Virtual power plant low-carbon bargaining optimization method based on differentiated multi-agent cooperative game

Abstract

The invention relates to the technical field of virtual power plant differentiated contribution distribution and distributed carbon transaction under a multi-agent cooperation framework, and provides a differentiated multi-agent cooperative game-based virtual power plant low-carbon bargaining optimization method, and the method comprises the steps: constructing a multi-virtual power plant-load aggregator MVPP-LA cooperation alliance framework; a wind power system, a photovoltaic system, a combined heat and power generation system and an energy storage system are integrated, supply and demand interaction is enhanced, and a carbon emission quota distribution mechanism, a carbon transaction mechanism framework, a cooperative objective function and an energy system operation constraint condition based on energy interaction are constructed. Cooperative benefits are quantified through a Nash bargaining model, and a distributed adaptive iterative algorithm is adopted for solving, so that fair benefit distribution is realized, and carbon emission and transaction cost are reduced. The method plays a positive role in promoting the utilization of renewable energy sources, improving the utilization efficiency of energy sources and enhancing the flexibility and stability of the system.

Description

Technical Field

[0001] The present invention relates to the technical field of differentiated contribution allocation of virtual power plants and distributed carbon trading under a multi-agent cooperation framework, and in particular to a low-carbon bargaining optimization method for virtual power plants based on differentiated multi-agent collaborative game. Background Art

[0002] With the growth of energy demand, the number of virtual power plants (VPPs) continues to increase. The energy interaction between multiple VPPs and the supply and demand interaction between multiple virtual power plants (MVPPs) and load aggregators (LAs) have attracted widespread attention. Compared with the independent operation of a single VPP, the collaborative operation of MVPPs can improve energy stability through multi-energy complementarity. However, since each VPP belongs to different stakeholders, the cooperation between different subjects involves information leakage. As a distributed optimization algorithm, the Alternating Direction Method of Multipliers (ADMM) only requires limited information to be exchanged between participants, which greatly protects privacy. At present, many studies have used ADMM to solve optimization problems in multi-agent cooperation, including: MVPP and operator collaborative operation optimization, energy trading between MVPP and distribution network, energy demand response between MVPP and LA, and energy flow optimization between power and heat networks in different VPPs. In this algorithm, the selection of the initial values ​​of Lagrange multipliers and penalty factors will affect the convergence speed and is affected by the complexity of the problem.

[0003] Usually, game theory is used to describe the interaction of interests among multiple subjects. Current research mostly uses cooperative games to characterize transactions and benefit distribution among multiple energy systems. One study proposed a cooperative game model for multi-park integrated energy systems suitable for industrial, commercial and residential areas. Other studies have considered the impact of cooperative games among various microgrids on the flexible consumption of regional renewable energy. The above studies mostly use the Nash bargaining model to achieve equal distribution of benefits. However, the variability of participants may lead to different contributions, so it is not fair to distribute benefits equally. For cooperative game models, it is particularly important to study a distribution mechanism based on differentiated contributions to ensure the fairness of benefit distribution.

[0004] As the carbon trading mechanism continues to improve, MVPP low-carbon operation has become a research hotspot. Many existing technologies focus on optimizing the existing centralized trading mechanism and its impact on carbon emissions. There are few studies on distributed carbon trading under the framework of multi-agent cooperation, and the coupling relationship between MVPP energy and carbon trading still needs to be further explored.

[0005] In terms of the allocation of Chinese Emission Allowance (CEA) to VPPs, most existing technologies use the baseline value method. Some existing technologies use the unit carbon emissions of power and heating equipment as the baseline value and calculate CEA based on the actual power generation. Other literature assumes that the purchased electricity comes from coal-fired units, and uses the unit carbon emissions of coal-fired power generation units as the baseline value. For heat and power coupling equipment such as cogeneration, CEA is calculated based on different baseline values ​​of electrical output and thermal output. However, in the case of MVPP cooperation, due to differences in different distributed energy resources (DER), each VPP contributes differently to total carbon emissions. Therefore, using a fixed baseline value cannot achieve a reasonable allocation of CEA.

[0006] In summary, in the existing technology, the initial value selection of Lagrange multipliers and penalty factors in distributed optimization algorithms (such as the alternating direction multiplier method, ADMM) has a significant impact on the convergence speed and performance of the algorithm. However, the selection of these initial values ​​often relies on experience or trial and error, and there is a lack of systematic methods to determine the optimal initial values. In existing cooperative game models, the Nash bargaining model is often used to achieve equal distribution of benefits, but this method ignores the differentiated contributions of participants to the growth of alliance benefits, resulting in unfair distribution of benefits. In particular, when the contribution levels of participants are different, the average distribution will dampen the enthusiasm of high contributors. Existing carbon trading mechanisms are mostly centralized, and less consideration is given to distributed carbon trading under the framework of multi-agent cooperation. In addition, although the ladder-type carbon trading mechanism has been proposed, there is insufficient research on its application effect in actual multi-agent cooperation and its coupling relationship with energy trading. Existing studies mostly use the baseline method to allocate carbon emission allowances (CEA), but this method fails to fully consider the differences in energy resources and supply characteristics of each VPP in the case of cooperation among multiple virtual power plants (MVPPs), resulting in unreasonable CEA allocation. Traditional distributed optimization algorithms often face problems such as slow convergence and low precision when solving large-scale complex problems. Especially in the framework of multi-agent cooperation, the practicality and universality of the algorithms are limited.

[0007] In view of the above problems, designing a low-carbon bargaining optimization method for virtual power plants based on differentiated multi-agent collaborative game has important practical significance. Summary of the invention

[0008] In order to solve the problems existing in the background technology, the present invention provides a virtual power plant low-carbon bargaining optimization method based on differentiated multi-agent collaborative game, which includes the following steps: S1. Build a multi-virtual power plant-load aggregator MVPP-LA cooperation alliance framework: Integrate wind turbines WT, photovoltaics PV, combined heat and power equipment CHP and energy storage systems ESS to form multiple virtual power plants MVPP; integrate users within the MVPP area through load aggregators LA to form an MVPP-LA cooperation alliance to enhance supply and demand interaction; S11. Construct a carbon emission quota CEA allocation mechanism based on energy interaction: Use the Shapley value method to quantify the marginal contribution of each VPP to the carbon emission of the cooperative alliance, and introduce energy interaction factors for correction to achieve a more reasonable CEA allocation; S12. Build a carbon trading mechanism framework: including distributed transactions between MVPPs and tiered transactions with the carbon market, determine the CEA transaction price and quantity through negotiation, and reduce carbon emissions and transaction costs; S13. Constructing the multi-virtual power plant-load aggregator MVPP-LA cooperation objective function: setting the profit maximization of each VPP as the goal, including operating income from selling energy to LA and operating costs, including DER power generation, energy-carbon trading, energy transmission, demand response DR instructions, etc.; LA represents the interests of all energy users in the MVPP power supply area and encourages users to reduce electricity consumption through the DR mechanism to achieve maximum benefits; S14. Construct energy system operation constraints, including: Cogeneration equipment constraints: Limit the power generation, heat output and operating efficiency of gas turbines GT and waste heat recovery boilers WHRB; Energy storage system charge and discharge constraints: Control the charge and discharge status, frequency and capacity of the energy storage system ESS to ensure energy balance; Wind and PV constraints: quantifying uncertainty using stochastic optimization methods to ensure a stable supply of renewable energy; Demand response (DR) constraints: limit the reduction of electricity and heat loads within the customer's energy comfort range; Energy-carbon trading volume constraints on VPP: set an upper limit on energy trading volume to avoid network congestion; VPP energy balance constraint: ensures the balance between VPP internal unit energy supply and external energy interaction; S2. Construct the Nash bargaining model of MVPP-LA cooperation: set the Nash bargaining breakdown point and quantify the benefits of multilateral cooperation for each participant; decompose the Nash bargaining model into two sub-problems: maximizing alliance benefits and optimal benefit allocation, and solve the optimal transaction volume and transaction price respectively; S3. Solve the differentiated multi-agent cooperative Nash bargaining model: Use an improved distributed adaptive iterative algorithm such as ADMM to solve the two sub-problems and protect the data privacy of each participant; improve the convergence speed and accuracy of the algorithm by dynamically adjusting the penalty factor; update the energy trading volume and Lagrange multiplier until the convergence conditions are met; according to the bargaining model based on differentiated contributions, achieve fair distribution of benefits for each participant; promote the use of renewable energy and reduce carbon emissions and transaction costs through the carbon emission quota CEA allocation and carbon trading mechanism.

