A parallel acceleration solving method for power system post-disaster recovery based on quantum computing

Abstract

The application belongs to the technical field of power system operation control, and relates to a parallel acceleration solving method for power system post-disaster recovery based on quantum calculation, which comprises the following steps: step 1, establishing a basic model of power system post-disaster recovery; step 2, according to a quantum agent Lagrange relaxation algorithm, decomposing the original optimization problem into various node sub-problems; and step 3, solving the post-disaster recovery model, continuously updating the Lagrange multiplier and the penalty factor in the iteration process, and realizing optimal solving under the interactive calculation of quantum calculation and classical calculation to obtain the post-disaster recovery scheme to guide the actual safe operation of the power system. According to the application, the problem is decomposed into multiple node-centered sub-problems, and the sub-problems are divided into discrete and continuous parts for solving, so that the solving difficulty is reduced. The application realizes the effects of parallel calculation and quantum embedding acceleration calculation, does not need to relax the problem for complete optimization, and guarantees better convergence.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system operation control, and relates to a parallel accelerated solution method for post-disaster recovery of a power system based on quantum computing. Background Art

[0002] After extreme disasters, it is crucial to quickly restore power to the power system after it has been damaged. Reasonable and efficient dispatch control and coordinated optimization are fundamental to ensuring the safe and stable operation of the power grid. As the scale of power systems continues to expand, the complexity and computational burden of optimization problems increase dramatically, necessitating the need for faster and more efficient solution technologies to address these challenges.

[0003] Quantum computing uses "qubits" as the basic unit of information and utilizes principles such as quantum superposition and entanglement to perform parallel computations. This can achieve exponential acceleration in certain optimization problems and is a key direction for future computing development. Currently, quantum computing hardware includes gate-based quantum computers, specialized quantum computers (such as coherent Bose quantum computers, coherent Ising machines, and quantum annealing computers), and quantum simulators based on classical computing. Although quantum hardware has entered the quantum intermediate-noise-scale (NISQ) era, the number of qubits in quantum processors is limited, and achieving large-scale, fault-tolerant, universal quantum computing still requires significant effort.

[0004] Current exploration of quantum computing applications primarily encompasses four areas: quantum combinatorial optimization, quantum parameter prediction, quantum artificial intelligence, and quantum linear algebra. Quantum combinatorial optimization aims to find the optimal solution among numerous possible solutions with greater efficiency and accuracy. Currently available quantum optimization algorithms include quantum approximate optimization, variational quantum eigensolvers, and Grover search algorithms. However, due to limitations in NISQ hardware and software, these algorithms cannot directly solve large-scale optimization problems. A more suitable solution approach for current development is hybrid quantum-classical computing, in which the global optimization problem is decomposed into subproblems solved interactively by classical and quantum computers.

[0005] Most power system post-disaster recovery optimization problems fall under mixed-integer linear programming models. To cleverly leverage the advantages of quantum computing, two technical approaches are proposed. The first is to scale up quantum hardware. Achieving this goal requires significant technological and engineering advances in quantum hardware. Although quantum annealing computers have been directly applied to commercial optimization problems, they still face scalability limitations. Another approach is to cleverly use decomposition and coordination algorithms. In recent years, domestic and foreign scholars have studied the application of decomposition algorithms in power systems, such as the alternating direction multiplier method, the proxy Lagrangian relaxation method, and the Benders decomposition algorithm. Existing scholars have made preliminary attempts to explore practical problems such as distributed unit combinations in power systems, day-ahead dispatch optimization, grid reconfiguration, and energy system energy management. Theoretical and experimental verification demonstrates the potential advantages of quantum computing in accelerating computing, which holds great promise and warrants further exploration.

[0006] However, in the above-mentioned existing studies, the use of the alternating direction multiplier method to solve the original optimization problem will not converge, and the Benders decomposition algorithm and the proxy Lagrangian relaxation method will have slow convergence speeds to varying degrees, which in turn affects the efficiency of solving the power system post-disaster recovery optimization problem. In addition, the existing quantum computing-related research in the power field is not yet detailed in the consideration of unit modeling and power line transmission constraints, and the scalability and adaptability of the model need to be adjusted.

