Cluster microgrid low-carbon cooperative scheduling method and device based on diffusion model

Abstract

The invention discloses a cluster microgrid low-carbon collaborative scheduling method and device based on a diffusion model. The method comprises the steps of new energy output scene generation, carbon flow modeling, carbon price mechanism design, distributed robust optimization model and Nash game model construction and solution, and final output of a day-ahead scheduling scheme. Wherein the diffusion model is adopted to generate a new energy power generation power scene, and under the background that the permeability of the new energy is continuously increased, the uncertainty characteristics of the output of the new energy can be described more accurately; through modeling learning of real historical data, the representativeness of the generated data is improved, and the authenticity of scene data is enhanced; the credibility improvement of the data provides a more stable input basis for the scheduling model, so that the reliability of an optimized scheduling result is remarkably improved; and finally, the safety and the economical efficiency of the power system in the case of new energy uncertainty are guaranteed, and effective support is provided for day-ahead scheduling.

Description

Technical Field

[0001] The present invention relates to the field of microgrid optimization scheduling, and in particular to a method and device for low-carbon collaborative scheduling of cluster microgrids based on a diffusion model. Background Art

[0002] A modern energy system centered around distributed energy and microgrid technologies is becoming a key enabler for optimizing power resource allocation and deepening market reform. As a key carrier of distributed energy, single microgrids offer advantages in improving energy efficiency and promoting renewable energy integration. However, their inherent lack of flexibility makes them unable to adapt to dynamic fluctuations in energy supply and demand. Furthermore, as the cost of renewable energy generation, such as wind and photovoltaic power, continues to decline, the uncertainty associated with high levels of renewable energy integration has significantly increased. The diverse power load, the random nature of renewable energy output, and the severe lack of historical operating data for newly built power plants pose dual challenges to the stable operation and optimized scheduling of microgrid systems. Against this backdrop, building a coordinated scheduling system for clustered microgrids, integrating distributed resources within a region, and establishing energy complementarity mechanisms has become an effective path to improving system economics, low carbon efficiency, and reliability. This interconnected architecture not only reduces dependence on the parent grid through resource sharing but also provides structural support for the large-scale integration of clean energy and reducing overall carbon emissions.

[0003] To address the complex uncertain scheduling issues in cluster microgrid systems, current research focuses on integrating advanced optimization algorithms with deep learning techniques. Distributed Robust Optimization (DRO) methods, by constructing fuzzy sets of probability distributions, circumvent the strong dependence of stochastic optimization on precise probability distributions while also overcoming the overly conservative limitations of traditional robust optimization. This approach offers unique advantages in balancing system economy and robustness. However, the accuracy of fuzzy set construction is highly dependent on the quality and availability of historical data, and the scarcity of data in newly built microgrids and high-volatility scenarios presents a key bottleneck restricting its application. Summary of the Invention

[0004] The present invention aims to address the deficiencies of the prior art and propose a method and device for low-carbon coordinated scheduling of cluster microgrids based on a diffusion model.

[0005] The object of the present invention is achieved through the following technical solution: a low-carbon coordinated scheduling method for cluster microgrids based on a diffusion model, the method comprising:

[0006] Based on historical photovoltaic and wind power data, a diffusion model is used to generate photovoltaic and wind power output scenarios for corresponding dates.

[0007] The carbon emission of the node is determined through the carbon flow model, and the carbon emission responsibility of the node is determined based on the node carbon emission and Shapely value method. The obtained node carbon emission responsibility is combined with the power cost of the microgrid as the total energy load;

[0008] According to the internal equipment of the microgrid, a microgrid operation optimization model is modeled with the minimum total energy load as the goal. The scenario data generated by the diffusion model is used as uncertainty data, and a Wasserstein-based DRO model is constructed based on the microgrid operation optimization model;

[0009] A Nash game model is constructed based on the power interaction between microgrids. The Nash game model is converted into a problem for solving the interaction power problem between microgrids and a problem for solving the interaction cost problem between microgrids. The alternating multiplier method is used to solve the two subproblems. When solving the interaction power problem between microgrids, the problem is converted into a main problem and subproblems. When solving the main problem, the embedded DRO model is converted using the convex mean transformation method and then solved.