[0009] Further, the specific process of step S11 is as follows: According to the carbon trading mechanism, if the actual carbon emissions are lower than the allocated carbon emission quota CEA, the excess can be sold on the carbon market; otherwise, the shortfall needs to be purchased on the market; the baseline method is the current CEA allocation method, which calculates CEA by fixing the industry carbon emission benchmark value; in order to solve the problem that the baseline method fails to consider the differences in energy resources and supply characteristics of each VPP and the impact of cooperation between different entities on CEA allocation in a cooperative alliance composed of multiple virtual power plants MVPP, the Shapley value method is used to quantify the marginal contribution of each VPP to the carbon emissions of the cooperative alliance; The traditional Shapley value method CEA allocation description is shown in formula (1): (1) Where: is the CEA allocation for VPPi, It is a combination of VPPi and other VPPi. is the carbon emission of set d after removing VPPi, is the occurrence probability of set D, is the total number of virtual power plants; In the traditional Shapley value method, the marginal contribution of different distributed energy DERs to carbon emissions in each VPP is quantified, but the impact of energy transactions between entities on the marginal contribution of carbon emissions is not considered. In order to achieve a more reasonable CEA allocation, the participation in energy interaction is used as a correction indicator to modify the Shapley value method. For VPPs with a large number of DERs, priority participation in energy transactions within the alliance can promote DER utilization. Therefore, for VPPs that actively participate in energy interactions, CEA allocation can be increased. The quantitative factor description of VPPi's participation in power and heat interactions is shown in formula (2): (2) Where: and are expressed as quantitative factors participating in the electrical and thermal interactions, respectively; and They are respectively represented as the power and heat traded by VPPi from VPPj at time t; Based on the entropy weight method, combined with formula (2), the comprehensive quantitative factor of energy interaction can be obtained, which is described as shown in formula (3): (3) Where: It is expressed as a comprehensive quantitative factor of energy interaction; and They are respectively expressed as the weights of VPPi participating in the interaction of electric power and heat; and are expressed as quantitative factors participating in the electrical and thermal interactions, respectively; The correction factor based on energy interaction is introduced to correct CEAs. After reorganizing equations (1) to (3), the corrected description is shown in equation (4): (4) Where: Expressed as a correction factor, Expressed as a comprehensive quantitative factor of energy interaction, It is expressed as the total carbon emissions of all VPP cooperation. Expressed as VPPi-adjusted CEA, is the CEA allocation for VPPi, is the total number of virtual power plants.

[0010] Further, the specific process of step S12 is as follows: Under the alliance cooperation framework, a carbon trading mechanism is established, including distributed transactions between MVPPs and tiered transactions with the carbon market; including the "electricity-heat-carbon" trading scenario of three virtual power plants, each VPP formulates a centralized trading plan on the carbon market by calculating daily carbon emissions and CEA allocation, and formulates a distributed CEA trading plan within the alliance, both of which are executed on the same time scale; distributed transactions within the alliance determine the price and quantity of CEA transactions between each VPPi through negotiation; CEAs traded in the carbon market are divided into multiple intervals, and the transaction price of each interval increases with the increase in quantity; The combined carbon trading mechanism model is shown in equations (5) to (7): (5) (6) (7) Where: It is expressed as the CEA transaction volume between VPPi and the carbon market; Expressed as VPPi-adjusted CEA; and They are respectively expressed as VPPi carbon emissions and carbon trading volume from VPPi to VPPj; It is expressed as the carbon trading cost between VPPi and other VPPs; It is expressed as the CEA transaction price between VPPi and VPPj; Expressed as VPPi and carbon market carbon transaction costs; Expressed as a benchmark carbon price; is the growth rate of carbon trading price; The length of the interval expressed as CEA trading volume; According to the relationship between carbon emissions and energy supply of coal-fired and gas-fired units, the actual carbon emissions of each VPP are described as shown in formula (8): (8) Where: Carbon emissions expressed as VPPi; and They are respectively expressed as the carbon emissions of purchased electricity and purchased heat of VPPi; It is expressed as the carbon emission of VPPi in CHP operation; and They are respectively represented as the electricity and heat purchased by VPPi from the upper network at time t; , and They are respectively expressed as the carbon emission coefficient of coal-fired units; , and They are respectively expressed as carbon emission coefficients of gas turbine units; Expressed as GT power output in VPPi at time t.

[0011] Further, the specific process of step S13 is as follows: In the MVPP-LA collaborative operation model, each VPP aims to maximize profits, and its objective function is shown in formulas (9) and (10). Among them, the operating income comes from selling energy to LA, while the operating costs include DER power generation, energy-carbon trading, energy transmission and DR instructions. Under the cooperation framework, LA represents the interests of all energy users in the MVPP power supply area. Through the DR mechanism incentive, LA reduces part of the power and thermal loads and negotiates with MVPP to achieve maximum benefits, which can be expressed as formula (11): (9) (10) (11) Where: It is expressed as the energy transaction cost between VPPi and LA; It is expressed as the energy transmission cost between VPPi and LA; Unit operating cost expressed as VPPi; It is expressed as the energy transmission cost between VPPi and other VPPs; It is expressed as the energy transaction cost between VPPi and the upper network; It is expressed as the carbon transaction cost between VPPi and the carbon market; It is expressed as the energy transaction cost between VPPi and other VPPs; DR compensation cost expressed as VPPi; It is expressed as the carbon trading cost between VPPi and other VPPs; and They are respectively expressed as the reduction of electrical load and thermal load of VPPi at time t; and They are respectively expressed as the average heat purchase and sale prices of the upper heat network; and They are respectively expressed as the price of electricity purchased and sold to the upper-level power grid; and They are respectively expressed as the compensation price for reducing electricity and heat load by VPPi; and They are respectively represented as the electricity and heat traded from VPPi to LA at time t; Expressed as the power transmission cost coefficient from VPPi to VPPj; As a combination of multiple energy supply units, the combined heat and power unit CHP converts and distributes electrical energy and thermal energy through a gas turbine (GT) and a waste heat recovery boiler (WHRB); GT is used to generate electricity, and WHRB provides thermal energy by recovering the flue gas emitted by GT; according to the operating characteristics of GT, the operating cost of CHP can be calculated by the relationship between electrical and thermal power output and fuel consumption; for ESS, the uncertainty and volatility of renewable energy output are stabilized by controlling its charging and discharging state, and it is assumed that its rental cost is linearly related to the stored energy; therefore, the operating cost formula of DER in VPP is shown in formula (12): (12) Where: Unit operating cost expressed as VPPi; It is represented by the GT power output in VPPi at time t; Expressed as ESS rental price; Expressed as natural gas price; Expressed as GT's operating efficiency; and They are respectively represented as the charging and discharging power in VPPi at time t.

