Active power distribution network source and load distribution method based on node elasticity and inertia distribution

Abstract

The invention belongs to the technical field of active power distribution network stable operation control, and discloses an active power distribution network source and load distribution method based on node elasticity and inertia distribution, and the method comprises the steps: obtaining an active power distribution network topological structure and power quality data; extracting node network topology measurement characteristics and node electrical characteristics of the active power distribution network; establishing an active power distribution network node importance evaluation model; calculating node inertia and elastic distribution of the active power distribution network, and establishing a source load distribution magnitude calculation model of the active power distribution network; constructing a source load distribution model to obtain a source load distribution matrix considering the overall strength and sparsity of the power grid; and carrying out source load distribution effect verification, carrying out load flow calculation on the active power distribution network after redistribution, and verifying whether the node power quality meets the standard requirement or not. According to the method, feature analysis is carried out on historical operation data of the active power distribution network, important nodes in the power grid are identified, and a source load distribution strategy is provided according to an importance identification result and a power grid disturbance condition.

Description

Technical Field

[0001] The present invention relates to the technical field of stable operation control of active distribution networks, and in particular to a method for allocating sources and loads of active distribution networks based on node elasticity and inertia distribution. Background Art

[0002] With the development of distributed energy technology, active distribution networks have become an indispensable infrastructure for economic development and daily life, and the public is highly dependent on the stable supply of electricity. However, the complexity and variability of the power system exposes it to various potential risks, especially in active distribution networks, where the output fluctuations of distributed energy make the system operation more unstable. In recent years, with the widespread application of renewable energy and distributed generation, the operation of distribution networks has become increasingly complex. When the network is disturbed, the traditional source-load allocation method has been difficult to adapt to the needs of modern power systems.

[0003] When disturbances occur in active distribution networks, unreasonable source-load distribution may lead to system instability and failure. How to achieve efficient and reliable source-load distribution in a complex power environment has become one of the key technologies to improve the stability and reliability of power systems. Existing technologies generally only consider the power value of nodes during normal operation, lack analysis of the overall network, and fail to fully consider the mutual influence and dynamic changes between nodes, making it difficult to effectively respond to frequent load changes and sudden operating conditions. Therefore, it is of great significance to improve the redistribution model and method and enhance the ability of active distribution networks to cope with disturbances. Summary of the invention

[0004] In view of the shortcomings of the above-mentioned prior art, the present invention proposes a source and load distribution method for an active distribution network based on node elasticity and inertia distribution, which aims to identify important nodes in the power grid by performing feature analysis on the historical operation data of the active distribution network, and propose a source and load distribution strategy based on the importance identification results and the power grid disturbance conditions, so as to provide technical support for reducing the load loss of the power grid caused by the disturbance.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for allocating sources and loads in an active power distribution network based on node elasticity and inertia distribution, comprising:

[0007] Step 1, obtaining the active distribution network topology and power quality data, and processing the original data into a standard data format;

[0008] Step 2: Based on the complex network structural hole theory, the network topology metric characteristics and node electrical characteristics of the active distribution network nodes are extracted, and the principal component features are extracted using the EFO function to form a reconstructed feature space;

[0009] Step 3, establish an active distribution network node importance evaluation model based on the improved I-PageRank method to obtain a ranked set of important nodes;

[0010] Step 4, based on the important node identification results of step 3, calculate the inertia and elasticity distribution of the active distribution network nodes, establish a calculation model for the source load distribution magnitude of the active distribution network, and obtain the source load distribution magnitude matrix;

[0011] Step 5, based on the source load distribution magnitude matrix obtained in step 4, a source load distribution model is constructed based on the SSPR method to obtain a source load distribution matrix that considers the overall strength and sparsity of the power grid;

[0012] Step 6, verify the source-load allocation effect, calculate the power flow of the redistributed active distribution network, and verify whether the node power quality meets the standard requirements. If it meets the standard requirements, issue a command; if not, return to step 4 to adjust the source load to be allocated.

[0013] Furthermore, the step 1 comprises:

[0014] The power quality related data of each node of the active distribution network is measured by the power monitoring system, and the obtained data is cleaned, missing value processed, denoised, standardized and normalized using the digital filtering method; at the same time, the generators, loads and new energy connected to the power grid are processed as network nodes. The processed power grid data includes the label matrix L storing the type information of the active distribution network nodes, the directed adjacency matrix of the power grid Matrix A is the directed adjacency matrix of the power grid, A.arcs[i][j] refers to the element in row i and column j in the directed adjacency matrix A of the power grid, the node current matrix I, the frequency matrix f, the voltage matrix V, the node active power matrix P, and the reactive power matrix Q.

