Method and system for rapid partitioning of power network after power outage based on coherent Ising machine

Abstract

The present invention discloses a method and system for rapid post-outage partitioning of power networks based on a coherent Ising machine. By constructing a Laplace matrix of the power network, building a sub-region partitioning model for parallel recovery after a network outage, converting the model into a binary quadratic unconstrained optimization model, solving the binary quadratic unconstrained optimization model using a coherent Ising machine, and converting the coherent Ising machine solution into a network partitioning result, the method provides a novel quantum computing-based solution to the problem of network partitioning after a power system outage. The method avoids the "curse of dimensionality" problem that may be encountered when using classical computers to solve the above partitioning problem, and proposes a quantum bit expansion method to address the problem of excessive number of variables during the solution process, thereby accelerating the solution of the power network partitioning problem.

Description

Technical Field

[0001] The present invention belongs to the field of application of quantum computing technology in power systems, and specifically relates to a method and system for rapid partitioning of power networks after a power outage based on a coherent Ising machine. Background Art

[0002] After a major blackout, implementing a parallel, zoned restoration approach can effectively reduce outage duration and rapidly restore normal grid operations, making it an effective means of enhancing the security and protection capabilities of new power systems. In the face of potential blackout risks, developing a rational, precise, and rapid restoration sub-regional division strategy is a prerequisite for ensuring the implementation of established emergency response plans.

[0003] With the development of new power systems, network structures are becoming increasingly complex, posing greater challenges to network partitioning. Furthermore, this development places higher demands on the real-time and precision of partitioning strategies. Existing partitioning methods struggle to meet these requirements due to factors such as the inability to directly solve the model, the need for manual selection of solutions, and the limitations of existing computing equipment. Summary of the Invention

[0004] The present invention addresses the problems in existing network partitioning methods, such as long solution times, the need for manual selection of solution results, and the inability to directly solve the model. It provides a method and system for rapid partitioning of power networks after power outages based on a coherent Ising machine. By constructing the Laplace matrix of the power network, constructing a sub-region partitioning model for parallel recovery after a network power outage, converting the model into a binary quadratic unconstrained optimization model, using a coherent Ising machine to solve the binary quadratic unconstrained optimization model, and converting the coherent Ising machine solution results into network partitioning results, the present invention provides a new solution based on quantum computing for the network partitioning problem after a power system power outage. The method of the present invention avoids the "curse of dimensionality" problem that may be encountered when using classical computers to solve the above partitioning problems, and proposes a quantum bit expansion method to address the problem of excessive number of variables in the solution process, thereby accelerating the solution of the power network partitioning problem.

[0005] To achieve the above-mentioned object, the technical solution adopted by the present invention is: a method for rapid partitioning of a power network after a power outage based on a coherent Ising machine, comprising the following steps:

[0006] S1. Construct Laplace matrix: The topology of the power network is regarded as a weighted undirected graph G(V,E), where V = {v1,...,v i ,...,v N} is the set of network nodes, E={(v i ,v j )|v i ∈V,v j∈V} is a set of lines, and the Laplace matrix for the power network is constructed as follows:

[0007] L=DW

[0008] Where L is the Laplace matrix, D is the degree matrix constructed based on the degree index of the importance of the nodes in the graph, and W is the adjacency matrix constructed based on the weight coefficient of the closeness of the connection between two nodes;

[0009] S2. Establishment of a partitioning model: Based on the Laplace matrix constructed in step S1, a partitioning model for the restoration sub-areas after a network power outage is constructed; the objective function of the partitioning model is:

[0010]

[0011] Among them, x k =[x 1,k ,...,x i,k ,...,x N,k ] T , x i,k ∈{0,1} is a binary variable describing the node partitioning result. If the node v i is divided into the kth sub-region, then x i,k =1, otherwise x i,k =0;

[0012] S3, model conversion: converting the post-power outage restoration sub-area partitioning model established in step S2 into a binary quadratic unconstrained optimization model;

[0013] S4. Model solution: Using a classical computer and a coherent Ising machine, the model converted in step S3 is jointly solved based on the quantum bit expansion method. The solution is converted into a partitioning result, and sub-regions are divided according to the partitioning result to verify the connectivity of each sub-region later.

