Bottom-up flexible resource hierarchical partitioning and dynamic aggregation method
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- STATE GRID SHANGHAI MUNICIPAL ELECTRIC POWER CO
- Filing Date
- 2024-12-06
- Publication Date
- 2026-06-09
Smart Images
Figure CN122174413A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a technology in the field of virtual power plants, specifically a bottom-up, flexible, hierarchical, partitioned, and dynamic aggregation method for resources. Background Technology
[0002] A static virtual power plant refers to a virtual power plant with a given combination of internal resources and topology, which is then optimized and scheduled according to different objectives. However, for different scenarios, the flexible resource combination of a static virtual power plant cannot guarantee optimality, cannot fully utilize the adjustable potential of each flexible resource, and may even lead to resource waste. Summary of the Invention
[0003] To address the aforementioned shortcomings of existing technologies, this invention proposes a bottom-up dynamic aggregation method for flexible resources with hierarchical partitioning. Based on the hierarchical partitioning control architecture of flexible resource virtual power plants, and considering the market participation scenarios of virtual power plants, this invention sets corresponding flexible resource aggregation standards and solves the corresponding hierarchical partitioning complementary aggregation scheme that takes into account the network topology.
[0004] This invention is achieved through the following technical solution:
[0005] This invention relates to a bottom-up method for hierarchical and partitioned dynamic aggregation of flexible resources, comprising: modeling the flexible resources within a virtual power plant, evaluating the adjustability of the flexible resources according to the adjustability assessment index, dividing the virtual power plant structure into a three-layer model of cloud-edge-terminal, namely, the virtual power plant cloud platform layer, the edge control layer, and the flexible resource terminal layer; and then aggregating them in a bottom-up manner to form the edge control layer structure and the cloud platform layer structure. Attached Figure Description
[0006] Figure 1 This is a flowchart of an implementation example;
[0007] Figure 2 A hierarchical and zoned control architecture for flexible resource virtual power plants;
[0008] Figure 3 The results are aggregated by hierarchical partitioning for flexible resource allocation. Detailed Implementation
[0009] like Figure 1 As shown in this embodiment, a bottom-up flexible resource hierarchical partitioning dynamic aggregation method is provided, including:
[0010] Step 1) Model the internal flexibility resources of the virtual power plant, namely distributed photovoltaic, distributed wind power, distributed energy storage, gas turbines, and building air conditioning, specifically including:
[0011] 1.1) Distributed photovoltaic modeling, specifically: in: Let k be the output power of the distributed photovoltaic system k under the external environment at time t; PV P represents the power loss factor caused by aging or damage to distributed photovoltaic panels, typically taken as 0.9. RP D represents the rated power of the distributed photovoltaic panel. t D represents the actual solar irradiance at the location of the distributed photovoltaic system at time t; STC Light intensity under standard test conditions (STC); T t PV Let T be the actual surface temperature of the distributed photovoltaic system at time t; STC The standard test temperature is 25℃. Where: C k,PV,t The cost of distributed photovoltaic energy consumption is represented by the curtailment penalty, where ο is the curtailment penalty coefficient. and These represent the actual and predicted output of distributed photovoltaic power during time period t, respectively.
[0012] 1.2) Distributed wind power modeling, specifically: in: Let be the output power of distributed wind power k under the external environment at time t; This refers to the rated power of distributed wind power. v in v out v rated These are wind speed, cut-in wind speed, cut-out wind speed, and rated operating wind speed, respectively. Where: C k,W,t γ represents the energy cost of distributed wind power, and γ is the wind curtailment penalty coefficient. and These represent the actual and predicted output of distributed wind power during time period t, respectively.
