A method for identifying vulnerable nodes in active distribution networks considering the impact of new energy

Abstract

The present invention discloses a method for identifying vulnerable nodes of an active distribution network taking into account the influence of new energy sources. The method samples a pre-constructed DG random output model and a load random model, obtains random output data of new energy sources and loads, determines a random output matrix based on the random output data, arranges the random output matrix, and obtains an arrangement matrix that minimizes the correlation of each random variable sampling value; inputs the data in the arrangement matrix into a pre-constructed node vulnerability index evaluation system model to obtain a node comprehensive vulnerability index; and uses a three-parameter interval number sorting method based on a Boolean matrix combined with a node comprehensive vulnerability index to identify vulnerable nodes. Advantages: The present invention can accurately and quickly identify vulnerable nodes in a power grid while taking into account the access of new energy sources to the distribution network, and effectively evaluate the vulnerability of the distribution network, which is beneficial for power grid operators to fully and deeply grasp the safety status of the distribution network and eliminate or mitigate the risks caused by vulnerability.

Description

Technical Field

[0001] The invention relates to a method for identifying vulnerable nodes in an active distribution network taking into account the influence of new energy, and belongs to the technical field of identifying vulnerable nodes in an active distribution network. Background Art

[0002] How to identify the risks of key equipment in the power system has become an important research topic to reduce the occurrence of major power outages. Nodes are the starting point and important convergence point of power grid energy transmission. A large number of facts have proved that the failure and shutdown of certain vulnerable nodes play a role in fueling the spread of power grid accidents. Identifying vulnerable nodes in the power grid not only helps to assess the current safety level of the system, but also grasps the changing trend of the safety level, which is of great significance for preventing cascading failures.

[0003] On the other hand, the penetration rate of new energy sources such as wind power and solar energy in the distribution network is increasing. However, due to the unique randomness and volatility of new energy sources, when the output of new energy power generation fluctuates, the distribution network flow will also change accordingly, and the identification of vulnerable nodes in the power grid often needs to rely on the flow results. Therefore, how to accurately and quickly identify vulnerable nodes in the power grid when new energy is connected to the distribution network is particularly urgent and important. Summary of the invention

[0004] The technical problem to be solved by the present invention is to overcome the defects of the prior art and provide a method for identifying vulnerable nodes in an active distribution network taking into account the influence of new energy, so as to accurately and quickly identify vulnerable nodes in the power grid when new energy is connected to the distribution network, thereby ensuring the safe and stable operation of the active distribution network.

[0005] In order to solve the above technical problems, the present invention provides a method for identifying vulnerable nodes in an active distribution network taking into account the impact of new energy.

[0006] Sampling the pre-built DG random output model and load random model to obtain random output data of new energy and load, determining the random output matrix according to the random output data, and arranging the random output matrix to obtain an arrangement matrix that minimizes the correlation of the sampled values ​​of each random variable;

[0007] The data in the arrangement matrix is ​​input into a pre-constructed node vulnerability index evaluation system model to obtain a node comprehensive vulnerability index; the pre-constructed node vulnerability index evaluation system model is a calculation model that comprehensively considers the inherent topology structure of the active distribution network and the uncertainty of DG, and takes into account the probability of node failure and the impact of changes in network topology and power flow on the system after it exits operation;

[0008] The vulnerable nodes are identified by using the three-parameter interval number sorting method based on Boolean matrix and the node comprehensive vulnerability index.

[0009] Furthermore, the pre-built node vulnerability index evaluation system model includes: a structural vulnerability index calculation model, a state vulnerability index calculation model and an index weight calculation model;

[0010] The structural vulnerability index calculation model is used to calculate network cohesion, network efficiency change rate and node electrical betweenness;

[0011] The state vulnerability index calculation model is used to calculate the improved power flow impact entropy and the improved minimum singular value change rate;

[0012] The index weight calculation model is used to calculate the weights of network cohesion, network efficiency change rate, node electrical betweenness, improved power flow impact entropy and improved minimum singular value change rate.