[0015] Furthermore, the step 2 comprises:

[0016] Step 2.1: Extract the feature set F of the network topology hole of the active distribution network node topo , which contains the following features: active distribution network node degree centrality matrix K, node betweenness centrality matrix C RB , the node proximity centrality matrix C AC , the expression is as follows:

[0017]

[0018]

[0019]

[0020] d(h,j,t)=log(α h,j (t),αj,h (t))

[0021]

[0022] in, is the out-degree of the active distribution network node h at time t, is the in-degree of node h at time t, n is the number of nodes in the active distribution network, a hj (t) is the value of the directed adjacency matrix A.arcs[h][j] of the power grid at time t, a jh (t) is the value of the directed adjacency matrix A.arcs[j][h] of the power grid at time t, σ sw (h,t) is the number of shortest paths from s to w passing through node h at time t, σ sw (t) is the number of shortest paths from s to w at time t, d(h,j,t) is the electrical distance from node h to node j at time t, α h,j (t) is the voltage sensitivity from node h to node j, α j,h (t) is the voltage sensitivity from node j to node h, U h (t) is the voltage of node h at time t, U j (t) is the voltage of node j at time t, P j (t) is the active power of node j at time t, Q j (t) is the reactive power of node j at time t;

[0023] Step 2.2: Extract the electrical feature set F of the active distribution network nodes ele , which includes the following features: Considering the real-time power flow tracking matrix P of distributed generation GL , network efficiency matrix E, node constraint coefficient matrix Co, the expressions are as follows:

[0024]

[0025]

[0026]

[0027] Co(h,t)=∑ k∈Γ(h) (p hk (t)+∑ j∈(Γ(h)IΓ(k)) (p hj (t)p jk (t))) 2

[0028]

[0029] Among them, P G(h, t) is the generator injection power at power node h at time t, P(h, t) is the injection power at node h at time t, [A d ] ij is the downstream distribution value from node i to node j, P L (k,t) is the power of load node k at time t, N G is the power node set, N L is the load node set, β i is the set of all nodes connected to node i and drawing power from node i, d(h,i,t) is the electrical distance between node h and node i at time t, P ij is the active power transmitted from node i to node j, P j is the total injected active power of node j, N is the set of active distribution network nodes, Γ(h) is the neighbor node of node h, Γ(k) is the neighbor node of node k, p hk (t) is the ratio of the power input from node h to node k to the total outflow power from node h, p hj (t) is the ratio of the power input from node h to node j to the total outflow power from node h, p jk (t) is the ratio of the power input from node j to node k to the total outflow power from node j, p ij (t) is the ratio of the power input from node i to node j to the total outflow power from node i, P ij (t) is the power injected from node i to node j at time t, P ig (t) is the power injected from node i to node g at time t;

[0030] Step 2.3, using the empirical orthogonal function EOF, the node network topology metric features and node electrical features with high correlation coefficients are reconstructed into feature space: the active distribution network node features are normalized and constructed into matrices X m×n , solve Get the principal components of the node feature vector, where It is the transpose of the eigenvector matrix; the variance contribution rates of the first n eigenvectors are accumulated, and the accumulation ends when the accumulated value exceeds the preset threshold γ. At the same time, the features involved in the accumulation are extracted and reconstructed into the feature matrix R o×n , where o is the number of reconstructed features.

[0031] Furthermore, the step 3 comprises:

[0032] Step 3.1: Use the improved k-shell method to decompose the active distribution network. Iteratively delete nodes from small to large according to the node in-degree to obtain the k-shell layer number K of the node. S (h, t), and record the number of iterations n(h, t) when the node is deleted; combined with the topological connection, calculate the number of active distribution network node cores KCore (h, t), the expression is as follows:

[0033] K Core (h,t)=K S (h,t)×K(h,t)+n(h,t)+μ i D(h,t)

[0034] Among them, K(h,t) is the degree of node h at time t, D(h,t) is the number of secondary neighbor nodes of the node, μ i is the influence coefficient;

[0035] Step 3.2, according to the reconstructed feature space R o×n Calculate the weight w between nodes ij , we get the active distribution network weight matrix W, which is expressed as follows:

[0036] w ij =(ω1*r 1,i +ω2*r 2,i +L+ω o *r o,i )*(ω1*r 1,j +ω2*r 2,j +L+ω o *r o,j )