[0010] The power of power grid equipment is adjusted according to the low-carbon dispatch plan output by the Nash game model.

[0011] According to another aspect of the specification, a low-carbon collaborative scheduling device for cluster microgrids based on a diffusion model is also provided, comprising a memory and one or more processors, wherein the memory stores executable code, and when the processor executes the executable code, the low-carbon collaborative scheduling method for cluster microgrids based on a diffusion model is implemented.

[0012] Beneficial effects of the present invention:

[0013] This paper introduces a diffusion model for renewable energy output within microgrids. This technology uses deep learning to extract and generalize historical data features to generate high-quality day-ahead wind and photovoltaic output scenarios. It also incorporates a carbon flow model to quantify the carbon emission responsibility of each node, assigning carbon emission responsibilities based on the node's carbon emissions.

[0014] This invention also considers the interaction of power between microgrids, constructs a Nash game model, and solves it using ADMM. By building a collaborative operation system for clustered microgrids that takes uncertainty into account, integrating distributed resources within the region and establishing an energy complementarity mechanism, it becomes an effective path to improving system reliability. This invention not only reduces dependence on the upper power grid through resource sharing but also provides structural support for the large-scale consumption of clean energy. Furthermore, through the carbon flow model, it helps to enhance electricity users' awareness of carbon responsibility and provides data support for their low-carbon demand response. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1A flow chart of a low-carbon collaborative scheduling method for cluster microgrids based on a diffusion model provided by an embodiment of the present invention;

[0016] Figure 2 A schematic diagram of a low-carbon collaborative scheduling device for cluster microgrids based on a diffusion model provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The specific embodiments of the present invention are further described in detail below with reference to the accompanying drawings.

[0018] like Figure 1 As shown, the present invention provides a low-carbon coordinated scheduling method for cluster microgrids based on a diffusion model, and the specific process includes:

[0019] S1. Applying the diffusion model to generate wind and solar output scenarios, and the node carbon responsibility strategy based on the carbon flow model;

[0020] S2. Based on the obtained wind and solar scenarios and node carbon responsibilities, a microgrid optimization model based on Wasserstein distance is constructed;

[0021] S3. Based on the single microgrid model, the cluster microgrid Nash game model is constructed by considering the power transaction between cluster microgrids, and it is solved iteratively through ADMM.

[0022] Said S1 comprises:

[0023] S1.1. Diffusion model to generate wind and solar output scenarios:

[0024] The diffusion model primarily consists of a forward diffusion process and a backward diffusion process. The forward diffusion process continuously adds noise to the original data until the data becomes pure, irregular noise. The backward diffusion process then gradually removes the noise to restore the historical data. During the training process, the diffusion model learns the probability distribution of the actual data and can generate scenarios that conform to the actual distribution. The forward process of the diffusion model is as follows:

[0025]

[0026] Where T is the total number of diffusion steps; β t ∈(0,1) is a constant that controls the intensity of the noise added at each step, and β0<β1<…<β t-1 <β t <…<β T .x T is the data corresponding to the original data after T diffusion; a t Indicates the proportion of original data retained in step t; It represents the product of the proportion of original data retained from step 1 to step t; x0 is the original data before diffusion. represents noise sampled from a standard normal distribution.

