Multi-microgrid active power distribution network collaborative optimization method based on light quantum acceleration

Abstract

The invention belongs to the technical field of power system coordinated optimization, and relates to a light quantum acceleration-based multi-microgrid active power distribution network coordinated optimization method, which comprises the following steps of: 1, constructing a light quantum acceleration-based multi-microgrid active power distribution network coordinated optimization framework according to an optimization target and an actual condition; 2, constructing a light quantum acceleration collaborative optimization model for the multi-microgrid active power distribution network; 3, discretizing the multi-microgrid active power distribution network collaborative operation model, and then establishing a quantum interpretable model on the basis; 4, performing problem decoupling aiming at the discretization model, using transmission line power deviation as a coupling relation of an upper main body and a lower main body, and then providing a precision adjustment method for a precision problem of a light quantum computer so as to identify an adaptive solver of the problem; according to the method, the modeling flexibility of mixed integer optimization is reserved, and the potential parallelism and acceleration capability of quantum calculation in processing large-scale combination optimization problems are fully exerted.

Description

Technical Field

[0001] The present invention belongs to the technical field of coordinated optimization of power systems and relates to a multi-microgrid active distribution network coordinated optimization method based on photon acceleration. Background Art

[0002] With the large-scale integration of renewable energy into power systems, constructing multi-microgrid active distribution systems (MMADS) consisting of multiple microgrids has become an important path to improving system flexibility and resilience. To achieve coordinated optimization between multiple microgrids and with the distribution system, academic research has extensively studied distributed coordinated optimization methods, such as the alternating direction method of multipliers (ADMM), the consensus algorithm (CA), and the target cascade algorithm (ATC). The ATC algorithm, due to its excellent convergence, simplified parameter settings, and parallel solution capabilities, has been widely used in the optimization of multi-level, multi-agent systems. Previous studies have constructed coordinated optimization models between microgrids and active distribution systems based on the ATC framework, achieving good computational efficiency in various planning and operation scenarios. As system scale expands and uncertainties increase, traditional algorithms face challenges in solving complex mixed-integer problems, such as heavy computational burdens and low solution efficiency. To this end, recent efforts have explored the parallel acceleration characteristics of quantum computing and its application in the field of power system optimization. Hybrid quantum-classical (HQC) algorithms, combining the strengths of quantum and classical optimization, have shown promising application prospects in a variety of power optimization problems, such as unit commitment and distribution network planning. By decomposing and coordinating large-scale problems within the HQC framework, they offer a viable path for the practical application of quantum computing in power systems.

[0003] However, existing MMADS coordinated optimization methods generally rely on classical optimization algorithms. When faced with large-scale, strongly coupled mixed-integer linear programming problems, they are prone to problems such as excessive computational time, local optima, and convergence failure, making them difficult to meet the demand for efficient decision-making in practical engineering. Although some research has attempted to incorporate quantum computing technology, the limited number of qubits and high noise levels in current quantum hardware have limited the ability to solve only small-scale problems, making it difficult to directly apply to complex power system optimization.

[0004] Therefore, a method that is compatible with the current medium-sized optical quantum hardware with medium noise and maintains efficient computing performance is needed to solve the above technical problems. Summary of the Invention

[0005] The technical solution adopted by the present invention to solve the technical problem is: a multi-microgrid active distribution network collaborative optimization method based on optical quantum acceleration, comprising the following steps:

[0006] Step 1: Based on the optimization objectives and actual conditions, a multi-microgrid active distribution network collaborative optimization framework based on photonic quantum acceleration is constructed to clarify the tasks in the photonic quantum computing environment.

[0007] Step 2: Construct a photonic quantum accelerated collaborative optimization model for multi-microgrid active distribution networks. The multi-layer modeling includes the objective functions, resource operation constraints, power balance constraints, and safe operation constraints of the active distribution network and each microgrid (MG).

[0008] Step 3: Discretize and simplify the multi-microgrid active distribution network collaborative operation model to reduce the complexity of the model and the number of quantum bits used. Then, build a quantum interpretable model based on this model to match the processing scale and format of current optical quantum devices.

[0009] Step 4: Decouple the problem for the discretized model, use the transmission line power deviation as the coupling relationship between the upper and lower entities, and then propose an accuracy adjustment method to address the accuracy problem of the optical quantum computer, so as to identify the adaptive solver for the problem; use the quantum target cascade analysis algorithm to build a process based on the basic steps of the target cascade algorithm, improve it, and embed the quantum algorithm.

[0010] Preferably, in step 1, the multi-microgrid active distribution network collaborative optimization framework based on photon acceleration collects the predicted output data of each load in real time through the measuring device and the information system, and synchronously transmits the predicted output data to the control center of the distribution network (ADS) and the multi-microgrid (MMG); the control center conducts multi-level collaborative optimization based on multi-agent characteristics and generates adjustment instructions, and the control center formulates the optimal operation strategy of the distribution network within the scope allowed by the operation constraints.