[0013] Furthermore, the network cohesion is obtained by the following formula:

[0014]

[0015] Where a(k) is the average shortest electrical distance in the network after node k is shrunk using the node shrinkage method, and n′ is the number of nodes in the network after shrinkage;

[0016]

[0017] Where, d ij is the shortest electrical distance between any two nodes i and j in the network after contraction, and V represents the set of all nodes in the network;

[0018] The network performance change rate is obtained by the following formula:

[0019]

[0020] In the formula, c(k) is the rate of change of network performance before and after node k fails, C(k) is the network performance after node k fails, and C is the network energy efficiency;

[0021]

[0022] Where G and D are the generator and load node sets respectively, N G and N D are the number of generators and load nodes respectively, min(P Gi′ ,P Dj′ ) is the smaller value of active power in the generator load node pair (i′, j′), P Gi′ is the active power of the generator node i′, P Di′ is the active power of load node j′;

[0023] The node electrical betweenness is obtained by the following formula:

[0024]

[0025] Where e(k) is the electrical betweenness of node k, is the flow carrying capacity on branch l when unit power is injected into the generator load node pair (i′, j′), E(k) is the flow change of node k when unit power is injected into the generator load node pair (i′, j′), Ω k is the set of lines directly connected to node k.

[0026] Furthermore, the improved tidal current impact entropy is obtained by the following formula:

[0027] l(k)=f(k)g(k)

[0028] In the formula, l(k) is the improved power flow impact entropy, f(k) is the node failure risk factor, g(k) is the power flow impact entropy of node k,

[0029] f(k)=e αF(k)

[0030] In the formula, α is the risk weighting coefficient, F(k) is the probability of voltage exceeding the limit at node k,

[0031]

[0032] In the formula, is the number of times the voltage at node k exceeds the upper limit, N(k) is the number of times the voltage at node k exceeds the lower limit, and N is the total number of sampling times

[0033]

[0034] In the formula, G j (k) is the power flow impact rate on line j caused by node k being out of operation, M is the total number of branches in the system,

[0035]

[0036] Where ΔP j (k) is the flow impact on line j caused by node k being out of operation due to a fault;

[0037]

[0038] In the formula, P′ j is the current power flow of line j after node k fails, is the initial power flow of line j;

[0039] The improved minimum singular value change rate is obtained by the following formula:

[0040] o(k)=f(k)h(k)

[0041] In the formula, o(k) is the improved minimum singular value change rate, h(k) is the minimum singular value change rate;

[0042]

[0043] In the formula, The minimum singular value of the Jacobian matrix of the system power flow before node k is removed, δ i,min (k) is the minimum singular value of the Jacobian matrix of the system power flow after node k is removed.

[0044] Furthermore, the calculation process of the indicator weight calculation model is:

[0045] The network cohesion, network efficiency change rate, node electrical betweenness and improved power flow impact entropy of each node are input into the stacked autoencoder neural network to determine the weight of each indicator.

[0046] Furthermore, the process of identifying vulnerable nodes by using the three-parameter interval number sorting method based on the Boolean matrix combined with the comprehensive vulnerability index of each node includes:

[0047] According to the value of the comprehensive vulnerability index of each node, the specific distribution of the comprehensive vulnerability index of each node is obtained, and the upper limit, expected value and lower limit of the comprehensive vulnerability index of each node are calculated according to the equal probability criterion. The upper limit, lower limit and expected value together constitute the value interval of the comprehensive vulnerability index of each node, and the comprehensive vulnerability index expressed in the form of interval is obtained;

[0048] A three-parameter interval number sorting method based on Boolean matrix is ​​used to sort the comprehensive vulnerability index expressed in interval form and identify vulnerable nodes.

