amplitude of node j; B is the total number of load nodes. The larger K V , the stronger the ability of the load nodes of the power system to withstand voltage disturbances, and the more stable the power system is.

[0088] Specifically, the load rate index is determined according to the transmission power of each branch of the power grid and the limit transmission capacity of each branch of the power grid.

[0089] It should be noted that the system load rate index refers to the ratio of the sum of the transmission powers of each branch to the sum of the limit transmission capacities of each branch, and it can reflect the probability of a power system failure.

[0090] In the embodiment of the present invention, the system load rate index L l is defined as:

[0091]

[0092] In the formula, F l is the active power flow of branch l, F lmax is the limit transmission capacity of branch l, and L is the total number of branches. The larger L l is, the greater the probability of failure and the more dangerous the operation of the power grid.

[0093] Specifically, the node voltage loss index is determined according to the voltages of each node after a power grid failure and the voltages of each node in the normal state of the power grid.

[0094] It should be noted that the node voltage loss index refers to the sum of the algebraic differences between the voltages of each node after a power grid failure and the node voltages in the normal state of the transmission grid, and it can reflect the volatility of the system voltage in terms of amplitude and phase angle after a failure occurs.

[0095] In the embodiment of the present invention, the node voltage loss index ΔU is defined as:

[0096]

[0097]

[0098] In the formula, ∠U a are respectively the amplitude and phase angle of the voltage of node a after a failure occurs; ∠U a0 are respectively the amplitude and phase angle of the voltage of node a during normal operation; U aN is the rated voltage of the network at this place; N is the number of network nodes. The larger ΔU is, the greater the deviation of the node voltage from the normal voltage after a failure occurs, and the more dangerous the system is.

[0099] It should be noted that these evaluation indicators can be measured from multiple aspects such as the topological structure of the power grid, power transmission, voltage stability, and load conditions. Through the electrical betweenness difference index, nodes and lines that have a greater impact on power transmission can be identified when the power grid structure changes, and these are often weak points; the load rate index can clarify which equipment is overloaded and prone to failure; the voltage margin and node voltage loss indexes can locate areas with poor voltage stability, and the power flow distribution index can detect unreasonable power distribution, providing a basis for targeted maintenance and improvement. Traditional evaluation indicators may only focus on certain characteristics of the power grid, while this set of indicators starts from multiple dimensions to achieve a comprehensive evaluation.

[0100] S2. Determine the initial weights of the respective evaluation indicators by using the analytic hierarchy process, and update the initial weights of the respective evaluation indicators by using the Bayesian network and self-learning method to obtain the weights of the respective evaluation indicators.

[0101] In one implementation, the determining the initial weights of the respective evaluation indicators by using the analytic hierarchy process includes:

[0102] Establish a hierarchical structure model according to the respective evaluation indicators.

[0103] Construct a judgment matrix according to the hierarchical structure model and the respective evaluation indicators.

[0104] Calculate the initial weights of the respective evaluation indicators according to the judgment matrix.

[0105] Verify the judgment matrix according to the consistency index. If the judgment matrix does not meet the preset conditions, adjust the judgment matrix and recalculate the initial weights of the respective evaluation indicators until the judgment matrix meets the preset conditions.

[0106] It should be noted that the analytic hierarchy process (AHP algorithm) is a method for multi-objective and multi-criteria decision-making analysis. It decomposes complex decision-making problems into multiple levels, including the goal level, criterion level, alternative level, etc. By establishing a hierarchical structure model, the problem is organized and hierarchized, enabling people to clearly analyze the various factors of the problem and their mutual relationships. Using the method of pairwise comparison, the importance of each element in the same level with respect to a certain criterion in the upper level is compared pairwise to construct a judgment matrix. Then, by calculating the eigenvector and eigenvalue of the judgment matrix, the relative weights of each element for this criterion are determined. Finally, by synthesizing the weights of each level, the total weights of each alternative in the alternative level for the goal level are calculated, so as to rank each alternative according to the total weights and provide a basis for decision-making.

