Virtual power plant dynamic aggregation method and device based on double-layer decision and satisfaction

Abstract

This invention relates to the field of power resource dispatching technology, specifically to a method and device for dynamic aggregation of virtual power plants based on two-layer decision-making and satisfaction. The method includes: using the virtual power plant as the upper-layer decision-maker in a two-layer decision-making architecture, and the corresponding entities of the participants as lower-layer decision-makers, to determine the aggregation scheme; wherein, the upper-layer decision-maker uses a two-layer late-acceptance hill-climbing algorithm combined with a proxy objective function to iteratively optimize internal transaction price information, and distributes it to the lower-layer decision-makers through a blockchain network; the lower-layer decision-makers listen to and obtain information through smart contracts, and when trigger conditions are met, call their local participant model to calculate the optimal operating scheme and calculate satisfaction information based on the Frynell satisfaction model, and upload it to the blockchain network for the upper-layer decision-maker to adjust the internal transaction price information of the dispatching cycle. This effectively improves the flexibility of the virtual power plant in dispatching power resources, thereby improving dispatching efficiency.

Description

Technical Field

[0001] This invention relates to the field of power resource dispatching technology, specifically to a method and device for dynamic aggregation of virtual power plants based on two-level decision-making and satisfaction. Background Technology

[0002] With the increasing penetration of renewable energy sources such as wind and solar power into the power grid, their inherent randomness and volatility pose significant challenges to the stable operation of the power system. Virtual power plants, as an effective technology for aggregating various distributed energy resources such as distributed energy sources, energy storage systems, and controllable loads, are considered key to improving the flexibility of the power system. However, existing virtual power plant operating models generally have several shortcomings.

[0003] First, the aggregation methods are relatively rigid. Many virtual power plants use static long-term contracts or fixed agreements to bind resources. This approach lacks flexibility and cannot adapt to real-time fluctuations in market electricity prices and dynamic changes in resource status, resulting in low resource utilization efficiency. Second, in terms of control mode, traditional virtual power plants mostly adopt centralized dispatch, with operators directly controlling the operation of all aggregated resources. This one-way, mandatory control model ignores the economic interests and decision-making autonomy of resource owners, seriously affecting their enthusiasm for participation and failing to utilize resource dispatch. Therefore, a new aggregation method is urgently needed that can achieve flexible and efficient resource aggregation while protecting the interests of all parties, thereby improving the overall effectiveness of resource dispatch. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a method, apparatus and equipment for dynamic aggregation of virtual power plants based on two-level decision-making and satisfaction, so as to overcome the problem of poor energy dispatching effect of current virtual power plants.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] Firstly, this application provides a method for dynamic aggregation of virtual power plants based on two-level decision-making and satisfaction, including: The virtual power plant is used as the upper-level decision-maker in the two-layer decision-making architecture, and the subjects corresponding to the participants are used as the lower-level decision-makers in the two-layer decision-making architecture, so as to construct a two-layer decision-making framework. The aggregation scheme is determined using the aforementioned two-layer decision-making framework; The participants include distributed energy resources and user-side power load resources; The upper-level decision-makers use a bilevel late acceptance hill climbing (b-LAHC) algorithm combined with a proxy objective function to iteratively optimize internal transaction price information, and then publish the internal transaction price information to the lower-level decision-makers through the blockchain network. The lower-level decision-makers automatically listen to and obtain the internal transaction price information through smart contracts deployed on local edge computing devices, and execute the following decentralized decision-making process locally: The internal transaction price information is compared with the expected electricity price preset by the local lower-level decision-makers. When the triggering conditions are met, the local participant model is invoked to calculate the optimal operation plan. The operation plan includes at least the power aggregated with the virtual power plant in each scheduling cycle. Satisfaction information is calculated based on the preset Frynell satisfaction model. The operation plan and satisfaction information are uploaded to the blockchain network. The upper-level decision-makers obtain the operational plans and satisfaction information from all lower-level decision-makers in the blockchain network, and adjust the internal transaction price information of the scheduling cycle based on the satisfaction information to drive the system to converge to the Pareto optimal state.

[0007] Furthermore, in some embodiments of this application, the distributed energy resources include distributed photovoltaic systems, distributed energy storage systems, and micro gas turbines; the user-side power load resources include interruptible loads, stationary loads, and transferable loads. In this context, the entity corresponding to the distributed energy storage system and the entity corresponding to the distributed photovoltaic system each correspond to different lower-level decision-makers; or the entity corresponding to the distributed energy storage system and the entity corresponding to the distributed photovoltaic system form a joint entity corresponding to the same lower-level decision-maker.

[0008] Furthermore, in some embodiments of this application, the internal transaction price information includes the virtual power plant's internal purchase price and the virtual power plant's internal selling price; When the entity corresponding to the distributed energy storage system and the entity corresponding to the distributed photovoltaic system correspond to different lower-level decision-makers, the optimization objective of the upper-level decision-maker is to maximize the revenue of the virtual power plant. The corresponding objective function is determined based on the virtual power plant's internal purchase price, virtual power plant's internal selling price, virtual power plant cost, and virtual power plant's power purchase and sale constraints in each time period of the scheduling cycle. The virtual power plant cost includes load cost, photovoltaic cost, micro gas turbine cost, energy storage system cost, and network cost. The virtual power plant's power purchase and sale constraints are that the purchase price within the virtual power plant for each time period is lower than the upper limit of the purchase price within the virtual power plant for that time period, and the sale price within the virtual power plant for each time period is higher than the upper limit of the sale price within the virtual power plant for that time period.

[0009] Furthermore, in some embodiments of this application, the lower-level decision-maker responds to the internal transaction price information based on a smart contract and executes a decision-making process based on the corresponding participant model to determine the lower-level decision-maker's operating plan, including: Step 1: Monitor digitally signed internal transaction price information published by upper-level decision-makers on the blockchain network; Step 2: Analyze the internal transaction price information to extract the internal buy price and internal sell price of the virtual power plant; Step 3: For power generation resources, the smart contract determines whether the purchase price within the virtual power plant is higher than the expected electricity price set by the corresponding lower-level decision-maker. If it is higher, the power to be aggregated with the virtual power plant is determined based on the corresponding participant model. For purchased electricity resources, the smart contract determines whether the selling price within the virtual power plant is lower than the expected electricity price set by the corresponding lower-level decision-maker. If it is lower, the power to be aggregated with the virtual power plant is determined based on the corresponding participant model. Among them, power generation resources include photovoltaic power generation resources, micro gas turbines and distributed energy storage system discharge power resources, and power purchase resources include user-side power load resources and distributed energy storage system charging power resources; Step 4: After the transaction is triggered, the locally stored participant model is invoked. The participants of the power generation resources aim to maximize their own benefits and solve for the optimal aggregated power sequence under the premise of satisfying local operating constraints. When invoking the local participant model, all parameters involving the privacy of lower-level decision-makers are calculated within the local edge computing device and are not uploaded to the blockchain network.