[0049] Furthermore, the process of using the three-parameter interval number sorting method based on the Boolean matrix to sort the comprehensive vulnerability index represented in the form of intervals and identifying the vulnerable nodes includes:

[0050] Assume that the interval number of the comprehensive vulnerability index of node i is The number of intervals of the comprehensive vulnerability index of node j is the lower limit of the comprehensive vulnerability index of node i, is the expected value, is the upper limit value; is the lower limit of the comprehensive vulnerability index of node j, is the expected value, is the upper limit value;

[0051] make Then ai Greater than a j Probability Let c ij =P(a i >a j );

[0052] Construct a Boolean matrix E = (e ij ) m×m , E is the interval number a1,a2,…,a m The sorting matrix of

[0053]

[0054] make Get the sorting vector λ=(λ1,λ2,…λ n ), λ i is the ranking value of the vulnerability level of node i;

[0055] According to λ i The number of intervals is sorted by the size, the vulnerability level of each node is determined, and the vulnerable nodes are identified.

[0056] The beneficial effects achieved by the present invention are:

[0057] The present invention can accurately and quickly identify vulnerable nodes in the power grid while taking into account the access of new energy to the distribution network, and effectively evaluate the vulnerability of the distribution network, which is beneficial for power grid operators to fully and deeply understand the security status of the distribution network and eliminate or mitigate the risks caused by vulnerability. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0059] Figure 2 is a general flow chart of a method for identifying vulnerable nodes in an active distribution network taking into account the impact of new energy sources provided in an embodiment;

[0060] Figure 3 It is a comprehensive vulnerability assessment index system for active distribution network nodes provided in the embodiment;

[0061] Figure 4 It is a wiring diagram of an IEEE39 node system including photovoltaic access provided in an embodiment;

[0062] Figure 5 are the upper limit, expected value and lower limit of the comprehensive vulnerability index value of the distribution network calculated in the embodiment. DETAILED DESCRIPTION

[0063] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0064] like Figure 1-3 As shown, a method for identifying vulnerable nodes in an active distribution network taking into account the impact of new energy sources comprises:

[0065] Step 1: Construct DG (distributed generation) random output model and load random model.

[0066] Step 2: Considering the inherent topology of the active distribution network and the uncertainty of DG, and taking into account the probability of node failure and the impact of changes in network topology and power flow on the system after its exit from operation, a node vulnerability index evaluation system is constructed.

[0067] Step 3: Use the stacked autoencoder neural network to obtain the indicator weights and establish the node comprehensive vulnerability index.

[0068] Step 4: The Latin hypercube sampling method is used to calculate the comprehensive vulnerability index of nodes, and the three-parameter interval number sorting method based on the Boolean matrix is ​​used to identify vulnerable nodes.

[0069] The step 1 is achieved by:

[0070] 1.1 Photovoltaic random output model

[0071] It is generally believed that light intensity belongs to Beta distribution, and its probability density function is as follows:

[0072]

[0073] In the formula, E represents the light intensity, E max is the maximum light intensity within a certain period of time, k′ and c′ are the shape parameters of the Beta distribution.

[0074] Photovoltaic output P V The relationship with light intensity E is as follows:

[0075] P V =EAη (2)

[0076] Where A is the area of ​​the photovoltaic array and η is the photoelectric conversion efficiency.

[0077] 1.2 Load random model

[0078] For the load random model, analysis of a large amount of historical actual data shows that the normal distribution model can approximately simulate its random volatility.

[0079] The probability density functions of the active and reactive power of the load are:

[0080]

[0081]

[0082] In the formula, μ P , μ Q are the mean values ​​of load active and reactive power, σ P , σ Q are their corresponding variances respectively.

[0083] The step 2 is achieved by:

[0084] 2.1 Construction of node structure vulnerability index

[0085] System structural vulnerability refers to the ability to maintain the integrity of the system after the change of its own composition structure under the influence of the randomness and volatility of new energy. Therefore, the structural vulnerability index mainly considers the topological structure and electrical characteristics of the power network, and selects important nodes in key positions in the power grid based on the complex network analysis method. The more important and influential the nodes are to the structure of the distribution network, the higher the vulnerability.

[0086] 2.1.1 Network Cohesion

[0087] The node network cohesion index is based on measuring the degree of damage to the overall connectivity of the network after deleting the node. The importance of the node is judged by determining the degree of damage to the system topology after the node is disconnected.