[0011] More preferably, after obtaining the forecast information, the control center of the multi-microgrid independently completes the internal resource optimization operation based on its own optimization goals and interaction deviations, implements a two-way coordinated control mechanism between the distribution network and the multi-microgrid, and promotes the operation of the distributed energy system.

[0012] Preferably, in step 2, the operation model of the active distribution network includes:

[0013]

[0014] In formulas (1) to (3), F ADS Indicates the improvement of system energy allocation efficiency. Indicates the operating energy consumption of MT, It represents the power balance adjustment item caused by the energy interaction with multiple microgrids; T represents the operation time span; Indicates the power of MT within the ADS range; n indicates the total number of MTs; c gas and L MT Respectively represent the natural gas characteristic parameters related to the MT operating status; η MT It represents the power generation efficiency of MT; and m represent the tie-line power and the total number of tie-lines; λ(t) represents the external power guidance signal set based on the time series; the model constraints of the active distribution network include:

[0015]

[0016] In formula (4) and formula (5), A∈{MTs,Tie-lines}, Indicates the load forecast value in ADS, and Represent the upper and lower limits of power respectively, Indicates the power of A within the ADS range.

[0017] Preferably, in step 2, the operation model of the microgrid includes:

[0018]

[0019] In formulas (6) to (11), F MG,j It means optimizing the operating efficiency of multi-source systems. Indicates the energy conversion consumption of MT, represents the electric energy exchange term with the distribution network, represents the charge and discharge loss of ESS, Indicates the load shedding regulation response impact, represents the control influence of DR adjustable load; c gas and L MT Respectively represent the natural gas characteristic parameters related to the MT operating status; η MT represents the power generation efficiency of MT, λ(t) represents the external power guidance signal set based on the time series, and Represent the power of MT and tie line in MG respectively, C ESS and Corresponding to the unit operation coefficient and power of ESS in MG, C load,s and Indicates load shedding coefficient and load shedding power, C DR and represent the operating coefficient and power of DR respectively;

[0020] The model constraints of the microgrid include:

[0021]

[0022] In formulas (12) to (18), Φ∈{ESS,DR,MTs,Tie-lines}, and Respectively represent the upper and lower limits of power within the MG, represents the power within Φ; E0, and Respectively represent the capacity, initial capacity, upper and lower limits of ESS; D DR and D load,s,max represent the total demand response load and maximum load shedding load, respectively.

[0023] Preferably, the step 3 specifically includes the following sub-steps:

[0024] Step 3-1: Model discretization and simplification. Given the accuracy limitations of noisy medium-scale quantum hardware, a partial linear discretization method is used to balance model fidelity and computational feasibility.

[0025] Step 3-2: Establishment of quantum interpretable model and conversion of scheduling discrete model based on ADS.

[0026] Preferably, the step 4 specifically includes the following sub-steps:

[0027] Step 4-1: Model decoupling: ADS and MMG are coupled through a tie line, connecting their respective optimization processes and hindering independent optimization;

[0028] Step 4-2: Accuracy adjustment: control the model accuracy to ensure computational feasibility;

[0029] Step 4-3, quantum target cascade analysis, integrates quantum computing into the ATC framework, fully utilizes the advantages of parameter tuning efficiency and parallel computing, and enhances the solution process of multi-microgrid active distribution network collaborative optimization.

[0030] More preferably, the step 4-2 specifically includes the following sub-steps:

[0031] Step 4-2-1: Calculate the maximum and minimum absolute values ​​of all non-zero elements in the QUBO matrix:

[0032] q abs,max =max(|q ij ||q ij ≠0) (45)

[0033] q abs,min =min(|q ij ||q ij ≠0) (46)

[0034] If q abs,max / q abs,min <τ extreme is violated, the matrix is ​​considered infeasible and the procedure terminates, where τ extreme represents the modeling limit;

[0035] Step 4-2-2: Transform the QUBO matrix into the equivalent Ising matrix M Ising ;make And the frequency function f(m ij ) is defined as m ij Then, the quantization resolution λ>0 is introduced and determined as follows:

[0036] λ∈{kΔt|k∈Z +} (47)

[0037]

[0038] Among them, Δt represents the minimum modeling step, λ is an integer multiple of Δt, Q up and Q down They are the upper and lower limits of the matrix processing of CPQC;

[0039] Step 4-2-3: Make Set the frequency threshold τ f and the ratio threshold τ p ,If Equation (49) is violated, the model is considered infeasible, and the process terminates;

[0040]

[0041] Step 4-2-4: After excluding incompatible elements and zero elements, the absolute value Ising matrix (|M Ising |) in [m abs,max , m abs,max ] is bounded, for M Ising The corresponding incompatible elements are smoothed:

[0042] m ij =sign(m ij )m abs,max ,λ|m ij |>Q up (50)

[0043] m ij =sign(m ij )m abs,min ,λ|m ij |<Q down (51)

[0044] Where, sign(·) is the sign function;

[0045] Step 4-2-5: By introducing a scaling factor (α), M Ising M' Ising =αM Ising Transform, and we get:

[0046] m′ abs,max =Q up or m′ abs,min =Q dowm (52)

[0047] Step 4-2-6: Evaluate whether the matrix meets the required accuracy criteria:

[0048] mod(m′ ij )∈[ε mod,min ,ε mod,max ] (53)

[0049] Where mod(·) represents the rounding function, ε mod,min and ε mod,max They represent the upper and lower bounds of rounding respectively; if the constraints are met, the Ising matrix can be imported into the optical quantum computer; otherwise, its accuracy is insufficient.

[0050] More preferably, the step 4-3 specifically includes the following sub-steps:

[0051] Step 4-3-1: Initialize the operating parameters of each system component and the maximum number of iterations k max , the initial value w of the penalty function multiplier j and y j , precision adjustment parameter ε extreme , ε mod,min , ε mod,max and coupling variables and Set the number of iterations k = 1 and start the QATC algorithm;

[0052] Step 4-3-2: Based on the coupled variable data, the distribution network control center constructs the QUBO problem with the objective function and the deviation from the MMG as the target.

[0053] Step 4-3-3: To ensure that the accuracy requirements are met, the accuracy adjustment strategy is implemented; if the accuracy requirements are not met, an optimization solver is first used to obtain a reliable solution; if the accuracy requirements are met, a quantum computer is used for the solution;

[0054] Step 4-3-4: The multi-microgrid control center establishes the QUBO model and implements the precision adjustment strategy to guide the selection of the appropriate solution method, and accelerates the processing speed by parallel computing across MGs;

[0055] Step 4-3-5: Evaluate the convergence criteria; Convergence Criterion I ensures that the power flow deviation on the tie line between the two entities meets the specified requirements, and Convergence Criterion II ensures that the difference in the MMADS objective function between two consecutive iterations meets the specified conditions. Convergence Criterion I is:

[0056]

[0057] Convergence criterion II is:

[0058]

[0059] In formulas (54) and (55), represents the overall optimization objective of MMADS, ε1 and ε2 represent the convergence limits;

[0060] Step 4-3-6: When all conditions in step 4-3-5 are met at the same time, output the result; otherwise, update the penalty function multiplier, set k = k + 1, and return to step 4-3-2; the updating principle of the penalty function multiplier is:

[0061]

[0062] In formula (56), β represents the update coefficient of the quadratic term of the penalty function multiplier.

[0063] The beneficial effects of the present invention are:

[0064] 1. The present invention constructs a unified optical quantum accelerated multi-microgrid active distribution network collaborative optimization model, and proposes a hierarchical modeling and decoupling optimization strategy for optical quantum computing architecture to address the problems of low computational efficiency and poor resource coordination ability of traditional optimization methods when dealing with large-scale discrete decision-making problems. Specifically, under a unified modeling framework, this patent collaboratively describes the objective functions and constraint relationships of the distribution network and multiple microgrids, and divides them into multiple sub-problems with decoupling characteristics based on the problem structure. For sub-problems that meet the accuracy constraints, they are converted into quadratic unconstrained optimization (QUBO) form and deployed on the optical quantum computing platform for parallel accelerated solution; for sub-problems that do not meet the embedding conditions, solvers such as GUROBI or CPLEX are used to efficiently process them in a classical environment. This method not only retains the modeling flexibility of mixed integer optimization, but also gives full play to the potential parallelism and acceleration capabilities of quantum computing in dealing with large-scale combinatorial optimization problems.

[0065] 2. The "quantum-classical collaborative optimization" mechanism proposed in this invention can effectively alleviate the exponential growth of computational complexity caused by the growth of problem scale. At the same time, in the context of the continuous improvement of quantum computing resources, this method has stronger scalability and foresight, and provides a feasible solution with engineering application prospects for achieving efficient coordination and optimization of distributed resources in new power systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 This is a schematic diagram of a photon acceleration framework of a multi-microgrid active distribution network collaborative optimization method based on photon acceleration of the present invention;

[0067] Figure 2 It is a schematic diagram of the method steps of the present invention. DETAILED DESCRIPTION

[0068] 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.