[0014] S5. Boundary node adjustment: Use the sub-Laplacian matrix to verify the connectivity of the sub-regions and merge the isolated nodes of the disconnected sub-regions. Finally, adjust the boundary nodes of each sub-region and verify the model constraints, and output the final partitioning results.

[0015] As an improvement of the present invention, the method for constructing the Laplace matrix in step S1 specifically includes the following steps:

[0016] S11: Define the weight coefficient w between two nodes ij Describe node v i and node v j The closeness of the relationship:

[0017]

[0018] Among them, Z ij For node v i and node v j Inter-impedance;

[0019] S12: Define node v i The degree index d i To describe the importance of nodes in the graph:

[0020]

[0021] S13: Using weight coefficient w ij Construct the adjacency matrix:

[0022]

[0023] S14: Utilization index d i Construct the degree matrix:

[0024]

[0025] As another improvement of the present invention, the model conversion in step S3 specifically includes the following steps:

[0026] S31: For the equality constraints in the model, use the square penalty function to convert them into augmented terms:

[0027]

[0028] in, is the penalty coefficient, h i (x) = b i is the corresponding equality constraint;

[0029] S32: For the inequality constraints in the model, slack variables are introduced to transform them into equality constraints, and then square penalty functions are used to transform them into augmented terms;

[0030] S321: If the model is medium-sized such as g i (x)≤c i The following transformation is used:

[0031]

[0032] in, is the slack variable;

[0033] S322: If the model is medium-sized such as d i ≤g i (x) constraint, the following conversion method is used:

[0034]

[0035] in, is the slack variable;

[0036] S33: Add the objective function to all augmentation terms to obtain the augmented model:

[0037]

[0038] Among them, H obj is the objective function, is the augmented term corresponding to the i-th inequality constraint;

[0039] S34: Augment continuous variables in the model Perform binary expansion to obtain a two-variable quadratic unconstrained optimization model:

[0040] minf(x)=x T Qx

[0041] Among them, x is the decision vector and Q is the coefficient matrix.

[0042] As another improvement of the present invention, the partition model of step S2 includes node unique partition constraint, sub-region node minimum number constraint, sub-region frequency stability constraint and sub-region backup recovery generator number constraint, wherein,

[0043] The node unique partition constraint is an equality constraint, which ensures that each node can only be divided into a unique sub-region, that is:

[0044]

[0045] The minimum number of nodes in the sub-region is an inequality constraint, which ensures that the number of nodes in each sub-region is balanced, that is:

[0046]

[0047] Among them, N k is the minimum number of nodes corresponding to the kth sub-region;

[0048] The sub-region frequency stability constraint is an inequality constraint, which ensures that the frequency fluctuation of each sub-region does not exceed the specified safety range, that is:

[0049]

[0050] Among them, P i and D i Node v i The generated active power and required active power at The unbalanced power upper limit to ensure that the frequency of the k-th sub-area does not exceed the limit;

[0051] The constraint on the number of backup recovery generators in the sub-region is an inequality constraint, which ensures that each sub-region has at least one backup recovery generator, namely:

[0052]

[0053] As another improvement of the present invention, the binary quadratic unconstrained optimization model of step S3 is:

[0054]

[0055] Among them, H only is the unique partition constraint of the transformed node, H least is the minimum number constraint of sub-region nodes after transformation, and is the sub-region power balance constraint after transformation, H back Restoring the generator quantity constraint for the converted sub-region backup.

[0056] As another improvement of the present invention, step S4 specifically includes the following steps:

[0057] S41: Determine whether the number of variables of the binary quadratic unconstrained optimization model output in step S3 is greater than the number of available quantum bits of the coherent Ising machine. If not, go to S42; otherwise, jump to S43;

[0058] S42: Input the coefficient matrix Q into the coherent Ising machine to find the optimal solution of the model;

[0059] S43: Use the quantum bit expansion method to solve the binary quadratic unconstrained optimization model output in step S3.