[0012] Further, the specific process of step S14 is as follows: S141. Establish cogeneration equipment constraints: The operation of gas turbine GT and waste heat recovery boiler WHRB is subject to hourly power generation limit. GT operation is limited by ramp-down rate, which is described as shown in equation (13): (13) Where: It is expressed as the electrical output of GT in VPPi at time t; It is expressed as the maximum power output of GT in VPPi; and They are respectively expressed as the maximum rising and falling rates of GT in VPP; It is expressed as the WHRB heat output in VPPi at time t; It is expressed as the maximum heat output of WHRB in VPPi at time t; S142. Establishing energy storage system charging and discharging constraints: In order to alleviate the impact of energy consumption and renewable energy fluctuations on VPP scheduling, it is necessary to control the charging and discharging state of ESS; at any time t, the ESS energy storage depends on the energy storage level at the previous time t~1, and the energy storage should remain consistent at the beginning and end of the cycle; the charging and discharging amount, frequency and energy storage capacity are all limited, and the charging and discharging operations cannot be performed simultaneously; the specific constraint conditions are described as shown in formula (14): (14) Where: and They are represented as binary variables that determine the charge and discharge state of ESS at time t, 1 indicates the charge and discharge state, and 0 indicates the non-charge and discharge state; and They are respectively expressed as the charging and discharging efficiency of ESS; It is represented as the energy stored in VPPi by ESS at time t; It is expressed as the degradation coefficient of ESS in VPPi; and They represent the minimum and maximum capacities of ESS in VPPi, respectively; and They are respectively represented as the maximum charge and discharge capacity of ESS in VPPi; It is expressed as the maximum number of discharge and charge times of ESS in VPPi; and They are respectively represented as the charging and discharging powers in VPPi at time t; S143. Establish wind power and photovoltaic constraints: Since wind power, photovoltaic power and load have random fluctuations in different time periods, a random optimization method is used to quantify uncertainty. Based on the Latin hypercube sampling scenario generation method, 500 scenarios are generated for each uncertainty factor, and 5 typical scenarios and their corresponding probabilities are selected. The probability of each scenario is multiplied by its predicted value and then summed to obtain the random optimization result. The specific constraint description is shown in formula (15): (15) Where: and They are represented as the stochastically optimized wind power and PV power generation of VPPi at time t, respectively; and Respectively represent the time t in the scenario and PV and wind power forecasts under VPPi; and Respectively represented as VPPi in scenario and probability; S144. Establishing Demand Response (DR) Constraints: Based on the DR mechanism, the electrical load and thermal load need to satisfy the equation (16), which can be limited to the customer's energy comfort range to a certain extent; (16) Where: and They are respectively represented as the random optimized electrical load and scenario of VPPi at time t. Power forecast for VPPi; and are the random optimal heat load and scenario of VPPi at time t, respectively. Heat load prediction of VPPi; and They are expressed as the maximum power and heat load reduction percentages of VPPi, respectively; and They are respectively expressed as the reduction of electrical load and thermal load of VPPi at time t; S145. Establish energy-carbon trading volume constraints on VPP: In order to avoid network congestion caused by MVPP energy trading, the present invention sets an upper limit constraint on the energy trading volume, and the energy-carbon interaction between MVPPs must comply with the balance constraint; the specific description is shown in formula (17): (17) Where: and They are respectively represented as the electricity and heat purchased by VPPi from the upper network at time t; and They represent the maximum power and heat purchased by VPP from the upper network respectively; and They are the maximum power and heat sold by VPP to the upper network respectively; and They are respectively expressed as the maximum power and heat exchanged between VPPs; and They are respectively represented as the electric power and heat traded by VPPi from VPPj at time t; It is expressed as the carbon trading volume from VPPi to VPPj; It is represented as the amount of electricity sold by VPPi to the upper network at time t; It is represented as VPPi selling heat power to the upper network at time t; S146. Establish VPP energy balance constraints: The VPP electrical and thermal balance, including the internal unit energy supply and the interaction with the LA, MVPP and upper network energy, is described as shown in Equation (18): (18) Where: It is represented as the amount of electricity traded from VPPi to LA at time t; It is represented by the amount of electricity purchased by VPPi from the upper network at time t; It is represented as the amount of electricity sold by VPPi to the upper network at time t; It is expressed as the heat traded from VPPi to LA at time t; It is expressed as the heat output of WHRB in VPPi at time t; It is represented as VPPi buying heat from the upper network at time t; It is represented as VPPi selling heat power to the upper network at time t; It is represented as the stochastic optimized wind power of VPPi at time t; Denoted as VPPi the electrical power traded from VPPj at time t.

[0013] Further, the specific process of step S2 is as follows: Assuming that MVPP and LA are independent entities in the cooperation, each participant hopes to find a balanced transaction strategy in the negotiation to maximize its own interests; therefore, the reasonable determination of transaction price and transaction volume becomes the focus of the MVPP-LA cooperation alliance; the traditional Nash bargaining model, as shown in formula (19), can quantify the benefits of multilateral cooperation for each participant; in the Nash bargaining process, the transaction volume of a VPP and the transaction price with other VPPs and LA are regarded as decision variables; and the Nash bargaining breaking point is a fixed value, at which point each VPP only trades with its upper network and interacts with LA alone in the Stackelberg game; at the Nash bargaining breaking point, VPP, as a leader, sets energy prices according to user energy demand, while LA, as a follower, adjusts its energy use according to VPP price signals; VPP and LA play continuous games with the goal of maximizing their respective interests until an equilibrium solution is reached; The traditional bargaining model is described as shown in formula (19): (19) Where: and are the Nash bargaining breaking points, which represent the benefits of LA and VPPi before cooperation; and They represent the increased benefits of LA and VPPi through bargaining respectively; From equations (9) to (18), we can see that the established Nash bargaining model is a non-convex nonlinear optimization problem. Equation (7) indicates that the step-type carbon trading cost is a piecewise function, which can be linearly constrained by using the big M method and introducing the convex transformation of Boolean variables. Equations (9) and (11) contain the product of two variables, transaction volume and transaction price, which are difficult to linearize using conventional methods. Therefore, in order to solve the non-convex model, it is decomposed into two sub-problems according to the overall strategy and individual interests of the cooperative game. These two sub-problems include the maximization and optimal allocation of the MVPP-LA alliance’s interests, which respectively determine the optimal interaction volume and interaction price within the alliance. By substituting the LA and MVPP objective functions and constraints, the overall cooperation benefit maximization subproblem can be obtained, as shown in formula (20): (20) Where: It is expressed as the energy transmission cost between VPPi and LA; Unit operating cost expressed as VPPi; It is expressed as the energy transmission cost between VPPi and other VPPs; It is expressed as the energy transaction cost between VPPi and the upper network; It is expressed as the carbon transaction cost between VPPi and the carbon market; It is expressed as the energy transfer cost between VPPi and other VPPs; Since formula (20) eliminates the transaction costs between entities, only the transaction volume of each entity can be obtained; substituting it into the original bargaining model (19), the optimal subproblem of alliance benefits can be obtained, which is described as shown in formula (21): (twenty one) Where: , , , , They represent the optimal solutions obtained by formula (20) respectively; the English character annotations in formula (21) are explained in formula (10); The traditional Nash bargaining model ignores the contribution of participants to the growth of alliance benefits, resulting in unfair distribution of benefits. To this end, this paper proposes a nonlinear mapping method based on exponential function to quantify the energy interaction contribution of each VPP in the alliance, and to evaluate the role of each VPP in Nash bargaining. The contribution quantification function is described as shown in formula (22): (twenty two) Where: and They represent the energy purchased and sold by VPPi within the alliance, respectively; It is expressed as the contribution factor of VPPi to MVPP cooperation; and They are respectively represented as the power and heat traded by VPPi from VPPj at time t; Based on the traditional bargaining formula (19), combined with formula (22), the contribution-based bargaining model of the benefit distribution sub-problem formula (21) is modified as shown in formula (23): (twenty three) Where: and are the Nash bargaining breaking points, which represent the benefits of LA and VPPi before cooperation; and They represent the increased benefits of LA and VPPi through bargaining respectively; It is expressed as the contribution factor of VPPi to MVPP cooperation; To ensure that MVPP and LA have priority in bargaining transactions within the alliance, the energy bargaining price should be within the range of the upper network purchase and sale energy price, and the CEA bargaining between MVPPs should be constrained, that is, not lower than the carbon market sales price and not higher than the carbon market purchase price; the description is shown in formula (24): (twenty four) Where: It is expressed as the electricity transaction price between VPPi and VPPj at time t; and They are respectively represented as the electricity purchase and electricity sales prices from the upper-level power grid at time t; and They are respectively expressed as the average heat purchase and sale price of the upper heat network; It is expressed as the heat transaction price between VPPi and VPPj at time t; It is expressed as the electricity transaction price of VPPi at time t; It is expressed as the heat transaction price between VPPi and LA at time t; Expressed as a benchmark carbon price; It is expressed as the CEA transaction price between VPPi and VPPj; Expressed as the carbon price corresponding to the carbon market trading volume trading range