[0107] Exemplarily, the weight value range of the power grid operation status evaluation index is calculated based on the AHP algorithm to obtain the initial weight of the transmission grid operation status evaluation index, including the following steps:

[0108] (1) Construct a judgment matrix:

[0109] Suppose it is now necessary to obtain the influence degree, that is, the weight, of n power grid operation status evaluation indicators D = {d 1 , d 2 , … d n} on a certain power grid fault event A. Take any two indicators d i , d j , and use c ij to represent the influence degree of d i , d j on A. Finally, form a comparison matrix C = [c ij n×n , where the comparison matrix needs to satisfy

[0110] Then, construct a judgment matrix through the comparison matrix. The constructed judgment matrix is:

[0111]

[0112] (2) Calculate the index weight:

[0113] Calculate the index weight by the method of eigenvalue and eigenvector. First, calculate the maximum eigenvalue β n of the judgment matrix B max :

[0114]

[0115] Then, normalize the maximum eigenvector λ max corresponding to the maximum eigenvalue β max to obtain the initial weight of each index. Through the repeated comparison between various indexes, a reasonable initial weight result can be obtained as much as possible on the basis of considering expert experience.

[0116] (3) Consistency test:

[0117] When generating the comparison matrix C = [c ij n×n , it is necessary to conduct a consistency test on it. If the consistency test is passed, it indicates that the judgment matrix B n is reasonably selected; otherwise, the reliability is reduced. The specific steps are as follows:

[0118] Calculate the consistency index CI:

[0119] ​​

[0120] Where n is the dimension of the comparison matrix.

[0121] Then, find the average random consistency index RI and calculate the consistency index ratio CR:

[0122]

[0123] When CR<0.1, the consistency test is passed; otherwise, the consistency test is not passed, and the judgment matrix is ​​modified until it passes the test.

[0124] When the judgment matrix B n When it is the optimal judgment matrix, the eigenvalue β of the judgment matrix max Equal to the dimension n of the judgment matrix, that is, β max =n; in other cases, β max >n. Therefore, the weight interval calculation is to solve the optimization problem:

[0125]

[0126] In the formula, w i is the weight of the i-th indicator, B n is the judgment matrix of formula (8), β max is the maximum eigenvalue of the judgment matrix calculated by formula (9), and n is the number of weights required.

[0127] Furthermore, the weights obtained by the AHP method are used as the initial solution, and the weights are iteratively updated by combining the Bayesian network and the self-learning method, and the maximum entropy is used as the objective function to obtain the final weights.

[0128] In one embodiment, the use of the Bayesian network and the self-learning method to update the initial weights of the various evaluation indicators to obtain the weights of the various evaluation indicators includes:

[0129] Using the initial weights of the evaluation indicators as node attributes to construct a Bayesian network;

[0130] Calculating the weight gradient on each node in the Bayesian network;

[0131] According to the weight gradient on each node, the weight on each node is iteratively updated using the post-gradient descent method until the maximum entropy principle is met, thereby obtaining the weights of each evaluation indicator; wherein, in each iteration, the iterated weights are normalized.

[0132] See also Figure 2 , Figure 2 This is a flowchart of solving the weight of an evaluation index provided by an embodiment of the present invention. Figure 2 As shown, for example, assuming there are random equally likely samples X1 , X 1 , …, X N , where \(w\) represents the initial weight obtained by the AHP algorithm, and \(a\) is a node attribute in the self-learning optimization process, that is, a node in the Bayesian network; \(P(a|X\) k ) represents the prior probability of node \(a\) under sample \(X\) k .

[0133] Regarding the value of \(w\), it is first assumed to be an equally likely event. The objective of the embodiment of the present invention is to have the maximum value, and \(P\) w (X\) k ) represents the comprehensive weight of \(X\) in the Bayesian network k .

[0134] The initial weight obtained by the AHP algorithm is introduced as the initial solution into the self-learning process, and iterative calculations are performed through the gradient descent method to finally converge to the optimal solution.

[0135] Among them, the self-learning steps are as follows:

[0136] (1) Gradient calculation, that is, calculate the weight on each node in the Bayesian network through the following formula:

[0137]

[0138] (2) Gradient iteration, that is, use the post-gradient descent method for gradient iteration through the following formula:

[0139]

[0140] In the formula, \(lr\) is the learning rate.

[0141] (3) Weight normalization, that is, perform normalization processing on the iterated weights to ensure that the feature weights of the same layer under the same node in the Bayesian network satisfy \(\sum w\) i = 1.

[0142] (4) End of self-learning. The end principle is to satisfy the maximum entropy principle. Calculate the entropy value after one iteration. When the maximum entropy principle is not satisfied, use the weight obtained in this round of iteration as the initial weight for the next round of iteration and recalculate; when the maximum entropy principle is satisfied, stop the calculation, and the output result is the required weight.