[0013] 1.3) Distributed energy storage modeling, specifically: the ratio between available energy and total energy capacity in the energy storage system, i.e., the state of charge (SOC) of distributed energy storage k at time t. Where: α ch and α dis These represent the charge / discharge efficiency of distributed energy storage; CAP represents the battery capacity of distributed energy storage. and Let be the charging and discharging power of distributed energy storage k during time period t. Since distributed energy storage cannot simultaneously charge and discharge at any given moment during operation, and there is an upper limit to the charging and discharging power of distributed energy storage, the constraints on the charging and discharging power of distributed energy storage include... Where: β k,tLet t be the 0-1 variable representing the charging and discharging state of distributed energy storage k at time t. A value of 1 indicates distributed energy storage is discharging, and a value of 0 indicates charging. and This represents the maximum charging and discharging power. Where: C k,ESS,t The energy cost of distributed energy storage; and The unit charging / discharging cost and maintenance cost of distributed energy storage are respectively considered.
[0014] 1.4) Gas turbine modeling, specifically: in: Let a be the output power of gas turbine k at time t; i and b i Ω represents the first-order coefficient and constant term of the piecewise linear function, respectively; ΔG(t,k) is the natural gas intake of gas turbine k at time t; i is the piecewise linear function segment number, Ω DG This is a set of segment numbers. The output constraints of the gas turbine include... in: and These are the minimum and maximum technical outputs allowed for the gas turbine, respectively. Where: a DG b DG and c DG This represents the cost coefficient for gas turbines.
[0015] 1.5) Building Air Conditioning Modeling: Due to the time-delay characteristic of indoor temperature changes, the building air conditioning model can be considered as energy storage, with heat as the energy storage medium. Taking the cooling characteristics of a building air conditioning system in summer as an example, when the air conditioner operates at a power level that just keeps the indoor temperature at the upper limit of the comfortable temperature range, further reducing the power level will cause the indoor temperature to exceed the comfortable temperature range. In this case, the adjustable capacity of the building air conditioner is 0, equivalent to no energy being stored inside the building. Conversely, when the air conditioner operates at a power level that just keeps the indoor temperature at the lower limit of the comfortable temperature range, if there is a need for flexible resource control, the power level of the building air conditioner can be reduced, equivalent to maximizing the energy stored inside the building. The expression for the building's stored energy is: in: Let be the energy stored in building k at time t. To ensure the building's air conditioning system can withstand the maximum electrical power; Where: C k,AC,t For the energy cost of building air conditioning, a AC b AC and c AC This is the energy cost coefficient for building air conditioning.
[0016] The aforementioned flexibility resource assessment indicators include:
[0017] 1) Maximum upward adjustment capability This represents the maximum upward shift of flexibility resources relative to the current operating point, and its external characteristic is to provide power support, corresponding to the upper limit of output of traditional units, specifically: Where: P k,t The output power of the flexible resource k at time t. Minimize the technical effort required for flexibility resource k.
[0018] 2) Maximum downward adjustment capability The maximum downward offset of flexibility resources relative to the current operating point corresponds to the lower limit of output of conventional units, specifically: in: The output limit of flexibility resource k.
[0019] 3) Response delay T R This refers to the time required for a flexible power source to fully reach the required power level after receiving a regulation command, corresponding to the start-up and shutdown time of a traditional unit. The expression includes... in: and These represent the response delays for upward and downward adjustments of the flexibility resource k, respectively; t ds,up and t ds,down These represent the moments when flexibility resources are adjusted upwards and downwards, respectively; t ss The moment when a response instruction is received for flexible resources.
[0020] 4) Response duration T D This refers to the longest time that the flexible resources can maintain the adjusted state after reaching the power regulation requirements, corresponding to the continuous start-up and shutdown time of traditional units, specifically: in: and These represent the response durations for upward and downward adjustments to the flexibility resource k, respectively; t de,up and t de,down These represent the longest periods of time that the flexibility resources maintain their adjusted state after being adjusted upwards and downwards, respectively.
[0021] 5) Adjustment rate R, divided into upward adjustment rate R k,up and downward adjustment rate R k,down , respectively, represent the maximum upward and downward adjustment power of flexible resources per unit time, corresponding to the ramp rate of traditional units, specifically:
[0022] 6) Response time probability γ when the battery is low LOLPLet be the probability that a flexible resource, after receiving an adjustment instruction, cannot reach the specified adjustment range due to uncertainties. Its expression is: Where: l represents the system operating state; p l The probability that the system is in a running state and the response power is insufficient; t l This refers to the duration during which the system experiences insufficient power while in operating state l.