[0010] Furthermore, in some embodiments of this application, the participant model includes a distributed energy storage model, a load model, a micro gas turbine model, and a distributed photovoltaic model; The constraints of the distributed energy storage model include charging and discharging power constraints and state of charge constraints. The constraints of the load model include chance constraints; The constraints of the micro gas turbine model include power constraints and ramping constraints; The constraints of the distributed photovoltaic model include power constraints.

[0011] Furthermore, in some embodiments of this application, the satisfaction information includes the expected electricity price, perceived value information, and perceived quality information of the corresponding lower-level decision-makers; Wherein, the expected electricity price is the price that triggers lower-level decision-makers to aggregate with the virtual power grid; the perceived value information is used to reflect the evaluation of lower-level decision-makers on the purchase price and the selling price within the virtual power plant; and the perceived quality information is used to reflect the evaluation of lower-level decision-makers on the benefits of participating in the aggregation of virtual power plants. The perceived value information includes a perceived value index, and the perceived quality information includes a perceived quality index. Specifically, for user-side power load resources, the perceived value index is determined based on the maximum and minimum purchase prices within the virtual power plant within a preset time range, as well as the current purchase price within the virtual power plant; and the perceived quality index is determined based on the maximum and minimum net costs of user-side power load resources within the target load resource range, as well as the current net costs of user-side power load resources. For distributed photovoltaic systems and micro gas turbines, the perceived value index is determined based on the maximum and minimum selling prices within the virtual power plant over a preset time range, as well as the current selling price within the virtual power plant; and the perceived quality index is determined based on the maximum and minimum net income within the target photovoltaic system and micro gas turbine range, as well as the current net income of the target photovoltaic system or micro gas turbine.

[0012] Furthermore, in some embodiments of this application, the upper-level decision-maker also adjusts its decision-making behavior based on the operating plan and the satisfaction information, including: Step 1: Obtain satisfaction information from each lower-level decision-maker in the blockchain network. The satisfaction information includes at least the perceived value index and perceived quality index of each lower-level decision-maker. Step 2: Based on the obtained perceived value index and perceived quality index, evaluate the incentive effect of the current internal transaction price on the lower-level decision-makers corresponding to power generation resources and the lower-level decision-makers corresponding to power purchase resources. Step 3: If the perceived value index of the lower-level decision-makers of power generation resources is lower than expected, it is determined that the current purchase price incentive within the virtual power plant is insufficient, and the purchase price within the virtual power plant will be increased in the next iteration; if the perceived value index is higher than expected, it is determined that the current purchase price within the virtual power plant is too high, and the purchase price within the virtual power plant will be decreased in the next iteration. Step 4: If the perceived quality index of the lower-level decision-makers in purchasing electricity resources is lower than expected, it is judged that the current internal selling price is too high, and the internal selling price of the virtual power plant will be reduced in the next iteration; if the perceived quality index is higher than expected, it is judged that the price is too low, and the internal selling price of the virtual power plant will be increased in the next iteration. Step 5: Repeat the above steps until the convergence condition is met. The convergence condition includes the change in the internal transaction price in multiple consecutive iterations being lower than a preset tolerance value, or reaching a preset maximum number of iterations.

[0013] Furthermore, in some embodiments of this application, the proxy objective function is used to estimate the net revenue of a virtual power plant corresponding to a set of internal transaction price schemes based on historical response data or load forecast data without calling the underlying smart contract, and the net revenue estimate does not include a satisfaction weighting factor.

[0014] Furthermore, in some embodiments of this application, the upper-level decision-maker employs a two-layer late acceptance hill-climbing algorithm combined with a proxy objective function to iteratively optimize internal transaction price information, and publishes the internal transaction price information to the lower-level decision-maker through the blockchain network, including: Step 1: Initialization and Greedy Descent An initial internal transaction price scheme is generated based on the time-of-use electricity price in the external electricity market; Define a proxy objective function, which is used to estimate the net revenue of a virtual power plant corresponding to a set of internal transaction price schemes based on historical response data or load forecast data without calling the underlying smart contract; The initial internal transaction price scheme is fine-tuned in a single time period, including randomly selecting a time period to increase or decrease the buying price by a preset step size, or to adjust the selling price synchronously. If the adjusted agent target value is better than the original value, the adjustment is accepted. This process is repeated until no improvement can be found, and a locally optimal price scheme is obtained. Step Two: Exploration at a Later Time Maintain a fixed-length historical record table. Each storage location in the historical record table is used to store the proxy objective function value of a specific iteration order during the historical iteration process. The historical record table is filled by multiplying the proxy objective function value of the current local optimal price solution by a preset noise factor during initialization. In each iteration, one of the following neighborhood operations is randomly selected to generate candidate price schemes: single-period price fine-tuning, price swapping between two adjacent periods, or overall price curve shifting; Calculate the proxy objective function value of the candidate price scheme and obtain the historical proxy objective value in the storage location corresponding to the current iteration index in the history table; If the proxy objective function value of the candidate price scheme is better than the proxy objective function value of the current scheme or better than the historical proxy objective value, then the candidate price scheme is accepted as the current scheme. Furthermore, if the proxy objective function value of the candidate price scheme is better than the preset multiple threshold of the current optimal proxy objective function value, a complete evaluation is initiated, including: sending the content transaction price information corresponding to the candidate price scheme to all lower-level smart contracts through the blockchain network, obtaining the real aggregate power and satisfaction information returned by each participant, and calculating the complete objective function value, which includes the actual revenue of the virtual power plant and a weighted penalty term based on the satisfaction information; If the complete objective function value is better than the historical best complete objective value, then update the optimal price plan; Repeat the above iterations until the termination condition is met, including reaching a preset maximum number of iterations or no improvement for multiple consecutive rounds; Step 3: Refining Once the termination condition is met, the optimal price solution will trigger a complete smart contract evaluation to obtain the final aggregation solution. The transaction price information corresponding to the final aggregation solution will then be published to the layer decision-makers through the blockchain network.

[0015] Secondly, this application also provides a virtual power plant dynamic aggregation device based on two-layer decision-making and satisfaction, including a processor and a memory, wherein the processor is connected to the memory: The processor is used to call and execute the program stored in the memory; The memory is used to store the program, which is at least used to execute the aforementioned virtual power plant dynamic aggregation method based on two-level decision-making and satisfaction.