[0069] refer to Figures 1-2 The multi-microgrid active distribution network collaborative optimization method based on optical quantum acceleration in this embodiment includes the following steps:

[0070] Step 1: Based on the optimization objectives and actual conditions, a multi-microgrid active distribution network collaborative optimization framework based on photonic quantum acceleration is constructed to clarify the tasks in the photonic quantum computing environment. (This framework includes the decomposition calculation principle of the hybrid photonic quantum-classical algorithm)

[0071] like Figure 1 As shown in Figure 1, this framework uses measurement devices and an information system to collect real-time forecasted output data for renewable energy and various loads, and simultaneously transmits this data to the control centers of the distribution network (ADS) and multi-microgrid (MMG). After obtaining the forecasted output information for distributed power sources and loads, the control center conducts multi-level collaborative optimization based on multi-agent characteristics. The distribution network control center comprehensively considers local operation and power interaction between multiple microgrids, coordinates resources such as photovoltaic (PV), wind power (WG), energy storage system (ESS), and gas turbine (MT), generates regulation instructions, and formulates the optimal operation strategy for the distribution network within the permitted operating constraints to achieve overall operational efficiency. Simultaneously, after obtaining forecast information, the multi-microgrid control center autonomously optimizes its internal resources based on its own optimization objectives and interaction deviations, thus realizing a two-way collaborative regulation mechanism between the distribution network and multiple microgrids, and promoting the efficient operation of the distributed energy system.

[0072] To fully leverage the potential advantages of quantum computing in the collaborative optimization of multi-microgrid active distribution networks, a distributed solution framework for the target cascade analysis algorithm is proposed. This method structurally decomposes and discretizes the original model, transforming it into a set of mutually coupled hybrid binary subproblems. Subproblems that meet the precision constraints of quantum hardware are deployed in an optical quantum computing environment for parallel optimization and solution; subproblems that do not meet the precision requirements are efficiently supplemented by a classical computing environment. The subproblems are linked through the power coupling relationship of the transmission lines, and multiple rounds of iterative coordinated calculations are performed to obtain the optimal solution that meets the global convergence conditions, thus achieving the efficient integration of quantum computing and classical optimization in the collaborative optimization of multi-microgrid active distribution networks.

[0073] For subproblems that do not meet quantum computing accuracy constraints, mature optimization solvers (such as Gurobi or CPLEX) can be directly invoked to obtain the global optimal solution in a mixed-integer linear programming (MILP) mode. For subproblems that meet accuracy requirements, they must be restructured into a form that can be embedded and deployed on quantum computing hardware, such as the quadratic unconstrained optimization problem (QUBO). During the quantum solution phase, users interact with the coherent optical quantum computing platform through a web user interface (UI). The platform consists of optical and electrical modules that work together to perform tasks such as modulation, interference, and measurement of photons, respectively. The system guides the evolution of the photon Hamiltonian toward its lowest energy state, ultimately producing a stable output represented by phase states (i.e., 0 and π), where the phase 0 and π states correspond to the spin states {-1, +1} of the photon qubit, respectively. This establishes a mapping between the decision variables {0, 1} and the photon qubit, achieving structural integration between the Ising model and the QUBO model, and ultimately obtaining the solution set for the optimization problem.

[0074] Step 2: Construct a photon quantum acceleration collaborative optimization model for multi-microgrid active distribution networks. The multi-layer modeling includes the objective functions, resource operation constraints, power balance constraints, and safe operation constraints of the active distribution network and each microgrid (MG). At the active distribution network level: the objective function is to achieve the improvement of system energy allocation efficiency, including the operating energy consumption of the local gas turbine (MT) and the multi-microgrid energy interaction power balance adjustment items; at the same time, consider power balance constraints and safety operation constraints such as component / line power over-limit. At the microgrid level: the operating efficiency of the multi-source system is the optimization goal, covering the energy conversion consumption of the MT, the power exchange items with the distribution network, the charging and discharging losses of the energy storage (ESS), the load shedding regulation response impact, and the control impact of the DR adjustable load when necessary; and consider power balance constraints, power over-limit constraints, and the operation constraints of various distributed resources.

[0075] (1) Active distribution network operation model

[0076] The optimization goal of active distribution network operation is to improve the energy allocation efficiency of the system (F ADS ), which includes the energy consumption of MT operation and power balance adjustment items caused by energy interaction with multiple microgrids For example (1)-(3):

[0077]

[0078] Where T represents the running time span; Indicates the power of MT within the ADS range; n indicates the total number of MTs; c gas and L MT Respectively represent the natural gas characteristic parameters related to the MT operating status; η MT It represents the power generation efficiency of MT; and m represent the tie line power and the total number of tie lines; λ(t) represents the external power guidance signal set based on the time series;

[0079] Model constraints include power balance constraints, tie-line and MT power limit constraints, such as (4)-(5):

[0080]

[0081] Where A∈{MTs,Tie-lines}, Indicates the load forecast value in ADS, and Indicates the upper and lower power limits respectively.