[0007] Therefore, a high-efficiency, low-difficulty, and good-convergence solution method is needed to solve the above technical problems. Summary of the Invention

[0008] The technical solution adopted by the present invention to solve the technical problem is: a parallel accelerated solution method for power system post-disaster recovery based on quantum computing, comprising the following steps:

[0009] Step 1: Establish a basic model for post-disaster recovery of the power system. The basic model includes: post-disaster recovery objective function, system balance and load constraint model, power transmission constraint model, and generator set operation constraint model;

[0010] Step 2: Based on the quantum agent Lagrangian relaxation algorithm, the original optimization problem is decomposed into various node sub-problems. In each node sub-problem, the discrete optimization part is first constructed into a quantized model for solution, and then the discrete variable solution value is passed to the continuous optimization part for solution;

[0011] Step 3: Use the quantum proxy Lagrangian relaxation algorithm to solve the post-disaster recovery model. During the iteration process, continuously update the Lagrangian multiplier, penalty factor, and intermediate variable optimization results of discrete optimization and continuous optimization sub-problems. Under the interactive calculation of quantum computing and classical computing, the optimal solution is achieved, and the post-disaster recovery plan is accelerated to guide the actual safe operation of the power system.

[0012] Preferably, in the step 1, the post-disaster recovery objective function is:

[0013]

[0014] In formula (1), denotes a set of time periods, u g,t denotes the start-stop 0 / 1 state variable of the gth unit at time t, denotes the active power output of the gth unit at time t, denotes the active load shedding of the ith node at time t, denotes the unit operation cost of the unit g at time t, denotes the unit start-stop cost of the unit g at time t, denotes the unit cost of load shedding of the node i at time t, denotes a set of nodes of the power system, denotes a set of generator units contained under the node i; the post-disaster recovery objective function specifically includes three parts: the first term is the load shedding cost, the second term is the generator unit operation cost, and the third term is the generator unit start-stop cost, and the model takes the minimum sum of the three costs as the target.

[0015] Preferably, in the step 1, the system balance and load constraint model is:

[0016]

[0017]

[0018] In formula (2), formula (3), denotes the active load of the ith node at time t; formula (2) is a system power balance model, and formula (3) is a node load shedding model.

[0019] Preferably, in the step 1, the power power transmission constraint model is:

[0020]

[0021]

[0022]

[0023]

[0024] In formula (4)-(7), P ij,t and denotes the active power passing through the branch ij and the maximum power transmission, M denotes a large enough constant, X ij denotes the reactance of the branch ij, and θi ,θ i,min ,θ i,max Represents the phase angle amplitude, minimum phase angle value, maximum phase angle value, z of the i-th node respectively. ij,t represents the operating status of the transmission branch ij at time t, Indicates that node j is a node in the upstream node set with node i as its node; Equation (4) is the upper and lower limit constraint model of power line power transmission, Equation (5) is the power flow constraint model of line ij, Equation (6) is the upper and lower limit constraint of node phase angle, and Equation (7) is the node power balance constraint.

[0025] Preferably, the generator set operation constraint model is:

[0026]

[0027]

[0028]

[0029]

[0030]

[0031]

[0032]

[0033]

[0034]

[0035] In formula (8) to formula (16), and Respectively represent the maximum and minimum values ​​of the unit's g output, RU g ,RD g They represent the ramp rate of power increase / decrease, RSU g ,RSD g They represent the ramp rate and y of unit g when starting / stopping respectively. g,t ,z g,t They represent the start / stop action variables of the g-th unit at time t, y g,max ,z g,max Respectively represent the maximum number of start / stop actions of the g-th unit, T g,on and T g,offThey represent the minimum duration of startup and shutdown respectively; Equation (8) is the upper and lower limit constraint model of the generator set output, Equations (9)-(10) are the generator set climbing constraint models, and Equations (11)-(16) are the generator set operating state constraint models.

[0036] Preferably, in step 2, the original optimization problem is decomposed into node sub-problems as follows:

[0037] The original optimization problem will be decomposed into the following sub-problems represented by nodes:

[0038]

[0039]

[0040] In formula (17) and formula (18), represents the Lagrange multiplier corresponding to time t at the kth iteration, q t represents the continuous variable added by the relaxation term at time t, c k and c P They represent the penalty term coefficients respectively; the relaxation term in formula (18)

[0041] More preferably, in step 2, the discrete optimization part is constructed into a quantized model for solution as follows:

[0042] The constrained optimization model is reconstructed into a binary unconstrained optimization model, and the constraints are converted into multiple penalty terms in the objective function; the details are as follows:

[0043]

[0044]

[0045]

[0046]

[0047]

[0048]

[0049]

[0050]

[0051]

[0052]

[0053]

[0054]

[0055]

[0056] In formula (19) to formula (31), H 2,i,t ,H 3,i,t ,H 4,i,t ,H 5,i,t The energy functions of the upper and lower limit constraint model of the generator output and the generator climbing constraint model, H 6,i,t ~H 9,i,t ,H 10,i ,H 11,i They represent the energy function of the generator set operation constraint model, P2~P 11 Represents the corresponding penalty term coefficient, δ k,g ,k=1,...,4 represents the relaxation term coefficient, Represents increasing state values ​​of discrete variables.