[0143] In summary, the weight is obtained by a combined subjective and objective method based on Bayesian and maximum entropy. The Bayesian network allows the incorporation of prior knowledge, which reflects people's subjective cognition and experience of the problem, while the maximum entropy principle is based on objective data and seeks the most uniform and unbiased probability distribution that conforms to the known data under no other constraints. The combination of the two can organically integrate subjective information and objective data, making the determination of the weight more comprehensive and reasonable.

[0144] S3. Generate power grid congestion event samples using an OPA model based on DC power flow optimization;

[0145] In one embodiment, the generating power grid congestion event samples using an OPA model based on DC power flow optimization includes:

[0146] Determine the DC power flow optimization objective function and construct an OPA model based on the DC power flow optimization objective function;

[0147] Collect the initial data of the power grid and set the initial operating state of the OPA model; wherein, the initial data includes the initial generator output, the load demand magnitude of each node, and the initial power grid topology;

[0148] Update the loads, generators, and capacities of some lines in the OPA model, simulate the power grid under different fault and operating conditions, and generate power grid congestion event samples.

[0149] Exemplarily, for a power system with n nodes and m branches, the injection power of the i-th node (generator i) of the system is p i , let p = (p 1 , p 2 , …, p n ) T , then the branch power flow The calculation formula is:

[0150]

[0151] In the formula, θ represents the node voltage phase angle; let the phase angle θ s of the slack node s = 0, is the injection power after removing node s, is the (n - 1) × (n - 1) node admittance matrix after removing the row and column where node s is located; M f is the admittance matrix associated with the branch - node; is the m × (n - 1) - dimensional matrix after removing the column related to node s.

[0152] When the power system is overloaded due to certain reasons, the system will automatically adjust the generator output according to the pre - plan to eliminate the overload, and even actively cut off some loads when necessary. This process can be described by the following optimization problem.

[0153] The power of the system generator i is p i , let p = (p 1 , p 2 , …, p n ) T , G is the set of generator nodes, and D is the set of load nodes.

[0154] Let the upper and lower power limits of generator i be Then the generator output constraint is:

[0155]

[0156] Let the load demand of node j be P dj (<0), then the load constraint of node j is:

[0157] P dj ≤p j ≤0, j ∈ D (17);

[0158] Let L be the set of lines, F l be the power flow of the l-th branch, F l max be the transmission capacity of the l-th branch, and denote F = (F 1 , F 2 , …, F m ) T . Considering the power flow direction, the branch capacity constraint is:

[0159] -F l max ≤F l ≤F l max , l ∈ L (18);

[0160] Considering the power balance equation and the branch power flow equation, the DC power flow optimization objective problem can be described as:

[0161]

[0162] In the formula, the variable to be solved is the node injection power p, c i represents the unit power generation cost of the i-th generator, w j represents the loss caused by removing a unit load at node j, which is generally large.

[0163] Furthermore, referring to Figure 3 , Figure 3 is a flowchart of a method for iteratively generating power grid congestion event samples using an OPA model based on power flow optimization provided by an embodiment of the present invention. As Figure 3As shown, the OPA model based on power flow optimization is used to simulate the faults on the k-th day. When initializing, k is set to 0. When k = 1, the initial generator output, load demand magnitude, and network structure of the power system are determined; in each loop update (k = k + 1), the load, generator, and part of the line capacity are updated, and the N-2 fault is simulated (the N-2 fault represents the operating state in which two components are simultaneously lost in the power system. Here, "N" represents the total number of components in the power system under normal operation, and "N - 2" means that two components are removed from the normal state. This is an extreme operating condition assumption used to evaluate the stability and reliability of the system under severe faults), and the DC optimization problem (19) is solved. Then, it is judged whether there are overloaded lines. If not, the load to be shed is calculated. If so, an overloaded line opening test is carried out; after the overloaded line opening test, it is judged whether there is a line opening. If so, a new network structure is formed, and then it returns to the loop update step. If not, the load to be shed is calculated, and it is judged whether k is equal to the set maximum loop count k max ; if not, it returns to the loop update step. If so, the process ends, and the corresponding power grid congestion event samples are generated. This flowchart is mainly used to simulate the response of the power system under specific fault conditions. By continuously iterating and judging, the operating state of the system, load shedding situation, etc. are analyzed to evaluate and optimize the reliability and stability of the power system.

