Optimized scheduling method and system for virtual power plant participating in super-large scale power grid

Abstract

The invention relates to an optimal scheduling method and system for a virtual power plant to participate in a super-large-scale power grid, and the method comprises the following steps: obtaining the related data of the virtual power plant and the super-large-scale power grid, and enabling the super-large-scale power grid to be provided with a plurality of units; based on the related data, constructing a full-time security constraint unit commitment model, and solving an initial unit start-stop state and initial output distribution of a virtual power plant and a unit by adopting a trained neural network model; merging according to similar time periods on the basis of the initial unit start-stop state and the output distribution of the virtual power plant and the unit; and based on the merging result, constructing and solving a security constraint economic dispatching model to obtain the optimal output of the virtual power plant and the unit, and completing the dispatching process. Compared with the prior art, the method has the advantages of improving scheduling flexibility and stability and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid optimization and dispatching, and in particular to a method and system for optimizing and dispatching a virtual power plant participating in an ultra-large-scale power grid. Background Art

[0002] With the acceleration of global energy transition and the widespread adoption of renewable energy, virtual power plants (VPPs) have garnered widespread attention as an innovative energy aggregation and management technology. VPPs integrate resources to form a flexibly dispatchable virtual entity, providing ancillary services and power support to the power grid. Optimizing the scheduling of VPPs is particularly crucial in ultra-large-scale power grids.

[0003] The complexity and dynamic nature of ultra-large-scale power grids pose significant challenges to traditional scheduling methods. Due to the intermittent and uncertain nature of distributed energy resources and the volatility of electricity demand, traditional centralized scheduling methods often struggle to meet real-time and accuracy requirements. The emergence of virtual power plants (VPPs) offers a new approach to addressing these issues. By integrating multiple resources, VPPs can achieve complementary and optimized resource allocation, improve energy efficiency, and enhance grid stability. However, due to the unique characteristics of ultra-large-scale power grids, optimized scheduling requires consideration of multiple factors, making existing scheduling methods highly complex and often inappropriate. This leads to insufficient grid stability and reliability, making them generally unsuitable for practical application.

[0004] Therefore, there is an urgent need for a scheduling method to solve the above problems and ensure the safe and stable operation of the power grid. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and system for optimizing the scheduling of virtual power plants participating in ultra-large-scale power grids with improved scheduling accuracy.

[0006] The purpose of the present invention can be achieved by the following technical solutions: A method for optimizing the dispatching of a virtual power plant in a super-large-scale power grid, comprising: Access data on virtual power plants and very large-scale power grids, where multiple units are deployed; Based on the relevant data, a full-time safety constraint unit combination model is constructed, and the trained neural network model is used to solve the initial unit start and stop status, the virtual power plant and the initial output distribution of the units; Based on the initial unit start and stop status, the virtual power plant and the output distribution of the units, merging them according to similar time periods; Based on the merging results, a safety-constrained economic dispatch model is constructed and solved to obtain the optimal output of the virtual power plant and units, completing the dispatch process.

[0007] Furthermore, the full-time safety-constrained unit commitment model takes the minimum total operating cost as the goal, and includes an objective function and corresponding constraints, wherein the objective function is: , in: Power generation cost of traditional units: , Traditional unit startup cost: , Virtual power plant costs: , Where, To minimize the total cost, For time period collection, For the crew collection, For traditional units j The total cost, For traditional units j The power generation cost curve, For the crew j In the period t The meritorious contribution, , indicating the unit j In the period t The start-stop state, For the crew j The start-stop cost, , indicating the unit j In the period t The startup action, is the overall cost of VPP, For VPP in the period t The net output, 、 、 is the weight coefficient, 、 、…、 is the weight coefficient, 、…、 is the set value, 、 is the minimum and maximum net output of VPP; Constraints include: Power balance constraints: , Where, For renewable energy i In the period t of efforts, For renewable energy collection, For the period t Total system load; Traditional unit constraints: 1) Output upper and lower limit constraints: , Where, 、 For the crew j The minimum and maximum output; 2) Climbing constraints: , Where, For the crew j The climbing rate, is the time period length; 3) Minimum running time constraint: , Where, For the crew j In the period t The minimum running time, For the crew j In the period The operating status of For the crew j In the period t The startup variable; 4) Minimum downtime constraint: , Where, For the crew j In the period t Minimum downtime, For the unit j In the period t The shutdown variable of 5) Start-stop logic constraints: , Virtual power plant constraints: 1) Net output range constraints: , Where, 、 is the minimum and maximum net output of VPP; 2) Climbing constraints: , Where, is the ramp rate of VPP; 3) Energy conservation constraints: , , Where, For the period t energy, 、 is the charge and discharge efficiency, is the time period length, 、 are the upper and lower limits of energy; Grid security constraints: 1) Power flow constraints: , Where, is the line current, is the set of power grid bus nodes, is the line-node power transfer distribution factor, To connect busbar n The collection of units, For VPP on bus n The weight of the output, is the internal busbar in time period t n Load value; 2) Line capacity constraints: , Where, is the transmission line capacity limit, A set of transmission lines.