[0057] More preferably, in step 2, the discrete variable solution value is passed to the continuous optimization part for solution as follows:

[0058] After the discrete problem is reconstructed, the number of qubits is calculated and mapped to the Q matrix, the Q matrix is ​​deployed in a quantum computer for solution, the evolution results are observed, and the Hamiltonian results are analyzed and output;

[0059] The processing of continuous optimization problems includes: after obtaining the solution of the discrete optimization problem, passing it as known parameters to the continuous optimization problem, and solving the continuous optimization problem on a classical computer.

[0060] Preferably, the step 3 specifically includes:

[0061] After all subproblems have been solved in the kth iteration, it is determined whether the convergence conditions are met. If the convergence conditions are met, the iteration ends and the optimal result is output. If the convergence conditions are not met, after checking whether the optimality conditions are met, the Lagrange multiplier is updated according to formula (32). If at least one subproblem meets the optimality conditions in one iteration, the iteration step size and penalty coefficient are updated according to formulas (33) to (35) based on the intermediate variable results of the two adjacent kth iterations. If the penalty coefficient is too large in the later stage, it is updated according to formula (36).

[0062]

[0063]

[0064]

[0065] ck+1 =c k β,β>1 (71)

[0066] c k+1 =c k β -1 ,β>1 (72)

[0067] In formula (33)-(36), α k represents the step size parameter, and β represents a preset constant parameter.

[0068] More optimally, the output includes generator set operating status, generator set output scheduling, node load shedding, and line flow data during the post-disaster recovery process. This output provides guidance to grid dispatchers, enabling them to quickly formulate a post-disaster recovery plan, effectively accelerating the recovery process through quantum computing.

[0069] The beneficial effects of the present invention are:

[0070] The present invention proposes a quantum proxy Lagrangian relaxation algorithm to solve the model, decomposing the problem into multiple node-centered sub-problems, and dividing the sub-problems into discrete parts and continuous parts for solution, thereby reducing the difficulty of solution. At the same time, the algorithm can achieve the effects of parallel computing and quantum embedded accelerated computing, without the need for complete optimization of the relaxation problem, and can ensure better convergence. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 This is a flow chart of a parallel accelerated solution method for post-disaster recovery of power systems based on quantum computing according to the present invention;

[0072] Figure 2 It is a practical application and implementation schematic diagram of the present invention. DETAILED DESCRIPTION

[0073] The following will provide a clear and complete description of the relevant technologies in the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0074] refer to Figures 1-2 The present invention provides a parallel accelerated solution method for post-disaster recovery of power systems based on quantum computing. The specific process includes steps 1 to 3. The flow chart is as follows: Figure 1 shown.

[0075] Step 1: Establish a basic model for post-disaster recovery of the power system, including: post-disaster recovery objective function, system balance and load constraint model, power transmission constraint model, generator set operation constraint model, as follows:

[0076] The post-disaster recovery objective function is shown in equation (1), which specifically includes three parts: the first term is the load shedding cost, the second term is the generator set operation cost, and the third term is the generator set start-stop cost. The model aims to minimize the sum of the three costs.

[0077]

[0078] In the formula, is a set of time periods, u g,t is a 0 / 1 state variable representing the start-stop of the gth unit at time t, represents the active output power of the gth unit at time t, is the active load shedding of the ith node at time t; represents the unit operation cost of unit g at time t; represents the unit start-stop cost of unit g at time t; represents the unit load shedding cost of node i at time t; is a set of power system nodes; is a set of generator units under node i.

[0079] The system balance and load constraint model is as follows: equation (2) is the system power balance model, and equation (3) is the node load shedding model.

[0080]

[0081]

[0082] In the formula, is the active load of the ith node at time t.

[0083] The power transmission constraint model is as follows: equation (4) is the upper and lower limit constraint model of power line power transmission, equation (5) is the line ij power flow constraint model, equation (6) is the node phase angle upper and lower limit constraint, and equation (7) is the node power balance constraint.