[0164] Exemplarily, the IEEE 39-bus system is selected as an example in the present invention, and the total system load is 6254 MW. Using the OPA model based on power flow optimization in the embodiment of the present invention, 6740 effective fault chains with an initial fault of N-2 fault are simulated and generated. Let the maximum error ε = 10 -4 ; initial After calculation, a total of 46 propagation lines and 38 vulnerable lines are identified.

[0165] Refer to Figure 4 , Figure 4 which is a schematic diagram of the propagation value and vulnerability value of each branch in the power system provided by the embodiment of the present invention. As Figure 4 shown, taking the IEEE 39-bus system as an example, there are a total of 46 branches. As can be seen from Figure 3 , not all branches are vulnerable lines, that is, some lines only serve as propagation lines rather than vulnerable lines and can only trigger other lines to fail as the initial N-2 fault, such as branches L2(1 - 39), L5(2 - 30), L10(5 - 6), etc.; some branches serve as both propagation lines and vulnerable lines, that is, such branches are not only prone to deteriorating the propagation of power grid congestion events, but also extremely likely to fail by themselves and cause line breaks, such as branches L1(1 - 2), L3(2 - 3), L42(26 - 27), L27(16 - 19), etc.

[0166] S4. According to the various evaluation indicators and corresponding weights, use the power grid congestion event samples to evaluate the operation state of the power grid and obtain the evaluation results;

[0167] Exemplarily, the embodiment of the present invention uses a subjective and objective method combining Bayesian and maximum entropy to determine the weights, and obtains that the weight of the Gini coefficient of electrical betweenness is 0.2038, the weight of the power flow distribution index is 0.1483, the weight of the voltage margin index is 0.1016, the weight of the system load rate index is 0.2894, and the weight of the node voltage loss index is 0.2569;

[0168] Each sample in the power grid congestion event samples contains the specific data of the electrical betweenness difference index, power flow distribution index, voltage margin index, load rate index, and node voltage loss index. For each sample, calculate according to the formula "evaluation score = index 1 × index 1 weight + index 1 × index 1 weight +... + index n × index n weight". Suppose the evaluation score of 40 - 45 is "there is a high risk", 45 - 50 is "there is a serious risk, which may trigger a cascading failure", and above 50 is "a cascading failure has occurred". When the evaluation scores of multiple consecutive samples or a certain proportion of samples are in the interval of "there is a serious risk, which may trigger a cascading failure" and above, and it is observed that the sample data shows a deteriorating trend, such as the continuous increase of the load rate and the gradual increase of the node voltage, etc., it is determined that there is a cascading failure in the power grid.

[0169] S5. If the evaluation result shows that there is a cascading failure in the power grid, then use the link analysis algorithm to identify the weak links of the power grid.

[0170] In an alternative embodiment, the using the link analysis algorithm to identify the weak links of the power grid includes:

[0171] Use the link analysis algorithm to determine the link relationship between the power grid branches, the propagation links of the power grid faults, and the vulnerable lines of the power grid; wherein, use the power grid branches as the nodes in the link analysis algorithm, use the propagation lines as the central nodes in the link analysis algorithm, and use the vulnerable lines as the weight nodes in the link analysis algorithm;

[0172] Use the samples of the power grid congestion events, and calculate the power grid branch set, the power grid fault propagation link set, and the power grid vulnerable line set according to the link relationship;

[0173] Initialize the propagation value and vulnerability value of each power grid branch, and perform iterative calculation on the central value of the propagation line and the weight value of the vulnerable line according to the propagation value and vulnerability value until the central value and weight value meet the preset convergence conditions;

[0174] Sort each power grid branch, and identify the weak links of the power grid according to the comprehensive index of weak links and the fault propagation correlation index, so as to obtain the weak links and the correlation of the weak links.

[0175] It should be noted that steps S1 to S4 complete the overall preliminary situation assessment of the power grid system. However, in the case of a cascading fault, a more in-depth and specific situation assessment is required. Therefore, based on a large number of power grid congestion event samples, the power grid congestion event state assessment is carried out, and the weak link identification based on the link analysis algorithm is performed.