[0088]

[0089] Where a(k) is the average shortest electrical distance in the network after node k is shrunk using the node shrinkage method, and d ij is the shortest electrical distance between any two nodes i and j in the network after contraction, that is, the sum of the impedance values ​​of the transmission lines on the power transmission path between nodes i and j is the smallest, and n′ is the number of nodes in the network after contraction. The network cohesion can be obtained by the following formula:

[0090]

[0091] Where: b(k) is the network cohesion of node k.

[0092] The network cohesion describes the degree to which the node is at the center of the network and its ability to maintain network connectivity. The greater the network cohesion of a node in the distribution network, the more important the node is.

[0093] 2.1.2 Network performance change rate

[0094] The network performance of the distribution network must reflect the topological structure of the network and take into account the electrical characteristics of the grid, so it is defined as:

[0095]

[0096] Where: C represents the network energy efficiency, G and D are the generator and load node sets respectively. G and N D are the number of generators and load nodes respectively, min(P Gi′ ,P Dj′ ) is the smaller value of active power in the generator load node pair (i′, j′), which represents the maximum power that can be transmitted between the node pair (i′, j′). Gi′ is the active power of the generator node i′, P Di′ is the active power of load node j′.

[0097] The connectivity of the power network is affected by each node. If a key node is lost, its connectivity and network performance will change. Therefore, the change rate of network performance before and after the removal of node k can be defined to measure the importance of the node, that is:

[0098]

[0099] Where: c(k) is the rate of change of network performance before and after node k fails, and C(k) is the network performance after node k fails.

[0100] The greater the change rate of the node's network efficiency, the greater the impact of the node's failure on the power transmission of the power grid, and the higher the importance of the node in the power grid.

[0101] 2.1.3 Node electrical betweenness

[0102] The node electrical betweenness organically combines the grid topology with the electrical parameters, making up for the defect of the traditional betweenness that ignores the electrical characteristics of the system, and is more in line with the operation status of the grid. It is defined as:

[0103]

[0104] Where: e(k) is the electrical betweenness of node k, is the power transfer distribution coefficient, that is, the flow carrying capacity on branch l when unit power is injected into the generator load node pair (i′, j′). E(k) is the flow change of node k when unit power is injected into the generator load node pair (i′, j′), Ω k is the set of lines directly connected to node k. The larger the electrical betweenness of a node is, the more important the node is.

[0105] 2.2 Construction of node status vulnerability index

[0106] The node status vulnerability index starts from the operating status of the distribution network, examines the degree of deviation of electrical quantities after a fault occurs, and characterizes the ability of the power grid to withstand interference or faults.

[0107] 2.2.1 Node failure risk factors

[0108] The voltage over-limit risk probability of each node can be calculated by the probabilistic power flow calculation method based on Latin hypercube sampling:

[0109]

[0110] Where: F(k) is the probability of voltage exceeding the limit at node k, is the number of times the node k voltage exceeds the upper limit, N (k) is the number of times the voltage at node k exceeds the lower limit, and N is the total number of sampling times.

[0111] The greater the probability of a node exceeding the limit calculated by the above formula, the more likely the node is to fail when the DG and load fluctuate. Therefore, the node failure risk factor is defined as:

[0112] f(k)=e αF(k) (11)

[0113] In the formula: α is the risk weighted coefficient, and α=2.56 is taken.

[0114] 2.2.2 Node power flow impact entropy

[0115] Assume that the power flow impact on line j caused by node k being out of operation due to a fault is:

[0116]

[0117] Where: P′ j is the current power flow of line j after node k fails, is the initial power flow of line j.

[0118] At this time, the power flow impact rate caused by node k's exit from operation to line j is:

[0119]

[0120] Where: M is the total number of branches in the system.

[0121] Using entropy theory, we can get the power flow impact entropy of node k as:

[0122]

[0123] After the node is disconnected, the smaller the power flow impact entropy is, the more likely it is that the power flow of the system is concentrated on certain branches, which may lead to overloaded branches and even cascading failures, seriously affecting the safety level of the system.