[0008] Furthermore, during the training process of the neural network model, the full-time safety constraint unit combination model is constructed as a graphical model, and based on the structure of the graphical model, a neural network model is selected and trained to obtain a trained neural network model.

[0009] Furthermore, the graph model is represented as G=(V,E,W), where V is a node set, which is represented as: , Where, is a graph node, representing a unit j In the period t The operating status of It is a graph node, which means VPP in the period t The net output, is a graph node, serving as a system-level coupling point, representing power balance, For the crew collection, Set for time period; E represents the edge set, including time edge, space edge and constraint edge, where the time edge includes the unit climbing constraint edge ( ) and the virtual power plant ramp constraint edge ( ), the spatial edge includes the coefficient of the unit output in the power balance ( ), the coefficient of virtual power plant output in power balance ( ) and power flow constraints ( ), , the constraint edges include minimum running time constraint edges and minimum downtime constraint edges; W represents the weight set, including node weight and edge weight, where the node weight includes traditional units j Total cost and VPP overall cost ,Edge weights include time edge weights, space edge weights and constraint edge weights.

[0010] Furthermore, the neural network model is a CNN model or a RNN model.

[0011] Furthermore, the step of merging according to similar time periods includes: a) Feature vector construction: Construct each period t The eigenvector of , expressed as: , Where, For the period t The total system load, is the number of online units, For the crew collection, For the crew j In the period t The initial start-stop state, is the proportion of traditional power supply, For the crew j In the period t Initial output distribution of b) Distance calculation: Calculate the current time period t distance : , Where, 、 For t During the period a Dot and b System load at the point, 、 for t Units within the period j exist a Dot and b The start and stop status of the point, 、 is an adjustable weight coefficient, + ; Based on the distance , get the similarity , expressed as: , Where, is the scaling factor; c) Typical situation clustering: Initialize the cluster centers: , Where K is the number of clusters, that is, the number of typical scenarios, For the k The center of the cluster, For the t period and Similarity of time periods; Allocation period: , Where, For the k The clustered time period set, For the The eigenvectors of cluster centers; Virtual Power Plant Participation Modifier: If the period t Virtual power plant participation , then force it to be an independent cluster , and update Otherwise, no action is taken. For virtual power plants in time period t Initial net output distribution; Update cluster centers: , Where, For the k The eigenvectors of cluster centers; Iterate until convergence: Repeat the steps of allocating time periods to updating cluster centers until the cluster allocation no longer changes and a cluster with k The merged period set of typical periods and k typical load values ​​in each typical period .

[0012] Furthermore, the step of merging according to similar time periods further includes: outputting a constraint activation mapping table by performing constraint validity judgment to avoid ignoring key path constraints after merging, wherein the specific step of outputting the constraint activation mapping table includes: For each period t and each linel , calculate the original power flow distribution: , Where, is the original tidal distribution, is the line-node power transfer distribution factor, To connect busbar n The collection of units, For VPP on bus n The initial component output, For the period t Internal busbar n load; Identify key constraint periods: , Where, is the key constraint period, is the capacity limit of the transmission line; By using the constraint activation logic, a constraint activation mapping table is output to avoid ignoring the critical path constraints after merging, wherein the constraint activation mapping table is expressed as: .