[0060] As a further improvement of the present invention, step S43 specifically includes the following steps:

[0061] S431: Solve the model N output in step S3 using a classical computer I times, obtain the initial solution pool and assign the optimal solution to

[0062] S432: Randomly select N in the initial solution pool s solutions, 2≤N S ≤N I , count the frequency of the number 1 in the i-th position of the selected solution, denoted as F i ;

[0063] S433: Calculate ξ i =|F i -N S / 2| size, extract the first N M Minimum ξi The corresponding variable x i Composing subproblems, solving the subproblems once using a quantum computer, and updating the solution to the initial solution pool in step S431;

[0064] S434: Repeat steps S431 to S433N E Second, select the best solution from the solution pool Output.

[0065] As a further improvement of the present invention, step S5 specifically includes the following steps:

[0066] S51: Output in step S434 Reconstruct the matrix into a dimension of N*K, where the number of rows corresponding to the elements of each column that are 1 is the grouping result corresponding to the node;

[0067] S52: Cut off the corresponding line in the network according to the grouping result in step S51;

[0068] S53: Verify whether the result of step S52 satisfies the sub-region connectivity constraint, that is, find the minimum non-trivial eigenvalue of the Laplacian matrix corresponding to a sub-region. If it is a negative value, it means that the sub-region is not connected, and jump to S54; if it is a positive value, take this result as the final partition result and jump to S55;

[0069] S54: merging isolated nodes and adjusting boundary nodes; the isolated node is a node that cannot be connected to any other node in the region through any path formed by the remaining nodes and edges in the region; the boundary node is a node connected at both ends of the cut line;

[0070] S55: Output the final partitioning result.

[0071] In order to achieve the above-mentioned purpose, the present invention also adopts a technical solution: a power network post-power outage rapid partitioning system based on a coherent Ising machine, comprising a computer program, which, when executed by a processor, implements the steps of any of the above-mentioned methods.

[0072] Compared with the existing technology, the present invention has the following technical advantages and effects: With the goal of minimizing the sum of the weights of the cut-off lines, the present invention considers the actual operational safety constraints of the power grid to construct a post-blackout network partition model. Based on the square penalty function, corresponding constraint processing methods are proposed for the equality and inequality constraints in the model. The constructed post-blackout network partition model is converted into a binary quadratic unconstrained optimization model that can be directly solved by an optical quantum computer, enabling it to be directly input into the optical quantum computer for solution. This overcomes the "curse of dimensionality" that traditional computers may suffer when faced with the same problem, and achieves accelerated solution of the power network partition problem. Furthermore, a quantum bit expansion method is proposed for the case where the scale of model variables is larger than the computable bits of the coherent Ising machine, effectively expanding the application scale of the model. To address the problem of sub-region disconnection that may arise after the solution, a processing step based on "sub-region connectivity verification - isolated node merging - boundary node adjustment - model constraint verification" is designed, completely solving the sub-region disconnection problem in the direct solution result, which is a major difficulty faced when applying the coherent Ising machine. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 It is a flowchart of the steps of the method for rapid partitioning of a power network after a power outage based on a coherent Ising machine of the present invention;

[0074] Figure 2 This is a schematic diagram of the network partitioning result of Example 1 of the present invention;

[0075] Figure 3 This is a schematic diagram of the network partitioning result of Example 2 of the present invention. DETAILED DESCRIPTION

[0076] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. It should be understood that the following specific embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention.

[0077] Example 1

[0078] This embodiment is applied to the New England 39-bus 10-generator system. The number of partitions is set to 2. Since the model only considers the active power balance of the partition, and the reactive imbalance can be balanced by the local reactive power compensation device, only the line resistance is considered. k Set to 10, Set to 0.