[0176] Exemplarily, taking the SALSA algorithm, a link analysis algorithm, as an example, the SALSA algorithm classifies nodes into Hub nodes (central nodes) and Authority nodes (weight nodes) according to the pointing and being pointed relationship: A Hub node is a node with an out-degree greater than 0, and an Authority node is a node with an in-degree greater than 0. At the same time, the Hub value (central value) and the Authority value (weight value) are used to measure the node quality. The link relationship is described mathematically as follows:

[0177]

[0178] In the formula, l = {l 1 , l 2 , …, l q} is the page set; the link relationship is G = (L hub , L aut , E), E is the edge between l h and l a , and the number of nodes of L hub and L aut are b1 and b2 respectively.

[0179] π hub 、π aut are obtained by iterative calculation, and the specific calculation method is:

[0180]

[0181]

[0182] When the convergence condition is met, the final Hub value and the Authority value can be obtained, which represent the quality metrics of the corresponding Hub nodes and Authority nodes. The convergence condition is:

[0183]

[0184]

[0185] where ε is the maximum error.

[0186] In the embodiment of the present invention, H ub represents the propagation value, and A ut represents the vulnerability value, and its calculation method is:

[0187]

[0188]

[0189] where W row is the row-normalized matrix of the link matrix W, and W col is the column-normalized matrix of the link matrix W.

[0190] Define the set of branches of the power transmission network system (with a total of s branches), and the set of fault chains of cascading failures L = {L 1 , L 2 , …, L t}, then Equation (20) can be specifically written as:

[0191]

[0192] where l h represents the propagation line, with a total of ; l a represents the vulnerable line, with a total of . represents the set of propagation lines, represents the set of vulnerable lines, represents the set of power system nodes. The probability link matrix is approximately calculated using the frequency of occurrence of the fault chain, that is:

[0193]

[0194] where represents the xth propagation line to the yth vulnerable line the number of occurrences. The larger the value, the xth propagation line to the yth vulnerable line the greater the probability.

[0195] Furthermore, calculate the propagation value and the vulnerability value, so as to iteratively calculate and corresponding to a certain branch or The larger the value, the higher the probability that the branch will be used as a propagation line or a vulnerable line, and the weaker it is in the fault chain of cascading failures.

[0196] To comprehensively evaluate the weakness of branches during cascading failures, an embodiment of the present invention also constructs a comprehensive index of weak links to comprehensively reflect the propagation value and vulnerability value of the branch:

[0197]

[0198] In the formula, α is the weight of, and 1 - α represents the weight of. For a certain branch, the corresponding The larger the value, the more critical the branch is in the fault chain of cascading failures, and the greater the probability of being called a weak link.

[0199] Furthermore, to illustrate the correlation of fault propagation, the present invention proposes a fault propagation correlation index The calculation method is:

[0200]

[0201] In the formula, represents the x,y-th line in the set of vulnerable lines ; represents that can all be affected by the same propagation line; represents the x-th and y-th lines in the vulnerable lines The probability that they can all be affected by the same propagation line, represents the probability that the y-th and x-th lines in the set of vulnerable lines can all be affected by the same propagation line; The larger the value of, the greater the probability that different vulnerable lines are affected by the same propagation line, and the greater the correlation between vulnerable lines.

[0202] Specifically, the link analysis algorithm is used to identify the weak links of the power grid, and the calculation steps are as follows:

[0203] Step 1: Use the OPA model based on power flow optimization to generate the fault chain of cascading failures as the calculation scenario.

[0204] Step 2: Rearrange the data according to Equation (27) to calculate the set of power grid branches the set of propagation links of power grid faults and the set of vulnerable lines of the power grid

[0205] Step 3: Calculate the propagation value and vulnerability value according to Equations (25) and (26), and further iteratively calculate the central value of the propagation line according to Equations (21) and (22) Weight value of vulnerable line Until the convergence condition is met.

[0206] Step 4: Calculate the comprehensive index of weak links and the index of fault propagation correlation according to formulas (29) and (30). Index of correlation with fault propagation And sort each branch to obtain the weak links of cascading faults and the correlation of weak links.