[0016] This invention relates to the field of power resource dispatching technology, specifically to a method and device for dynamic aggregation of virtual power plants based on a two-layer decision-making and satisfaction model. The method includes: constructing the virtual power plant as the upper-layer decision-maker in a two-layer Stackelberg game architecture, and constructing the participants of distributed energy resources and user-side power load resources as the lower-layer decision-makers to determine a dynamic aggregation scheme. Specifically, the upper-layer decision-maker, based on external power market transaction information and its own profit maximization objective, combines a proxy objective function and a two-layer late-acceptance hill-climbing algorithm to generate internal buy-in prices and sell-out prices within the virtual power plant in each dispatching cycle, and publishes the digitally signed internal transaction price information through a blockchain network. The lower-layer decision-makers automatically monitor and obtain price information through smart contracts deployed on local edge devices, analyze the internal buy / sell prices of the virtual power plant and compare them with a preset expected electricity price. If a trigger condition is met, the local participant model is invoked to solve for the optimal aggregated power. Simultaneously, the perceived value index and perceived quality index are calculated based on the Frynell satisfaction model. Finally, the power sequence and the anonymized satisfaction index are digitally signed and uploaded to the blockchain. Upper-level decision-makers collect satisfaction information from all lower-level decision-makers via the blockchain and adjust transaction prices within a cycle based on perceived value and perceived quality indices: if the perceived value index of participants in power generation resource acquisition is low, the purchase price is increased; if the perceived quality index of participants in power purchase resource acquisition is low, the selling price is decreased. This closed-loop feedback is repeated until convergence to a Pareto optimal state. This invention significantly improves the flexibility, privacy protection, and scheduling efficiency of virtual power plant aggregation through a decentralized response mechanism driven by smart contracts and a satisfaction-based closed-loop control algorithm. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the virtual power plant dynamic aggregation method based on two-layer decision-making and satisfaction provided in an embodiment of the present invention.

[0019] Figure 2 This is a schematic diagram of the structure of a virtual power plant dynamic aggregation device based on two-layer decision-making and satisfaction provided in an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other implementation methods obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0021] Figure 1 This is a flowchart illustrating the virtual power plant dynamic aggregation method based on two-layer decision-making and satisfaction provided in an embodiment of the present invention. Please refer to [link / reference]. Figure 1 This embodiment may include the following steps: S101. The virtual power plant is used as the upper-level decision-maker in the two-layer decision-making architecture, and the subjects corresponding to the participants are used as the lower-level decision-makers in the two-layer decision-making architecture, so as to construct a two-layer decision-making framework.

[0022] S102. Determine the aggregation scheme through a two-level decision-making framework.

[0023] The participants include distributed energy resources and user-side power load resources. Upper-level decision-makers employ a two-layer late-acceptance hill-climbing algorithm combined with a proxy objective function to optimize internal transaction price information within each scheduling cycle. This internally signed transaction price information is then published to lower-level decision-makers via the blockchain network. Lower-level decision-makers automatically monitor and acquire this internal transaction price information through smart contracts deployed on local edge computing devices. They compare the internal transaction price with their preset expected electricity price, and when trigger conditions are met, they invoke their local participant model to calculate the optimal operating plan. This plan includes the power aggregated with virtual power plants in each scheduling cycle. Satisfaction information is calculated based on the Frynell satisfaction model, and the operating plan and satisfaction information are digitally signed and uploaded to the blockchain network. Based on the operating plan and satisfaction information, upper-level decision-makers iteratively adjust the internal transaction price information for each scheduling cycle to drive the system to converge to a Pareto optimal state.

[0024] Specifically, Virtual Power Plants (VPPs), as an emerging resource aggregation and management platform, theoretically enhance overall system flexibility and optimize operational efficiency by integrating distributed resources to participate in the electricity market and system regulation. However, existing VPP operating mechanisms generally rely on fixed agreements or long-term contracts to bind resources. This rigid binding not only raises the barrier to resource access but also limits the dynamic response capability of resources, making it difficult to adapt to the rapid changes in user-side resources. Furthermore, traditional VPPs often employ centralized control methods, which, while facilitating unified scheduling, also incur higher communication and coordination costs, affecting effective collaboration and flexible scheduling among multiple parties, and reducing the participation enthusiasm of the resource owners.

[0025] In this application, the VPP communicates with the power system's control center and trading center, and interacts with the power grid to achieve its full participation in power dispatch by aggregating user-side power load resources and distributed energy resources (DERs), including distributed photovoltaic systems, micro gas turbines, and distributed energy storage systems.

[0026] Specifically, VPP provides an aggregation platform for DERs and loads, i.e., user-side power load resources. Each participant, i.e., the owner of the DERs and loads, dynamically participates in the aggregation of VPP according to its own interests.

[0027] In practical applications, VPPs participate in power dispatch in the electricity market as price takers. Based on the purchase and sale prices of electricity in the external electricity market, they set internal transaction price information for virtual power plant aggregation participants. This includes transaction information based on the external electricity market and preset optimization objectives. In each dispatch cycle, they determine and publish internal transaction price information to lower-level decision-makers.

[0028] In this process, each participant's entity signs a smart contract with the VPP. Through the smart contract, the participant's entity can respond to the internal transaction price information determined by the VPP and decide whether to participate in the VPP's aggregation (or directly trade with the external electricity market) in each time period of the scheduling cycle. The participant's entity only exchanges price information, operational transaction details included in the operation plan, and satisfaction information with the VPP. It does not involve the privacy data of the participant's entity. At the same time, the participant's assets and power interface with the grid do not need to be changed, thereby reducing the transformation cost and the threshold and cost of aggregation with the VPP.

[0029] Furthermore, in this application, the satisfaction information is determined based on the Frynell model and consists of three parts: expected electricity price, perceived quality information, and perceived value information. The expected electricity price represents the price that the participant expects to aggregate with the VPP. In practical applications, the triggering condition for a participant to aggregate with the VPP is that the internal transaction price information within the VPP meets their expected electricity price requirement. The perceived value information reflects the participant's evaluation of the internal transaction price information determined by the VPP. The perceived quality information reflects the participant's evaluation of the benefits brought by VPP aggregation.

[0030] In practical applications, each participant can automatically determine its own operating plan through smart contracts based on the VPP's internal transaction price information and the buying and selling prices of the external electricity market. This plan includes the power it aggregates with the virtual power plant or the power it trades directly with the external electricity market in each scheduling cycle, and it feeds back the operating plan and satisfaction information to the VPP.

[0031] Based on this, after obtaining feedback on the operation plans and satisfaction information of various resources (i.e., the distributed energy resources and user-side power load resources mentioned above), the VPP adjusts its decision-making behavior, including comprehensively considering factors such as satisfaction information and network usage costs, aggregating internal resources and formulating trading strategies for the external power market, as well as adjusting the internal trading price information for each time period in the scheduling cycle.

[0032] Furthermore, in some embodiments of this application, the two-layer decision-making framework is constructed based on Stackelberg game theory.

[0033] It is understandable that the dynamic aggregation process of the aforementioned virtual power plant can be divided into two stages: the pricing management stage and the power management stage.

[0034] First, during the pricing management phase, the VPP (specifically, the VPP operator) uses a two-layer late acceptance hill-climbing algorithm to set the internal purchase price and internal selling price of the virtual power plant within the VPP based on the external market electricity price.

[0035] Then, during the power management phase, once the purchase price and sale price within the VPP's virtual power plant are determined, each participating entity automatically responds through locally deployed smart contracts, submitting operational plans and satisfaction information for each time period to the VPP operator.

[0036] Understandably, in aggregation models built on Stackelberg game theory, the payoff of participating entities depends on the electricity or power they exchange with the VPP (Virtual Power Provider) in different time periods. The determination of the power a participant contributes to aggregation in different time periods is primarily influenced by the VPP's internal transaction price information and the market-provided electricity price. The aggregation method provided in this application automates and protects privacy by having participating entities sign smart contracts with the VPP and deploying these contracts on edge devices. This allows participating entities to flexibly decide whether to participate in VPP aggregation within a specific time period based on different parties' electricity price information and their own needs. Simultaneously, each participating entity provides feedback on satisfaction information to the VPP, which iteratively adjusts its internal pricing behavior accordingly, thus forming a Stackelberg game structure with learning capabilities. In this structure, the VPP acts as the leader (upper-level decision-maker), and the participating entities are followers (lower-level decision-makers).