[0023] 7) Expected response time γ for insufficient power LOLE The expected value for the time during which the adjustable capacity deficit of flexible resources is greater than or equal to the reserve capacity is expressed as: Wherein: the day is divided into I time periods; T i Let C be the number of time periods contained in the i-th time period; i L represents the theoretical maximum adjustment capacity of the flexibility resource in the i-th time period; i,t p[X≥C] represents the peak load power in the i-th time period; i -L i,t This refers to the probability that the available capacity deficit of flexible resources is greater than or equal to the reserve capacity.
[0024] 8) Response to insufficient power supply expectation value γ EENS Indicators 6) and 7) represent the probability of insufficient response power and the expected duration of insufficient response power, respectively. This analysis focuses on the quantification of insufficient response power deficit, representing the expected power supply value reduced due to uncertainties affecting flexibility resources and resulting in insufficient response. Its expression is: Where: N is the number of simulated hours; L I C represents the load power in the I-th simulation hour; I P represents the theoretical maximum adjustability of the flexibility resources during the I-th simulation hour. I (X) represents the probability of insufficient response power during the I-th simulation hour.
[0025] Step 2) Divide the virtual power plant structure into the following categories: Figure 2 The cloud-edge-device model shown consists of three layers: the virtual power plant cloud platform layer, the edge control layer, and the flexible resource terminal layer. The flexible resource terminal layer comprises controllers for various flexible resources. Each controller reports the adjustable capability parameters of the flexible resources and decomposes and executes the scheduling instructions issued by the edge control layer. The edge control layer is a resource aggregation formed by multiple flexible resources within the same area, uniformly regulated by a single edge controller. The virtual power plant cloud platform layer, located in the cloud as the virtual power plant operation server, coordinates the control of edge controllers in different areas, processes the adjustable information and operational status reported by the edge control layer, interacts with the trading center and dispatch center, receives market trading instructions, and decomposes and distributes them.
[0026] The unified control refers to: considering the impact of geographical location and multiple scenario objectives, dividing flexible resources into different regions, each controlled by an edge controller. This edge controller is responsible for identifying the external characteristics of the flexible resource aggregate and reporting feasible domain information, as well as optimally decomposing and forwarding control commands issued from the upper layer (virtual power plant cloud platform layer). Its advantage lies in utilizing the computing resources of the edge controller's location to significantly alleviate the computational pressure on the virtual power plant.
[0027] Step 3) Aggregate flexibility resources in a bottom-up manner to form an edge control layer structure and a cloud platform layer structure, specifically including:
[0028] 3.1) Construct the optimal structure of the edge control layer, specifically including:
[0029] i) The network topology is represented by an undirected graph G = (N, B), where N is the set of network nodes and B is the set of branches. The power flow distribution in a power network is the set of injected power at each node. t∈T, where: t is the system running time, T is the total number of time points, i b Node numbering; the set of load power of each node in the power network: in: For node i b The power of the connected load.
[0030] ii) Form the optimal structure of the flexible resource aggregate to form the edge control layer of the virtual power plant. Let the set of aggregate partitioning results be: V(E,F) e ), Where: E represents the set of flexible resource aggregates within the system after aggregation; F e Let the set of nodes contained within the e-th flexible resource aggregate be denoted as follows: The variable is 0-1. It determines the region where node i is located. When the value is 1, node i is located in region e.
[0031] iii) An improved modularity metric is used to describe the internal communication cost of the aggregate, specifically: in: To improve the modularity index for internal communication costs within the flexible resource aggregate e; ij The communication cost between the flexible resources accessed by node i and node j, (x i ,y i Let ξ be the x and y coordinates of node i, and ζ be the communication cost per unit distance. e Cost of configuring edge controllers; αij β is the sum of the communication costs between all flexible resources within the aggregate; i This is the sum of the communication costs between the flexibility resource at node i within the aggregate and other internal flexibility resources.