[0079] A fast partitioning method for power network after power outage based on coherent Ising machine, such as Figure 1 As shown, the following steps are included:

[0080] Step S1: constructing the Laplace matrix of the power network;

[0081] S11. Assume that the goal of network partitioning is to divide N nodes into K sub-areas. Consider the power grid topology as a weighted undirected graph G(V,E), where V = {v1,...,v i ,...,v N} is the set of network nodes, E={(v i ,v j )|v i ∈V,v j ∈V} is the set of lines;

[0082] S12. Define the weight coefficient w between two nodes ij Describe node v i and node v j The closeness of the relationship:

[0083]

[0084] Among them, Z ij For node v i and node v j Inter-impedance;

[0085] S13. In order to describe the importance of nodes in the graph, we define node v i The degree is:

[0086]

[0087] S14, using weight coefficient w ij Construct the adjacency matrix, noting that this matrix is ​​symmetric:

[0088]

[0089] S15, Utilization Index d i Construct the degree matrix, noting that this matrix is ​​diagonal:

[0090]

[0091] S16. Construct a Laplace matrix using the adjacency matrix and degree matrix constructed in steps S14 and S15:

[0092] L=DW.

[0093] Step S2: Constructing a sub-area partitioning model for parallel restoration after network power outage. The objective function of the sub-area partitioning model for parallel restoration after network power outage is:

[0094]

[0095] Among them, x k =[x 1,k ,…,x i,k,…,x N,k ] T .x i,k ∈{0,1} is a binary variable describing the node partitioning result. If the node v i is divided into the kth sub-region, then x i,k =1, otherwise x i,k =0.

[0096] The partitioning model includes node unique partitioning constraints, sub-area node minimum number constraints, sub-area frequency stability constraints, and sub-area backup recovery generator number constraints, among which,

[0097] Node unique partition constraint: The purpose of this constraint is to ensure that each node can only be divided into a unique sub-region:

[0098]

[0099] Minimum number of nodes in each sub-region: This constraint aims to ensure that the number of nodes in each sub-region is as balanced as possible:

[0100]

[0101] Among them, N k is the minimum number of nodes corresponding to the kth sub-region;

[0102] Sub-area frequency stability constraint: This constraint is intended to ensure that the frequency fluctuation of each sub-area does not exceed the specified safety range:

[0103]

[0104] Among them, P i and D i Node v i The generated active power and required active power at The unbalanced power upper limit to ensure that the frequency of the k-th sub-area does not exceed the limit;

[0105] Sub-region backup recovery generator quantity constraint: The purpose of this constraint is to ensure that each sub-region has at least one backup recovery generator:

[0106]

[0107] The objective function and four constraints of the partitioning model are used to construct a sub-area partitioning model for parallel restoration after network power outage.

[0108] Step S3: The model in step S2 is converted into a binary quadratic unconstrained optimization model. The specific construction method is as follows:

[0109] S31. For the equality constraints in the model, use the square penalty function to transform them into augmented terms:

[0110]

[0111] in, is the penalty coefficient, h i (x) = b i is the corresponding equality constraint;

[0112] S32. For the inequality constraints in the model, first introduce slack variables to convert them into equality constraints, and then use the square penalty function to convert them into augmented terms, specifically:

[0113] When the model is medium-sized such as g i (x)≤c i The following transformation is used:

[0114]

[0115] in, is the slack variable;

[0116] When the model is medium-sized such as d i ≤g i (x) constraint, the following conversion method is used:

[0117]

[0118] in, is the slack variable;

[0119] S33. Add the objective function to all augmentation terms to obtain the augmented model:

[0120]

[0121] Among them, H obj is the objective function, is the augmented term corresponding to the i-th inequality constraint;

[0122] S34. Augment the continuous variables in the model Perform binary expansion to obtain a two-variable quadratic unconstrained optimization model:

[0123] minf(x)=x T Qx

[0124] Among them, x is the decision vector and Q is the coefficient matrix.