[0014] Further, the specific process of step S3 is as follows: S31. Solve the sub-problem of maximizing alliance benefits: From constraints (16) and (18), we can see that the maximum coalition benefit subproblem contains energy and carbon interaction coupling variables: , , , , ; Introduce auxiliary variables , , , , , , Decouple atomic problems; when equation (25) is satisfied, it indicates that all participants have reached a consensus on the bargaining; (25) Where: and They are respectively represented as the electric power and heat purchased by LA from VPPi at time t; and They are respectively represented as the electrical power and heat output by VPPi to LA at time t; It is expressed as the carbon trading volume from VPPi to VPPj at time t; Based on the basic principles of distributed optimization algorithm, Lagrangian multipliers and penalty factors are introduced to create the following augmented Lagrangian functions for LA and MVPP, respectively, as shown in Equations (26) and (27): (26) (27) Where: , , , , denote the Lagrange multipliers respectively; is the penalty factor; the English character annotations of formulas (26) and (27) are explained in formulas (13) to (18); S32. Solve the optimal benefit allocation sub-problem: The optimal benefit allocation subproblem includes coupling variables , , , , introduce the corresponding auxiliary variables , , , , , , Decoupling; The relationship between variables is described as shown in formula (28): (28) Where: and They are respectively represented as the price of electricity and heat purchased by LA from MVPP at time t; and They are respectively represented as the price at which MVPP sells power and heat to LA at time t; It is expressed as the CEA transaction price between VPPi and VPPj; Equation (23) is logarithmically transformed into a convex optimization problem, and the LA and MVPP augmented Lagrangian functions are established as shown in Equations (29) and (30): (29) (30) Where: , , , , It means that the optimal solution is derived from equations (26) and (27); , , , , represents the Lagrange multiplier; It is expressed as a penalty factor; the English character annotations of formulas (29) and (30) are explained in formula (24); S33, solution based on distributed adaptive iterative algorithm: In order to protect the data privacy of MVPP and LA during the bargaining process, the improved ADMM is used to solve the two sub-problems in a distributed manner; since the bargaining model proposed in this invention involves a large number of subjects, an update mechanism is proposed on the basis of the traditional fixed penalty factor ADMM, so that the penalty factor can be dynamically adjusted during the iteration process; let κ and σ be the scale factors, and the penalty factor The adaptive update method is described as shown in formula (31): (31) Where: and They are expressed as raw residual and double residual respectively; is the number of iterations; κ and σ are scale factors, Expressed as a penalty factor; The variable update equation and convergence condition of the two-level subproblem of “Benefit maximization-optimal benefit allocation” based on the distributed adaptive DP-ADMM iterative algorithm are described as Equation (32) and Equation (33), respectively. For the initial value of the variable, the energy transaction volume and Lagrange multiplier are set to 0, and the penalty factor is set to 10. −2 , the iterative update description of coupling variables and auxiliary variables is shown in formula (32): (32) Where: , , , , They are respectively represented as the Lagrange multipliers iterated to the dth time; , , , , They are respectively represented as iterations to the (d+1)th Lagrange multiplier; is the penalty factor; and They are respectively represented as the power and heat purchased by LA from VPPi at the dth iteration at time t; and They are respectively represented as the power and heat output by VPPi to LA at the dth iteration at time t; It is expressed as the carbon trading volume from VPPi to VPPj at time t; , , They are , , Introducing auxiliary variables, and They are respectively the power and heat traded by VPPi from VPPj at time t, , , They are respectively represented by the introduction of auxiliary variables at the dth iteration at time t, see formula (25); They respectively represent the introduction of auxiliary variables at the (d+1)th iteration; It is represented as the first LA optimal iteration variable parameter set; and They are represented as the optimal iterative variable parameter sets of VPPi and VPPj respectively; and They are respectively represented as the power and heat traded by VPPi from VPPj at time t; The dual variable and penalty factor update formula and the convergence condition of the variables are described as shown in formula (33): (33) Where: , , , , They are respectively represented as the Lagrange multipliers iterated to the dth time; , , , , They are respectively represented as the iteration to the d+1th Lagrange multiplier; Represented as the penalty factor for iteration to the dth time; and They are respectively represented as the power and heat purchased by LA from VPPi at the dth iteration at time t; and They are respectively represented as the power and heat output by VPPi to LA at the dth iteration at time t; It is expressed as the carbon trading volume from VPPi to VPPj at time t; , , They respectively represent the introduction of auxiliary variables at the dth iteration at time t, and the comments are shown in formula (25).

[0015] The beneficial effects achieved by the invention are: The present invention, by introducing a differentiated benefit distribution mechanism, a CEA distribution mechanism based on energy interaction, a carbon trading mechanism framework, a multi-virtual power plant-load aggregator MVPP-LA cooperation objective function, energy system operation constraints, and an algorithm for solving a differentiated multi-agent cooperation Nash bargaining model, has brought about the effects of promoting the use of renewable energy, reducing carbon emissions and transaction costs, achieving fair benefit distribution, improving the algorithm convergence speed and accuracy, and enhancing the flexibility and stability of the system, as follows: First, the present invention is different from the average distribution in the traditional Nash bargaining model. The bargaining model based on differentiated contributions realizes a fairer distribution of benefits among different entities. This distribution method can encourage each virtual power plant (VPP) to participate more actively in energy interaction and carbon trading, and improve the overall cooperation efficiency.

[0016] Second, by adopting the Shapley value method and introducing the energy interaction factor for correction, a more reasonable CEA allocation is achieved. This allocation method fully considers the differences in energy resources and supply characteristics of each VPP, as well as the impact of cooperation between different entities on CEA allocation, thereby promoting the use of renewable energy.

[0017] Third, the present invention designs distributed transactions between MVPPs and tiered transactions with the carbon market, which helps reduce carbon emissions and transaction costs by negotiating the transaction price and quantity of CEA. At the same time, this trading mechanism also promotes regional resource complementarity and improves energy efficiency.

[0018] Fourth, the present invention sets the profit maximization of each VPP as the goal, and takes into account the operating income and operating costs (such as DER power generation, energy-carbon trading, energy transmission, demand response DR instructions, etc.), so that each VPP can consider the interests of the overall cooperative alliance while pursuing its own interests. The setting of this objective function helps to enhance the interaction between supply and demand and improve the overall cooperation benefits.