[0027] Reverse diffusion process of the diffusion model:

[0028]

[0029] p θ (x t-1 ∣x t ) means given the current state x t , predict the state x at the previous moment t-1 The conditional probability distribution of is generated by a neural network controlled by the parameter θ; Represents Gaussian distribution. μ θ (x t ,t) represents the mean of the posterior Gaussian distribution, ε θ (x t ,t) represents the noise added at time t, which is an unknown learnable parameter. For example, in actual situations, the historical data Xi of photovoltaic and wind power output and its corresponding weather information for one or more years are input, and the model is trained based on this data. After the trained model is trained, the weather information of a certain day is input to generate the corresponding photovoltaic and wind power output scenarios. is the nth sample corresponding to the qth uncertainty of microgrid i at time t;

[0030] S1.2. Node carbon responsibility strategy based on the carbon flow model:

[0031] The calculation of carbon emission flow is mainly based on two basic principles: proportional allocation and energy aggregation. The proportional allocation principle states that the amount of electricity flowing out of a node is allocated by the amount of electricity flowing into the node in each branch according to the proportion of each branch's electricity to the total amount of electricity flowing in. Its expression is:

[0032]

[0033] Among them, N + represents the set of all branches flowing into node i; N - represents the set of all branches out of node i; P ij,t,u represents the amount of branch ij from branch ui at time t; P ui,t represents the power from branch ui at time t; P ij,t is the active power flowing from node i to node j. The energy sharing principle states that when active power from different branches flows into the same node, the carbon emission intensity of the node is equal to the weighted average of the carbon emission intensities of all inflow branches. This principle can be expressed as:

[0034]

[0035] in, represents the carbon emission intensity of node i at time t; ρ ui,t represents the carbon emission flow rate of branch ui at time t; ρ g,t represents the carbon emission intensity of the generator at time t; P G,t Represents the generator output electric power at time t.

[0036]

[0037] in, represents the shapely value of node i at time t; N represents the set of all participants in the game, that is, the set of all microgrids or nodes; n represents the total number of participants in set N; |S| represents the number of elements in subset S; S represents the set excluding node i; E(S∪{i}) represents the carbon emissions when node i is included; E(S) represents the carbon emissions when node i is excluded. Subsequently, the carbon responsibility of each microgrid is derived based on its contribution value. The allocation of carbon emission costs can be expressed as:

[0038]

[0039] in: is the total carbon emission responsibility; λ1, λ2, λ3 are the carbon emission responsibility coefficients under different levels; is the carbon emissions exceeding the unaccounted carbon quota; E1 and E2 are the step lengths of carbon trading; represents the carbon cost shared by node i at time t. Finally, after obtaining the carbon cost of the node, the unit carbon responsibility of the node can be obtained according to the load at time t:

[0040]

[0041] Among them, p i,t represents the node carbon responsibility cost of node i at time t; is the node load of node i at time t.

[0042] In the real world, for example, within a distribution network with a specific topology, the corresponding power flow calculation formula is used to determine the corresponding distribution network's power flow results, providing information such as voltage, active power, and reactive power at each node. Based on these power flow calculation results, the carbon emissions of each node are calculated using the aforementioned proportional allocation and energy sharing principles. Based on the carbon emissions of each node, the overall carbon emissions of the entire distribution network are determined, allowing for the overall carbon emissions responsibility to be determined through a tiered carbon trading mechanism. The carbon emissions responsibility of each node is then calculated using the Shapely value method.

[0043] The S2 includes:

[0044] S2.1. Constructing a microgrid operation optimization model;

[0045]

[0046] in, is the power reduction of microgrid i at time t; is the maximum curtailed power of microgrid i; is the transferable load of microgrid i at time t; and are the minimum transferable load and the maximum transferable load of microgrid i, respectively;

[0047]

[0048] W i,min ≤W i,t ≤W i,max (16)

[0049] in, and Respectively represent the upper limits of the charging and discharging power of the energy storage system; and Respectively represent the charging and discharging power of the energy storage system; W i,t Represents the energy storage state at time t; W i,t+1 Indicates the energy storage state at time t+1; W i,min and W i,max Indicates the lower and upper limits of energy storage capacity; η cha and η dis Indicates the charging efficiency and discharging efficiency of energy storage;

[0050]