[0013] Furthermore, the safety-constrained economic dispatch model includes an objective function and corresponding constraints, wherein the objective function is based on a typical period. k The minimum operating cost is the goal, which is expressed as: , Where, Typical period k Minimum operating cost, For traditional units j The power generation cost curve, For traditional units j During typical periods k The meritorious contribution, For the overall quotation of VPP, For VPP during typical hours k The net output, Assemble for the crew; Constraints include: Power balance constraints: , Where, For renewable energy i The predicted output, For renewable energy collection, Typical period k Typical load values; Unit output constraints: , Where, For the crew j The initial start-stop state, 、 For the crew j The minimum and maximum output; Virtual power plant constraints: (1) Net output range constraints: , Where, 、 is the minimum and maximum net output of VPP; (2) Energy conservation constraint: , , Where, Typical period k energy, 、 is the charge and discharge efficiency, is the time period length, 、 are the upper and lower limits of energy; Grid security constraints: (1) Power flow constraints: , Where, is the line current, is the set of power grid bus nodes, is the line-node power transfer distribution factor, To connect busbar n The collection of units, For VPP on bus n The weight of the output, For typical periods k busbar n Load value; (2) Line capacity constraints: , Where, is the transmission line capacity limit, A collection of transmission lines.

[0014] Furthermore, the security-constrained economic dispatch model is solved by calling the pandapower solver.

[0015] The present invention also provides a virtual power plant participating in an optimized dispatching system for a super-large-scale power grid, comprising: Data acquisition module: used to obtain relevant data of virtual power plants and ultra-large-scale power grids, where multiple units are installed in the ultra-large-scale power grid; Initial state and output acquisition module: used to output the initial unit start and stop state, the virtual power plant and the initial output distribution of the units based on the relevant data and using the trained neural network model; Merging module: used to merge according to similar time periods based on the initial unit start and stop status, virtual power plants and unit output distribution; Scheduling module: used to build and solve the security-constrained economic scheduling model based on the merging results, obtain the optimal output of the virtual power plant and units, and complete the scheduling process.

[0016] Compared with the prior art, the present invention has the following beneficial effects:

[0017] (1) The present invention constructs and solves a full-time safety constraint unit combination model to obtain the initial start-stop status and output distribution, and merges time periods by mining similarities, reducing the scale of problem solving, and solving the technical problem of low scheduling accuracy caused by high scheduling calculation complexity when virtual power plants participate in ultra-large-scale power grids.

[0018] (2) By converting the complex SCUC model into a graph-structured expression, the present invention can provide a learnable topological structure for the neural network, which not only accelerates the solution of the SCUC model but also does not lose the optimization accuracy.

[0019] (3) The present invention reduces the number of optimization periods by merging similar periods, reduces the scale of the problem, and ensures the rationality and safety of the scheduling optimization results, thereby improving the technical problems of complex scheduling, low accuracy and low safety caused by virtual power plants participating in ultra-large-scale power grids.

[0020] (4) The present invention constructs a model with the minimum operating cost as the goal, achieving dual optimization of economy and scheduling accuracy, which is of great significance for real application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 Schematic diagram of the method of the present invention. DETAILED DESCRIPTION

[0022] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0023] Example 1

[0024] This embodiment provides a method for optimizing the scheduling of a virtual power plant participating in a super-large-scale power grid. Figure 1 As shown, the method includes the following steps: S1. Obtain relevant data of a virtual power plant and a very large-scale power grid, wherein the very large-scale power grid has multiple units.

[0025] The relevant data of the virtual power plant and the ultra-large-scale power grid obtained in this embodiment include the total system load of the virtual power plant and the ultra-large-scale power grid. , the power and net output quotation of virtual power plants (VPPs), the status of traditional units in ultra-large-scale power grids, power generation costs, start-up and shutdown costs, etc.

[0026] S2. Based on the relevant data, a full-time safety constraint unit combination model is constructed, and the trained neural network model is used to solve the initial unit start and stop status, the virtual power plant and the initial output distribution of the unit.