[0082] (2) Microgrid operation model

[0083] The operation goal in MG is to optimize the operation efficiency of the multi-source system (F MG,j ), which includes the energy conversion consumption of MT Electric energy exchange with the distribution network ESS charging and discharging losses Impact of load shedding regulation response and DR adjustable load control impact As shown in the following formula,

[0084]

[0085] Where, and Respectively represent the power of MT and tie line in MG; C ESS and Corresponding to the unit operation coefficient and power of ESS in MG; C load,s and Indicates load shedding coefficient and load shedding power; CDR and They represent the operating coefficient and power of DR respectively.

[0086] The constraints of the model are shown in (12)-(18), including ESS, DR, MT, tie line operation constraints, power balance constraints and load shedding constraints.

[0087]

[0088]

[0089] Where, Φ∈{ESS,DR,MTs,Tie-lines}, and Represents the upper and lower power limits within the MG; E0, and Represent the capacity, initial capacity, upper and lower limits of ESS respectively; D DR and D load,s,max represent the total demand response load and maximum load shedding load, respectively.

[0090] In fact, this model also supports the consideration of more factors, such as electric vehicles, mobile energy storage or other distributed power sources, etc., which can be achieved by simply replacing the corresponding constraints of this model.

[0091] Step 3: Discretize and simplify the collaborative operation model of the multi-microgrid active distribution network to reduce the complexity of the model and the number of quantum bits used. Then, build a quantum interpretable model based on this to match the current processing scale and format of optical quantum devices.

[0092] Step 3-1: Model discretization and simplification

[0093] Given the accuracy limitations of noisy intermediate-scale (NISQ) quantum hardware, a partial linear discretization method is used to balance model fidelity and computational feasibility. Within the ADS framework, and The discretization of is as follows:

[0094]

[0095]

[0096] Where d(·) represents the discretization process, and It is a binary variable introduced to promote the discretization of variables in ADS. and represents the step size associated with the corresponding binary variable.

[0097] In the MG framework, the variables and The discretization is as follows:

[0098]

[0099] Where, and It is a binary variable introduced to facilitate the discretization of MG internal variables. and represents the step size associated with the corresponding binary variable. Corresponds to the initial capacity after discretization.

[0100] In this stage, a discretized multi-microgrid active distribution network cooperative operation model is established, and the coupling between the ADS and the multi-microgrid (MMG) is realized through the power exchange of the tie line. In order to simplify the model, a redundant constraint identification method is introduced, in which the tie line power limit in the ADS is determined by (5) and (19) as follows:

[0101]

[0102] Redundant constraints are eliminated by comparing lower and upper bounds at each time step, often without the need for slack variables. This allows constraints related to DR, MT, and tie lines to be adjusted or omitted to simplify the model without affecting the optimal solution.

[0103] Step 3-2: Establishing a quantum interpretable model

[0104] In order to express the problem in the QUBO format compatible with quantum computers, the ADS-based scheduling discrete model conversion process is as follows:

[0105]

[0106] Where, and represent the corresponding penalty factors; and corresponds to the Hamiltonian; and represents the remaining time set after eliminating redundant constraints; and is the slack variable; and represents the length of the corresponding slack variable; and Indicates the maximum conservatism of the corresponding constraint.

[0107] Similarly, the quantum interpretable model of MG is defined as follows:

[0108]

[0109] The terms are consistent with the previous explanations. This phase established a quantum interpretable model of MMADS, laying the foundation for the subsequent implementation of the HQC algorithm.

[0110] In step 4, the problem is decoupled for the discretized model, using transmission line power deviation as the coupling relationship between the upper and lower entities. Then, addressing the accuracy issues of optical quantum computers, an effective precision adjustment method is proposed to identify an appropriate solver for the problem. The Quantum Target Cascade Analysis (QATC) algorithm construction process improves upon the basic steps of the target cascade algorithm and embeds a quantum algorithm.

[0111] Step 4-1: Model decoupling

[0112] ADS and MMG are coupled through tie lines, which connect their respective optimization processes and hinder their independent optimization. In order to adapt to the ATC framework, the tie line power is modeled as a virtual load in ADS. Modeled as virtual power in MMG This achieves decoupled optimization and iterates until convergence. An augmented Lagrangian penalty term is introduced into the ADS objective to account for the deviation between virtual load and power. The resulting quantum interpretable model is as follows:

[0113]

[0114] Where w j (t) and y j (t) denote the linear and quadratic coefficients of the penalty function, respectively. The penalty function adjusts the optimization process by introducing linear and quadratic deviations in the objective function.

[0115] Similarly, the resulting MG quantum interpretable model is as follows:

[0116]

[0117] Therefore, equation (42) characterizing the operation of ADS and equation (43) describing the operation of MG are reformulated to be compatible with quantum computing architecture and can be solved in an alternating iterative manner.