[0125] In this embodiment, the partition model includes a node unique partition constraint, a sub-area node minimum number constraint, a sub-area frequency stability constraint, and a sub-area backup recovery generator number constraint. Therefore, when the partition model is converted into a binary quadratic unconstrained optimization model, the node unique partition constraint is converted into:

[0126]

[0127] in, is the corresponding penalty coefficient;

[0128] The minimum number constraint of sub-region nodes is converted into:

[0129]

[0130] in, is the corresponding penalty coefficient, is the slack variable;

[0131] The sub-region frequency stability constraint is transformed into:

[0132]

[0133] in, is the corresponding penalty coefficient, is the slack variable;

[0134] The constraint on the number of backup recovery generators in the sub-region is converted to:

[0135]

[0136] in, is the corresponding penalty coefficient, is the slack variable;

[0137] The final output of the binary quadratic unconstrained optimization model is:

[0138]

[0139] Step S4: using a coherent Ising machine to solve the binary quadratic unconstrained optimization model outputted above;

[0140] S41, judging whether the number of variables in the binary quadratic unconstrained optimization model outputted in step S3 is greater than the number of available quantum bits of the coherent Ising machine; if not, jumping to S42; otherwise, jumping to S43;

[0141] S42, inputting the coefficient matrix Q into the coherent Ising machine to obtain the optimal solution of the model;

[0142] S43. Solve the model output in step S3 using a quantum bit expansion method; the quantum bit expansion method is specifically:

[0143] S431, using a classical computer to solve the model N output in step S34 I times, obtain the initial solution pool and assign the optimal solution to

[0144] S432, randomly select N in the initial solution pool s (2≤N S ≤N I ) solutions, count the frequency of the number 1 in the i-th position of the selected solution, and record it as F i ;

[0145] S433, calculate ξ i =|F i -N S / 2| size, extract the first N M Minimum ξ i The corresponding variable x i Composing subproblems, solving the subproblems once using a quantum computer, and updating the results of this solution to the initial solution pool in step S431;

[0146] S434, repeat steps S431 to S433N E Second, select the best solution from the solution pool Output.

[0147] Step S5: Convert the coherent Ising machine solution into a network partition result, specifically:

[0148] S51, output in step S434 Reconstruct the matrix into a dimension of N*K, where the number of rows corresponding to the elements of each column that are 1 is the grouping result corresponding to the node;

[0149] S52, cutting off the corresponding line in the network according to the grouping result in step S51;

[0150] S53: Verify whether the result of step S52 satisfies the sub-region connectivity constraint. The connectivity of the sub-region can be expressed as: finding the minimum non-trivial eigenvalue of the Laplacian matrix corresponding to a sub-region. If the minimum non-trivial eigenvalue is negative, it means that the sub-region is not connected, and jump to S54; if the minimum non-trivial eigenvalue is positive, the result is used as the final partition result, and jump to S55.

[0151] S54: merging isolated nodes and adjusting boundary nodes. The merging isolated nodes and adjusting boundary nodes include:

[0152] S541. Merge the isolated node into the nearest sub-region. An isolated node can be defined as: a node in a region that cannot be connected to any other node in the region through any path consisting of nodes and edges in the region;

[0153] S542: Determine whether the result of merging the isolated nodes in step S541 satisfies the various constraints constructed by the partitioning model. If so, jump to S55; if not, adjust the boundary nodes to satisfy them. The boundary nodes can be defined as nodes that are missing from the final graph obtained in step S541 compared with the power grid topology graph in step S11.

[0154] S55. Output the final partitioning result.

[0155] The optimal solution of this embodiment is converted into a partition result, such as Figure 2 In this example, the partitioning result obtained in step S51 is: line 2-25, line 3-18, and line 14-15 are removed. According to the above result, the lines are removed and the sub-region Laplacian matrix is ​​constructed. The minimum non-trivial eigenvalues ​​of the two sub-region Laplacian matrices are both positive, and the partitioning result meets the model constraints. No adjustment is required and it can be output as the final partitioning result.

[0156] Example 2

[0157] This embodiment is applied to the IEEE 118 standard node system verification, the number of partitions is set to 3, N is set to 20, k When N is set to 20 and the quantum bit expansion method proposed in this paper is used for solution, I Set to 20, N S Set to 10, N E Set to 5.