[0084]

[0085]

[0086]

[0087]

[0088] In the formula, Pij,t and is the active power and maximum transmission power passing through branch ij; M is a sufficiently large constant; X ij is the reactance of branch ij; θ i ,θ i,min ,θ i,max are the phase angle amplitude, minimum phase angle value, and maximum phase angle value of the i-th node respectively. ij,t represents the operating status of the transmission branch ij at time t, Indicates that node j is a node in the set of nodes with node i as its upstream node.

[0089] The generator set operation constraint model is as follows: Equation (8) is the upper and lower limit constraint model of the generator set output, Equations (9)-(10) are the generator set climbing constraint models, and Equations (11)-(16) are the generator set operation state constraint models.

[0090]

[0091]

[0092]

[0093]

[0094]

[0095]

[0096]

[0097]

[0098]

[0099] Where, and are the maximum and minimum values ​​of the unit g output, RU g ,RD g RSU is the ramp rate of power increase / decrease during normal operation of generator set g. g ,RSD g Ramp rate when starting / stopping unit g, y g,t ,z g,t is the variable representing the start / stop action of the g-th unit at time t, y g,max ,z g,max It represents the maximum number of start / stop actions of the g-th unit, T g,on and T g,off Respectively represent the minimum duration of startup and shutdown.

[0100] Step 2: According to the quantum agent Lagrangian relaxation algorithm, the original optimization problem is decomposed into various node sub-problems. In each node sub-problem, the discrete optimization part is first constructed into a quantized model for solution, and then the discrete variable solution value is passed to the continuous optimization part for solution.

[0101] The original problem will be decomposed into the following sub-problems represented by nodes:

[0102]

[0103] Where, is the Lagrange multiplier corresponding to time t in the kth iteration, q t is the continuous variable added as relaxation term at time t. k and c P is the penalty coefficient.

[0104] Relaxation term It can be written as follows,

[0105]

[0106] The above node sub-problems are further divided into discrete optimization problems and continuous optimization problems. The treatment for discrete problems is to reconstruct the constrained optimization model into a binary unconstrained optimization (QUBO) model and transform the constraints into multiple penalty terms in the objective function. The details are as follows:

[0107]

[0108]

[0109]

[0110]

[0111]

[0112]

[0113]

[0114]

[0115]

[0116]

[0117]

[0118]

[0119]

[0120] Where H 2,i,t ,H 3,i,t ,H 4,i,t ,H 5,i,t is the energy function of the generator set output upper and lower limit constraint model (8)-(10) and the generator set ramp constraint model, H 6,i,t ~H 9,i,t ,H 10,i ,H 11,i is the energy function of the generator set operation constraint model, P2~P 11 is the corresponding penalty term coefficient, δ k,g ,k=1,...,4 is the relaxation term coefficient, is the increasing discrete variable state value.

[0121] After the discrete problem is reconstructed, the number of quantum bits is calculated and mapped to the Q matrix, the Q matrix is ​​deployed in the quantum computer for solution, the evolution results are observed, and the Hamiltonian results are analyzed and output.

[0122] The approach to continuous optimization problems is: after obtaining the solution to the discrete optimization problem, it is passed as a known parameter to the continuous optimization problem, and the continuous optimization problem is solved on a classical computer.

[0123] Step 3: Use the quantum proxy Lagrangian relaxation algorithm to solve the post-disaster recovery model, continuously update parameters such as the Lagrangian multiplier and penalty factor during the iteration process, achieve the optimal solution through the interactive calculation of quantum computing and classical computing, and accelerate the acquisition of post-disaster recovery plans to guide the actual safe operation of the power system.

[0124] Specifically, at the kth iteration, after all subproblems have been solved, the convergence condition is determined to be met. If so, the iteration ends and the optimal result is output. If the convergence condition is not met, after verifying whether the optimality condition is met, the Lagrange multiplier is updated according to Equation (32). If at least one subproblem in an iteration meets the optimality condition, the iteration step size and penalty coefficient are updated according to (33)-(35) based on the intermediate variable results of the two adjacent kth iterations. If the penalty coefficient is too large in the later stage, it is updated according to Equation (36).

[0125]

[0126]

[0127]

[0128] c k+1 =c kβ,β>1 (107)

[0129] c k+1 =c k β -1 ,β>1 (108)

[0130] Where, α k is the step size parameter, and β is a preset constant parameter.