[0032] iv) The optimal structure optimization model for the virtual power plant edge control layer includes... st Where: the decision variable is V(E,F) e The optimal virtual power plant edge control layer partitioning result set V(E,F) is obtained. e ) * The constraint is that the communication cost index of the flexible resources combined into an aggregate meets the maximum aggregation cost limit.
[0033] 3.2) Constructing the optimal structure of the cloud platform layer: Based on the above optimization results of the virtual power plant edge control layer structure, the behavior of aggregating flexible resource aggregates to form virtual power plants is modeled as a dynamic alliance game problem, and solved using the merging and splitting method. The optimal alliance of the aggregates is the virtual power plant cloud platform layer structure. When the alliance game C = (E, v), |E| = E is the total number of members, i.e., the number of flexible resource aggregates in the edge control layer. The function v is called the characteristic function, which maps the alliance S (i.e., the virtual power plant S) to a real number, corresponding to the total expected benefits that members in the alliance S can obtain through cooperation. Its calculation formula is: Wherein: the inner layer represents the game-theoretic bidding behavior of alliance S, p rofit (S) is the utility function of the alliance S.
[0034] The principle of the alliance includes considering two non-overlapping alliances S1 and S2, which cooperate to form a larger alliance when the following conditions are met:
[0035] 1) The characteristic function of the major league is greater than the sum of the characteristic functions of the minor leagues. This criterion is called the overall league criterion. v(S1∪S2)>v(S1)+v(S2);
[0036] 2) At least one member gains a greater benefit in the major leagues without causing other members to suffer a loss, i.e., the Pareto criterion, specifically:
[0037] 3.3) Establish a market fit index for flexible resource virtual power plants as the alliance utility function, specifically: in: Let S be the market fit index for the virtual power plant. The formula combines relevant flexible resource adjustability indicators, indicating that the higher the adjustable upper and lower limits, the shorter the response time, and the longer the duration of the virtual power plant's market participation, the higher its market fit. and These are the normalized values of the corresponding indicators of internal flexibility resources of virtual power plant k.
[0038] Based on specific practical experiments, the IEEE-118 node test system network structure was selected as the system topology. When the system contains various flexible resources and traditional units, the nodes where the resources are located and their parameter ratings are shown in Tables 1 and 2. In the tables, α and β are the first-order and second-order cost coefficients of the resource, respectively, and the unit distance communication cost ζ is 50,000 / km; the edge controller configuration cost ζ... e The price is 100,000 yuan per set.
[0039] Table 1 shows the flexibility resources and parameters included in the system.
[0040] Table 2 lists the traditional generating units included in the system and their parameters.
[0041] Based on the proposed improvement of modularity index considering the internal communication cost of flexible resource aggregates according to the present invention, the optimal structure of the edge control layer is obtained as shown in Table 3.
[0042] Table 3 Optimal Structure Results of Edge Control Layer
[0043] As shown in Table 3, to ensure optimal controller configuration cost and flexible resource communication cost in the edge control layer, the 14 flexible resources in the system are aggregated into 8 aggregates, resulting in an optimal total cost of 3.0778 million yuan. For comparison with the centralized control mode, the edge control layer is set to include one controller without aggregation. In this case, the total controller configuration cost and flexible resource communication cost is 4.7905 million yuan. This aggregation method can save 35.75% of communication costs. It can be concluded that the optimal aggregation method for the edge control layer proposed in this invention, which considers the internal communication cost of flexible resource aggregates and improves the modularity index, can effectively reduce the total communication cost and improve the overall modularity, thus facilitating cost savings for virtual power plants in the market environment.
[0044] After obtaining the optimal structure of the virtual power plant edge control layer, it is then aggregated to the upper layer to form a virtual power plant. According to the dynamic alliance game method proposed in this invention, the split-merge method is used to solve the problem, and the results are shown in Table 4.