[0037] Furthermore, in the embodiments of this application, the entity corresponding to the distributed energy storage system and the entity corresponding to the distributed photovoltaic system can each correspond to different lower-level decision-makers; and the entity corresponding to the distributed energy storage system and the entity corresponding to the distributed photovoltaic system can also form a joint entity corresponding to the same lower-level decision-maker. The two schemes are described below: First, the entities corresponding to distributed energy storage systems and distributed photovoltaic systems correspond to different lower-level decision-makers. That is, when the distributed energy storage system (ESS) is directly operated by the VPP operator, the optimization objective of the upper-level decision-maker, i.e., the VPP, is to maximize the revenue of the virtual power plant. The corresponding objective function is determined based on the virtual power plant's internal purchase price, virtual power plant's internal selling price, virtual power plant cost, and virtual power plant's power purchase and sale constraints in each time period of the scheduling cycle. Its objective function is specifically expressed as follows: (1) (2) (3) (4) (5) (6) In the formula, represents the total revenue of the VPP, where max indicates the maximum; T represents the scheduling period, and t is the time period within the scheduling period. and VPP in The price of buying and selling electricity in the external electricity market during a given period; and VPP in The power that is constantly sold and bought in the external electricity market; Indicates the sampling time interval; , , , and These are respectively the load cost, photovoltaic cost, micro gas turbine cost, energy storage system cost, and network cost (for the sake of a more direct and simpler formula description, in the formula of this application, load corresponds to load, PV corresponds to distributed photovoltaic, GT corresponds to micro gas turbine, and ESS corresponds to distributed energy storage system).

[0038] in, Indicates load Trading power with VPP, The number of loads aggregated by VPP within the time period t; and They are respectively photovoltaic Trading power and time period with VPP The number of photovoltaic cells generated by VPP polymerization; and Gas turbine Trading power and time period with VPP The number of micro-units for VPP aggregation; , , The satisfaction index, which represents the feedback from participants in load, photovoltaic and micro gas turbines (the specific calculation formula is shown in Formula 52), allows VPP operators to better illustrate the relationship between the current situation and their ability to generate future revenue by using the satisfaction index as a measurable asset. and These are the ratios between current returns and potential future returns; The cost of energy storage systems, in practical applications, mainly consists of the attenuation costs caused by discharging and charging. It is a time period The energy storage capacity of VPP polymerization; This refers to the internal balanced power of the VPP; For grid connection electricity price; This represents the internal purchase price of the virtual power plant determined by VPP for time period t; This represents the internal selling price of the virtual power plant determined by VPP for the time period t.

[0039] It is understandable that the power generation resources participating in VPP aggregation mainly include photovoltaic power generation resources, micro gas turbines, and the discharge power resources of distributed energy storage systems; its energy consumption side, i.e., the electricity purchase resources, consists of user-side power load resources, i.e., user load demand and energy storage system charging power. Indicates VPP during the time period The power of interaction with the electricity market, then when VPPs need to purchase electricity from the external electricity market to meet their demand; conversely, when At that time, VPP can sell it to external markets, which is consistent with the formula above. and The specific calculation formula is as follows: (7) (8) (9) (10) (11) (12) In the formula, , These represent the saleable power and the required power to be purchased within the VPP, respectively, that can be traded with the external electricity market. It should be noted that the mixed-integer quadratic programming (MIQP) model used in this application cannot inherently address the sign of the optimization variable P. t vpp,tr The sign of the variable is determined by the properties of the objective function and the variables in the constraints. Therefore, in this application, additional 0-1 variables are introduced to construct the constraints, thereby calculating... and ;in, and These represent VPP in the time period. The electricity sales status and electricity purchase status are both Boolean variables; and in the above formula, "sale" represents electricity sales and "purchase" represents electricity purchase. and Energy storage time period The charging power and discharging power.

[0040] The internal balance of the virtual power plant is specifically represented as follows: (13) The specific constraints on the purchase and sale price of electricity for virtual power plants are as follows: (14) In the formula, and Time period The upper limit of the electricity sales price and purchase price within the virtual power plant means that the purchase price within the virtual power plant for each time period is lower than the upper limit of the purchase price within the virtual power plant for the corresponding time period, and the selling price within the virtual power plant for each time period is lower than the upper limit of the sales price within the virtual power plant for the corresponding time period.

[0041] Furthermore, in this application, the lower-level decision-maker responds to internal transaction price information based on smart contracts and executes a decision-making process based on the corresponding participant model to determine the strategy for participating in virtual power plant aggregation in each scheduling cycle, including: Step 1: Monitor digitally signed internal transaction price information published by upper-level decision-makers on the blockchain network; Step 2: Analyze the internal transaction price information and extract the internal buy price and sell price of the virtual power plant; Step 3: For power generation resources (including photovoltaic power generation resources, micro gas turbines, and distributed energy storage system discharge power resources), the smart contract determines whether the purchase price within the virtual power plant is higher than the expected electricity price set by the corresponding lower-level decision-maker. If it is higher, the power to be aggregated with the virtual power plant is determined based on the corresponding participant model. For purchased electricity resources (including user-side power load resources and distributed energy storage system charging power resources), the smart contract determines whether the selling price within the virtual power plant is lower than the expected electricity price set by the corresponding lower-level decision-maker. If it is lower, the power to be aggregated with the virtual power plant is determined based on the corresponding participant model. Step 4: After the transaction is triggered, the locally stored participant model is invoked. The participants of the power generation resources aim to maximize their own benefits and solve for the optimal aggregated power sequence under the premise of satisfying local operating constraints. When invoking the local participant model, all parameters involving the privacy of lower-level decision-makers are calculated in the local edge computing device and are not uploaded to the blockchain network. In practical applications, the game-theoretic decision-making strategy of the participants involves formulating operational plans as followers within each time period of the scheduling cycle to maximize their respective gains. Specifically, the three types of loads (interruptible loads, stationary loads, and transferable loads) can be managed by a single aggregator, while micro gas turbines and distributed photovoltaics are managed by their respective aggregators. Under this mechanism, each participant can independently choose to aggregate with a VPP or trade directly with the electricity market within each time period.

[0042] In this scheme, the participant models mentioned above specifically include a distributed energy storage model, a load model, a micro gas turbine model, and a distributed photovoltaic model. The constraints of the distributed energy storage model include charge / discharge power constraints and state of charge constraints; the constraints of the load model include opportunity constraints; the constraints of the micro gas turbine model include power constraints and ramping constraints; and the constraints of the distributed photovoltaic model include power constraints. The specific models are described below: For distributed energy storage systems, the objective function is to maximize operational benefits, specifically expressed as: (15) (16) In the formula, This indicates the operational benefits of a distributed energy storage system; These represent the discharge efficiency of the battery cell (correspondingly, in the formula below). (This indicates the charging efficiency of the battery cell). and They represent the first The rated capacity and unit battery cost of an ESS.