[0131] The flow chart is as attached Figure 1 The output of the optimal results includes: generator unit operating status during the post-disaster recovery process, generator unit output scheduling results, node load shedding results, and line power flow results. This optimal output provides guidance to grid dispatchers, allowing them to quickly formulate post-disaster recovery plans, realizing the benefits of quantum computing in accelerating the post-disaster recovery process.

[0132] Example

[0133] This embodiment takes wind disaster as an example. Figure 2 As shown, after the model and algorithm are built, in order to realize the interaction between quantum computing and classical computing to accelerate the post-disaster recovery of the power system, the actual application process and schematic scheme are as follows:

[0134] Step 1: After a typhoon disaster occurs, the initial fault state parameters are confirmed immediately as input data for the above model. Each power node confirms the connection status between the classical computer and the quantum computer.

[0135] Step 2: Each power node interacts with the power dispatching and control center to understand the fault repair process at each moment, and forms a sub-optimization problem based on the power node. After the discrete problem in the sub-optimization problem is converted into a quantum model, it is first transmitted from a personal PC to the quantum cloud platform and the actual quantum computer is called to return the calculation results. The intermediate results are then passed to the continuous problem for further optimization and solution.

[0136] Step 3: Perform iterative solution according to the quantum agent Lagrangian relaxation algorithm. After the convergence conditions are met, output the generator unit operating status results, generator unit output dispatch results, each node load shedding results, and line flow results during the post-disaster recovery process. After confirming that the optimization results are safe and reasonable, issue actual power dispatch commands to control the operation of the power system and support rapid post-disaster recovery.

[0137] Finally, after the power system's post-disaster recovery is complete, the computational efficiency and performance of quantum computing will be evaluated, and the post-disaster power dispatch experience and plan evaluation work will be summarized. For discrete problems, this invention reconstructs the constrained optimization model into a binary unconstrained optimization (QUBO) model, providing a reference for converting post-disaster recovery problem models into quantum models. The quantum proxy Lagrangian relaxation algorithm process of this invention enables parallel computation of subproblems and accelerated solution through quantum computing.

[0138] In summary, the present invention proposes a quantum proxy Lagrangian relaxation algorithm to solve the model, decomposes the problem into multiple node-centered sub-problems, and divides the sub-problems into discrete parts and continuous parts for solution, which reduces the difficulty of solution. At the same time, the algorithm can achieve the effects of parallel computing and quantum embedded accelerated computing, without the need for complete optimization of the relaxation problem, and can ensure better convergence. Therefore, the present invention has broad application prospects in the fields of rapid power supply restoration after disaster damage to the power system, reasonable and efficient dispatching control and coordinated optimization.

[0139] It should be emphasized that the above are only preferred embodiments of the present invention and do not limit the present invention in any form. Any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A parallel accelerated solution method for post-disaster recovery of power systems based on quantum computing, characterized in that: The following steps are involved: Step 1: Establish a basic model for post-disaster recovery of the power system, which includes: a post-disaster recovery objective function, a system balance and load constraint model, a power transmission constraint model, and a generator set operation constraint model; Step 2: Based on the quantum agent Lagrangian relaxation algorithm, the original optimization problem is decomposed into various node sub-problems. In each node sub-problem, the discrete optimization part is first constructed into a quantized model for solution, and then the discrete variable solution value is passed to the continuous optimization part for solution; Step 3: Solve the post-disaster recovery model using the quantum proxy Lagrangian relaxation algorithm. During the iteration process, the Lagrangian multiplier, penalty factor, and intermediate variable optimization results of the discrete and continuous optimization subproblems are continuously updated. The optimal solution is achieved through the interaction of quantum and classical computing, accelerating the development of post-disaster recovery plans to guide the actual safe operation of the power system. In step 1, the post-disaster recovery objective function is: In formula (1), T represents the time period set, u g,t represents the start / stop 0 / 1 state variable of the g-th unit at time t, represents the active output power of the g-th unit at time t, represents the active load shedding of the i-th node at time t, represents the unit operating cost of unit g at time t, represents the unit start-up and shutdown cost of unit g at time t, represents the unit cost of load shedding of node i at time t, B represents the set of power system nodes, G i represents the set of generators contained under node i; In step 2, the discrete variable solution value is passed to the continuous optimization part for solution as follows: After the discrete problem is reconstructed, the number of qubits is calculated and mapped to the Q matrix, the Q matrix is ​​deployed in a quantum computer for solution, the evolution results are observed, and the Hamiltonian results are analyzed and output; The processing of continuous optimization problems includes: after obtaining the solution of the discrete optimization problem, passing it as known parameters to the continuous optimization problem, and solving the continuous optimization problem on a classical computer.