[0037] where ω1,ω2L,ω o is the benign adjustment factor of the node corresponding feature, r 1,i ,r 2,i ,L,r o,i is the eigenvalue of the corresponding node i in the reconstruction matrix, r 1,j ,r 2,j ,L,r o,j is the eigenvalue of the corresponding node j in the reconstruction matrix;

[0038] Step 3.3, calculate the PR value of the node, the calculation formula is as follows:

[0039]

[0040] Where i is the number of iterations of PR value calculation, d is the damping factor, PR(h) i is the PR value of node h in the i-th iteration, and the decay coefficient is the number of cores K of node j Core (j), p jh (t) is the ratio of the power input from node j to node h to the total outflow power from node j, w jh is the weight between node j and node h;

[0041] Step 3.4, according to the label matrix L of the node type information, the active distribution network node set is divided into a power node set, a load node set, and an intermediate node set. According to the PR value of the node, the power node set, load node set, and intermediate node set of the active distribution network are sorted in importance to determine the key nodes in the power grid.

[0042] Furthermore, the step 4 comprises:

[0043] Step 4.1, first calculate the grid node inertia and elasticity value, the expression is as follows:

[0044]

[0045] Among them, Hs(h) is the inertia value of the active distribution network node, which is used to characterize the inertia of the node; f is the corresponding node frequency, per unit value; ΔP h is the increment of unbalanced active power of node h caused by disturbance; Re(h) is the elasticity value of the node, R o (h, t) is the electrical function retention rate of node h, specifically the load retention rate and source retention rate of the node, t e is the disturbance occurrence time, t pr Calculate the deadline for node elasticity, R true (h, t) is the actual value of the electrical function, R nom (h, t) is the electrical function value under normal operation;

[0046] Step 4.2: Identify the faulty nodes after the active distribution network is disturbed and the nodes whose source load exceeds their maximum capacity, and obtain the node set N to be allocated assign ; Determine the source load distribution amount according to the inertia and elasticity of the power grid nodes, and obtain the target adjacency matrix Λ after redistribution.

[0047] Furthermore, the specific steps of step 4.2 are as follows:

[0048] Step 4.2.1, traverse the set of nodes to be assigned N assign For the node q that needs to be redistributed, traverse its neighboring nodes g and calculate the redistributed source load ΔF between nodes g , the expression is as follows:

[0049]

[0050] Among them, F p is the source load of the node to be allocated under normal operating conditions, Γ p is the neighbor node of node p, α1, α2 are elasticity and inertia weight adjustment factors, d pgis the electrical distance between node p and node g, Hs(g) and Hs(k) are the inertia values ​​of nodes g and k respectively, Re(g) and Re(k) are the elastic values ​​of nodes g and k respectively;

[0051] Step 4.2.2, after calculating the redistribution amount of all nodes, determine whether all nodes trigger the fault condition, that is, whether the node source load exceeds its maximum capacity. If it exceeds, put the node into the set N assign In the process, redistribution is performed again, otherwise the calculation of the redistribution amount is terminated, and the target adjacency matrix Λ after the active distribution network redistribution is obtained.

[0052] Furthermore, the step 5 comprises:

[0053] Step 5.1, determining the allocable node matrix Z based on the network topology of the active distribution network;

[0054] Step 5.2: Calculate the initial source-load transfer matrix Ψ based on the original active power matrix P of the active distribution network and the target adjacency matrix Λ after redistribution initial ;

[0055] Step 5.3: For the transfer amount Ψ between node i and node j i,j , iteratively calculate the optimal amount of source load distribution Δω i,j,k,l , and update the transfer amount Ψ between the corresponding nodes in turn i,j ,Ψ k,l ,Ψ i,l ,Ψ k,j ,Ψ i，j represents the transfer amount between node i and node j, Ψ k，l represents the transfer amount between node k and node l, Ψ i，l represents the transfer amount between node i and node l, Ψ k，j Represents the transfer amount between node k and node j. Iteratively calculate the transfer amount of all nodes to obtain the final source-load transfer matrix Ψ, which is expressed as follows:

[0056] Ψ initial =P-Λ

[0057]

[0058]

[0059] Among them, z k,l ,z i,l ,z k,j are the corresponding elements of the assignable node matrix Z, which respectively indicate whether nodes k and l are assignable, whether nodes i and l are assignable, and whether nodes k and j are assignable.