[0158] The sub-region division was performed according to the steps described in Example 1 and repeated 5 times. The final result was as follows. Figure 3 As shown, the optimal partitioning result differs from the reference optimal partitioning result only at nodes 40, 41, 42, and 98. While the worst partitioning result removes fewer lines, the objective function value differs significantly from the optimal reference value. However, both results satisfy network connectivity and operational constraints.

[0159] In summary, our method, by constructing a fast partitioned binary quadratic unconstrained optimization model for power grid outages and solving it using a coherent Ising machine, can effectively overcome the "curse of dimensionality" that classical computers suffer when solving such problems, thereby improving computational efficiency.

[0160] It should be noted that the above content merely illustrates the technical idea of ​​the present invention and cannot be used to limit the scope of protection of the present invention. For ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications all fall within the scope of protection of the claims of the present invention.

Claims

1. A fast partitioning method for power network after power outage based on coherent Ising machine, characterized by: The steps include: S1. Construct Laplace matrix: The topology of the power network is regarded as a weighted undirected graph G(V,E), where V = {v1,...,v i ,...,v N } is the set of network nodes, E={(v i ,v j )|v i ∈V,v j ∈V} is a set of lines, and the Laplace matrix for the power network is constructed as follows: L=DW Where L is the Laplace matrix, D is the degree matrix constructed based on the degree index of the importance of the nodes in the graph, and W is the adjacency matrix constructed based on the weight coefficient of the closeness of the connection between two nodes; S2. Establishment of a partitioning model: Based on the Laplace matrix constructed in step S1, a partitioning model for the restoration sub-areas after a network power outage is constructed; the objective function of the partitioning model is: Among them, x k =[x 1,k ,...,x i,k ,...,x N,k ] T , x i,k ∈{0,1} is a binary variable describing the node partitioning result. If the node v i is divided into the kth sub-region, then x i,k =1, otherwise x i,k =0; S3, model conversion: converting the post-power outage restoration sub-area partitioning model established in step S2 into a binary quadratic unconstrained optimization model; S4. Model solution: Using a classical computer and a coherent Ising machine, the model converted in step S3 is jointly solved based on the quantum bit expansion method, and the solution result is converted into a partition result, and the sub-region is divided according to the partition result; S5. Boundary node adjustment: Use the sub-Laplacian matrix to verify the connectivity of the sub-regions, merge the isolated nodes of the disconnected sub-regions, adjust the boundary nodes of each sub-region and verify the model constraints, and output the final partitioning results.

2. The method for rapid partitioning of a power network after a power outage based on a coherent Ising machine according to claim 1, characterized in that: The method for constructing the Laplace matrix in step S1 specifically includes the following steps: S11: Define the weight coefficient w between two nodes ij Describe node v i and node v j The closeness of the relationship: Among them, Z ij For node v i and node v j Inter-impedance; S12: Define node v i The degree index d i To describe the importance of nodes in the graph: S13: Using weight coefficient w ij Construct the adjacency matrix: S14: Utilization index d i Construct the degree matrix:

3. The method for rapid partitioning of a power network after a power outage based on a coherent Ising machine according to claim 1, characterized in that: The model conversion in step S3 specifically includes the following steps: S31: For the equality constraints in the model, use the square penalty function to convert them into augmented terms: in, is the penalty coefficient, h i (x) = b i is the corresponding equality constraint; S32: For the inequality constraints in the model, slack variables are introduced to transform them into equality constraints, and then square penalty functions are used to transform them into augmented terms; S321: If the model is medium-sized such as g i (x)≤c i The following transformation is used: in, is the slack variable; S322: If the model is medium-sized such as d i ≤g i (x) constraint, the following conversion method is used: in, is the slack variable; S33: Add the objective function to all augmentation terms to obtain the augmented model: Among them, H obj is the objective function, is the augmented term corresponding to the i-th inequality constraint; S34: Augment continuous variables in the model Perform binary expansion to obtain a two-variable quadratic unconstrained optimization model: minf(x)=x T Qx Among them, x is the decision vector and Q is the coefficient matrix.