[0027] S21. Constructing a full-time safety-constrained unit commitment (SCUC) model This embodiment aims to minimize the total operating cost over the entire time period (including 12 hours, a day, a week, etc.) and constructs a full-time safety constraint unit combination model. The objective function is: , in: Power generation cost of traditional units: , Traditional unit startup cost: , Virtual power plant costs: , Where, To minimize the total cost, For time period collection, For the crew collection, For traditional units j The total cost, For traditional units j The power generation cost curve, For the crew j In the period t The meritorious contribution, , indicating the unit j In the period t The start-stop state, For the crew j The start-stop cost, , indicating the unit j In the period t The startup action, is the overall cost of VPP, For VPP in the periodt The net output, 、 、 is the weight coefficient, 、 、…、 is the weight coefficient, 、…、 is the set value, 、 is the minimum and maximum net output of VPP; The key constraints are: Power balance constraints: , Where, For renewable energy i In the period t of efforts, For renewable energy collection, For the period t Total system load; Traditional unit constraints: 1) Output upper and lower limit constraints: , Where, 、 For the crew j The minimum and maximum output; 2) Climbing constraints: , Where, For the crew j The climbing rate, is the time period length; 3) Minimum running time constraint: , Where, For the crew j In the period t The minimum running time, For the crew j In the period The operating status of For the crew j In the period t The startup variable; 4) Minimum downtime constraint: , Where, For the crew j In the period t Minimum downtime, For the unitj In the period t The shutdown variable of 5) Start-stop logic constraints: , Virtual power plant constraints: 1) Net output range constraints: , Where, 、 is the minimum and maximum net output of VPP; 2) Climbing constraints: , Where, is the ramp rate of VPP; 3) Energy conservation constraints: , , Where, For the period t energy, 、 is the charge and discharge efficiency, is the time period length, 、 are the upper and lower limits of energy; Grid security constraints: 1) Power flow constraints: , Where, is the line current, is the set of power grid bus nodes, is the line-node power transfer distribution factor, To connect busbar n The collection of units, For VPP on bus n The weight of the output, is the internal busbar in time period t n Load value; 2) Line capacity constraints: , Where, is the transmission line capacity limit, A set of transmission lines.

[0028] S22. Build a graph model This embodiment constructs a graphical model based on the aforementioned SCUC model, visually representing the complex SCUC model constraints graphically for easier understanding and analysis. By converting the SCUC model into a structured representation, a learnable topology is provided for the neural network, accelerating the solution of the SCUC model without sacrificing optimization accuracy.

[0029] Through the graph structure, the independence between constraints can be identified, thereby decomposing the problem and simplifying the solution.

[0030] Graph structure definition G=(V,E,W): Node Set : Graph Node :unit j In the period t The operating status of Graph Node :VPP in time period t The net output of Graph Node : As a system-level coupling point, it represents power balance.

[0031] Edge set E: Time edge: including unit climbing constraint edge ( ) and the virtual power plant ramp constraint edge ( ); Space side: including the coefficient of unit output in power balance ( ), the coefficient of virtual power plant output in power balance ( ) and power flow constraints ( ), ; Constraint edges: include minimum running time constraint edges and minimum downtime constraints.

[0032] Weight set W: Node weight: including traditional units j Total cost and VPP overall cost ; Edge weights: include time edge weights, spatial edge weights, and constraint edge weights.

[0033] S23. Neural Network Model Solving Graph Model This embodiment uses a CNN neural network model or an RNN neural network model to solve historical solutions and predict the initial solution to the current problem.

[0034] Training neural network: Input a large number of historical scene features X (load curve , real-time electricity prices of historical data, etc.), output label Y: initial unit start and stop status, initial output distribution of virtual power plants and units.

[0035] The trained neural network is used to solve and predict the graphical model to obtain the initial start and stop status of the unit in the current period t , Initial output distribution of virtual power plants , initial output distribution of the unit .

[0036] S3. Based on the initial unit start and stop status, the virtual power plant and the output distribution of the units, the units are merged according to similar time periods.

[0037] This embodiment reduces the number of optimization time periods by merging similar time periods.

[0038] S31. Feature vector construction Construct each period t The eigenvector of , expressed as: , Where, For the period t The total system load, is the number of online units, For the crew collection, For the crew j In the period t The initial start-stop state, is the proportion of traditional power supply, For the crew j In the period t Initial output distribution of S32. Distance calculation Calculate the current time period t distance : , Where, 、 For t During the period a Dot and b The system of points meets, 、 for t Units within the period j exist a Dot and b The start and stop status of the point, 、 is an adjustable weight coefficient, + ; Based on the distance , get the similarity , expressed as: , Where, is the scaling factor; S33. Typical Situation Clustering Initialize the cluster centers: , Where K is the number of clusters, that is, the number of typical scenarios, For the k The center of the cluster, For the t period and Similarity of time periods; Allocation period: , Where, For the k The clustered time period set, For the The eigenvectors of cluster centers; Virtual Power Plant Participation Modifier: If the period t Virtual power plant participation , then force it to be an independent cluster , and update Otherwise, no action is taken. For virtual power plants in time period t Initial net output distribution; Update cluster centers: , Where, For the k The eigenvectors of cluster centers; Iterate until convergence: Repeat the steps of allocating time periods to updating cluster centers until the cluster allocation no longer changes and a cluster with k The merged period set of typical periods and k typical load values ​​in each typical period .