[0124] 2.2.3 Node minimum singular value change rate

[0125] For a power system with p independent nodes and q PV nodes, the polar coordinate form of the power flow equation is:

[0126]

[0127] Where: P i and Q i The active and reactive power injected into node i, U i and U j is the voltage amplitude of node i, G ij and B ij is the element of the admittance matrix, θ ij is the phase angle difference between nodes, i = 1, 2, …, p and j = 1, 2, …, p.

[0128] The Taylor series expansion of equation (17) is used to obtain the Jacobian matrix J, and the singular value decomposition is performed on it to obtain:

[0129]

[0130] Where: J∈P c×c , V1 and U1 are both c×c orthogonal matrices, Λ is the singular value δ i (i=1,2,…,c), v i and u i are δ in V1 and U1 respectively i The corresponding column vector.

[0131] The minimum singular value δ of the power flow Jacobian matrix J i,min It can characterize the relative degree of system voltage stability. The smaller the minimum singular value, the more unstable the system voltage; conversely, the more stable the system voltage is. Therefore, the change rate of the minimum singular value before and after the node k is removed is defined as:

[0132]

[0133] Where: The minimum singular value of the Jacobian matrix of the system power flow before node k is removed, δ i,min (k) is the minimum singular value of the Jacobian matrix of the system power flow after node k is removed.

[0134] The smaller the rate of change of the minimum singular value after a node exits operation, the more important the node is.

[0135] 2.2.4 Considering the possibility and consequences of node disconnection, the improved power flow impact entropy index is defined as:

[0136] l(k)=f(k)g(k) (18)

[0137] The improved minimum singular value change rate indicator is defined as:

[0138] o(k)=f(k)h(k) (19)

[0139] In the present invention, step (3) is achieved by the following method:

[0140] 3.1 Determine indicator weights

[0141] The indicator value of each node is input into the stacked autoencoder neural network to determine the weight of each indicator.

[0142] 3.2 Establishing comprehensive node vulnerability index

[0143] The network cohesion index, network efficiency change rate index, node electrical betweenness index, and the improved power flow impact entropy index and the improved minimum singular value change rate index proposed for the state of the power grid after the node exits operation are combined in step (2), and the comprehensive vulnerability index of node k is defined as:

[0144] θ(k)=ω b b(k)+ω c c(k)+ω e e(k)+ω l l(k)+ω o o(k) (20)

[0145] Where: ω represents the weight of each indicator.

[0146] In the present invention, step 4 is achieved by the following method:

[0147] The node comprehensive vulnerability index established in step 3 can only obtain a deterministic index calculation value when calculated under a certain method, which cannot reflect the uncertainty of random fluctuations of renewable energy output and load under normal operation of the power grid. Therefore, in order to take into account the impact of uncertainty in renewable energy output and load, the probabilistic flow method is used to calculate the node comprehensive vulnerability index. As a simulation method for solving probabilistic flow, the Latin hypercube sampling method has greatly improved its computational efficiency compared with the Monte Carlo sampling method, and can provide rich information of output random variables, so that the calculated comprehensive vulnerability index can better meet the requirements statistically.

[0148] 4.1 Latin Supersampling

[0149] The main idea of ​​Latin Hypercube Sampling is based on the inverse function conversion method. This method is divided into the following two steps: ① Sampling: Sampling each input random variable to ensure that the random distribution area can be completely covered by the sampling points. ② Arrangement: Changing the arrangement order of the sampled values ​​of each random variable to minimize the correlation of the sampled values ​​of each random variable.

[0150] 4.1.1 Sampling

[0151] Assume that there are K random input variables in the problem to be solved, X k (k=1,2,…,K) is any random input variable, and its cumulative probability distribution function can be expressed as Y k =F k (X k ), let X kn is the nth sampling value of the kth random variable, then it can be expressed as:

[0152]

[0153] Where n = 1, 2, ..., N. N is the total number of sampling times.