[0045] Table 4 Optimal Aggregation Results of Flexible Resource Virtual Power Plant Cloud Platform Layer
[0046] Analysis of the optimal aggregation results of the flexible resource virtual power plant cloud platform layer shows that virtual power plant 3, composed of aggregates 4 and 5, has the highest alliance utility function value and the highest market fit of 0.1018. This is because it contains two gas turbine units, enabling 24-hour continuous response. To verify the superiority of the method of this invention, all possible alliance structures of flexible resource aggregates to form virtual power plants were traversed, and the sum of the alliance utility function values was less than the results of the example. Thus, the optimal structure of the hierarchical and partitioned virtual power plant composed of flexible resources is completed. The system topology and the hierarchical and partitioned aggregation structure of flexible resources are as follows: Figure 2 As shown, the formation of this structure not only reduces the overall communication cost of the virtual power plant and improves the overall fit of flexible resources in the market, but also enables cloud-edge collaborative computing, using the local computing resources of the edge controller to distribute the computing pressure of the virtual power plant.
[0047] Compared with existing technologies, this method considers communication costs and market fit indicators to aggregate flexible resources from top to bottom, resulting in an optimal hierarchical and partitioned structure for virtual power plants. This improves the overall fit of flexible resources in the market and also enables cloud-edge collaborative computing, using the local computing resources of edge controllers to distribute the computing pressure of virtual power plants.
[0048] The above-described specific implementations can be partially adjusted by those skilled in the art in different ways without departing from the principles and purpose of the present invention. The scope of protection of the present invention is defined by the claims and is not limited to the above-described specific implementations. All implementation schemes within the scope of the claims are bound by the present invention.
Claims
1. A bottom-up flexible resource hierarchical partitioning dynamic aggregation method, characterized in that, include: After modeling the internal flexibility resources of the virtual power plant and evaluating the adjustability of the flexibility resources according to the adjustability assessment index, the virtual power plant structure is divided into a three-layer model of cloud-edge-terminal, namely the virtual power plant cloud platform layer, edge control layer and flexibility resource terminal layer; then, it is aggregated in a bottom-up manner to form the edge control layer structure and the cloud platform layer structure.
2. The bottom-up flexible resource hierarchical partitioning and dynamic aggregation method according to claim 1, characterized in that, specifically include: Step 1) Model the internal flexibility resources of the virtual power plant, namely distributed photovoltaic, distributed wind power, distributed energy storage, gas turbine and building air conditioning; Step 2) The virtual power plant structure is divided into three layers: cloud-edge-device mode, namely the virtual power plant cloud platform layer, edge control layer, and flexible resource terminal layer. Among them: the flexible resource terminal layer consists of controllers for various flexible resources. Each controller reports the adjustable capability parameters of the flexible resources and decomposes and executes the scheduling instructions issued by the edge control layer; the edge control layer is a resource aggregation formed by multiple flexible resources in the same area, which is uniformly controlled by an edge controller; the virtual power plant cloud platform layer is the virtual power plant operation server located in the cloud. It coordinates the edge controllers in different areas, processes the adjustable information and operating status reported by the edge control layer, and interacts with the trading center and dispatch center to receive market trading instructions and decompose and issue them. Step 3) Aggregate flexibility resources in a bottom-up manner to form an edge control layer structure and a cloud platform layer structure.
3. The bottom-up flexible resource hierarchical partitioning dynamic aggregation method according to claim 2, characterized in that, The distributed photovoltaic modeling mentioned above specifically includes: in: Let k be the output power of the distributed photovoltaic system k under the external environment at time t; PV P represents the power loss factor caused by aging or damage to distributed photovoltaic panels. RP D represents the rated power of the distributed photovoltaic panel. t D represents the actual solar irradiance at the location of the distributed photovoltaic system at time t; STC Light intensity under standard test conditions (STC); T t PV Let T be the actual surface temperature of the distributed photovoltaic system at time t; STC Standard test temperature; C k,PV,t The cost of distributed photovoltaic energy consumption is represented by the curtailment penalty, where ο is the curtailment penalty coefficient. and These represent the actual and predicted output of distributed photovoltaic power during time period t, respectively.