[0118] Step 4-2: Accuracy adjustment

[0119] Optical quantum computers only support Ising matrices with 8-bit integers in the range [-127, 128]. Therefore, strict control of model accuracy is required to ensure computational feasibility. To facilitate subsequent discussion, the generalized QUBO matrix corresponding to the quantum interpretable model is shown below:

[0120]

[0121] Step 4-2-1: Calculate the maximum and minimum absolute values ​​of all non-zero elements in the QUBO matrix:

[0122] q abs,max =max(|q ij ||q ij ≠0) (45)

[0123] q abs,min =min(|q ij ||q ij ≠0) (46)

[0124] If q abs,max / q abs,min <τ extreme is violated, the matrix is ​​considered infeasible and the procedure terminates, where τ extreme Represents the modeling limit.

[0125] Step 4-2-2: Transform the QUBO matrix into an equivalent Ising matrix (M Ising ).make And the frequency function f(m ij ) is defined as m ij Then, the quantization resolution λ>0 is introduced and determined as follows:

[0126] λ∈{kΔt|k∈Z +} (47)

[0127]

[0128] Where Δt represents the minimum modeling step size, and λ is usually selected as an integer multiple of Δt and determined by appropriate enumeration. up and Q down They are the upper and lower limits of the matrix processing of CPQC

[0129] Step 4-2-3: Make Set the frequency threshold (τ f ) and the ratio threshold (τ p ), if the following is violated, the model is considered infeasible and the process terminates.

[0130]

[0131] Step 4-2-4: After excluding incompatible elements and zero elements, the absolute value Ising matrix (|M Ising |) in [m abs,max , m abs,max ] is bounded, as defined by equations (50) and (51). IsingThe corresponding incompatible elements are smoothed as follows:

[0132] m ij =sign(m ij )m abs,max ,λ|m ij |>Q up (50)

[0133] m ij =sign(m ij )m abs,min ,λ|m ij |<Q dowm (51)

[0134] Where sign(·) is the sign function.

[0135] Step 4-2-5: By introducing a scaling factor (α), M Ising M' Ising =αM Ising Transformation, we get the following relationship:

[0136] m′ abs,max =Q up or m′ abs,min =Q down (52)

[0137] Step 4-2-6: Evaluate whether the matrix meets the required accuracy criteria as shown below,

[0138] mod(m′ ij )∈[ε mod,min ,ε mod,max ] (53)

[0139] Where mod(·) represents the rounding function, ε mod,min and ε mod,max If the constraints are met, the Ising matrix can be imported into an optical quantum computer; otherwise, its accuracy is insufficient.

[0140] Step 4-3: Quantum target cascade analysis

[0141] Integrating quantum computing into the ATC framework fully leverages its advantages in parameter tuning efficiency and parallel computing, thereby enhancing the solution process for multi-microgrid active distribution network collaborative optimization. The specific implementation steps are as follows:

[0142] Step 4-3-1: Initialize the operating parameters of each system component, the maximum number of iterations (k max ), the initial value of the penalty function multiplier (w j and y j), precision adjustment parameter (ε extreme , ε mod,min , ε mod,max ) and coupling variables and Set the number of iterations k = 1 and start the QATC algorithm.

[0143] Step 4-3-2: Based on the coupled variable data, the distribution network control center constructs the QUBO problem in the form of equation (42) with the objective function and the deviation from the MMG as the target.

[0144] Step 4-3-3: To ensure that the accuracy requirements are met, the accuracy adjustment strategy proposed in the previous section is adopted. If the accuracy requirements are not met, the optimization solver is first used to obtain a reliable solution. Conversely, if the accuracy requirements are met, the quantum computer is used for the solution.

[0145] Step 4-3-4: The multi-microgrid control center establishes the QUBO model based on formula (43), executes the precision adjustment strategy to guide the selection of the appropriate solution method, and speeds up the processing by parallel computing across MGs.

[0146] Step 4-3-5: Evaluate the convergence criteria. Convergence criterion I (defined in formula (54)) ensures that the power flow deviation on the tie line between the two entities meets the specified requirements. Convergence criterion II (defined in formula (55)) ensures that the difference in the MMADS objective function between two consecutive iterations meets the specified conditions.

[0147]

[0148]

[0149] Where, represents the overall optimization objective of MMADS, and ε1 and ε2 represent the convergence limits.

[0150] Step 4-3-6: When all conditions in step 4-3-5 are met at the same time, output the result; otherwise, update the penalty function multiplier, set k = k + 1, and return to step 4-3-2. The updating principle of the penalty function multiplier is as follows:

[0151]

[0152] Where β represents the update coefficient of the quadratic term of the penalty function multiplier.