2. A parallel accelerated solution method for power system post-disaster recovery based on quantum computing according to claim 1, characterized in that: In step 1, the system balance and load constraint model is: In formula (2) and formula (3), represents the active load of the i-th node at time t; Formula (2) is the system power balance model, and Formula (3) is the node load shedding model.

3. The method for parallel acceleration of power system post-disaster recovery based on quantum computing according to claim 2 is characterized in that: In step 1, the electric power transmission constraint model is: In formulas (4)-(7), P ij,t and represents the active power and maximum transmission power passing through branch ij, M represents a sufficiently large constant, X ij represents the reactance of branch ij, θ i ,θ i,min ,θ i,max Represents the phase angle amplitude, minimum phase angle value, maximum phase angle value, z of the i-th node respectively. ij,t represents the operating state of the transmission branch ij at time t, j∈L(i,j) represents the node j in the set of upstream nodes with node i as the node; Formula (4) is the upper and lower limit constraint model of power transmission of power line, Formula (5) is the power flow constraint model of line ij, Formula (6) is the upper and lower limit constraint of node phase angle, and Formula (7) is the node power balance constraint.

4. The method for parallel acceleration of power system post-disaster recovery based on quantum computing according to claim 3 is characterized in that: The generator set operation constraint model is: In formulas (8) to (16), and Respectively represent the maximum and minimum values ​​of the unit's g output, RU g ,RD g They represent the ramp rate of power increase / decrease during normal operation of generator set g, RSU g ,RSD g They represent the ramp rate and y of the unit when starting / stopping. g,t ,z g,t They represent the start / stop action variables of the g-th unit at time t, y g,max ,z g,max Respectively represent the maximum number of start / stop actions of the g-th unit, T g,on and T g,off They represent the minimum duration of startup and shutdown respectively; Equation (8) is the upper and lower limit constraint model of the generator set output, Equations (9)-(10) are the generator set climbing constraint models, and Equations (11)-(16) are the generator set operating state constraint models.

5. The method for parallel acceleration of power system post-disaster recovery based on quantum computing according to claim 4 is characterized in that: In step 2, the original optimization problem is decomposed into various node sub-problems as follows: The original optimization problem will be decomposed into the following sub-problems represented by nodes: In formula (17) and formula (18), represents the Lagrange multiplier corresponding to time t at the kth iteration, q t represents the continuous variable added by the relaxation term at time t, c k and c P They represent the penalty term coefficients respectively; the relaxation term in formula (18) 6. A parallel accelerated solution method for power system post-disaster recovery based on quantum computing according to claim 5, characterized in that: In step 2, the discrete optimization part is constructed into a quantized model for solution as follows: The constrained optimization model is reconstructed into a binary unconstrained optimization model, and the constraints are converted into multiple penalty terms in the objective function; the details are as follows: In formula (19) to formula (31), H 2,i,t ,H 3,i,t ,H 4,i,t ,H 5,i,t The energy functions of the generator set output upper and lower limit constraint model and the generator set ramp constraint model, H 6,i,t ~H 9,i,t ,H 10,i ,H 11,i They represent the energy function of the generator set operation constraint model, P2~P 11 Represents the corresponding penalty coefficient, δ k,g ,k=1,...,4 represents the relaxation term coefficient, Represents increasing state values ​​of discrete variables.

7. The method for parallel acceleration of power system post-disaster recovery based on quantum computing according to claim 6 is characterized in that: The step 3 specifically includes: After all subproblems have been solved in the kth iteration, it is determined whether the convergence conditions are met. If the convergence conditions are met, the iteration ends and the optimal result is output. If the convergence conditions are not met, after checking whether the optimality conditions are met, the Lagrange multiplier is updated according to formula (32). If at least one subproblem meets the optimality conditions in one iteration, the iteration step size and penalty coefficient are updated according to formulas (33) to (35) based on the intermediate variable results of the two adjacent kth iterations. If the penalty coefficient is too large in the later stage, it is updated according to formula (36). c k+1 =c k ·β,β>1 (35) c k+1 =c k ·b -1 ,β>1 (36) In formula (33)-(36), α k represents the step size parameter, and β represents a preset constant parameter.

8. The method for parallel acceleration of power system post-disaster recovery based on quantum computing according to claim 7 is characterized in that: The output optimal results include: generator set operating status results during the post-disaster recovery process, generator set output scheduling results, load shedding results at each node, and line flow results.

Images