[0154] After all random input variables are sampled, the sampled values ​​of each random variable are randomly arranged into a row of the matrix. Then all the sampled values ​​form a K×N order sampling matrix, which can be expressed as:

[0155]

[0156] Since the elements in the above sampling matrix are randomly arranged, the correlation between the sampling values ​​of each random variable is random and uncontrollable. Therefore, it is also necessary to reduce the correlation between each random variable in the sampling matrix through arrangement.

[0157] 4.1.2 Sorting method based on Cholesky decomposition

[0158] 4.1.2.1 Initializing the permutation matrix L KN , each row of which consists of a random permutation of integers 1, 2, ..., N. The permutation matrix L KN It is a K×N matrix, and the element value of each row represents the sampling matrix X KN The arrangement position of the corresponding row elements.

[0159] 4.1.2.2 Calculation of the permutation matrix L KN The correlation coefficient matrix between each row ρ L ,ρ L is a K×K matrix, which can be expressed as:

[0160]

[0161] Where ρ ij is the permutation matrix L KN Correlation coefficient between row i and row j. ρ ij It can be calculated by the following formula:

[0162]

[0163] Where: L ik is the permutation matrix L KN The value of the element in row i and column k, is the permutation matrix L KN The mean of the elements in the i-th row.

[0164] 4.1.2.3 It can be proved that the correlation coefficient matrix ρ L is a positive symmetric matrix, so using the Cholesky decomposition method to decompose it can obtain a real non-singular lower triangular matrix D, and it satisfies DD T =ρ L .

[0165] 4.1.2.4 Since D is non-singular, its inverse matrix exists. Combined with the original permutation matrix L KN A new permutation matrix G can be constructed KN .

[0166] G KN =D -1 L KN (25)

[0167] 4.1.2.5 Using the permutation matrix G KN The elements of the sampling matrix X are arranged in descending order to indicate KN The arrangement position of the elements in the sampling matrix X KN Rearrange the elements in .

[0168] 4.2 Based on the DG random output model and load random model established in step (1), random sampling is performed using the Latin hypercube sampling method. The sampling results are the active power of random loads and renewable energy. The sampling results are substituted into the power flow equation of formula (17) for power flow calculation to obtain the node voltage and branch power flow. Finally, the comprehensive vulnerability index value of each node can be calculated based on the sampling results, node voltage and branch power flow.

[0169] 4.3 The Latin hypercube sampling method in step 4.2 can obtain the comprehensive vulnerability index value of the nodes within the set sampling scale, and then obtain the specific distribution of the comprehensive vulnerability index value of each node. According to the equal probability criterion, the upper limit, expected value and lower limit of the comprehensive vulnerability index value of each node can be calculated. The upper and lower limits and the expected value together constitute the value range of the comprehensive vulnerability index value of each node, thereby reflecting the vulnerability level of each node in the form of interval instead of the determined value.

[0170] 4.4 In order to characterize the vulnerability level of each node under uncertainty, it is necessary to use a three-parameter interval number sorting method based on a Boolean matrix to sort the required interval-form comprehensive vulnerability index.

[0171] 4.4.1 Let any two intervals be Be two interval numbers, let

[0172] but

[0173] 4.4.2 Constructing the Boolean Matrix E = (eij)m × m, call E the interval number a1,a2,…,a m The sorting matrix of , where:

[0174] 4.4.3 Order Get the sorting vector λ=(λ1,λ2,…λ n ).

[0175] 4.4.4 According to λ i The number of intervals is sorted by the size of , and the vulnerability level of each node is determined, so as to identify the vulnerable nodes.

[0176] like Figure 4 As shown, the present invention uses a standard IEEE39 node system for example verification, serial numbers 1-39 represent the 1st to 39th nodes, and a photovoltaic with a rated capacity of 100MW is connected at node 21. The light intensity obeys Beta distribution, and the shape parameters k′=0.2274, c′=1.2995. The load obeys normal distribution, the mean is the rated power of the system load, and the variance is 20% of the mean.