[0051] Among them, P ij,t is the interaction power between microgrids i and j at time t; is the power purchased by microgrid i from its superior at time t; and are the minimum and maximum values ​​of the interaction power between microgrid i and microgrid j, respectively; and are the minimum and maximum values ​​of electricity purchased by microgrid i from the upper grid;

[0052]

[0053] in, is the load of microgrid i at time t; is the photovoltaic output of microgrid i at time t; is the wind power output of microgrid i at time t; the microgrid operation optimization model is specifically as follows:

[0054]

[0055] is the total energy load of microgrid i; is the demand response cost of microgrid i; Responsibility for carbon emissions from microgrid i; is the cost of electric energy interaction for microgrid i; For network costs; The charging and discharging operating costs of energy storage; The cost of purchasing electricity; is the operating cost coefficient of energy storage; i,t is the unit cost of microgrid i purchasing electricity from the upper grid; p i,t is the unit carbon emission cost of microgrid i purchasing electricity from the upper grid; p i-j,t is the unit cost of electric energy transaction between microgrid i and microgrid j; is the unit transmission cost between microgrid i and microgrid j; K DRT represents the unit compensation cost of the transferable load; K DRC It represents the unit compensation cost of load that can be reduced.

[0056] S2.2. Construct the DRO model of the microgrid based on the wind and solar output scenario generated by the diffusion model:

[0057] The microgrid operation optimization model is modeled as a DRO model, and its compact form is:

[0058]

[0059] x i {P ij,t}

[0060]

[0061] stAx a

[0062] By i b

[0063] Dy i Cx i

[0064] Among them, x i represents the decision variable of the first stage, which is the electric energy interaction variable between microgrids; u i is an uncertainty quantity, which indicates the uncertainty of new energy processing within the microgrid; U i represents the uncertainty fuzzy set; y iis the set of other decision variables; the matrix A is x i Inequality constraints related to y; matrices B and D are related to y i ,u i Related equality and inequality constraints. Methods for solving Wasserstein's two-stage distribution robustness include the convex mean method, affine transformation, strong duality transformation, and polyhedron approximation. Since affine transformation requires a linear relationship between the uncertain subject and the decision, but this linear relationship cannot be guaranteed in practice; strong duality transformation and polyhedron approximation are less efficient, this paper uses the convex mean method to transform the model. The convex optimization transformation form is:

[0065]

[0066] After the above transformation, the power balance constraint can be converted into the following form:

[0067]

[0068] Among them, U i is the uncertainty set of microgrid i after conversion; N res is the total number of samples of internal uncertainty of microgrid i; is the fluctuation value of the nth sample corresponding to the qth uncertainty of microgrid i at time t; Q is the number of uncertainties within the microgrid;

[0069] Based on the above convex mean transformation, the objective function can be further equivalent to:

[0070]

[0071] Where u i is an uncertain variable; i ∈G(x i ,u i ) represents the feasible region of the second-stage decision variables; is the coefficient of the inner min function matrix; is the coefficient of the outer min function. After the above transformation, the Wasserstein two-stage robust model is converted into the corresponding two-stage robust model.

[0072] The S3 includes:

[0073] S3.1. Consider the power interaction between cluster microgrids and construct a Nash game model:

[0074] In a cluster microgrid, each microgrid operates as an independent entity and achieves energy sharing through power interaction to reduce operating costs. Therefore, designing a reasonable trading mechanism not only helps to improve overall benefits, but also ensures the fairness of income distribution. Nash bargaining theory provides theoretical support for multi-agent cooperative games. It ensures that each participant's strategy can make the best response to the choices of other entities, thereby ultimately achieving a stable equilibrium and ensuring the realization of the collaborative optimization goals of the cluster microgrid system. The Nash bargaining model of the cluster microgrid cooperative game can be expressed as:

[0075]