[0060] Compared with the prior art, the present invention has the following beneficial effects:

[0061] 1) The present invention establishes a calculation formula for the characteristics of active distribution network nodes taking into account complex network structure holes, comprehensively integrating the topological structure characteristics and electrical characteristics of the nodes, solving the problem that traditional methods ignore the complex relationships between nodes when analyzing the importance of power grid nodes, thereby providing a more accurate and comprehensive node importance assessment feature index.

[0062] 2) The present invention establishes an active distribution network node importance assessment model based on the improved I-PageRank method. The model can deeply analyze the relative importance of each node in the power grid and fully consider the mutual influence between nodes and network structure characteristics. The K-shell algorithm node core number is introduced as the attenuation coefficient in the traditional PageRank algorithm, which further improves the performance of the model in identifying nodes.

[0063] 3) The present invention establishes a calculation model for the source load distribution level of an active distribution network, combines the results of active distribution network node importance identification, takes into account the elasticity and inertia distribution of the network, calculates the probability of source load transfer between nodes, and determines the source load distribution level, which can improve the stability and reliability of the system while maintaining as many source loads as possible.

[0064] 4) A source load distribution model based on the SSPR method is established. When complex scenarios such as multi-node source load distribution, multiple redistribution of a single node, and multiple redistribution of multiple nodes occur in the active distribution network, the model can calculate the solution that optimizes the network strength and sparsity while achieving distribution. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0066] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] See also Figure 1 , a source and load distribution method of an active distribution network based on node elasticity and inertia distribution, comprising:

[0068] Step 1: Based on the active distribution network topology of IEEE118 nodes, including 8 photovoltaic nodes, 5 wind turbine nodes, 3 synchronous generator nodes, and 18 load nodes, obtain the active distribution network topology structure and power quality data, and process the original data into a standard data format, including:

[0069] The power quality related data of each node of the active distribution network is measured by the power monitoring system, and the obtained data is cleaned, missing value processed, denoised, standardized and normalized using the digital filtering method; at the same time, the generators, loads and new energy connected to the power grid are processed as network nodes. The processed power grid data includes the label matrix L storing the type information of the active distribution network nodes, the directed adjacency matrix of the power grid Matrix A is the directed adjacency matrix of the power grid, A.arcs[i][j] refers to the element in row i and column j in the directed adjacency matrix A of the power grid, the node current matrix I, the frequency matrix f, the voltage matrix V, the node active power matrix P, and the reactive power matrix Q.

[0070] Step 2: Based on the complex network structural hole theory, extract the network topology metric characteristics and node electrical characteristics of the active distribution network nodes, use the EFO function to extract the principal component features, and form a reconstructed feature space, including:

[0071] Step 2.1: Extract the node network topology hole feature set F for the active distribution network topo , which contains the following features: active distribution network node degree centrality matrix K, node betweenness centrality matrix C RB , the node proximity centrality matrix C AC , the specific calculation is as follows:

[0072]

[0073]

[0074]

[0075] d(h,j,t)=log(α h,j (t),α j,h (t))

[0076]

[0077] in, is the out-degree of the active distribution network node h at time t, is the in-degree of node h at time t, n is the number of nodes in the active distribution network, a hj (t) is the value of the directed adjacency matrix A.arcs[h][j] of the power grid at time t, a jh(t) is the value of the directed adjacency matrix A.arcs[j][h] of the power grid at time t, σ sw (h,t) is the number of shortest paths from s to w passing through node h at time t, σ sw (t) is the number of shortest paths from s to w at time t, d(h,j,t) is the electrical distance from node h to node j at time t, α h,j (t) is the voltage sensitivity from node h to node j, α j,h (t) is the voltage sensitivity from node j to node h, U h (t) is the voltage of node h at time t, U j (t) is the voltage of node j at time t, P j (t) is the active power of node j at time t, Q j (t) is the reactive power of node j at time t;

[0078] Step 2.2: Extract the node electrical feature set F for the active distribution network ele , which includes the following features: Considering the real-time power flow tracking matrix P of distributed generation GL , network efficiency matrix E, node constraint coefficient matrix Co, the specific calculation is as follows:

[0079]

[0080]

[0081]

[0082] Co(h,t)=∑ k∈Γ(h) (p hk (t)+∑ j∈(Γ(h)IΓ(k)) (p hj (t)p jk (t))) 2

[0083]