[0177] 2000 groups of photovoltaic power generation and active and reactive power of load were sampled by Latin hypercube sampling, and the sampling results were respectively brought into the distribution network data for power flow calculation. Five indicators, including node degree, network efficiency change rate, network cohesion, improved power flow impact entropy and improved minimum singular value change rate, were calculated, and the data were brought into the automatic encoder neural network to calculate the weight value of each indicator. Furthermore, the comprehensive vulnerability index value of the distribution network was calculated.

[0178] At this time, the upper limit, expected value and lower limit of the calculated comprehensive vulnerability index of the distribution network are as follows: Figure 5 shown.

[0179] On this basis, the upper and lower limits and the expected value together constitute the value range of the comprehensive vulnerability index value of each node. On this basis, the three-parameter interval number sorting method based on the Boolean matrix is ​​used to identify vulnerable nodes. The identification results are shown in the following table:

[0180] Comparison of node vulnerability assessment results

[0181]

[0182]

[0183] As can be seen from the table above, the results of the three evaluation methods are highly consistent, that is, six of the top ten nodes obtained by the method proposed in this patent overlap with the first three methods, such as nodes 4, 16, and 17. These nodes are located at the core of the power grid and are important transmission hubs. Once a fault occurs and they are out of operation, the system will experience a large flow transfer, which will have a great impact on the line, easily causing line overload, and even causing the power grid to enter a self-organized critical state, triggering a disaster.

[0184] In addition, the patented method takes into account the impact of new energy access on vulnerability, reflecting the principle that the uncertainty after new energy is connected to the grid has a greater impact on the vicinity. For example, after node 21 is connected to new energy, its nearby nodes 16 and 22 and subsequent affected nodes will become very vulnerable.

[0185] The method of the present invention constructs a scientific and comprehensive vulnerability evaluation index based on the structure and status of the distribution network. It can accurately and quickly identify vulnerable nodes in the power grid and effectively evaluate the vulnerability of the distribution network while taking into account the access of new energy to the distribution network, thereby reducing the probability of accident safety risks.

[0186] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for identifying vulnerable nodes in an active distribution network taking into account the impact of new energy, characterized in that: Sampling the pre-built DG random output model and load random model to obtain random output data of new energy and load, determining the random output matrix according to the random output data, and arranging the random output matrix to obtain an arrangement matrix that minimizes the correlation of the sampled values ​​of each random variable; The data in the arrangement matrix is ​​input into a pre-constructed node vulnerability index evaluation system model to obtain a node comprehensive vulnerability index; the pre-constructed node vulnerability index evaluation system model is a calculation model that comprehensively considers the inherent topology structure of the active distribution network and the uncertainty of DG, and takes into account the probability of node failure and the impact of changes in network topology and power flow on the system after it exits operation; The vulnerable nodes are identified by using the three-parameter interval number sorting method based on Boolean matrix and the node comprehensive vulnerability index.

2. The method for identifying vulnerable nodes in an active distribution network taking into account the impact of new energy sources according to claim 1 is characterized in that: The pre-built node vulnerability index evaluation system model includes: a structural vulnerability index calculation model, a state vulnerability index calculation model and an index weight calculation model; The structural vulnerability index calculation model is used to calculate network cohesion, network efficiency change rate and node electrical betweenness; The state vulnerability index calculation model is used to calculate the improved power flow impact entropy and the improved minimum singular value change rate; The index weight calculation model is used to calculate the weights of network cohesion, network efficiency change rate, node electrical betweenness, improved power flow impact entropy and improved minimum singular value change rate.

3. The method for identifying vulnerable nodes in an active distribution network taking into account the impact of new energy sources according to claim 2 is characterized in that: The network cohesion is obtained by the following formula: Where a(k) is the average shortest electrical distance in the network after node k is shrunk using the node shrinkage method, and n′ is the number of nodes in the network after shrinkage; Where, d ij is the shortest electrical distance between any two nodes i and j in the network after contraction, and V represents the set of all nodes in the network; The network performance change rate is obtained by the following formula: In the formula, c(k) is the rate of change of network performance before and after node k fails, C(k) is the network performance after node k fails, and C is the network energy efficiency; Where G and D are the generator and load node sets respectively, N G and N D are the number of generators and load nodes respectively, min(P Gi′ ,P Dj′ ) is the smaller value of active power in the generator load node pair (i′, j′), P Gi′ is the active power of the generator node i′, P Di′ is the active power of load node j′; The node electrical betweenness is obtained by the following formula: Where e(k) is the electrical betweenness of node k, is the flow carrying capacity on branch l when unit power is injected into the generator load node pair (i′, j′), E(k) is the flow change of node k when unit power is injected into the generator load node pair (i′, j′), Ω k is the set of lines directly connected to node k.