[0039] S34. Constraint Validity Judgment For each period t and each line l , calculate the original power flow distribution: , Where, is the original tidal distribution, is the line-node power transfer distribution factor, To connect busbar n The collection of units, For VPP on bus n The initial component output, For the period t Internal busbar n load; Identify key constraint periods: , Where, is the key constraint period, is the capacity limit of the transmission line; By using the constraint activation logic, a constraint activation mapping table is output to avoid ignoring the critical path constraints after merging, wherein the constraint activation mapping table is expressed as: .

[0040] , indicating the line l During typical periods k Constraints need to be activated. By judging the validity of constraints, a constraint activation mapping table is output to identify potential blocking lines to avoid ignoring key line constraints after merging.

[0041] This step reduces the problem size while ensuring the rationality and security of the scheduling optimization results by merging similar time periods.

[0042] S4. Based on the merged results, a security-constrained economic dispatch model is constructed and solved to obtain the optimal output of the virtual power plant and units, completing the dispatch process.

[0043] The Security Constrained Economic Dispatch (SCED) model includes the typical period k The objective function and corresponding constraints with the minimum operating cost as the goal are: , Where, for k The minimum operating cost of the period, For traditional units j The power generation cost curve, For traditional units j During typical periods k The meritorious contribution, For the overall quotation of VPP, For VPP during typical hours k The net output, Assemble for the crew; Constraints include: Power balance constraints: , Where, For renewable energy i The predicted output, For renewable energy collection, Typical period k Typical load values; Unit output constraints: , Where, For the crew j The initial start-stop state, 、 For the crew j The minimum and maximum output; Virtual power plant constraints: (1) Net output range constraints: , Where, 、 is the minimum and maximum net output of VPP; (2) Energy conservation constraint: , , Where, Typical period k energy, 、 is the charge and discharge efficiency, is the time period length, 、 are the upper and lower limits of energy; Grid security constraints: (1) Power flow constraints: , Where, is the line current, is the set of power grid bus nodes, is the line-node power transfer distribution factor, To connect busbar n The collection of units, For VPP on bus n The weight of the output, For typical periods k busbar n Load value; (2) Line capacity constraints: , Where, is the transmission line capacity limit, A collection of transmission lines.

[0044] After obtaining a good initial solution, by speeding up the call of the pandapower solver to solve the SCED model, the optimal 0 / 1 set of virtual power plants and unit start and shutdown plans is obtained, thereby realizing the scheduling optimization of virtual power plants participating in ultra-large-scale power grids.

[0045] Example 2

[0046] This embodiment provides a system for optimizing the scheduling of a virtual power plant participating in a super-large-scale power grid, including: Data acquisition module: used to obtain relevant data of virtual power plants and ultra-large-scale power grids, where multiple units are installed in the ultra-large-scale power grid; Initial state and output acquisition module: used to output the initial unit start and stop state, the virtual power plant and the initial output distribution of the units based on the relevant data and using the trained neural network model;

[0047] Merging module: used to merge according to similar time periods based on the initial unit start and stop status, virtual power plants and unit output distribution; Scheduling module: used to build and solve the security-constrained economic scheduling model based on the merging results, obtain the optimal output of the virtual power plant and units, and complete the scheduling process.

[0048] The rest is the same as in Example 1.

[0049] If the above functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0050] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk drives, CD-ROMs, optical storage devices, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention may be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0051] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0052] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0053] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0054] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0055] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A method for optimizing the scheduling of a virtual power plant participating in a super-large-scale power grid, characterized in that: include: Access data on virtual power plants and very large-scale power grids, where multiple units are deployed; Based on the relevant data, a full-time safety constraint unit combination model is constructed, and the trained neural network model is used to solve the initial unit start and stop status, the virtual power plant and the initial output distribution of the units; Based on the initial unit start and stop status, the virtual power plant and the output distribution of the units, merging them according to similar time periods; Based on the merging results, a safety-constrained economic dispatch model is constructed and solved to obtain the optimal output of the virtual power plant and units, completing the dispatch process.