[0084] Among them, P G (h, t) is the generator injection power at power node h at time t, P(h, t) is the injection power at node h at time t, [A d ] ij is the downstream distribution value from node i to node j, P L (k,t) is the power of load node k at time t, N G is the power node set, N L is the load node set, P ij is the active power transmitted from node i to node j, P jis the total active power injected into node j, N is the set of active distribution network nodes, Γ(h) is the neighbor node of node h, Γ(k) is the neighbor node of node k, phk(t) is the ratio of the power injected into node k by node h to the total outflow power of node h, p hj (t) is the ratio of the power input from node h to node j to the total outflow power from node h, p jk (t) is the ratio of the power input from node j to node k to the total outflow power from node j, p ij (t) is the ratio of the power input from node i to node j to the total outflow power from node i, P ij (t) is the power injected from node i to node j at time t, P ig (t) is the power injected from node i to node g at time t;

[0085] Step 2.3, using the empirical orthogonal function EOF, the node network topology metric features and node electrical features with high correlation coefficients are reconstructed into feature space: the features of the active distribution network nodes are normalized and constructed into matrices X m×n , solve Get the principal components of the node feature vector, where It is the transpose of the eigenvector matrix; the variance contribution rates of the first n eigenvectors are accumulated, and when the accumulated value exceeds the preset threshold γ, the process ends. At the same time, the features involved in the accumulation are extracted and reconstructed into the feature matrix R o×n , where o is the number of reconstructed features.

[0086] Step 3: For the active distribution network, a node importance evaluation model based on the improved I-PageRank method is established to obtain a ranking set of important nodes, including:

[0087] Step 3.1: Use the improved k-shell method to decompose the active distribution network. Iteratively delete nodes from small to large according to the node in-degree to obtain the k-shell layer number K of the node. S (h, t), and record the number of iterations n(h, t) when the node is deleted; combined with the topological connection, calculate the number of active distribution network node cores K Core (h, t), the expression is as follows:

[0088] K Core (h,t)=K S (h,t)×K(h,t)+n(h,t)+μ i D(h,t)

[0089] Among them, K(h,t) is the degree of node h at time t, D(h,t) is the number of secondary neighbor nodes of the node, μ i is the influence coefficient;

[0090] Step 3.2, according to the reconstructed feature space R o×n Calculate the weight w between nodes ij , we get the active distribution network weight matrix W, which is calculated as follows:

[0091] w ij =(ω1*r 1,i +ω2*r 2,i +L+ω o *r o,i )*(ω1*r 1,j +ω2*r 2,j +L+ω o *r o,j )

[0092] where ω1,ω2L,ω o is the benign adjustment factor of the node corresponding feature, r 1,i ,r 2,i ,L,r o,i is the eigenvalue of the corresponding node i in the reconstruction matrix, r 1,j ,r 2,j ,L,r o,j is the eigenvalue of the corresponding node j in the reconstruction matrix;

[0093] Step 3.3, calculate the PR value of the node, the calculation formula is as follows:

[0094]

[0095] Where i is the number of iterations of PR value calculation, d is the damping factor, PR(h) i is the PR value of node h in the i-th iteration, and the decay coefficient is the number of cores K of node j Core (j);

[0096] Step 3.4, according to the label matrix L of the node type information, the active distribution network node set is divided into a power node set, a load node set, and an intermediate node set. According to the PR value of the node, the power node set, load node set, and intermediate node set of the active distribution network are sorted in importance to determine the key nodes in the power grid.

[0097] Step 4: Assume that the output of the photovoltaic node fluctuates violently due to the sudden change of irradiance during operation, resulting in a significant drop in the voltage of load node 3. For this source-side disturbance, based on the important node identification results of step 3, the inertia and elasticity distribution of the active distribution network nodes are calculated, and a calculation model for the source-load distribution magnitude of the active distribution network is established to obtain the source-load distribution magnitude matrix. The specific steps are as follows:

[0098] Step 4.1, first calculate the grid node inertia and elasticity value, as follows:

[0099]

[0100] Among them, Hs(h) is the inertia value of the active distribution network node, which is used to characterize the inertia of the node; f is the corresponding node frequency, per unit value; ΔP h is the increment of unbalanced active power of node h caused by disturbance; Re(h) is the elasticity value of the node, R o (h, t) is the electrical function retention rate of node h, specifically the load retention rate and source retention rate of the node, t e is the disturbance occurrence time, t pr Calculate the deadline for node elasticity, R true (h, t) is the actual value of the electrical function, R nom (h, t) is the electrical function value under normal operation;

[0101] Step 4.2: Identify the faulty nodes after the active distribution network is disturbed and the nodes whose source load exceeds their maximum capacity, and obtain the node set N to be allocated assign ; Determine the source load distribution amount according to the inertia and elasticity of the power grid nodes, and obtain the target adjacency matrix Λ after redistribution. The specific steps are as follows:

[0102] Step 4.2.1, traverse the set of nodes to be assigned N assign For the node q that needs to be redistributed, traverse its neighboring nodes g and calculate the redistributed source load ΔF between nodes g , the expression is as follows:

[0103]

[0104] Among them, F p is the source load of the node to be allocated under normal operating conditions, Γ p is the neighbor node of node p, α1, α2 are elasticity and inertia weight adjustment factors, d pg is the electrical distance between node p and node g, Hs(g) and Hs(k) are the inertia values ​​of nodes g and k respectively, Re(g) and Re(k) are the elastic values ​​of nodes g and k respectively;

[0105] Step 4.2.2, after calculating the allocation of all nodes to be allocated, determine whether all nodes trigger the fault condition, that is, whether the node source load exceeds its maximum capacity. If it exceeds, put the node into the set N assign In the process, redistribution is performed again, otherwise the calculation of the redistribution amount is terminated, and the target adjacency matrix Λ after the active distribution network redistribution is obtained.

[0106] Step 5: Based on the source load distribution magnitude matrix obtained in step 4, a source load distribution model is constructed based on the SSPR method to obtain a source load distribution matrix that considers the overall strength and sparsity of the power grid, including:

[0107] Step 5.1, determining the allocable node matrix Z based on the network topology of the active distribution network;

[0108] Step 5.2: Calculate the initial source-load transfer matrix Ψ based on the original active power matrix P of the active distribution network and the target adjacency matrix Λ after redistribution initial ;

[0109] Step 5.3: For the transfer amount Ψ between node i and node j i,j , iteratively calculate the optimal amount of source load distribution Δω i,j,k,l , and update the transfer amount Ψ between the corresponding nodes in turn i,j ,Ψ k,l ,Ψ i,l ,Ψ k,j ,Ψ i，j represents the transfer amount between node i and node j, Ψ k，l represents the transfer amount between node k and node l, Ψ i，l represents the transfer amount between node i and node l, Ψ k，j Represents the transfer amount between node k and node j. Iteratively calculate the transfer amount of all nodes to obtain the final source-load transfer matrix Ψ, which is expressed as follows:

[0110] Ψ initial =P-Λ

[0111]

[0112]

[0113] Among them, z k,l ,z i,l ,z k,j are the corresponding elements of the assignable node matrix Z, which respectively indicate whether nodes k and l are assignable, whether nodes i and l are assignable, and whether nodes k and j are assignable. It indicates that only the assignable node set can participate in the optimization reallocation process.

[0114] Step 6, verify the source-load allocation effect, calculate the power flow of the redistributed active distribution network, and verify whether the node power quality meets the standard requirements. If it meets the standard requirements, issue a command; if not, return to step 4 to adjust the source load to be allocated.

[0115] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for allocating sources and loads in an active power distribution network based on node elasticity and inertia distribution, characterized in that: include: Step 1, obtaining the active distribution network topology and power quality data, and processing the original data into a standard data format; Step 2: Based on the complex network structural hole theory, the network topology metric characteristics and node electrical characteristics of the active distribution network nodes are extracted, and the principal component features are extracted using the EFO function to form a reconstructed feature space; Step 3, establish an active distribution network node importance evaluation model based on the improved I-PageRank method to obtain a ranked set of important nodes; Step 4, based on the important node identification results of step 3, calculate the inertia and elasticity distribution of the active distribution network nodes, establish a calculation model for the source load distribution magnitude of the active distribution network, and obtain the source load distribution magnitude matrix; Step 5, based on the source load distribution magnitude matrix obtained in step 4, a source load distribution model is constructed based on the SSPR method to obtain a source load distribution matrix that considers the overall strength and sparsity of the power grid; Step 6, verify the source-load allocation effect, calculate the power flow of the redistributed active distribution network, and verify whether the node power quality meets the standard requirements. If it meets the standard requirements, issue a command; if not, return to step 4 to adjust the source load to be allocated.

2. The method for allocating sources and loads of an active power distribution network based on node elasticity and inertia distribution according to claim 1, characterized in that: The step 1 comprises: The power quality related data of each node of the active distribution network is measured by the power monitoring system, and the obtained data is cleaned, missing value processed, denoised, standardized and normalized using the digital filtering method; at the same time, the generators, loads and new energy connected to the power grid are processed as network nodes. The processed power grid data includes the label matrix L storing the type information of the active distribution network nodes, the directed adjacency matrix of the power grid Matrix A is the directed adjacency matrix of the power grid. A.arcs[i][j] refers to the element in row i and column j in the directed adjacency matrix A of the power grid, the node current matrix I, the frequency matrix f, the voltage matrix V, the node active power matrix P, and the reactive power matrix Q.