[0076] Where, Indicates the breaking point of negotiations; represents the cost of participating in the collaboration; it can be broken down into the energy interaction subproblem SP1 and the electricity price interaction subproblem SP2. Specifically, in a real-world environment, within a specific distribution network, for example, there are industrial microgrids, commercial microgrids, and residential microgrids. Due to the different types of microgrids, their electricity usage patterns vary. These microgrids can be organized into a microgrid cluster to accommodate the renewable energy output within the microgrid. First, based on power flow and carbon flow models, the unit carbon responsibility signal of each node is derived. Then, the devices within the microgrid, such as photovoltaic equipment, energy storage equipment, and transferable loads, are modeled. Based on the carbon responsibility signals and renewable energy output of each node, a Nash game model is used to calculate the inter-microgrid exchange power and interaction costs. Specifically, each microgrid first communicates its transaction power. When each microgrid has determined the exchange power required to meet its own carbon reduction goals based on its own carbon responsibility, renewable energy output, and energy storage capacity, the power calculation concludes. Based on the exchange power results, each microgrid then calculates the interaction cost.

[0077] S3.2. Solution to subproblem SP1 of the Nash game model:

[0078] Solution to subproblem SP1:

[0079]

[0080] Where k is the number of iterations of ADMM; and is the electric power interacting with other microgrids and the electric power sold to the upper grid at the kth iteration; and is the Lagrange multiplier of the kth iteration; ρ ij and ρ bs are the corresponding penalty parameters respectively.

[0081] To solve the above equation, the CCG algorithm is used to decompose it into the master problem (MP) and subproblem (SP) for iteration. The MP is:

[0082]

[0083] Among them, y l is the solution of the subproblem after iteration l times; is the value of the uncertain variable u in the worst scenario obtained after iterating l times; A, B, C, D, I u is the coefficient matrix of the variables under the corresponding constraints; a, b are the constant coefficient vector matrices;

[0084] The subproblem SP is:

[0085]

[0086] Among them, x * The value of x obtained by solving the upper MP problem is passed as a known parameter to the lower SP problem; the optimal value obtained by solving the main problem MP is LB; the optimal value obtained by solving the subproblem SP is UB, then the convergence condition of CCG is:

[0087] |UB-LB|≤κ(35)

[0088] κ is the convergence condition of the CCG model algorithm. When the CCG algorithm converges, the obtained interaction power is sent to the corresponding microgrid sub-model and iterated until the convergence condition of the ADMM algorithm is met.

[0089] S3.3. Solution to subproblem SP2 of the Nash game model:

[0090] Solution to subproblem SP2:

[0091]

[0092] The Lagrange multiplier representing the price coupling constraint between microgrid i and microgrid j, ρ m represents the corresponding penalty parameter;

[0093] Corresponding to the aforementioned embodiment of a low-carbon coordinated scheduling method for cluster microgrids based on a diffusion model, the present invention also provides an embodiment of a low-carbon coordinated scheduling device for cluster microgrids based on a diffusion model.

[0094] See also Figure 2An embodiment of the present invention provides a low-carbon collaborative scheduling device for a cluster microgrid based on a diffusion model, comprising a memory and one or more processors. The memory stores executable code, and when the processor executes the executable code, it is used to implement a low-carbon collaborative scheduling method for a cluster microgrid based on a diffusion model in the above embodiment.

[0095] The embodiment of the low-carbon collaborative dispatching device for cluster microgrids based on a diffusion model provided by the present invention can be applied to any device with data processing capabilities, and the device with data processing capabilities can be a device or apparatus such as a computer. The device embodiment can be implemented through software, or through hardware or a combination of software and hardware. Taking software implementation as an example, as a device in a logical sense, it is formed by the processor of any device with data processing capabilities in which it is located reading the corresponding computer program instructions in the non-volatile memory into the memory for execution. From the hardware level, if Figure 2 As shown, it is a hardware structure diagram of any device with data processing capability where a cluster microgrid low-carbon collaborative scheduling device based on a diffusion model provided by the present invention is located. Figure 2 In addition to the processor, memory, network interface, and non-volatile memory shown, any device with data processing capabilities in which the apparatus in the embodiment is located may also include other hardware, generally based on the actual functions of the device with data processing capabilities, which will not be described in detail.