[0019] Fifth, the present invention ensures the safe and stable operation of the energy system by constructing energy system operation constraints including cogeneration equipment constraints, energy storage system charging and discharging constraints, wind power and photovoltaic constraints, demand response DR constraints, energy-carbon trading volume constraints on VPP, and VPP energy balance constraints. At the same time, these constraints also provide an important basis for optimizing energy configuration and scheduling.

[0020] Sixth, the present invention solves the problem by setting the Nash bargaining breakdown point and decomposing it into two sub-problems: maximizing the benefits of the alliance and allocating the optimal benefits. The Nash bargaining model proposed by the present invention can quantify the benefits of multilateral cooperation for each participant and ensure the maximization of the overall benefits of the cooperative alliance. The application of this model helps to improve the efficiency and fairness of cooperative negotiations.

[0021] Seventh, the present invention adopts a distributed adaptive iterative algorithm to solve the differentiated multi-agent cooperative Nash bargaining model, protects the data privacy of each participant, and improves the convergence speed and accuracy of the algorithm by dynamically adjusting the penalty factor. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A schematic diagram of the MVPP-LA alliance cooperation framework is shown; Figure 2 The combined relationship and process diagram of the "electricity-heat-carbon" multi-energy flow trading scenarios of three virtual power plants are shown; Figure 3 A schematic diagram of the distributed optimization process of low-carbon bargaining for virtual power plants in the MVPP-LA collaborative game model is shown. DETAILED DESCRIPTION

[0023] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the present invention. In addition, the forms of the various structures recorded in the following embodiments are merely illustrative, and the present invention is not limited to the various structures recorded in the following embodiments. All other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0024] Reference Figure 1 - Figure 3 The present invention provides a virtual power plant low-carbon bargaining optimization method based on differentiated multi-agent collaborative game, and the specific steps are as follows: S1. Build the MVPP-LA cooperation alliance framework: Each VPP coordinates the production and supply of regional energy by integrating wind turbines (WT), photovoltaics (PV), combined heating and power (CHP) and energy storage systems (ESS). In order to enhance the interaction between supply and demand, MVPP integrates users in its energy supply area through LA to form a multi-virtual power plant-load aggregator MVPP-LA cooperative alliance. Under the management of the load aggregator LA, users participate in demand response (DR) and reduce electricity consumption through price incentives. Under the multi-subject cooperation model, MVPP trades electricity and heat with LA to meet its energy needs. In addition, MVPPs also conduct energy-carbon transactions to achieve regional resource complementarity. The cooperation framework structure of the multi-virtual power plant-load aggregator MVPP-LA alliance is as follows: Figure 1 shown.

[0025] S11. CEA allocation mechanism based on energy interaction: According to China's current carbon trading mechanism, if the actual carbon emissions are lower than the allocated carbon emission allowance (CEA), the excess can be sold on the carbon market; otherwise, the shortfall needs to be purchased on the market. The baseline method is the current CEA allocation method, which calculates CEA by fixing the industry carbon emission benchmark value. However, for a cooperative alliance composed of multiple virtual power plants (MVPPs), the baseline method fails to take into account the differences in energy resources and supply characteristics of each VPP and the impact of cooperation between different entities on CEA allocation. To solve these problems, the present invention adopts the Shapley value method to quantify the marginal contribution of each VPP to the carbon emissions of the cooperative alliance.

[0026] The traditional Shapley value method CEA allocation description is shown in formula (1): (1) Where: is the CEA allocation for VPPi, It is a combination of VPPi and other VPPi. is the carbon emission of set d after removing VPPi, is the occurrence probability of set D, is the total number of virtual power plants;

[0027] In the traditional Shapley value method, the marginal contribution of different distributed energy DERs to carbon emissions in each VPP is quantified. However, this method does not consider the impact of energy transactions between entities on the marginal contribution of carbon emissions. Therefore, in order to achieve a more reasonable CEA allocation, participation in energy interaction is used as a correction indicator to correct the Shapley value method. For VPPs with a large number of DERs, priority participation in energy transactions within the alliance can promote DER utilization. Therefore, for VPPs that actively participate in energy interactions, CEA allocation can be increased. The quantitative factor description of VPPi's participation in power and heat interactions is shown in formula (2): (2) Where: and are expressed as quantitative factors participating in the electrical and thermal interactions, respectively; and are respectively the power and heat traded by VPPi from VPPj at time t.

[0028] Based on the entropy weight method, combined with formula (2), the comprehensive quantitative factor of energy interaction can be obtained, which is described as shown in formula (3): (3) Where: It is expressed as a comprehensive quantitative factor of energy interaction; and They are respectively expressed as the weights of VPPi participating in the interaction of electric power and heat; and They are expressed as quantitative factors participating in the electrical and thermal interactions, respectively.

[0029] The correction factor based on energy interaction is introduced to correct CEAs. After reorganizing equations (1) to (3), the corrected description is shown in equation (4): (4) Where: Expressed as a correction factor, Expressed as a comprehensive quantitative factor of energy interaction, It is expressed as the total carbon emissions of all VPP cooperation. Expressed as VPPi-adjusted CEA, is the CEA allocation for VPPi, is the total number of virtual power plants.

[0030] S12. Constructing a carbon trading mechanism framework: Under the alliance cooperation framework, a method combining carbon trading mechanisms is proposed, including distributed trading between MVPPs and tiered trading with the carbon market. Taking the "electricity-heat-carbon" trading scenario of three virtual power plants as an example, the mechanism combination relationship and trading process are as follows: Figure 2 As shown in the figure. Each VPP formulates a centralized trading plan in the carbon market by calculating daily carbon emissions and CEA allocations, and formulates a distributed CEA trading plan within the alliance, both of which are executed on the same time scale. Unlike the tiered pricing in the centralized carbon trading market, the distributed trading within the alliance determines the price and quantity of CEA transactions between each VPPi through negotiation. CEAs traded in the carbon market are divided into multiple intervals, and the transaction price of each interval increases as the quantity increases.

[0031] The combined carbon trading mechanism model is shown in equations (5) to (7): (5) (6) (7) Where: It is expressed as the CEA transaction volume between VPPi and the carbon market; Expressed as VPPi-adjusted CEA; and They are respectively expressed as VPPi carbon emissions and carbon trading volume from VPPi to VPPj; It is expressed as the carbon trading cost between VPPi and other VPPs; It is expressed as the CEA transaction price between VPPi and VPPj; Expressed as VPPi and carbon market carbon transaction costs; Expressed as a benchmark carbon price; is the growth rate of carbon trading price; The length of the interval expressed as CEA trading volume;

[0032] According to the relationship between carbon emissions and energy supply of coal-fired and gas-fired units, the actual carbon emissions of each VPP are described as shown in formula (8): (8) Where: Carbon emissions expressed as VPPi; and They are respectively expressed as the carbon emissions of purchased electricity and purchased heat of VPPi; It is expressed as the carbon emission of VPPi in CHP operation; and They are respectively represented as the electricity and heat purchased by VPPi from the upper network at time t; , and They are respectively expressed as the carbon emission coefficient of coal-fired units; , and They are respectively expressed as carbon emission coefficients of gas turbine units; Expressed as GT power output in VPPi at time t.