[0207] In summary, the method for identifying weak links in a power grid provided by the embodiments of the present invention first constructs a complete set of quantitative evaluation index systems by selecting evaluation indexes of the power grid operation state, which can comprehensively capture the operation state of the power system, so as to realize the accurate evaluation of the power grid security state. At the same time, an innovative Bayesian network and maximum entropy weight self-adaptive method are adopted in determining the weights. This method effectively integrates subjective and objective factors, making the weight distribution more in line with the actual situation. In the aspects of power grid congestion event simulation and weak link identification, the OPA model based on power flow optimization and the SALSA algorithm are combined, enabling the embodiments of the present invention to more deeply explore potential weak links in the power grid and providing strong technical support for the operation and maintenance of the power grid. Through the method for identifying weak links in a power grid provided by the embodiments of the present invention, power grid congestion events can be effectively predicted and prevented, greatly reducing the risk of large blackout accidents, which is of great significance for ensuring the safe and stable operation of the power system.

[0208] Based on the above method items, the embodiments of the present invention correspondingly provide embodiments of the system item.

[0209] See Figure 5 , Figure 5 which is the structural block diagram of a power grid weak link identification system provided by the embodiments of the present invention. The power grid weak link identification system includes:

[0210] Index selection module 21, used to select each evaluation index of the power grid operation state;

[0211] Index weight calculation module 22, used to determine the initial weights of the respective evaluation indexes by using the analytic hierarchy process, and update the initial weights of the respective evaluation indexes by using the Bayesian network and self-learning method to obtain the weights of the respective evaluation indexes;

[0212] Sample generation module 23, used to generate power grid congestion event samples by using the OPA model based on DC power flow optimization;

[0213] Evaluation module 24, used to evaluate the power grid operation state according to the respective evaluation indexes and corresponding weights by using the power grid congestion event samples to obtain an evaluation result;

[0214] A weak-link identification module 25, which is used to identify the weak links of the power grid by using a link analysis algorithm if the evaluation result shows that there are cascading failures in the power grid.

[0215] In one embodiment, the step of determining the initial weights of the respective evaluation indicators by using the analytic hierarchy process includes:

[0216] Establishing a hierarchical structure model according to the respective evaluation indicators;

[0217] Constructing a judgment matrix according to the hierarchical structure model and the respective evaluation indicators;

[0218] Calculating the initial weights of the respective evaluation indicators according to the judgment matrix;

[0219] Verifying the judgment matrix according to the consistency index. If the judgment matrix does not meet the preset conditions, adjusting the judgment matrix and recalculating the initial weights of the respective evaluation indicators until the judgment matrix meets the preset conditions.

[0220] In one embodiment, the step of updating the initial weights of the respective evaluation indicators by using a Bayesian network and a self-learning method to obtain the weights of the respective evaluation indicators includes:

[0221] Taking the initial weights of the respective evaluation indicators as node attributes to construct a Bayesian network;

[0222] Calculating the weight gradient on each node in the Bayesian network;

[0223] According to the weight gradient on each node, using the back gradient descent method to iteratively update the weights on each node until the maximum entropy principle is satisfied to obtain the weights of the respective evaluation indicators; wherein, in each iteration, the iterated weights are normalized.

[0224] In one embodiment, the sample generation module 23 is specifically configured to:

[0225] Determine a DC power flow optimization objective function, and build an OPA model based on the DC power flow optimization objective function;

[0226] Collect power grid data and set the initial operating state of the OPA model; wherein, the power grid data includes the initial generator output, the load demand magnitude of each node, and the initial power grid topology;

[0227] Update the loads, generators, and partial line capacities in the OPA model to simulate the power grid under different fault and operating conditions, and generate samples of power grid congestion events.

[0228] In one embodiment, the weak link identification module 25 is specifically configured to:

[0229] Use a link analysis algorithm to determine the link relationship between power grid branches, the propagation links of power grid faults, and the vulnerable lines of the power grid; wherein, the power grid branches are used as nodes in the link analysis algorithm, the propagation lines are used as central nodes in the link analysis algorithm, and the vulnerable lines are used as weighted nodes in the link analysis algorithm;

[0230] Use the samples of the power grid congestion events to calculate the power grid branch set, the power grid fault propagation link set, and the power grid vulnerable line set according to the link relationship;

[0231] Initialize the propagation value and vulnerability value of each power grid branch, and perform iterative calculations on the central value of the propagation line and the weighted value of the vulnerable line according to the propagation value and vulnerability value until the central value and the weighted value meet the preset convergence conditions;

[0232] Sort the power grid branches, and identify the weak links of the power grid according to the comprehensive weak link index and the fault propagation correlation index, and obtain the weak links and the correlation of the weak links.