[0043] Its constraints include charging and discharging power constraints and state of charge constraints.

[0044] Specifically, the charging and discharging power constraint is expressed as follows: (17) (18) (19) In the formula, Indicates the first The rated power of each ESS; and Energy storage time period The charging power and discharging power; and They represent the first An ESS in a time period The discharge and charge states are both Boolean variables.

[0045] The charge state constraint is specifically expressed as follows: (20) (twenty one) (twenty two) In the formula, , , , , , They represent the first An ESS in a time period The state of charge (SOC), initial SOC value, final SOC value, lower limit of SOC, upper limit of SOC, and self-loss rate.

[0046] For user-side power load resources, in practical applications, load aggregators can combine demand responses from multiple users by reducing or shifting load. The objective function is to minimize the user-side electricity cost, specifically expressed as: (twenty three) In the formula, This represents the net cost of the load aggregator; Indicates the first The load during the time period Trading power with external markets; Indicates the first The response cost of a load is used to reflect the expenses incurred by the load in adjusting its electricity usage in response to fluctuations in electricity prices.

[0047] Based on this, this application considers the uncertainty of the load by introducing opportunity constraints, as detailed below: (twenty four) In the formula, and Other means the first The load during the time period Actual load values ​​and forecast values; This indicates that the load is at time [time]. The prediction error; This indicates the confidence level at which the constraint holds. It represents the probability of an event occurring.

[0048] Based on this, the following constraints can be added for different load types: When load When the load is interruptible, the following additional constraints apply: (25) (26) (27) (28) (29) In the formula, Indicates the first The load during the time period Reduced interruptible load; This represents the percentage of the maximum interruptible load. This indicates the transaction option status. When its value is 1, it means that the load chooses to perform aggregated transactions with VPP; otherwise, it chooses to trade with external markets. This indicates the expected electricity price for the load to participate in VPP transactions; and These are the interruption cost coefficients.

[0049] When load For fixed loads, which do not respond to electricity prices, the following constraints are added: (30) (31) (32) (33) For transferable loads, since they are equipped with their own energy storage systems, daily load transfer can be achieved through the charging and discharging of the energy storage. Therefore, their operating constraints are as shown in formulas (34-39), and they must also meet the energy storage operating constraints in formulas (17-22) above. It should be noted that the capacity of the self-owned energy storage system is usually small, mainly used to provide uninterrupted power or reduce electricity costs.

[0050] (34) (35) (36) (37) (38) (39) in, and They represent the first Each load's own energy storage system during the time period The charging power and discharging power; Indicates the first Each load's own energy storage system during the time period Net charge / discharge power, This represents the lifespan loss cost of the energy storage system, and its calculation method is the same as that of formula (16).

[0051] For micro gas turbines, the objective function aims to maximize their net revenue, as shown below: (40) (41) In the formula, This indicates the net revenue of a micro gas turbine. Indicates the first Taiwanese micro gas turbines during the time period Power traded with external electricity markets; Indicates its time period Operating costs , and These are the operating cost coefficients for micro gas turbines.

[0052] Its constraints include power constraints and ramping constraints.

[0053] The power constraint is expressed as follows: (42) (43) (44) In the formula, For the first The rated power of a micro gas turbine; This indicates the trading option status for the micro gas turbine. When its value is 1, it means that the turbine is selected to trade with a VPP; otherwise, it is traded with an external electricity market. This indicates the expected electricity price for micro gas turbines participating in VPP transactions.

[0054] The climbing constraint is expressed as: (45) In the formula, This represents the maximum ramp power of the micro gas turbine.

[0055] For the distributed photovoltaic model, the operating strategy is to maximize the operating revenue, and the objective function is as follows: (46) In the formula, It is the net income from distributed photovoltaic power. It is the first Each photovoltaic unit at time The power of interaction with the market.

[0056] In this application, for the aforementioned distributed photovoltaic model, opportunity constraints are introduced to account for the uncertainty of photovoltaic power generation, as detailed below: (47) In the formula, For the first Each photovoltaic unit at time Maximum power output, For the first Each photovoltaic unit at time Predicted power output; Indicates that photovoltaics are at a certain time. The output prediction error.

[0057] Based on this, a power constraint is added, specifically as follows: (48) (49) (50) (51) In the formula, This is the selected transaction status. When it is 1, it indicates a transaction with a virtual power plant; otherwise, it indicates a transaction with the market. This is the expected price for photovoltaic (PV) companies participating in virtual power plant transactions.

[0058] Furthermore, in this application, the perceived value information includes a perceived value index. For user-side power load resources, the perceived value index is determined based on the maximum and minimum purchase prices within a preset time range of the virtual power plant, and the current purchase price within the virtual power plant; and a perceived quality index is determined based on the maximum and minimum net costs of user-side power load resources within the target load resource range, and the current net cost of user-side power load resources. For distributed photovoltaic systems and micro gas turbines, the perceived value index is determined based on the maximum and minimum selling prices within a preset time range of the virtual power plant, and the current selling price within the virtual power plant; and a perceived quality index is determined based on the maximum and minimum net revenue within the target photovoltaic system and micro gas turbine range, and the current net revenue of the target photovoltaic system or micro gas turbine. Additionally, in this application, a satisfaction index can be calculated based on the above two indices and preset coefficients. In some embodiments, the two indices can also be combined to generate a comprehensive index. The specific calculation formula is as follows: (52) (53) (54) (55) (56) In the formula, It is a composite index generated by combining two indices. , These are the perceived value index and the perceived quality index (it can be understood that in these two indices, H represents specific distributed energy resources or user-side power load resources, such as distributed photovoltaic systems (PV), micro gas turbines (GT), or user-side power load resources (load)). and This is the proportionality coefficient. Represents the mean. This represents the net cost of the load or the net benefit of the photovoltaic and gas turbine systems. As shown in formulas 53-56 above, in this application, a separate calculation formula is used for user-side power load resources when calculating the perceived value index and perceived quality index, while the same formula can be used for distributed photovoltaic systems and micro gas turbines.

[0059] It should be noted that the above calculation of the perceived value index is performed within different preset time ranges (such as the current scheduling cycle). The selling price and buying price within the virtual power plant are different for each time period. The maximum and minimum values ​​mentioned above are the maximum and minimum values ​​for the entire time period. and This refers to the value at the current time point; and when calculating the perceived quality index, comparisons are made across different distributed photovoltaic systems and micro gas turbines. Therefore, the above maximum and minimum values ​​represent the maximum and minimum values ​​among different distributed photovoltaic systems and micro gas turbines (micro gas turbines are not considered when calculating distributed photovoltaic systems, and distributed photovoltaic systems are not considered when calculating micro gas turbines). This refers to the value of the resource currently being analyzed (i.e., a micro gas turbine or a distributed photovoltaic system).