4. The method for rapid partitioning of a power network after a power outage based on a coherent Ising machine according to claim 3, characterized in that: The partition model of step S2 includes node unique partition constraint, sub-region node minimum number constraint, sub-region frequency stability constraint and sub-region backup recovery generator number constraint, wherein, The node unique partition constraint is an equality constraint, which ensures that each node can only be divided into a unique sub-region, that is: The minimum number of nodes in the sub-region is an inequality constraint, which ensures that the number of nodes in each sub-region is balanced, that is: Among them, N k is the minimum number of nodes corresponding to the kth sub-region; The sub-area power balance constraint is an inequality constraint, which ensures that the frequency fluctuation of each sub-area does not exceed the specified safety range, that is: Among them, P i and D i Node v i The generated active power and required active power at The unbalanced power upper limit to ensure that the frequency of the k-th sub-area does not exceed the limit; The constraint on the number of backup recovery generators in the sub-region is an inequality constraint, which ensures that each sub-region has at least one backup recovery generator, namely:

5. The method for rapid partitioning of a power network after a power outage based on a coherent Ising machine according to claim 4, characterized in that: The binary quadratic unconstrained optimization model of step S3 is: Among them, H only is the unique partition constraint of the transformed node, H least is the minimum number constraint of sub-region nodes after transformation, and is the sub-region power balance constraint after transformation, H back Restoring the generator quantity constraint for the converted sub-region backup.

6. The method for rapid partitioning of a power network after a power outage based on a coherent Ising machine according to claim 1, characterized in that: The step S4 specifically includes the following steps: S41: Determine whether the number of variables of the binary quadratic unconstrained optimization model output in step S3 is greater than the number of available quantum bits of the coherent Ising machine. If not, go to S42; otherwise, jump to S43; S42: Input the coefficient matrix Q into the coherent Ising machine to find the optimal solution of the model; S43: Use the quantum bit expansion method to solve the binary quadratic unconstrained optimization model output in step S3.

7. The method for rapid partitioning of a power network after a power outage based on a coherent Ising machine according to claim 6, characterized in that: The step S43 specifically includes the following steps: S431: Solve the model N output in step S3 using a classical computer I times, obtain the initial solution pool and assign the optimal solution to S432: Randomly select N in the initial solution pool s solutions, 2≤N S ≤N I , count the frequency of the number 1 in the i-th position of the selected solution, denoted as F i ; S433: Calculate ξ i =|F i -N S / 2| size, extract the first N M Minimum ξ i The corresponding variable x i Combining subproblems, solving the subproblems once using a quantum computer, and updating the solution to the initial solution pool in step S431; S434: Repeat steps S431 to S433N E Second, select the best solution from the solution pool Output.

8. The method for rapid partitioning of a power network after a power outage based on a coherent Ising machine according to claim 7, characterized in that: The step S5 specifically includes the following steps: S51: Output in step S434 Reconstruct the matrix into a dimension of N*K, where the number of rows corresponding to the elements of each column that are 1 is the grouping result corresponding to the node; S52: Cut off the corresponding line in the network according to the grouping result in step S51; S53: Verify whether the result of step S52 satisfies the sub-region connectivity constraint, that is, find the minimum non-trivial eigenvalue of the Laplacian matrix corresponding to a sub-region. If it is a negative value, it means that the sub-region is not connected, and jump to S54; if it is a positive value, take this result as the final partition result and jump to S55; S54: merging isolated nodes and adjusting boundary nodes; the isolated node is a node that cannot be connected to any other node in the region through any path formed by the remaining nodes and edges in the region; the boundary node is a node connected at both ends of the cut line; S55: Output the final partitioning result.

9. A system for rapid partitioning of power networks after a power outage based on a coherent Ising machine, comprising a computer program, characterized in that: When the computer program is executed by a processor, the steps of any one of the above methods are implemented.

Images