[0233] It should be noted that a power grid weak link identification system provided by an embodiment of the present invention is used to execute all the process steps of a power grid weak link identification method in the above embodiment. The working principles and beneficial effects of the two correspond one by one, so they will not be elaborated here.

[0234] An embodiment of the present invention also provides a terminal device, as Figure 6 shown, which is a structural block diagram of a preferred embodiment of a terminal device provided by the present invention. The terminal device includes a processor 31, a memory 32, and a computer program stored in the memory 32 and configured to be executed by the processor 31. When the processor 31 executes the computer program, it implements the power grid weak link identification method described in any of the above embodiments.

[0235] In addition, an embodiment of the present invention also provides a computer-readable storage medium, which includes a stored computer program. When the computer program runs, it controls the device where the computer-readable storage medium is located to execute the power grid weak link identification method described in any of the above embodiments.

[0236] When the processor 31 executes the computer program, it implements the steps in the above embodiments of the power grid weak link identification method. For example Figure 1All steps of the power grid weak link identification method shown. Alternatively, when the processor 31 executes the computer program, it realizes the functions of each module in the above-mentioned power grid weak link identification system embodiment. For example Figure 5 The functions of each module of the power grid weak link identification system shown.

[0237] Preferably, the computer program can be divided into one or more modules / units, and the one or more modules / units are stored in the memory 32 and executed by the processor 31 to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, and these instruction segments are used to describe the execution process of the computer program in the terminal device.

[0238] The processor 31 can be a central processing unit (CPU), or can also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor, or the processor 31 can also be any conventional processor. The processor 31 is the control center of the terminal device and connects various parts of the terminal device through various interfaces and lines.

[0239] The memory 32 mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function, etc., and the data storage area can store relevant data, etc. In addition, the memory 32 can be a high-speed random access memory, or can also be a non-volatile memory, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc., or the memory 32 can also be other volatile solid-state storage devices.

[0240] It should be noted that the above-mentioned terminal device may include, but is not limited to, a processor and a memory. Those skilled in the art can understand that Figure 6 The structure block diagram shown is only an example of the structure of the above-mentioned terminal device and does not constitute a limitation on the structure of the above-mentioned terminal device. The above-mentioned terminal device may include more or fewer components than those shown, or combine some components, or have different components.

[0241] The above are the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. A method for identifying weak links in a power grid, characterized in that: include: Select various evaluation indicators of the power grid operation status; The initial weights of the evaluation indicators are determined by using a hierarchical analysis method, and the initial weights of the evaluation indicators are updated by using a Bayesian network and a self-learning method to obtain the weights of the evaluation indicators. The OPA model based on DC power flow optimization is used to generate grid congestion event samples; According to the evaluation indicators and corresponding weights, the power grid congestion event samples are used to evaluate the power grid operation status to obtain an evaluation result; If the assessment results show that there are cascading failures in the power grid, the link analysis algorithm is used to identify the weak links in the power grid.

2. The method for identifying weak links in a power grid according to claim 1, characterized in that: The initial weights of the evaluation indicators are updated by using the Bayesian network and the self-learning method to obtain the weights of the evaluation indicators, including: Using the initial weights of the evaluation indicators as node attributes to construct a Bayesian network; Calculating the weight gradient on each node in the Bayesian network; According to the weight gradient on each node, the weight on each node is iteratively updated using the post-gradient descent method until the maximum entropy principle is met, thereby obtaining the weights of each evaluation indicator; wherein, in each iteration, the iterated weights are normalized.

3. The method for identifying weak links in a power grid according to claim 1, characterized in that: The OPA model based on DC power flow optimization is used to generate grid congestion event samples, including: Determine a DC power flow optimization objective function, and build an OPA model based on the DC power flow optimization objective function; Collecting grid data and setting the initial operation state of the OPA model; wherein the grid data includes initial generator output, load demand size of each node and initial grid topology; The loads, generators and some line capacities in the OPA model are updated to simulate the grid under different fault and operating conditions and generate samples of grid congestion events.

4. The method for identifying weak links in a power grid according to claim 3, characterized in that: The DC power flow optimization objective function is: Where G represents the set of generator nodes, are the upper and lower limits of the power of generator i, p i represents the power of generator i, and D represents the load node set, p j represents the load of node j, p dj represents the load demand of node j, and P dj ≤p j ≤0,j∈D;c i represents the unit power generation cost of the i-th generator, w j represents the loss caused by the removal of unit load at node j; L is the line set, F l is the flow of the lth branch, F l max is the transmission capacity of the lth branch, F=(F1,F2,…,F m ) T ; p is the node injection power, p=(p1,p2,…,p n ) T ; A represents the conversion matrix from node injection power to branch power flow.