4. The method for identifying vulnerable nodes in an active distribution network taking into account the impact of new energy sources according to claim 2, characterized in that: The improved tidal current impulse entropy is obtained by the following formula: l(k)=f(k)g(k) In the formula, l(k) is the improved power flow impact entropy, f(k) is the node failure risk factor, g(k) is the power flow impact entropy of node k, f(k)=e αF(k) In the formula, α is the risk weighting coefficient, F(k) is the probability of voltage exceeding the limit at node k, In the formula, is the number of times the node k voltage exceeds the upper limit, N (k) is the number of times the voltage at node k exceeds the lower limit, and N is the total number of sampling times In the formula, G j (k) is the power flow impact rate on line j caused by node k being out of operation, M is the total number of branches in the system, Where ΔP j (k) is the flow impact on line j caused by node k being out of operation due to a fault; In the formula, P′ j is the current power flow of line j after node k fails, is the initial power flow of line j; The improved minimum singular value change rate is obtained by the following formula: o(k)=f(k)h(k) In the formula, o(k) is the improved minimum singular value change rate, h(k) is the minimum singular value change rate; In the formula, The minimum singular value of the Jacobian matrix of the system power flow before node k is removed, δ i,min (k) is the minimum singular value of the Jacobian matrix of the system power flow after node k is removed.

5. The method for identifying vulnerable nodes in an active distribution network taking into account the impact of new energy sources according to claim 2, characterized in that: The calculation process of the indicator weight calculation model is as follows: The network cohesion, network efficiency change rate, node electrical betweenness and improved power flow impact entropy of each node are input into the stacked autoencoder neural network to determine the weight of each indicator.

6. The method for identifying vulnerable nodes in an active distribution network taking into account the impact of new energy sources according to claim 1, characterized in that: The process of identifying vulnerable nodes by using the three-parameter interval number sorting method based on the Boolean matrix combined with the comprehensive vulnerability index of each node includes: According to the value of the comprehensive vulnerability index of each node, the specific distribution of the comprehensive vulnerability index of each node is obtained, and the upper limit, expected value and lower limit of the comprehensive vulnerability index of each node are calculated according to the equal probability criterion. The upper limit, lower limit and expected value together constitute the value interval of the comprehensive vulnerability index of each node, and the comprehensive vulnerability index expressed in the form of interval is obtained; A three-parameter interval number sorting method based on Boolean matrix is ​​used to sort the comprehensive vulnerability index expressed in interval form and identify vulnerable nodes.

7. The method for identifying vulnerable nodes in an active distribution network taking into account the impact of new energy sources according to claim 6, characterized in that: The process of using the three-parameter interval number sorting method based on the Boolean matrix to sort the comprehensive vulnerability index represented in the form of intervals and identifying the vulnerable nodes includes: Assume that the interval number of the comprehensive vulnerability index of node i is The number of intervals of the comprehensive vulnerability index of node j is the lower limit of the comprehensive vulnerability index of node i, is the expected value, is the upper limit value; is the lower limit of the comprehensive vulnerability index of node j, is the expected value, is the upper limit value; make Then a i Greater than a j Probability Let c ij =P(a i >a j ); Construct a Boolean matrix E = (e ij ) m×m , E is the interval number a1,a2,…,a m The sorting matrix of make Get the sorting vector λ=(λ1,λ2,…λ n ), λ i is the ranking value of the vulnerability level of node i; According to λ i The number of intervals is sorted by the size, the vulnerability level of each node is determined, and the vulnerable nodes are identified.

Images