2. The method for optimizing the scheduling of a virtual power plant participating in a super-large-scale power grid according to claim 1, characterized in that: The full-time safety-constrained unit commitment model aims to minimize the total operating cost, and includes an objective function and corresponding constraints. The objective function is as follows: ， in: Power generation cost of traditional units: ， Traditional unit startup cost: ， Virtual power plant costs: ， Where, To minimize the total cost, For time period collection, For the crew collection, For traditional units j The total cost, For traditional units j The power generation cost curve, For the crew j In the period t The meritorious contribution, , indicating the unit j In the period t The start-stop state, For the crew j The start-stop cost, , indicating the unit j In the period t The startup action, is the overall cost of VPP, For VPP in the period t The net output, 、 、 is the weight coefficient, 、 、…、 is the weight coefficient, 、…、 is the set value, 、 is the minimum and maximum net output of VPP; Constraints include: Power balance constraints: ， Where, For renewable energy i In the period t of efforts, For renewable energy collection, For the period t Total system load; Traditional unit constraints: 1) Output upper and lower limit constraints: ， Where, 、 For the crew j The minimum and maximum output; 2) Climbing constraints: ， Where, For the crew j The climbing rate, is the time period length; 3) Minimum running time constraint: ， Where, For the crew j In the period t The minimum running time, For the crew j In the period The operating status of For the crew j In the period t The startup variable; 4) Minimum downtime constraint: ， Where, For the crew j In the period t Minimum downtime, For the unit j In the period t The shutdown variable of 5) Start-stop logic constraints: ， Virtual power plant constraints: 1) Net output range constraints: ， Where, 、 is the minimum and maximum net output of VPP; 2) Climbing constraints: ， Where, is the ramp rate of VPP; 3) Energy conservation constraints: ， ， Where, For the period t energy, 、 is the charge and discharge efficiency, is the time period length, 、 are the upper and lower limits of energy; Grid security constraints: 1) Power flow constraints: ， Where, is the line current, is the set of power grid bus nodes, is the line-node power transfer distribution factor, To connect busbar n The collection of units, For VPP on bus n The weight of the output, is the internal busbar in time period t n Load value; 2) Line capacity constraints: ， Where, is the transmission line capacity limit, A set of transmission lines.

3. The method for optimizing the scheduling of a virtual power plant participating in a super-large-scale power grid according to claim 2, characterized in that: During the training process of the neural network model, the full-time safety constraint unit combination model is constructed as a graphical model, and based on the structure of the graphical model, a neural network model is selected and trained to obtain a trained neural network model.

4. The method for optimizing the scheduling of a virtual power plant participating in a super-large-scale power grid according to claim 3, characterized in that: The graph model is represented as G=(V,E,W), where V is a node set, which is represented as: ， Where, is a graph node, representing a unit j In the period t The operating status of It is a graph node, which means VPP in the period t The net output, is a graph node, serving as a system-level coupling point, representing power balance, For the crew collection, Set for time period; E represents the edge set, including time edge, space edge and constraint edge, where the time edge includes the unit climbing constraint edge ( ) and the virtual power plant ramp constraint edge ( ), the spatial edge includes the coefficient of the unit output in the power balance ( ), the coefficient of virtual power plant output in power balance ( ) and power flow constraints ( ), , the constraint edges include minimum running time constraint edges and minimum downtime constraint edges; W represents the weight set, including node weight and edge weight, where the node weight includes traditional units j Total cost and VPP overall cost ,Edge weights include time edge weights, space edge weights and constraint edge weights.

5. The method for optimizing the scheduling of a virtual power plant participating in a super-large-scale power grid according to claim 1, characterized in that: The neural network model is a CNN model or an RNN model.