[0096] The implementation process of the functions and effects of each unit in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and will not be repeated here.

[0097] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial description of the method embodiments. The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present invention. A person of ordinary skill in the art can understand and implement the present invention without inventive work.

[0098] An embodiment of the present invention further provides a computer-readable storage medium having a program stored thereon. When the program is executed by a processor, the method for low-carbon coordinated scheduling of cluster microgrids based on a diffusion model in the above embodiment is implemented.

[0099] The computer-readable storage medium may be an internal storage unit of any device with data processing capabilities described in any of the aforementioned embodiments, such as a hard disk or memory. The computer-readable storage medium may also be an external storage device of any device with data processing capabilities, such as a plug-in hard disk, a smart media card (SMC), an SD card, a flash card, etc. equipped on the device. Furthermore, the computer-readable storage medium may also include both an internal storage unit and an external storage device of any device with data processing capabilities. The computer-readable storage medium is used to store the computer program and other programs and data required by any device with data processing capabilities, and may also be used to temporarily store data that has been output or is to be output.

[0100] The present invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the low-carbon collaborative scheduling method of cluster microgrids based on a diffusion model.

[0101] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the contents disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present application that follow the general principles of the present application and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered merely as exemplary, and the true scope and spirit of the present application are indicated by the claims.

[0102] It should be understood that the above general description and the detailed description that follows are exemplary and explanatory only and do not limit the present application. The present application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes may be made without departing from the scope of the present application. The scope of the present application is limited only by the appended claims.

Claims

1. A low-carbon coordinated scheduling method for cluster microgrids based on a diffusion model, characterized in that: The method includes: Based on historical photovoltaic and wind power data, a diffusion model is used to generate photovoltaic and wind power output scenarios for corresponding dates. The carbon emission of the node is determined through the carbon flow model, and the carbon emission responsibility of the node is determined based on the node carbon emission and Shapely value method. The obtained node carbon emission responsibility is combined with the power cost of the microgrid as the total energy load; According to the internal equipment of the microgrid, a microgrid operation optimization model is modeled with the minimum total energy load as the goal. The scenario data generated by the diffusion model is used as uncertainty data, and a Wasserstein-based DRO model is constructed based on the microgrid operation optimization model; A Nash game model is constructed based on the power interaction between microgrids. The Nash game model is converted into a problem for solving the interaction power problem between microgrids and a problem for solving the interaction cost problem between microgrids. The alternating multiplier method is used to solve the two subproblems. When solving the interaction power problem between microgrids, the problem is converted into a main problem and subproblems. When solving the main problem, the embedded DRO model is converted using the convex mean transformation method and then solved. The power of power grid equipment is adjusted according to the low-carbon dispatch plan output by the Nash game model.

2. The low-carbon coordinated scheduling method for cluster microgrids based on a diffusion model according to claim 1 is characterized in that: The photovoltaic and wind power output scenarios for corresponding dates generated by the diffusion model specifically include: Forward process of the diffusion model: Where T is the total number of diffusion steps; β t ∈(0,1) is a constant that controls the intensity of the noise added at each step, and β0<β1<…<β t-1 <β t <…<β T ;x T is the data corresponding to the original data after T diffusion; a t Indicates the proportion of original data retained in step t; It represents the product of the proportion of the original data retained from step 1 to step t; x0 is the original data before diffusion; represents noise sampled from a standard normal distribution; Reverse diffusion process of the diffusion model: μ θ (x t ,t) represents the mean of the posterior Gaussian distribution, ε θ (x t ,t) represents the noise added at time t, which is an unknown learnable parameter.