[0060] Furthermore, as mentioned above, in the virtual power plant dynamic aggregation method based on two-level decision-making and satisfaction provided in this application, the entity corresponding to the distributed energy storage system and the entity corresponding to the distributed photovoltaic system can also be a consortium corresponding to the same lower-level decision-maker. That is, the distributed energy storage system and the distributed photovoltaic system jointly participate in the virtual power plant. In this scenario, their operating strategy is to respond to the internal transaction price information determined by the VPP in order to maximize the benefits of the consortium.

[0061] Specifically, compared to the scenario where the entities corresponding to the distributed energy storage system and the distributed photovoltaic system correspond to different lower-level decision-makers, for the virtual power plant, the variables related to energy storage in the above formulas 1-14 can be directly ignored. At the same time, the models (i.e., including objective functions and constraints) corresponding to the user-side power load resources and micro gas turbines adopt the same variables or models as above, while the models of the above distributed energy storage system and distributed photovoltaic system are transformed into a joint model of photovoltaic-energy storage system. The objective function is shown in formula 5, and the constraints of the above formulas 17-22 are retained, as well as the constraints of photovoltaic-energy storage system and external transactions are added.

[0062] The objective function of the joint photovoltaic-energy storage system model is specifically expressed as follows: (57) in, Net income from photovoltaic-energy storage systems; and These represent the revenue from the photovoltaic-energy storage system from the external electricity market and from the virtual power plant, respectively. For the internal power balance of the photovoltaic-energy storage system.

[0063] For photovoltaic-energy storage systems, external trading constraints specifically include system output constraints and trading constraints.

[0064] It should be noted that in this scenario, because the charging power of energy storage (i.e., distributed energy storage system) may be greater than the power generation of photovoltaic (i.e., distributed photovoltaic system), external charging power is required. Therefore, the power demand of energy storage and the associated photovoltaic output constraints are as follows (by introducing opportunity constraints to address the uncertainty of photovoltaics): (58) (59) (60) (61) In the formula, It is an intermediate variable for the net purchase and sale of electricity in photovoltaic-energy storage systems. When it is negative, it means that electricity needs to be purchased from virtual power plants or the market. and These respectively represent the photovoltaic-energy storage system in Electricity sold and bought at the same time; and The variables represent the states of electricity sales and purchases, and are 0-1 variables.

[0065] The joint venture entity for the photovoltaic-energy storage system can choose to trade with the electricity market or with virtual power plants, with specific constraints as follows: (62) (63) (64) (65) In the formula, This indicates the revenue generated by the consortium of photovoltaic-energy storage systems and the aggregation of VPP (Virtual Power Plant). This indicates the revenue generated by the joint venture of the photovoltaic-energy storage system in transactions with the external electricity market; and It is a 0-1 variable used to indicate the transaction status between the joint entity of the photovoltaic-energy storage system and the VPP. A value of 1 indicates a transaction with a virtual power plant, while a value of 0 indicates a transaction with the electricity market. This indicates the expected electricity price for a consortium of photovoltaic-energy storage system participants in virtual power plant transactions.

[0066] The model also includes internal balance constraints for the photovoltaic-energy storage system, specifically represented as follows: (66) The corresponding internal equilibrium constraints of the VPP are expressed as follows: (67) Furthermore, based on the above model, the internal transaction price optimization problem of upper-level decision-makers has complex characteristics such as high dimensionality, nonlinearity, and multi-period coupling. If the lower-level smart contract is called for a complete evaluation in each iteration, it will incur huge overhead due to blockchain communication and distributed computing, and traditional local search methods are prone to getting trapped in suboptimal pricing. To address these difficulties, this application provides a two-layer late acceptance hill-climbing algorithm (b-LAHC). This algorithm uses a proxy objective function to quickly screen solutions (without triggering smart contracts), escapes local extrema through a late acceptance mechanism, has fixed parameters for good robustness, and can converge to Pareto optimality in conjunction with the satisfaction closed loop. The algorithm consists of the following three stages.

[0067] The first phase involves initialization and greedy descent. Based on the time-of-use electricity price in the external electricity market, an initial internal trading price scheme is generated, including an initial internal buy-in price and sell-out price sequence. A proxy objective function is defined, which is used to quickly estimate the net revenue of the virtual power plant corresponding to a set of price schemes without calling the underlying smart contract. The expression for the proxy objective function is: (1) in, , This represents the initial price generated. , , These are user-side load, photovoltaic, and gas turbine data, respectively, obtained based on historical data or load forecasts. The constraints of the proxy objective are the constraints of the upper-level objective function. This proxy objective function does not contain any satisfaction weighting factors or lower-level smart contracts, therefore its calculation speed is much faster than the full objective function value.

[0068] During the greedy descent phase, the current price strategy is fine-tuned in a single time period: a time period is randomly selected, and the determined buy price is increased or decreased by a preset step size, or the sell price is adjusted simultaneously. If the adjusted proxy objective value is better than the original value, the adjustment is accepted. This process is repeated until no further improvement can be found, resulting in the proxy objective function value under the locally optimal price strategy.

[0069] The second phase involves late-stage exploration. A fixed-length historical record table is maintained, with the length denoted as . During initialization, each position in the record table is filled with the proxy objective function value of the locally optimal price scheme in the first stage multiplied by a random factor uniformly distributed within the range [0.98-1.02]. A threshold is then set. And the maximum number of iterations. Perform the following operations in each iteration: Step 1: Randomly select a neighboring operation (single-period price fine-tuning, price swap between two adjacent periods, or overall price curve shift) to generate candidate price schemes.

[0070] Step 2: Calculate the surrogate target value for the candidate price scheme and obtain the historical value corresponding to the current iteration index from the history table. The corresponding position in the history table is the current iteration number divided by... The remainder.

[0071] Step 3: If the proxy target value of the candidate price plan is better than the proxy target value of the current plan, or better than the historical value in the corresponding storage location in the history table, then accept the candidate price plan as the new current plan, and store its proxy target value in the same storage location in the history table (overwriting the original value).

[0072] Step 4: If the proxy target value of the candidate price scheme is better than the current optimal proxy target value and the threshold. The product of these factors triggers a full evaluation: the candidate price scheme is sent to all lower-level smart contracts via the blockchain network, the actual aggregation power and satisfaction information returned by each participant are obtained, and the complete objective function value is calculated according to the complete two-layer dynamic aggregation framework. Step 5: If the current complete objective function value is better than the historical best complete objective value, then update the optimal price scheme and its corresponding complete objective value. Repeat the above iterations until the preset maximum number of iterations is reached or there is no improvement for several consecutive rounds.

[0073] The third phase triggers a mandatory full smart contract evaluation of the optimal price scheme obtained in the second phase, in which all lower-level decision-makers provide real responses. During the evaluation, the lower-level smart contract automatically verifies the operational constraints of each participant (including upper and lower limits of energy storage state of charge, charging and discharging power limits, load interruptibility / transferability constraints, gas turbine ramping constraints, photovoltaic output range, etc.). The upper-level decision-makers verify the global power balance constraints based on the returned results, and finally output an aggregated scheme that satisfies all physical constraints.