[0033] S13. Constructing the MVPP-LA cooperation objective function In the MVPP-LA collaborative operation model, each VPP aims to maximize profits, and its objective function is shown in formulas (9) and (10). Among them, operating income comes from selling energy to LA, while operating costs include DER power generation, energy-carbon trading, energy transmission and DR instructions. Under the cooperation framework, LA represents the interests of all energy users in the MVPP power supply area. Through the DR mechanism incentive, LA reduces part of the power and thermal loads and negotiates with MVPP to achieve maximum benefits, which can be expressed as formula (11): (9) (10) (11) Where: It is expressed as the energy transaction cost between VPPi and LA; It is expressed as the energy transmission cost between VPPi and LA; Unit operating cost expressed as VPPi; It is expressed as the energy transmission cost between VPPi and other VPPs; It is expressed as the energy transaction cost between VPPi and the upper network; It is expressed as the carbon transaction cost between VPPi and the carbon market; It is expressed as the energy transaction cost between VPPi and other VPPs; DR compensation cost expressed as VPPi; It is expressed as the carbon trading cost between VPPi and other VPPs; and They are respectively expressed as the reduction of electrical load and thermal load of VPPi at time t; and They are respectively expressed as the average heat purchase and sale prices of the upper heat network; and They are respectively expressed as the price of electricity purchased and sold to the upper-level power grid; and They are respectively expressed as the compensation price for reducing electricity and heat load by VPPi; and They are respectively represented as the electricity and heat traded from VPPi to LA at time t; Expressed as the power transmission cost coefficient from VPPi to VPPj.

[0034] As a combination of multiple energy supply units, the combined heat and power unit CHP converts and distributes electrical energy and thermal energy through a gas turbine (GT) and a waste heat recovery boiler (WHRB). Among them, GT is used for power generation, and WHRB provides thermal energy by recovering the flue gas emitted by GT. According to the operating characteristics of GT, the operating cost of CHP can be calculated by the relationship between electrical and thermal power output and fuel consumption. For ESS, the uncertainty and volatility of renewable energy output are stabilized by controlling its charging and discharging state, and it is assumed that its rental cost is linearly related to the stored energy. Therefore, the operating cost formula of DER in VPP is shown in formula (12): (12) Where: Unit operating cost expressed as VPPi; It is represented by the GT power output in VPPi at time t; Expressed as ESS rental price; Expressed as natural gas price; Expressed as GT's operating efficiency; and They are respectively represented as the charging and discharging power in VPPi at time t.

[0035] S14. Constructing energy system operation constraints (1) Cogeneration equipment constraints: The operation of gas turbine GT and waste heat recovery boiler GT and WHRB are both subject to hourly power generation constraints. GT operation is limited by the ramp-down rate, which is described as shown in equation (13): (13) Where: It is expressed as the electrical output of GT in VPPi at time t; It is expressed as the maximum power output of GT in VPPi; and They are respectively expressed as the maximum rising and falling rates of GT in VPP; It is expressed as the WHRB heat output in VPPi at time t; Expressed as the maximum heat output of WHRB in VPPi at time t.

[0036] (2) Energy storage system charging and discharging constraints: In order to alleviate the impact of energy consumption and renewable energy fluctuations on VPP scheduling, it is necessary to control the charging and discharging state of ESS. At any time t, the ESS energy storage depends on the energy storage level at the previous time t~1, and the energy storage should remain consistent at the beginning and end of the cycle. The charging and discharging amount, frequency and energy storage capacity are all restricted, and charging and discharging operations cannot be performed simultaneously. The specific constraint conditions are described as shown in formula (14): (14) Where: and They represent the binary variables that determine the charge and discharge state of ESS at time t, 1 represents the charge and discharge state, and 0 represents the non-charge and discharge state; and They are respectively expressed as the charging and discharging efficiency of ESS; It is represented as the energy stored in VPPi by ESS at time t; It is expressed as the degradation coefficient of ESS in VPPi; and They represent the minimum and maximum capacities of ESS in VPPi, respectively; and They are respectively represented as the maximum charge and discharge capacity of ESS in VPPi; It is expressed as the maximum number of discharge and charge times of ESS in VPPi; and They are respectively represented as the charging and discharging powers in VPPi at time t.

[0037] Wind power and photovoltaic constraints: Since wind power, photovoltaic power and load have random fluctuations in different time periods, a random optimization method is used to quantify uncertainty, based on the Latin hypercube sampling scenario generation method. For each uncertainty factor, 500 scenarios are generated, and 5 typical scenarios and their corresponding probabilities are selected. The probability of each scenario is multiplied by its predicted value and then summed to obtain the random optimization result. The specific constraint description is shown in formula (15): (15) Where: and They are represented as the stochastically optimized wind power and PV power generation of VPPi at time t, respectively; and Respectively represent the time t in the scenario and PV and wind power forecasts under VPPi; and Respectively represented as VPPi in scenario and probability.

[0038] (4) Demand response (DR) constraints: Based on the DR mechanism, the electrical load and thermal load must satisfy the equation (16), which can be limited to the customer's energy comfort range to a certain extent.

[0039] (16) Where: and They are respectively represented as the random optimized electrical load and scenario of VPPi at time t. Power forecast for VPPi; and are the random optimal heat load and scenario of VPPi at time t, respectively. Heat load prediction of VPPi; and They are expressed as the maximum power and heat load reduction percentages of VPPi, respectively; and They are respectively represented as the reduction in electrical load and thermal load of VPPi at time t.

[0040] (5) Energy-carbon trading volume constraints on VPP: In order to avoid network congestion caused by MVPP energy trading, the present invention sets an upper limit constraint on the energy trading volume, and the energy-carbon interaction between MVPPs must meet the balance constraint. The specific description is shown in formula (17): (17) Where: and They are respectively represented as the electricity and heat purchased by VPPi from the upper network at time t; and They represent the maximum power and heat purchased by VPP from the upper network respectively; and They are respectively represented as the maximum power and heat sold by VPP to the upper network at time t; and They are respectively represented as the maximum power and heat exchanged between VPPs at time t; and They are respectively represented as the electric power and heat traded by VPPi from VPPj at time t; It is expressed as the carbon trading volume from VPPi to VPPj; It is represented as the amount of electricity sold by VPPi to the upper network at time t; It means that VPPi sells thermal power to the upper network at time t.

[0041] (6) VPP energy balance constraint: The VPP electrical and thermal balance, including the internal unit energy supply and the interaction with the LA, MVPP and upper network energy, is described as shown in Equation (18): (18) Where: It is represented as the amount of electricity traded from VPPi to LA at time t; It is represented by the amount of electricity purchased by VPPi from the upper network at time t; It is represented as the amount of electricity sold by VPPi to the upper network at time t; It is expressed as the heat traded from VPPi to LA at time t; It is expressed as the heat output of WHRB in VPPi at time t; It is represented as VPPi buying heat from the upper network at time t; It is represented as VPPi selling heat power to the upper network at time t; It is represented as the stochastic optimized wind power of VPPi at time t; Denoted as VPPi the electrical power traded from VPPj at time t.

[0042] S2, Nash bargaining model of MVPP-LA cooperation: Assuming that MVPP and LA are independent entities in the cooperation, each participant hopes to find a balanced transaction strategy in the negotiation to maximize its own interests. Therefore, the reasonable determination of transaction prices and transaction volumes has become the focus of the MVPP-LA cooperation alliance. The traditional Nash bargaining model, as shown in formula (19), can quantify the benefits of multilateral cooperation for each participant. In the Nash bargaining process, the transaction volume of a VPP and the transaction price with other VPPs and LA are regarded as decision variables. The Nash bargaining breaking point is a fixed value, at which point each VPP only trades with its upper network and interacts with LA alone in the Stackelberg game. At the Nash bargaining breaking point, VPP, as a leader, sets energy prices according to user energy demand, while LA, as a follower, adjusts its energy use according to VPP price signals. VPP and LA play continuous games with the goal of maximizing their respective interests until an equilibrium solution is reached.

[0043] The traditional bargaining model is described as shown in formula (19): (19) Where: and are the Nash bargaining breaking points, which represent the benefits of LA and VPPi before cooperation; and They represent the increased benefits of LA and VPPi through bargaining respectively.