3. The low-carbon coordinated scheduling method for cluster microgrids based on a diffusion model according to claim 1, characterized in that: It is characterized in that Determining the carbon emissions of a node using a carbon flow model includes: calculating the carbon emissions of the node using a proportional allocation principle and an energy aggregation principle; The proportional sharing principle includes: the power outflow of a node is allocated by the power of each branch flowing into the node according to the proportion of each branch power to the total power inflow; The energy aggregation principle includes: when active power from different branches is integrated into the same node, the carbon emission intensity of the node is equal to the weighted average of the carbon emission intensities of all the inflow branches.

4. The low-carbon coordinated scheduling method for cluster microgrids based on a diffusion model according to claim 1, characterized in that: The method of determining the carbon emission responsibility of a node based on the node carbon emission and Shapely value includes: Calculate shapely values: in, represents the shapely value of node i at time t; S represents the set that does not include node i; E(S∪{i}) represents the carbon emissions when node i is included; E(S) represents the carbon emissions when node i is not included; Then, the carbon emission responsibility is determined based on the contribution value of each microgrid: in: is the total carbon emission responsibility; λ1, λ2, λ3 are the carbon emission responsibility coefficients under different levels; is the carbon emissions exceeding the irresponsible carbon quota; E1 and E2 are the lengths of the carbon emission ladder; represents the carbon emission responsibility shared by node i at time t. Finally, after obtaining the carbon emission responsibility of the node, the carbon emission responsibility per unit time of the node is calculated according to the load at time t: Among them, p i,t represents the carbon emission responsibility of node i at time t; is the node load of node i at time t.

5. The low-carbon coordinated scheduling method for cluster microgrids based on a diffusion model according to claim 1, characterized in that: It is characterized in that The modeling microgrid operation optimization model includes: in, is the power reduction of microgrid i at time t; is the maximum curtailed power of microgrid i; is the transferable load of microgrid i at time t; and are the minimum transferable load and the maximum transferable load of microgrid i, respectively; IN i,min ≤W i,t ≤W i,max in, and Respectively represent the upper limits of the charging and discharging power of the energy storage system; and Respectively represent the charging and discharging power of the energy storage system; W i,t Represents the energy storage state at time t; W i,min and W i,max Indicates the lower and upper limits of energy storage capacity; η cha and η dis Indicates the charging efficiency and discharging efficiency of energy storage; Among them, P ij,t is the interaction power between microgrids i and j at time t; is the power purchased by microgrid i from its superior at time t; and are the minimum and maximum values ​​of the interaction power between microgrid i and microgrid j, respectively; and are the minimum and maximum values ​​of electricity purchased by microgrid i from the upper grid; in, is the load of microgrid i at time t; is the photovoltaic output of microgrid i at time t; is the wind power output of microgrid i at time t; The microgrid operation optimization model is specifically as follows: is the total energy load of microgrid i; is the demand response cost of microgrid i; Responsibility for carbon emissions from microgrid i; is the cost of electric energy interaction for microgrid i; For network costs; The charging and discharging operating costs of energy storage; The cost of purchasing electricity; is the operating cost coefficient of energy storage; i,t is the unit electricity cost of microgrid i purchasing electricity from the upper grid; p i,t is the unit carbon emission cost of microgrid i purchasing electricity from the upper grid; p i-j,t is the unit cost of electric energy transaction between microgrid i and microgrid j; is the unit transmission cost between microgrid i and microgrid j, K DRT represents the unit compensation cost of the transferable load; K DRC It represents the unit compensation cost of load that can be reduced.