[0074] The virtual power plant dynamic aggregation method based on two-layer decision-making and satisfaction provided in this application is based on a two-layer dynamic aggregation mechanism constructed by Stackelberg game theory. It adopts a two-layer late acceptance hill climbing algorithm combined with a proxy objective function to efficiently optimize the upper-layer pricing. The proxy objective function uses historical or predictive data to quickly estimate the net revenue of the pricing scheme without frequently triggering the lower-layer smart contract, which greatly reduces the computational overhead. By using virtual power plants as the dominant players in the game, internal transaction price information is formulated and published in each scheduling cycle. This guides various distributed energy resources and user-side power load resources to dynamically decide whether to join the virtual power plant for aggregation and trading based on their own operating costs, external electricity market prices, and expected revenue. This effectively breaks through the existing one-way control mode of virtual power plant operation, which is mainly based on centralized scheduling and passive participant response. It significantly improves the flexibility of resource access and the adaptability of aggregation strategies. At the same time, by introducing satisfaction information as a key indicator to measure the trading willingness and aggregation effect of participants, it comprehensively considers users' electricity price expectations, perceived value in the trading process, and actual service quality. Information interaction with the virtual power plant is carried out through smart contracts to ensure that users participate in dynamic games while maintaining privacy. This not only enhances users' predictability and trust in the aggregation results but also helps to build a more sustainable multi-party collaborative environment. Moreover, for energy storage systems, two operating strategies are designed to further improve flexibility and greatly enhance the effectiveness of virtual power plant power resource scheduling.

[0075] Based on the same inventive concept, the present invention also provides a virtual power plant dynamic aggregation device based on two-layer decision-making and satisfaction, for implementing the above method embodiments. Figure 2 This is a schematic diagram of the structure of a virtual power plant dynamic aggregation device based on two-layer decision-making and satisfaction provided in an embodiment of the present invention, as shown below. Figure 2 As shown, the virtual power plant dynamic aggregation device based on two-layer decision-making and satisfaction in this embodiment includes a processor 21 and a memory 22, with the processor 21 connected to the memory 22. The processor 21 is used to call and execute the program stored in the memory 22; the memory 22 is used to store the program, which is at least used to execute the virtual power plant dynamic aggregation method based on two-layer decision-making and satisfaction in the above embodiments.

[0076] The specific implementation scheme of the virtual power plant dynamic aggregation device based on two-layer decision-making and satisfaction provided in this application embodiment can refer to the implementation scheme of the virtual power plant dynamic aggregation method based on two-layer decision-making and satisfaction in any of the above embodiments, and will not be repeated here.

[0077] It is understood that the same or similar parts in the above embodiments can be referred to each other, and the contents not described in detail in some embodiments can be referred to the same or similar contents in other embodiments.

[0078] It should be noted that in the description of this invention, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, in the description of this invention, unless otherwise stated, "a plurality of" means at least two.

[0079] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

[0080] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0081] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it includes one or a combination of the steps of the method embodiments.

[0082] Furthermore, the functional units in the various embodiments of the present invention can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0083] The storage media mentioned above can be read-only memory, disk, or optical disk, etc.

[0084] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0085] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for dynamic aggregation of virtual power plants based on two-level decision-making and satisfaction, characterized in that, include: The virtual power plant is used as the upper-level decision-maker in the two-layer decision-making architecture, and the subjects corresponding to the participants are used as the lower-level decision-makers in the two-layer decision-making architecture, so as to construct a two-layer decision-making framework. The aggregation scheme is determined using the aforementioned two-layer decision-making framework; The participants include distributed energy resources and user-side power load resources; The upper-level decision-makers use a two-layer late acceptance hill-climbing algorithm combined with a proxy objective function to iteratively optimize the internal transaction price information, and publish the internal transaction price information to the lower-level decision-makers through the blockchain network. The lower-level decision-makers automatically listen to and obtain the internal transaction price information through smart contracts deployed on local edge computing devices, and execute the following decentralized decision-making process locally: The internal transaction price information is compared with the expected electricity price preset by the local lower-level decision-makers. When the triggering conditions are met, the local participant model is invoked to calculate the optimal operation plan. The operation plan includes at least the power aggregated with the virtual power plant in each scheduling cycle. Satisfaction information is calculated based on the preset Frynell satisfaction model. The operation plan and satisfaction information are uploaded to the blockchain network. The upper-level decision-makers obtain the operational plans and satisfaction information from all lower-level decision-makers in the blockchain network, and adjust the internal transaction price information of the scheduling cycle based on the satisfaction information to drive the system to converge to the Pareto optimal state.

2. The virtual power plant dynamic aggregation method based on two-level decision-making and satisfaction as described in claim 1, characterized in that, The distributed energy resources include distributed photovoltaic systems, distributed energy storage systems, and micro gas turbines; the user-side power load resources include interruptible loads, stationary loads, and transferable loads. In this context, the entity corresponding to the distributed energy storage system and the entity corresponding to the distributed photovoltaic system each correspond to different lower-level decision-makers; or the entity corresponding to the distributed energy storage system and the entity corresponding to the distributed photovoltaic system form a joint entity corresponding to the same lower-level decision-maker.

3. The virtual power plant dynamic aggregation method based on two-level decision-making and satisfaction as described in claim 2, characterized in that, The internal transaction price information includes the internal purchase price and the internal selling price of the virtual power plant. When the entity corresponding to the distributed energy storage system and the entity corresponding to the distributed photovoltaic system correspond to different lower-level decision-makers, the optimization objective of the upper-level decision-maker is to maximize the revenue of the virtual power plant. The corresponding objective function is determined based on the virtual power plant's internal purchase price, virtual power plant's internal selling price, virtual power plant cost, and virtual power plant's power purchase and sale constraints in each time period of the scheduling cycle. The virtual power plant cost includes load cost, photovoltaic cost, micro gas turbine cost, energy storage system cost, and network cost. The virtual power plant's power purchase and sale constraints are that the purchase price within the virtual power plant for each time period is lower than the upper limit of the purchase price within the virtual power plant for that time period, and the sale price within the virtual power plant for each time period is higher than the upper limit of the sale price within the virtual power plant for that time period.

4. The virtual power plant dynamic aggregation method based on two-level decision-making and satisfaction as described in claim 3, characterized in that, The lower-level decision-makers respond to the internal transaction price information based on smart contracts and execute a decision-making process based on the corresponding participant model to determine the lower-level decision-makers' operating plan, including: Step 1: Monitor digitally signed internal transaction price information published by upper-level decision-makers on the blockchain network; Step 2: Analyze the internal transaction price information to extract the internal buy price and internal sell price of the virtual power plant; Step 3: For power generation resources, the smart contract determines whether the purchase price within the virtual power plant is higher than the expected electricity price set by the corresponding lower-level decision-maker. If it is higher, the power to be aggregated with the virtual power plant is determined based on the corresponding participant model. For purchased electricity resources, the smart contract determines whether the selling price within the virtual power plant is lower than the expected electricity price set by the corresponding lower-level decision-maker. If it is lower, the power to be aggregated with the virtual power plant is determined based on the corresponding participant model. Among them, power generation resources include photovoltaic power generation resources, micro gas turbines and distributed energy storage system discharge power resources, and power purchase resources include user-side power load resources and distributed energy storage system charging power resources; Step 4: After the transaction is triggered, the locally stored participant model is invoked. The participants of the power generation resources aim to maximize their own benefits and solve for the optimal aggregated power sequence under the premise of satisfying local operating constraints. When invoking the local participant model, all parameters involving the privacy of lower-level decision-makers are calculated within the local edge computing device and are not uploaded to the blockchain network.