3. The method for allocating sources and loads of an active power distribution network based on node elasticity and inertia distribution according to claim 1, characterized in that: The step 2 comprises: Step 2.1: Extract the feature set F of the network topology hole of the active distribution network node topo , which contains the following features: active distribution network node degree centrality matrix K, node betweenness centrality matrix C RB , the node proximity centrality matrix C AC , the expression is as follows: d(h,j,t)=log(α h,j (t),α j,h (t)) in, is the out-degree of the active distribution network node h at time t, is the in-degree of node h at time t, n is the number of nodes in the active distribution network, a hj (t) is the value of the directed adjacency matrix A.arcs[h][j] of the power grid at time t, a jh (t) is the value of the directed adjacency matrix A.arcs[j][h] of the power grid at time t, σ sw (h,t) is the number of shortest paths from s to w passing through node h at time t, σ sw (t) is the number of shortest paths from s to w at time t, d(h,j,t) is the electrical distance from node h to node j at time t, α h,j (t) is the voltage sensitivity from node h to node j, α j,h (t) is the voltage sensitivity from node j to node h, U h (t) is the voltage of node h at time t, U j (t) is the voltage of node j at time t, P j (t) is the active power of node j at time t, Q j (t) is the reactive power of node j at time t; Step 2.2: Extract the electrical feature set F of the active distribution network nodes ele , which includes the following features: Considering the real-time power flow tracking matrix P of distributed generation GL , network efficiency matrix E, node constraint coefficient matrix Co, the expressions are as follows: Co(h,t)=∑ k∈Γ(h) (p hk (t)+∑ j∈(Γ(h)IΓ(k)) (p hj (t) p jk (t))) 2 Among them, P G (h, t) is the generator injection power at power node h at time t, P(h, t) is the injection power at node h at time t, [A d ] ij is the downstream distribution value from node i to node j, P L (k,t) is the power of load node k at time t, N G is the power node set, N L is the load node set, β i is the set of all nodes connected to node i and drawing power from node i, d(h,i,t) is the electrical distance between node h and node i at time t, P ij is the active power transmitted from node i to node j, P j is the total injected active power of node j, N is the set of active distribution network nodes, Γ(h) is the neighbor node of node h, Γ(k) is the neighbor node of node k, p hk (t) is the ratio of the power input from node h to node k to the total outflow power from node h, p hj (t) is the ratio of the power input from node h to node j to the total outflow power from node h, p jk (t) is the ratio of the power input from node j to node k to the total outflow power from node j, p ij (t) is the ratio of the power input from node i to node j to the total outflow power from node i, P ij (t) is the power injected from node i to node j at time t, P ig (t) is the power injected from node i to node g at time t; Step 2.3, using the empirical orthogonal function EOF, the node network topology metric features and node electrical features with high correlation coefficients are reconstructed into feature space: the active distribution network node features are normalized and constructed into matrices X m×n , solve Get the principal components of the node feature vector, where It is the transpose of the eigenvector matrix; the variance contribution rates of the first n eigenvectors are accumulated, and when the accumulated value exceeds the preset threshold γ, the process ends. At the same time, the features involved in the accumulation are extracted and reconstructed into the feature matrix R o×n , where o is the number of reconstructed features.