4. The bottom-up flexible resource hierarchical partitioning dynamic aggregation method according to claim 2, characterized in that, The distributed wind power modeling mentioned above specifically includes: in: Let be the output power of distributed wind power k under the external environment at time t; This refers to the rated power of distributed wind power. v in v out v rated These are wind speed, cut-in wind speed, cut-out wind speed, and rated operating wind speed, respectively. Where: C k,W,t γ represents the energy cost of distributed wind power, and γ is the wind curtailment penalty coefficient. and These represent the actual and predicted output of distributed wind power during time period t, respectively.
5. The bottom-up flexible resource hierarchical partitioning dynamic aggregation method according to claim 2, characterized in that, The distributed energy storage modeling mentioned above specifically refers to the ratio between available energy and total energy capacity in the energy storage system, i.e., the state of charge of distributed energy storage k during time period t. Where: α ch and α dis These represent the charge / discharge efficiency of distributed energy storage; CAP represents the battery capacity of distributed energy storage. and Let be the charging and discharging power of distributed energy storage k during time period t. Since distributed energy storage cannot be in a state of charging and discharging simultaneously during operation, and there is an upper limit to the charging and discharging power of distributed energy storage, the constraints on the charging and discharging power of distributed energy storage include: Where: β k,t Let t be the 0-1 variable representing the charging and discharging state of distributed energy storage k at time t. A value of 1 indicates distributed energy storage is discharging, and a value of 0 indicates charging. and This represents the maximum charging and discharging power. Where: C k,ESS,t The energy cost of distributed energy storage; and The unit charging / discharging cost and maintenance cost of distributed energy storage are respectively considered.
6. The bottom-up flexible resource hierarchical partitioning dynamic aggregation method according to claim 2, characterized in that, The gas turbine modeling described above is specifically as follows: i∈Ω DG ,in: Let a be the output power of gas turbine k at time t; i and b i Ω represents the first-order coefficient and constant term of the piecewise linear function, respectively; ΔG(t,k) is the natural gas intake of gas turbine k at time t; i is the piecewise linear function segment number, Ω DG Given a set of segment numbers, the output constraints of the gas turbine include... in: and These are the minimum and maximum technical outputs allowed for the gas turbine, respectively. Where: a DG b DG and c DG This represents the cost coefficient for gas turbines.
7. The bottom-up flexible resource hierarchical partitioning dynamic aggregation method according to claim 2, characterized in that, The aforementioned building air conditioning modeling, i.e., the building energy storage expression: in: Let be the energy stored in building k at time t. To ensure the building's air conditioning system can withstand the maximum electrical power; Where: C k,AC,t For the energy cost of building air conditioning, a AC b AC and c AC This is the energy cost coefficient for building air conditioning.
8. The bottom-up flexible resource hierarchical partitioning dynamic aggregation method according to claim 2, characterized in that, The aforementioned flexibility resource assessment indicators include: 1) Maximum upward adjustment capability Where: P k,t The output power of the flexible resource k at time t. Minimize the technical effort required for flexibility resource k; 2) Maximum downward adjustment capability in: The upper limit of output for the flexibility resource k; 3) Response latency in: and These represent the response delays for upward and downward adjustments of the flexibility resource k, respectively; t ds,up and t ds,down These represent the moments when flexibility resources are adjusted upwards and downwards, respectively; t ss The moment when a flexible resource receives a response instruction; 4) Response duration in: and These represent the response durations for upward and downward adjustments to the flexibility resource k, respectively; t de,up and t de,down These represent the longest periods of time that the flexibility resources maintain their adjusted state after being adjusted upwards and downwards, respectively. 5) Adjustment rate R, divided into upward adjustment rate R k,up and downward adjustment rate R k,down , respectively, represent the maximum upward and downward adjustment power of flexible resources per unit time, corresponding to the ramp rate of traditional units, specifically: 6) Response time probability of insufficient battery Where: l represents the system operating state; p l The probability that the system is in a running state and the response power is insufficient; t l The duration of the response when the system is in running state l and there is insufficient power; 7) Expected response time when battery is low Wherein: the day is divided into I time periods; T i Let C be the number of time periods contained in the i-th time period; i L represents the theoretical maximum adjustment capacity of the flexibility resource in the i-th time period; i,t p[X≥C] represents the peak load power in the i-th time period; i -L i,t [This refers to the probability that the adjustable capacity deficit of flexible resources is greater than or equal to the reserve capacity.] 8) Response to insufficient power consumption expectations Where: N is the number of simulated hours; L I C represents the load power in the I-th simulation hour; I P represents the theoretical maximum adjustability of the flexibility resources during the I-th simulation hour. I (X) represents the probability of insufficient response power during the I-th simulation hour.