[0044] From equations (9) to (18), we can see that the established Nash bargaining model is a non-convex nonlinear optimization problem, where equation (7) indicates that the step-type carbon trading cost is a piecewise function, which can be used to obtain linear constraints by using the big M method and introducing the convex transformation of Boolean variables. Equations (9) and (11) contain the product of two variables, transaction volume and transaction price, which are difficult to linearize using conventional methods. Therefore, in order to solve the non-convex model, it is decomposed into two sub-problems according to the overall strategy and individual interests of the cooperative game. These two sub-problems include the maximization and optimal allocation of the MVPP-LA alliance's interests, which respectively determine the optimal interaction volume and interaction price within the alliance.

[0045] By substituting the LA and MVPP objective functions and constraints, the overall cooperation benefit maximization subproblem can be obtained, as shown in formula (20): (20) Where: It is expressed as the energy transmission cost between VPPi and LA; Unit operating cost expressed as VPPi; It is expressed as the energy transmission cost between VPPi and other VPPs; It is expressed as the energy transaction cost between VPPi and the upper network; It is expressed as the carbon transaction cost between VPPi and the carbon market; Expressed as the energy transfer cost between VPPi and other VPPs.

[0046] Since formula (20) eliminates the transaction costs between entities, only the transaction volume of each entity can be obtained. Substituting it into the original bargaining model (19), the optimal subproblem of alliance benefits can be obtained, which is described as shown in formula (21): (twenty one) Where: , , , , They respectively represent the optimal solutions obtained by formula (20); the English character annotations in formula (21) are explained in formula (10).

[0047] The traditional Nash bargaining model ignores the contribution of participants to the growth of alliance benefits, resulting in unfair distribution of benefits. To this end, this paper proposes a nonlinear mapping method based on exponential functions to quantify the energy interaction contribution of each VPP in the alliance, and to evaluate the role of each VPP in Nash bargaining. The contribution quantification function is described as shown in formula (22): (twenty two) Where: and They represent the energy purchased and sold by VPPi within the alliance, respectively; It is expressed as the contribution factor of VPPi to MVPP cooperation; and They are respectively the power and heat traded by VPPi from VPPj at time t.

[0048] Based on the traditional bargaining formula (19), combined with formula (22), the contribution-based bargaining model of the benefit distribution sub-problem formula (21) is modified as shown in formula (23): (twenty three) Where: and are the Nash bargaining breaking points, which represent the benefits of LA and VPPi before cooperation; and They represent the increased benefits of LA and VPPi through bargaining respectively; It is expressed as the contribution factor of VPPi to MVPP cooperation; To ensure that MVPP and LA have priority in bargaining transactions within the alliance, the energy bargaining price should be within the range of the upper network's energy purchase and sale prices, and the CEA bargaining between MVPPs should be constrained, that is, not lower than the carbon market sales price and not higher than the carbon market purchase price. The description is shown in formula (24): (twenty four) Where: It is expressed as the electricity transaction price between VPPi and VPPj at time t; and They are respectively represented as the electricity purchase and electricity sales prices from the upper-level power grid at time t; and They are respectively expressed as the average heat purchase and sale price of the upper heat network; It is expressed as the heat transaction price between VPPi and VPPj at time t; It is expressed as the electricity transaction price of VPPi at time t; It is expressed as the heat transaction price between VPPi and LA at time t; Expressed as a benchmark carbon price; It is expressed as the CEA transaction price between VPPi and VPPj; Expressed as the carbon price corresponding to the trading range of carbon market trading volume.

[0049] S3. Solve the differentiated multi-agent cooperation Nash bargaining model: S31. Sub-problem of maximizing alliance benefits: From constraints (16) and (18), we can see that the maximum coalition benefit subproblem contains energy and carbon interaction coupling variables: , , , , ; Introduce auxiliary variables , , , , , , Decouple atomic problems; when equation (25) is satisfied, it indicates that all participants have reached a consensus on the bargaining.

[0050] (25) Where: and They are respectively represented as the electric power and heat purchased by LA from VPPi at time t; and They are respectively represented as the electrical power and heat output by VPPi to LA at time t; Expressed as the carbon trading volume from VPPi to VPPj at time t.

[0051] Based on the basic principles of distributed optimization algorithm, Lagrangian multipliers and penalty factors are introduced to create the following augmented Lagrangian functions for LA and MVPP, respectively, as shown in Equations (26) and (27): (26) (27) Where: , , , , denote the Lagrange multipliers respectively; is the penalty factor; the English character annotations of formulas (26) and (27) are explained in formulas (13) to (18).

[0052] S32. Solve the optimal benefit allocation sub-problem: The optimal benefit allocation subproblem includes coupling variables , , , , introduce the corresponding auxiliary variables , , , , , , Decoupling; The relationship between variables is described as shown in formula (28): (28) Where: and They are respectively represented as the price of electricity and heat purchased by LA from MVPP at time t; and They are respectively represented as the price at which MVPP sells power and heat to LA at time t; Expressed as the CEA transaction price between VPPi and VPPj.

[0053] Equation (23) is logarithmically transformed into a convex optimization problem, and the LA and MVPP augmented Lagrangian functions are established as shown in Equations (29) and (30): (29) (30) Where: , , , , It means that the optimal solution is derived from equations (26) and (27); , , , , represents the Lagrange multiplier; Represents the penalty factor; other English character annotations are explained in formula (24).

[0054] S33, solution based on distributed adaptive iterative algorithm: In order to protect the data privacy of MVPP and LA during the bargaining process, the improved ADMM is used to solve the two sub-problems in a distributed manner; since the bargaining model proposed in this invention involves a large number of subjects, an update mechanism is proposed on the basis of the traditional fixed penalty factor ADMM, so that the penalty factor can be dynamically adjusted during the iteration process; let κ and σ be the scale factors, and the penalty factor The adaptive update method is described as shown in formula (31): (31) Where: and They are expressed as raw residual and double residual respectively; is the number of iterations; κ and σ are scale factors, It is represented as a penalty factor.

[0055] The variable update equation and convergence condition of the two-level subproblem of “Benefit maximization-optimal benefit allocation” based on the distributed adaptive DP-ADMM iterative algorithm are described as Equation (32) and Equation (33), respectively. For the initial value of the variable, the energy transaction volume and Lagrange multiplier are set to 0, and the penalty factor is set to 10. −2 , the iterative update description of coupling variables and auxiliary variables is shown in formula (32): (32) Where: , , , , They are respectively represented as the Lagrange multipliers iterated to the dth time; , , , , They are respectively represented as iterations to the (d+1)th Lagrange multiplier; is the penalty factor; and They are respectively represented as the power and heat purchased by LA from VPPi at the dth iteration at time t; and They are respectively represented as the power and heat output by VPPi to LA at the dth iteration at time t; It is expressed as the carbon trading volume from VPPi to VPPj at time t; , , They are , , Introducing auxiliary variables, and They are respectively the power and heat traded by VPPi from VPPj at time t, , , They are respectively represented by the introduction of auxiliary variables at the dth iteration at time t, see formula (25); They respectively represent the introduction of auxiliary variables at the (d+1)th iteration; It is represented as the first LA optimal iteration variable parameter set; and They are represented as the optimal iterative variable parameter sets of VPPi and VPPj respectively; and They are respectively represented as the power and heat traded by VPPi from VPPj at time t; The dual variable and penalty factor update formula and the convergence condition of the variables are described as shown in formula (33): (33) Where: , , , , They are respectively represented as the Lagrange multipliers iterated to the dth time; , , , , They are respectively represented as the iteration to the d+1th Lagrange multiplier; Represented as the penalty factor for iteration to the dth time; and They are respectively represented as the power and heat purchased by LA from VPPi at the dth iteration at time t; and They are respectively represented as the power and heat output by VPPi to LA at the dth iteration at time t; It is expressed as the carbon trading volume from VPPi to VPPj at time t; , , They respectively represent the introduction of auxiliary variables at the dth iteration at time t, and the comments are shown in formula (25).