4. The method for allocating sources and loads of an active power distribution network based on node elasticity and inertia distribution according to claim 1, characterized in that: The step 3 comprises: Step 3.1: Use the improved k-shell method to decompose the active distribution network. Iteratively delete nodes from small to large according to the node in-degree to obtain the k-shell layer number K of the node. S (h, t), and record the number of iterations n(h, t) when the node is deleted; combined with the topological connection, calculate the number of active distribution network node cores K Core (h, t), the expression is as follows: K Core (h,t)=K S (h,t)×K(h,t)+n(h,t)+μ i D(h,t) Among them, K(h,t) is the degree of node h at time t, D(h,t) is the number of secondary neighbor nodes of the node, μ i is the influence coefficient; Step 3.2, according to the reconstructed feature space R o×n Calculate the weight w between nodes ij , we get the active distribution network weight matrix W, which is expressed as follows: w ij =(ω1*r 1,i +ω2*r 2,i +L+ω o *r o,i )*(ω1*r 1,j +ω2*r 2,j +L+ω o *r o,j ) where ω1,ω2L,ω o is the benign adjustment factor of the node corresponding feature, r 1,i ,r 2,i ,L,r o,i is the eigenvalue of the corresponding node i in the reconstruction matrix, r 1,j ,r 2,j ,L,r o,j is the eigenvalue of the corresponding node j in the reconstruction matrix; Step 3.3, calculate the PR value of the node, the calculation formula is as follows: Where i is the number of iterations of PR value calculation, d is the damping factor, PR(h) i is the PR value of node h in the i-th iteration, and the decay coefficient is the number of cores K of node j Core (j), p jh (t) is the ratio of the power input from node j to node h to the total outflow power from node j, w jh is the weight between node j and node h; Step 3.4, according to the label matrix L of the node type information, the active distribution network node set is divided into a power node set, a load node set, and an intermediate node set. According to the PR value of the node, the power node set, load node set, and intermediate node set of the active distribution network are sorted in importance to determine the key nodes in the power grid.

5. The method for allocating sources and loads of an active power distribution network based on node elasticity and inertia distribution according to claim 1, characterized in that: The step 4 comprises: Step 4.1, first calculate the grid node inertia and elasticity value, the expression is as follows: Among them, Hs(h) is the inertia value of the active distribution network node, which is used to characterize the inertia of the node; f is the corresponding node frequency, per unit value; ΔP h is the increment of unbalanced active power of node h caused by disturbance; Re(h) is the elasticity value of the node, R o (h, t) is the electrical function retention rate of node h, specifically the load retention rate and source retention rate of the node, t e is the disturbance occurrence time, t pr Calculate the deadline for node elasticity, R true (h, t) is the actual value of the electrical function, R nom (h, t) is the electrical function value under normal operation; Step 4.2: Identify the faulty nodes after the active distribution network is disturbed and the nodes whose source load exceeds their maximum capacity, and obtain the node set N to be allocated assign ; Determine the source load distribution amount according to the inertia and elasticity of the power grid nodes, and obtain the target adjacency matrix Λ after redistribution.

6. The method for allocating sources and loads of an active power distribution network based on node elasticity and inertia distribution according to claim 5, characterized in that: The specific steps of step 4.2 are as follows: Step 4.2.1, traverse the set of nodes to be assigned N assign For the node q that needs to be redistributed, traverse its neighboring nodes g and calculate the redistributed source load ΔF between nodes g , the expression is as follows: Among them, F p is the source load of the node to be allocated under normal operating conditions, Γ p is the neighbor node of node p, α1, α2 are elasticity and inertia weight adjustment factors, d pg is the electrical distance between node p and node g, Hs(g) and Hs(k) are the inertia values ​​of nodes g and k respectively, Re(g) and Re(k) are the elastic values ​​of nodes g and k respectively; Step 4.2.2, after calculating the redistribution amount of all nodes, determine whether all nodes trigger the fault condition, that is, whether the node source load exceeds its maximum capacity. If it exceeds, put the node into the set N assign In the process, redistribution is performed again, otherwise the calculation of the redistribution amount is terminated, and the target adjacency matrix Λ after the active distribution network redistribution is obtained.

7. The method for allocating sources and loads of an active power distribution network based on node elasticity and inertia distribution according to claim 1, characterized in that: The step 5 comprises: Step 5.1, determining the allocable node matrix Z based on the network topology of the active distribution network; Step 5.2: Calculate the initial source-load transfer matrix Ψ based on the original active power matrix P of the active distribution network and the target adjacency matrix Λ after redistribution initial ; Step 5.3: For the transfer amount Ψ between node i and node j i,j , iteratively calculate the optimal amount of source load distribution Δω i,j,k,l , and update the transfer amount Ψ between the corresponding nodes in turn i,j ,Ψ k,l ,Ψ i,l ,Ψ k,j ,Ψ i，j represents the transfer amount between node i and node j, Ψ k，l represents the transfer amount between node k and node l, Ψ i，l represents the transfer amount between node i and node l, Ψ k，j Represents the transfer amount between node k and node j. Iteratively calculate the transfer amount of all nodes to obtain the final source-load transfer matrix Ψ, which is expressed as follows: P initial =P-Λ Among them, z k,l ,z i,l ,z k,j are the corresponding elements of the assignable node matrix Z, which respectively indicate whether nodes k and l are assignable, whether nodes i and l are assignable, and whether nodes k and j are assignable.

Images