9. The bottom-up flexible resource hierarchical partitioning dynamic aggregation method according to claim 2, characterized in that, Step 3) specifically includes: 3.1 Constructing the Optimal Structure of the Edge Control Layer: Based on the above optimization results of the virtual power plant edge control layer structure, the behavior of aggregating flexible resource aggregates to form a virtual power plant is modeled as a dynamic alliance game problem, and solved using the merging and splitting method. The optimal alliance of the aggregates is the cloud platform layer structure of the virtual power plant. When the alliance game C=(E,v), |E|=E is the total number of members, that is, the number of flexible resource aggregates in the edge control layer, and the function v is called the characteristic function. The alliance S, that is, the virtual power plant S, is mapped to a real number, which corresponds to the sum of expected benefits obtained by the members in the alliance S through cooperation. Its calculation formula is: Wherein: the inner layer represents the game-theoretic bidding behavior of alliance S, p rofit (S) is the utility function of the alliance S; 3.2 Constructing the optimal structure for the cloud platform layer, specifically including: i) The network topology is represented by an undirected graph G = (N, B), where N is the set of network nodes, B is the set of branches, and the power flow distribution in the power network is the set of injected power at each node. t∈T, where: t is the system running time, T is the total number of time points, i b Node numbering; the set of load power of each node in the power network: in: For node i b The power of the connected load; ii) Form the optimal structure of the flexible resource aggregate to form the edge control layer of the virtual power plant. Let the set of aggregate partitioning results be: V(E,F) e ), Where: E represents the set of flexible resource aggregates within the system after aggregation; F e Let be the set of nodes contained within the e-th flexible resource aggregate, and if i∈F e , The variable is 0-1, used to determine the region where node i is located. When its value is 1, node i is located in region e. iii) An improved modularity metric is used to describe the internal communication cost of the aggregate, specifically: in: To improve the modularity index for internal communication costs within the flexible resource aggregate e; ij The communication cost between the flexible resources accessed by node i and node j, (x i ,y i Let ξ be the x and y coordinates of node i, and ζ be the communication cost per unit distance. e Cost of configuring edge controllers; α ij β is the sum of the communication costs between all flexible resources within the aggregate; i This is the sum of the communication costs between the flexibility resource at node i within the aggregate and other internal flexibility resources; iv) The optimal structure optimization model for the virtual power plant edge control layer includes... st Where: the decision variable is V(E,F) e The optimal virtual power plant edge control layer partitioning result set V(E,F) is obtained. e ) * The constraint is that the communication cost index of the flexible resources combined into an aggregate meets the maximum aggregation cost limit.
10. The bottom-up flexible resource hierarchical partitioning dynamic aggregation method according to claim 2, characterized in that, The principles of the alliance include: considering two non-overlapping alliances S1 and S2, a larger alliance is formed through cooperation when the following conditions are met: 1) The characteristic function of the major league is greater than the sum of the characteristic functions of the minor leagues. This criterion is called the overall criterion for the leagues, v(S1∪S2)>v(S1)+v(S2); 2) At least one member gains a greater benefit in the major leagues without causing other members to suffer a loss, i.e., the Pareto criterion, specifically: 3.3) Establish a market fit index for flexible resource virtual power plants as the alliance utility function, specifically: in: Let S be the market fit index for the virtual power plant. The formula combines relevant flexible resource adjustability indicators, indicating that the higher the adjustable upper and lower limits, the shorter the response time, and the longer the duration of the virtual power plant's market participation, the higher its market fit. and These are the normalized values of the corresponding indicators of internal flexibility resources of virtual power plant k.