5. The method for identifying weak links in a power grid according to claim 1, characterized in that: The use of a link analysis algorithm to identify weak links in the power grid includes: A link analysis algorithm is used to determine the link relationship between the power grid branch, the propagation link of the power grid fault and the vulnerable line of the power grid; wherein the power grid branch is used as a node in the link analysis algorithm, the propagation line is used as a central node in the link analysis algorithm, and the vulnerable line is used as a weight node in the link analysis algorithm; Using the samples of the power grid congestion events, and calculating the set of power grid branches, the set of propagation links of power grid faults, and the set of vulnerable lines of the power grid according to the link relationship; Initializing the propagation value and the vulnerability value of each power grid branch, and iteratively calculating the center value of the propagation line and the weight value of the vulnerable line according to the propagation value and the vulnerability value, until the center value and the weight value meet the preset convergence condition; The various power grid branches are sorted, and the weak links of the power grid are identified according to the comprehensive indicators of the weak links and the indicators of the correlation of the fault propagation, so as to obtain the weak links and the correlation of the weak links.

6. The method for identifying weak links in a power grid according to claim 5, characterized in that: The comprehensive index of weak links is calculated by the following formula: in, It represents the comprehensive index of weak links. represents the center value of the propagation line, α represents The weight of Represents the weight value of the vulnerable line, 1-α represents The weight of The fault propagation correlation index is calculated by the following formula: in, represents the fault propagation correlation index, The larger the value of is, the greater the probability that different vulnerable lines are affected by the same propagation line is, and the greater the correlation between vulnerable lines is; a Indicates a fragile line. They represent the xth and yth lines in the vulnerable lines respectively; express can all be affected by the same propagation line; Indicates the xth and yth lines in the vulnerable line The probability that both can be affected by the same propagation line, It represents the probability that both the yth and xth lines in the vulnerable line set can be affected by the same propagation line.

7. The method for identifying weak links in a power grid according to claim 1, characterized in that: The evaluation indicators include an electrical intermediate difference indicator, a power flow distribution indicator, a voltage margin indicator, a load factor indicator, and a node voltage loss indicator; The electrical betweenness difference index is determined by using the Gini coefficient; The power flow distribution index is determined according to the limit transmission capacity of each branch of the power grid, the power flow distribution and the number of branches of the power grid; The voltage margin index is determined according to the voltage amplitude of the power grid load node; The load rate index is determined according to the transmission power of each branch of the power grid and the limit transmission capacity of each branch of the power grid; The node voltage loss index is determined according to the voltage of each node after a power grid failure occurs and the voltage of each node in a normal state of the power grid.

8. The method for identifying weak links in a power grid according to claim 1, characterized in that: The method of using the analytic hierarchy process to determine the initial weights of the evaluation indicators includes: Establishing a hierarchical model according to each of the evaluation indicators; Constructing a judgment matrix according to the hierarchical structure model and the various evaluation indicators; Calculating the initial weights of the various evaluation indicators according to the judgment matrix; The judgment matrix is ​​verified according to the consistency index. If the judgment matrix does not meet the preset conditions, the judgment matrix is ​​adjusted and the initial weights of the evaluation indicators are recalculated until the judgment matrix meets the preset conditions.

9. A power grid weak link identification system, characterized in that: include: An indicator selection module is used to select various evaluation indicators of the power grid operation status; An indicator weight calculation module is used to determine the initial weight of each evaluation indicator by using a hierarchical analysis method, and to update the initial weight of each evaluation indicator by using a Bayesian network and a self-learning method to obtain the weight of each evaluation indicator; A sample generation module, used to generate grid congestion event samples using an OPA model based on DC power flow optimization; An evaluation module, configured to evaluate the operation status of a power grid using the power grid congestion event sample according to the evaluation indicators and corresponding weights, and obtain an evaluation result; The weak link identification module is used to identify the weak links of the power grid by using a link analysis algorithm if the evaluation result shows that there are cascading failures in the power grid.

10. A terminal device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the method for identifying weak links in a power grid as claimed in any one of claims 1 to 8 is implemented.

Images