5. The virtual power plant dynamic aggregation method based on two-level decision-making and satisfaction as described in claim 4, characterized in that, The participant models include distributed energy storage models, load models, micro gas turbine models, and distributed photovoltaic models; The constraints of the distributed energy storage model include charging and discharging power constraints and state of charge constraints. The constraints of the load model include chance constraints; The constraints of the micro gas turbine model include power constraints and ramping constraints; The constraints of the distributed photovoltaic model include power constraints.

6. The virtual power plant dynamic aggregation method based on two-level decision-making and satisfaction as described in claim 1, characterized in that, The satisfaction information includes the expected electricity price, perceived value information, and perceived quality information of the corresponding lower-level decision-makers. Wherein, the expected electricity price is the price that triggers lower-level decision-makers to aggregate with the virtual power grid; the perceived value information is used to reflect the evaluation of lower-level decision-makers on the purchase price and the selling price within the virtual power plant; and the perceived quality information is used to reflect the evaluation of lower-level decision-makers on the benefits of participating in the aggregation of virtual power plants. The perceived value information includes a perceived value index, and the perceived quality information includes a perceived quality index. Specifically, for user-side power load resources, the perceived value index is determined based on the maximum and minimum purchase prices within the virtual power plant within a preset time range, as well as the current purchase price within the virtual power plant; and the perceived quality index is determined based on the maximum and minimum net costs of user-side power load resources within the target load resource range, as well as the current net costs of user-side power load resources. For distributed photovoltaic systems and micro gas turbines, the perceived value index is determined based on the maximum and minimum selling prices within the virtual power plant over a preset time range, as well as the current selling price within the virtual power plant; and the perceived quality index is determined based on the maximum and minimum net income within the target photovoltaic system and micro gas turbine range, as well as the current net income of the target photovoltaic system or micro gas turbine.

7. The virtual power plant dynamic aggregation method based on two-level decision-making and satisfaction as described in claim 2, characterized in that, The upper-level decision-makers also adjust their decision-making behavior based on the operational plan and the satisfaction information, including: Step 1: Obtain satisfaction information from each lower-level decision-maker in the blockchain network. The satisfaction information includes at least the perceived value index and perceived quality index of each lower-level decision-maker. Step 2: Based on the obtained perceived value index and perceived quality index, evaluate the incentive effect of the current internal transaction price on the lower-level decision-makers corresponding to power generation resources and the lower-level decision-makers corresponding to power purchase resources. Step 3: If the perceived value index of the lower-level decision-makers of power generation resources is lower than expected, it is determined that the current purchase price incentive within the virtual power plant is insufficient, and the purchase price within the virtual power plant will be increased in the next iteration; if the perceived value index is higher than expected, it is determined that the current purchase price within the virtual power plant is too high, and the purchase price within the virtual power plant will be decreased in the next iteration. Step 4: If the perceived quality index of the lower-level decision-makers in purchasing electricity resources is lower than expected, it is judged that the current internal selling price is too high, and the internal selling price of the virtual power plant will be reduced in the next iteration; if the perceived quality index is higher than expected, it is judged that the price is too low, and the internal selling price of the virtual power plant will be increased in the next iteration. Step 5: Repeat the above steps until the convergence condition is met. The convergence condition includes the change in the internal transaction price in multiple consecutive iterations being lower than a preset tolerance value, or reaching a preset maximum number of iterations.

8. The virtual power plant dynamic aggregation method based on two-level decision-making and satisfaction as described in claim 1, characterized in that, The proxy objective function is used to estimate the net revenue of a virtual power plant corresponding to a set of internal transaction price schemes based on historical response data or load forecast data without calling the underlying smart contract, and the net revenue estimate does not include a satisfaction weighting factor.

9. The virtual power plant dynamic aggregation method based on two-level decision-making and satisfaction as described in claim 1, characterized in that, The upper-level decision-makers employ a two-layer late-acceptance hill-climbing algorithm combined with a proxy objective function to iteratively optimize internal transaction price information, and then publish this information to lower-level decision-makers via the blockchain network, including: Step 1: Initialization and Greedy Descent An initial internal transaction price scheme is generated based on the time-of-use electricity price in the external electricity market; Define a proxy objective function, which is used to estimate the net revenue of a virtual power plant corresponding to a set of internal transaction price schemes based on historical response data or load forecast data without calling the underlying smart contract; The initial internal transaction price scheme is fine-tuned in a single time period, including randomly selecting a time period to increase or decrease the buying price by a preset step size, or to adjust the selling price synchronously. If the adjusted agent target value is better than the original value, the adjustment is accepted. This process is repeated until no improvement can be found, and a locally optimal price scheme is obtained. Step Two: Exploration at a Later Time Maintain a fixed-length historical record table. Each storage location in the historical record table is used to store the proxy objective function value of a specific iteration order during the historical iteration process. The historical record table is filled by multiplying the proxy objective function value of the current local optimal price solution by a preset noise factor during initialization. In each iteration, one of the following neighborhood operations is randomly selected to generate candidate price schemes: single-period price fine-tuning, price swapping between two adjacent periods, or overall price curve shifting; Calculate the proxy objective function value of the candidate price scheme and obtain the historical proxy objective value in the storage location corresponding to the current iteration index in the history table; If the proxy objective function value of the candidate price scheme is better than the proxy objective function value of the current scheme or better than the historical proxy objective value, then the candidate price scheme is accepted as the current scheme. Furthermore, if the proxy objective function value of the candidate price scheme is better than the preset multiple threshold of the current optimal proxy objective function value, a complete evaluation is initiated, including: sending the content transaction price information corresponding to the candidate price scheme to all lower-level smart contracts through the blockchain network, obtaining the real aggregate power and satisfaction information returned by each participant, and calculating the complete objective function value, which includes the actual revenue of the virtual power plant and a weighted penalty term based on the satisfaction information; If the complete objective function value is better than the historical best complete objective value, then update the optimal price plan; Repeat the above iterations until the termination condition is met, including reaching a preset maximum number of iterations or no improvement for multiple consecutive rounds; Step 3: Refining Once the termination condition is met, the optimal price solution will trigger a complete smart contract evaluation to obtain the final aggregation solution. The transaction price information corresponding to the final aggregation solution will then be published to the layer decision-makers through the blockchain network.

10. A virtual power plant dynamic aggregation device based on two-layer decision-making and satisfaction, characterized in that, It includes a processor and a memory, wherein the processor is connected to the memory: The processor is used to call and execute the program stored in the memory; The memory is used to store the program, which is at least used to execute the virtual power plant dynamic aggregation method based on two-level decision-making and satisfaction as described in any one of claims 1-9.

Images