6. A cluster microgrid low-carbon coordinated scheduling method based on a diffusion model according to claim 5, characterized in that: The construction of the Wasserstein-based DRO model includes: The microgrid operation optimization model is modeled as a DRO model, and its compact form is: Among them, x i represents the decision variable of the first stage, which is the electric energy interaction variable between microgrids; u i is an uncertainty quantity, which indicates the uncertainty of new energy processing within the microgrid; U i represents the uncertainty fuzzy set; y i is the set of other decision variables; the matrix A is x i Inequality constraints related to y; matrices B and D are related to y i ,u i The relevant equality constraints and inequality constraints are transformed using the convex mean method. The convex optimization transformation form is: After the above conversion, the power balance constraint is converted into the following form: Among them, U i is the uncertainty set of microgrid i after conversion; N res is the total number of samples of internal uncertainty of microgrid i; is the fluctuation value of the nth sample corresponding to the qth uncertainty of microgrid i at time t; is the nth sample corresponding to the qth uncertainty of microgrid i at time t; Q is the number of uncertainties within the microgrid; Based on the above convex mean transformation, the objective function is further equivalent to: Where u i is an uncertain variable; i ∈G(x i ,u i ) represents the feasible domain of the second-stage decision variables; after the above transformation, the Wasserstein two-stage distribution robustness is converted into the corresponding two-stage robust model.

7. The method for low-carbon coordinated scheduling of cluster microgrids based on a diffusion model according to claim 1, characterized in that: It is characterized in that The Nash game model constructed based on the electric energy interaction between microgrids includes: Where, Indicates the breaking point of negotiations; Represents the total energy load of microgrid i as the cost of participating in cooperation; it is decomposed into the power interaction problem and the electricity price interaction problem.

8. The low-carbon coordinated scheduling method for cluster microgrids based on a diffusion model according to claim 7 is characterized in that: The solution to the interactive power problem between microgrids includes: The objective function is: The constraints are the carbon emission responsibility per unit time of the node and the internal equipment model of the microgrid, which includes energy storage equipment, photovoltaic equipment, wind power equipment, transferable load and curtailable load; Where k is the number of iterations of ADMM; P ij,t is the interaction power between microgrids i and j at time t, is the power purchased by microgrid i from the upstream at time t, and is the electric power interacting with other microgrids and the electric power sold to the upper grid at the kth iteration; and is the Lagrange multiplier of the kth iteration; ρ ij and ρ bs are the corresponding penalty parameters respectively; The solution process includes: using CCG algorithm to decompose the first sub-problem into the main problem and the third sub-problem and iterate; The main issues include: Among them, y l is the solution of the subproblem after iteration l times; is the value of the uncertain variable u in the worst scenario obtained after iterating l times; A, B, C, D, I u is the coefficient matrix of the variables under the corresponding constraints; a, b are the constant coefficient vector matrices; Sub-questions include: Among them, x i Expressed as the electric energy interaction variable between microgrids; u i is the uncertainty of new energy processing within the microgrid; U i represents the uncertainty fuzzy set; y i is the set of other decision variables; x * The value of x solved for the upper MP problem is passed as a known parameter to the lower SP problem; the optimal value obtained by solving the main problem MP is LB; the optimal value obtained by solving the subproblem SP is UB, then the convergence condition of CCG is: |UB-LB|≤κ κ is the convergence condition of the CCG model algorithm. When the CCG algorithm converges, the obtained interaction power is sent to the corresponding microgrid sub-model to continue iterating until the convergence condition of the ADMM algorithm is met.

9. The method for low-carbon coordinated scheduling of cluster microgrids based on a diffusion model according to claim 7, characterized in that: It is characterized in that The second sub-question includes: The Lagrange multiplier representing the price coupling constraint between microgrid i and microgrid j, ρ m represents the corresponding penalty parameter, is the demand response cost of microgrid i; Responsibility for carbon emissions from microgrid i; is the cost of electric energy interaction for microgrid i; The charging and discharging operating costs of energy storage; The cost of purchasing electricity.

10. A low-carbon coordinated dispatching device for cluster microgrids based on a diffusion model, comprising a memory and one or more processors, wherein the memory stores executable code, characterized in that: When the processor executes the executable code, a low-carbon coordinated scheduling method for cluster microgrids based on a diffusion model according to any one of claims 1 to 9 is implemented.

Images