6. The method for optimizing the scheduling of a virtual power plant participating in a super-large-scale power grid according to claim 1, characterized in that: The step of merging according to similar time periods includes: a) Feature vector construction: Construct each period t The eigenvector of , expressed as: ， Where, For the period t The total system load, is the number of online units, For the crew collection, For the crew j In the period t The initial start-stop state, is the proportion of traditional power supply, For the crew j In the period t Initial output distribution of b) Distance calculation: Calculate the current time period t distance : ， Where, 、 For t During the period a Dot and b System load at the point, 、 for t Units within the period j exist a Dot and b The start and stop status of the point, 、 is an adjustable weight coefficient, + ; Based on the distance , get the similarity , expressed as: ， Where, is the scaling factor; c) Typical situation clustering: Initialize the cluster centers: ， Where K is the number of clusters, that is, the number of typical scenarios, For the k The center of the cluster, For the t period and Similarity of time periods; Allocation period: ， Where, For the k The clustered time period set, For the The eigenvectors of cluster centers; Virtual Power Plant Participation Modifier: If the period t Virtual power plant participation , then force it to be an independent cluster , and update Otherwise, no action is taken. For virtual power plants in time period t Initial net output distribution; Update cluster centers: ， Where, For the k The eigenvectors of cluster centers; Iterate until convergence: Repeat the steps of allocating time periods to updating cluster centers until the cluster allocation no longer changes and a cluster with k The merged period set of typical periods and k typical load values ​​in each typical period .

7. The method for optimizing the scheduling of a virtual power plant participating in a super-large-scale power grid according to claim 6, characterized in that: The step of merging according to similar time periods further includes: outputting a constraint activation mapping table by performing constraint validity judgment to avoid ignoring key path constraints after merging, wherein the specific step of outputting the constraint activation mapping table includes: For each period t and each line l , calculate the original power flow distribution: ， Where, is the original tidal distribution, is the line-node power transfer distribution factor, To connect busbar n The collection of units, For VPP on bus n The initial component output, For the period t Internal busbar n load; Identify key constraint periods: ， Where, is the key constraint period, is the capacity limit of the transmission line; By using the constraint activation logic, a constraint activation mapping table is output to avoid ignoring the critical path constraints after merging, wherein the constraint activation mapping table is expressed as: 。 8. The method for optimizing the scheduling of a virtual power plant participating in a super-large-scale power grid according to claim 1, characterized in that: The safety-constrained economic dispatch model includes an objective function and corresponding constraints, wherein the objective function is based on a typical period k The minimum operating cost is the goal, which is expressed as: ， Where, Typical period k Minimum operating cost, For traditional units j The power generation cost curve, For traditional units j During typical periods k The meritorious contribution, For the overall quotation of VPP, For VPP during typical hours k The net output, Assemble for the crew; Constraints include: Power balance constraints: ， Where, For renewable energy i The predicted output, For renewable energy collection, Typical period k Typical load values; Unit output constraints: ， Where, For the crew j The initial start-stop state, 、 For the crew j The minimum and maximum output; Virtual power plant constraints: (1) Net output range constraints: ， Where, 、 is the minimum and maximum net output of VPP; (2) Energy conservation constraint: ， ， Where, Typical period k energy, 、 is the charge and discharge efficiency, is the time period length, 、 are the upper and lower limits of energy; Grid security constraints: (1) Flow constraints: ， Where, is the line current, is the set of power grid bus nodes, is the line-node power transfer distribution factor, To connect busbar n The collection of units, For VPP on bus n The weight of the output, For typical periods k busbar n Load value; (2) Line capacity constraints: ， Where, is the transmission line capacity limit, A collection of transmission lines.

9. The method for optimizing the scheduling of a virtual power plant participating in a super-large-scale power grid according to claim 1, characterized in that: The security-constrained economic dispatch model is solved by calling the pandapower solver.

10. A virtual power plant participating in an optimized dispatching system for a super-large-scale power grid, characterized in that: include: Data acquisition module: used to obtain relevant data of virtual power plants and ultra-large-scale power grids, where multiple units are installed in the ultra-large-scale power grid; Initial state and output acquisition module: used to output the initial unit start and stop state, the virtual power plant and the initial output distribution of the units based on the relevant data and using the trained neural network model; Merging module: used to merge according to similar time periods based on the initial unit start and stop status, virtual power plants and unit output distribution; Scheduling module: used to build and solve the security-constrained economic scheduling model based on the merging results, obtain the optimal output of the virtual power plant and units, and complete the scheduling process.

Images