[0153] In summary, the present invention collaboratively describes the objective functions and constraint relationships of the distribution network and multiple microgrids under a unified modeling framework, and divides the problem into multiple sub-problems with decoupling characteristics based on the problem structure; for sub-problems that meet the accuracy constraints, they are converted into the quadratic unconstrained optimization (QUBO) form and deployed on the optical quantum computing platform for parallel accelerated solution; for sub-problems that do not meet the embedding conditions, solvers such as GUROBI or CPLEX are used to efficiently process them in a classical environment; the present invention not only retains the modeling flexibility of mixed integer optimization, but also gives full play to the potential parallelism and acceleration capabilities of quantum computing in processing large-scale combinatorial optimization problems.

[0154] 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 multi-microgrid active distribution network collaborative optimization method based on photon acceleration, characterized in that: The following steps are involved: Step 1: Based on the optimization objectives and actual conditions, a multi-microgrid active distribution network collaborative optimization framework based on photonic quantum acceleration is constructed to clarify the tasks in the photonic quantum computing environment. Step 2: Construct a photon quantum accelerated collaborative optimization model for the multi-microgrid active distribution network. The multi-layer modeling includes the objective functions of the active distribution network and each microgrid, resource operation constraints, power balance constraints, and safe operation constraints. Step 3: Discretize and simplify the multi-microgrid active distribution network collaborative operation model to reduce the complexity of the model and the number of quantum bits used. Then, build a quantum interpretable model based on this model to match the processing scale and format of current optical quantum devices. Step 4: Decouple the problem for the discretized model, use the transmission line power deviation as the coupling relationship between the upper and lower entities, and then propose an accuracy adjustment method to address the accuracy problem of the optical quantum computer, so as to identify the adaptive solver for the problem; use the quantum target cascade analysis algorithm to build a process based on the basic steps of the target cascade algorithm, improve it, and embed the quantum algorithm.

2. The method for collaborative optimization of multi-microgrid active distribution networks based on photon acceleration according to claim 1 is characterized in that: In step 1, the multi-microgrid active distribution network collaborative optimization framework based on photon acceleration collects the predicted output data of each load in real time through the measurement device and the information system, and synchronously transmits the predicted output data to the control center of the distribution network and the multi-microgrid; the control center performs multi-level collaborative optimization based on multi-agent characteristics and generates adjustment instructions, and the control center formulates the optimal operation strategy of the distribution network within the scope allowed by the operation constraints.

3. The method for collaborative optimization of multi-microgrid active distribution networks based on photon acceleration according to claim 2 is characterized in that: After obtaining the prediction information, the control center of the multi-microgrid independently completes the internal resource optimization operation based on its own optimization goals and interaction deviations, implements a two-way coordinated control mechanism between the distribution network and the multi-microgrid, and promotes the operation of the distributed energy system.

4. The method for collaborative optimization of multi-microgrid active distribution networks based on photon acceleration according to claim 1 is characterized in that: In step 2, the operation model of the active distribution network includes: In formulas (1) to (3), F ADS Indicates the improvement of system energy allocation efficiency. Indicates the operating energy consumption of MT, It represents the power balance adjustment item caused by the energy interaction with multiple microgrids; T represents the operation time span; Indicates the power of MT within the ADS range; n indicates the total number of MTs; c gas and L MT Respectively represent the natural gas characteristic parameters related to the MT operating status; η MT It represents the power generation efficiency of MT; and m represent the tie line power and the total number of tie lines; λ(t) represents the external power guidance signal set based on the time series; The model constraints of the active distribution network include: In formula (4) and formula (5), A∈{MTs,Tie-lines}, Indicates the load forecast value in ADS, and Represent the upper and lower limits of power respectively, Indicates the power of A within the ADS range.

5. The method for collaborative optimization of multi-microgrid active distribution networks based on photon acceleration according to claim 1 is characterized in that: In step 2, the operation model of the microgrid includes: In formulas (6) to (11), F MG,j It means optimizing the operating efficiency of multi-source systems. Indicates the energy conversion consumption of MT, represents the electric energy exchange term with the distribution network, represents the charge and discharge loss of ESS, Indicates the load shedding regulation response impact, represents the control influence of DR adjustable load; c gas and L MT Respectively represent the natural gas characteristic parameters related to the MT operating status; η MT represents the power generation efficiency of MT, λ(t) represents the external power guidance signal set based on the time series, and Represent the power of MT and tie line in MG respectively, C ESS and Corresponding to the unit operation coefficient and power of ESS in MG, C load,s and Indicates load shedding coefficient and load shedding power, C DR and represent the operating coefficient and power of DR respectively; The model constraints of the microgrid include: In formulas (12) to (18), Φ∈{ESS,DR,MTs,Tie-lines}, and Respectively represent the upper and lower limits of power within the MG, represents the power within Φ; E0, and Respectively represent the capacity, initial capacity, upper and lower limits of ESS; D DR and D load,s,max represent the total demand response load and maximum load shedding load, respectively.

6. The method for collaborative optimization of multi-microgrid active distribution networks based on photon acceleration according to claim 1 is characterized in that: The step 3 specifically includes the following sub-steps: Step 3-1: Model discretization and simplification. Given the accuracy limitations of noisy medium-scale quantum hardware, a partial linear discretization method is used to balance model fidelity and computational feasibility. Step 3-2: Establishment of quantum interpretable model and conversion of scheduling discrete model based on ADS.

7. The method for collaborative optimization of multi-microgrid active distribution networks based on photon acceleration according to claim 1 is characterized in that: The step 4 specifically includes the following sub-steps: Step 4-1: Model decoupling: ADS and MMG are coupled through a tie line, connecting their respective optimization processes and hindering independent optimization; Step 4-2: Accuracy adjustment: control the model accuracy to ensure computational feasibility; Step 4-3, quantum target cascade analysis, integrates quantum computing into the ATC framework, fully utilizes the advantages of parameter tuning efficiency and parallel computing, and enhances the solution process of multi-microgrid active distribution network collaborative optimization.

8. The method for collaborative optimization of multi-microgrid active distribution networks based on photon acceleration according to claim 7 is characterized in that: The step 4-2 specifically includes the following sub-steps: Step 4-2-1: Calculate the maximum and minimum absolute values ​​of all non-zero elements in the QUBO matrix: q abs,max =max(q ij ||q ij ≠0) (45) q abs,min =min(|q ij ||q ij ≠0) (46) If q abs,max / q abs,min <τ extreme is violated, the matrix is ​​considered infeasible and the procedure terminates, where τ extreme represents the modeling limit; Step 4-2-2: Transform the QUBO matrix into the equivalent Ising matrix M Ising ;make And the frequency function f(m ij ) is defined as m ij Then, the quantization resolution λ>0 is introduced and determined as follows: λ∈{kΔ|k∈Z + } (47) Among them, Δt represents the minimum modeling step, λ is an integer multiple of Δt, Q up and Q down They are the upper and lower limits of the matrix processing of CPQC; Step 4-2-3: Make Set the frequency threshold τ f and the ratio threshold τ p ,If Equation (49) is violated, the model is considered infeasible, and the process terminates; Step 4-2-4: After excluding incompatible elements and zero elements, the absolute value Ising matrix (|M Ising |) in [m abs,max , m abs,max ] is bounded, for M Ising The corresponding incompatible elements are smoothed: m ij =sign(m ij )m abs，max ,λ|m ij |>Q up (50) m ij =sign(m ij )m abs，min ,λ|m ij |<Q down (51) Where, sign(·) is the sign function; Step 4-2-5: By introducing a scaling factor (α), M Ising M' Ising =αM Ising Transform, and we get: m′ abs,max =Q up or m′ abs,min =Q down (52) Step 4-2-6: Evaluate whether the matrix meets the required accuracy criteria: mod(m′ ij) ∈[e mod,min ,he mod,max ] (53) Where mod(·) represents the rounding function, ε mod,min and ε mod,max They represent the upper and lower bounds of rounding respectively; if the constraints are met, the Ising matrix can be imported into the optical quantum computer; otherwise, its accuracy is insufficient.

9. The method for collaborative optimization of multi-microgrid active distribution networks based on photon acceleration according to claim 7 is characterized in that: The step 4-3 specifically includes the following sub-steps: Step 4-3-1: Initialize the operating parameters of each system component and the maximum number of iterations k max , the initial value w of the penalty function multiplier j and y j , precision adjustment parameter ε extreme , ε mod,min , ε mod,max and coupling variables and Set the number of iterations k = 1 and start the QATC algorithm; Step 4-3-2: Based on the coupled variable data, the distribution network control center constructs the QUBO problem with the objective function and the deviation from the MMG as the target. Step 4-3-3: To ensure that the accuracy requirements are met, the accuracy adjustment strategy is implemented; if the accuracy requirements are not met, an optimization solver is first used to obtain a reliable solution; if the accuracy requirements are met, a quantum computer is used for the solution; Step 4-3-4: The multi-microgrid control center establishes the QUBO model and implements the precision adjustment strategy to guide the selection of the appropriate solution method, and accelerates the processing speed by parallel computing across MGs; Step 4-3-5: Evaluate the convergence criteria; Convergence Criterion I ensures that the power flow deviation on the tie line between the two entities meets the specified requirements, and Convergence Criterion II ensures that the difference in the MMADS objective function between two consecutive iterations meets the specified conditions. The convergence criterion I is: The convergence criterion II is: In formulas (54) and (55), represents the overall optimization objective of MMADS, ε1 and ε2 represent the convergence limits; Step 4-3-6: When all conditions in step 4-3-5 are met at the same time, output the result; Otherwise, update the penalty function multiplier, set k = k + 1, and return to step 4-3-2; the updating principle of the penalty function multiplier is: In formula (56), β represents the update coefficient of the quadratic term of the penalty function multiplier.

Images