[0056] The distributed optimization solution process of the "alliance benefit maximization - optimal benefit allocation" two-level sub-problem model of multi-virtual power plants and load aggregators MVPP-LA is as follows Figure 3 shown.

[0057] The present invention designs and constructs a differentiated benefit distribution mechanism, a CEA distribution mechanism based on energy interaction, a carbon trading mechanism framework, a multi-virtual power plant-load aggregator MVPP-LA cooperation objective function, energy system operation constraints, and an algorithm for solving a differentiated multi-agent cooperation Nash bargaining model, etc., to promote the utilization of renewable energy, reduce carbon emissions and transaction costs, achieve fair benefit distribution, improve the algorithm convergence speed and accuracy, and enhance the system flexibility and stability.

[0058] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A virtual power plant low-carbon bargaining optimization method based on differentiated multi-agent collaborative game, characterized in that: It includes the following steps: S1. Build a multi-virtual power plant-load aggregator MVPP-LA cooperation alliance framework: Integrate wind turbines WT, photovoltaics PV, combined heat and power equipment CHP and energy storage systems ESS to form multiple virtual power plants MVPP; integrate users within the MVPP area through load aggregators LA to form an MVPP-LA cooperation alliance to enhance supply and demand interaction; S11. Construct a carbon emission quota CEA allocation mechanism based on energy interaction: Use the Shapley value method to quantify the marginal contribution of each VPP to the carbon emission of the cooperative alliance, and introduce energy interaction factors for correction to achieve a more reasonable CEA allocation; S12. Build a carbon trading mechanism framework: including distributed transactions between MVPPs and tiered transactions with the carbon market, determine the transaction price and quantity of carbon emission quotas CEA through negotiation, and reduce carbon emissions and transaction costs; S13. Constructing the multi-virtual power plant-load aggregator MVPP-LA cooperation objective function: setting the profit maximization of each VPP as the goal, including operating income from selling energy to LA and operating costs, including distributed energy DER generation, energy-carbon trading, energy transmission, demand response DR instructions, etc.; LA represents the interests of all energy users in the MVPP power supply area and encourages users to reduce electricity consumption through the DR mechanism to achieve maximum benefits; S14. Construct energy system operation constraints, including: Cogeneration equipment constraints: Limit the power generation, heat output and operating efficiency of gas turbines GT and waste heat recovery boilers WHRB; Energy storage system charge and discharge constraints: Control the charge and discharge status, frequency and capacity of the energy storage system ESS to ensure energy balance; Wind and PV constraints: quantifying uncertainty using stochastic optimization methods to ensure a stable supply of renewable energy; Demand response (DR) constraints: limit the reduction of electricity and heat loads within the customer's energy comfort range; Energy-carbon trading volume constraints on VPP: set an upper limit on energy trading volume to avoid network congestion; VPP energy balance constraint: ensures the balance between VPP internal unit energy supply and external energy interaction; S2. Construct the Nash bargaining model of MVPP-LA cooperation: set the Nash bargaining breakdown point and quantify the benefits of multilateral cooperation for each participant; decompose the Nash bargaining model into two sub-problems: maximizing alliance benefits and optimal benefit allocation, and solve the optimal transaction volume and transaction price respectively; S3. Solve the differentiated multi-agent cooperative Nash bargaining model: Use the improved distributed adaptive iterative algorithm ADMM to solve the two sub-problems and protect the data privacy of each participant; improve the convergence speed and accuracy of the algorithm by dynamically adjusting the penalty factor; update the energy trading volume and Lagrange multiplier until the convergence conditions are met; according to the bargaining model based on differentiated contributions, realize fair distribution of benefits for each participant; promote the use of renewable energy and reduce carbon emissions and transaction costs through the carbon emission quota CEA allocation and carbon trading mechanism.

2. A virtual power plant low-carbon bargaining optimization method based on differentiated multi-agent collaborative game as claimed in claim 1, characterized in that: The specific process of step S11 is as follows: According to the carbon trading mechanism, if the actual carbon emissions are lower than the allocated carbon emission quota CEA, the excess can be sold on the carbon market; otherwise, the shortfall needs to be purchased on the market; the baseline method is the current CEA allocation method, which calculates CEA by fixing the industry carbon emission benchmark value; in order to solve the problem that the baseline method fails to consider the differences in energy resources and supply characteristics of each VPP and the impact of cooperation between different entities on CEA allocation in a cooperative alliance composed of multiple virtual power plants MVPP, the Shapley value method is used to quantify the marginal contribution of each VPP to the carbon emissions of the cooperative alliance; The traditional Shapley value method CEA allocation description is shown in formula (1): （1） Where: is the CEA allocation for VPPi, It is a combination of VPPi and other VPPi. is the carbon emission of set d after removing VPPi, is the occurrence probability of set D, is the total number of virtual power plants; In the traditional Shapley value method, the marginal contribution of different distributed energy DERs to carbon emissions in each VPP is quantified, but the impact of energy transactions between entities on the marginal contribution of carbon emissions is not considered. In order to achieve a more reasonable CEA allocation, the participation in energy interaction is used as a correction indicator to modify the Shapley value method. For VPPs with a large number of DERs, priority participation in energy transactions within the alliance can promote DER utilization. Therefore, for VPPs that actively participate in energy interactions, CEA allocation can be increased. The quantitative factor description of VPPi's participation in power and heat interactions is shown in formula (2): （2） Where: and are expressed as quantitative factors participating in the electrical and thermal interactions, respectively; and They are respectively represented as the power and heat traded by VPPi from VPPj at time t; Based on the entropy weight method, combined with formula (2), the comprehensive quantitative factor of energy interaction can be obtained, which is described as shown in formula (3): （3） Where: It is expressed as a comprehensive quantitative factor of energy interaction; and They are respectively expressed as the weights of VPPi participating in the interaction of electric power and heat; and are expressed as quantitative factors participating in the electrical and thermal interactions, respectively; The correction factor based on energy interaction is introduced to correct CEAs. After reorganizing equations (1) to (3), the corrected description is shown in equation (4): （4） Where: Expressed as a correction factor, Expressed as a comprehensive quantitative factor of energy interaction, It is expressed as the total carbon emissions of all VPP cooperation. Expressed as VPPi-adjusted CEA, is the CEA allocation for VPPi, is the total number of virtual power plants.

3. A virtual power plant low-carbon bargaining optimization method based on differentiated multi-agent collaborative game as claimed in claim 2, characterized in that: The specific process of step S12 is as follows: Under the alliance cooperation framework, a carbon trading mechanism is established, including distributed transactions between MVPPs and tiered transactions with the carbon market; including three virtual power plants "electricity-heat-carbon" trading scenarios, each VPP calculates daily carbon emissions and CEA allocations, formulates a centralized trading plan in the carbon market, and formulates a distributed CEA trading plan within the alliance, both of which are executed on the same time scale; distributed transactions within the alliance determine the price and quantity of CEA transactions between each VPPi through negotiation; CEAs traded in the carbon market are divided into multiple intervals, and the transaction price of each interval increases with the increase in quantity; The combined carbon trading mechanism model is shown in equations (5) to (7): （5） （6） （7） Where: It is expressed as the CEA transaction volume between VPPi and the carbon market; Expressed as VPPi-adjusted CEA; and They are respectively expressed as VPPi carbon emissions and carbon trading volume from VPPi to VPPj; It is expressed as the carbon trading cost between VPPi and other VPPs; It is expressed as the CEA transaction price between VPPi and VPPj; Expressed as VPPi and carbon market carbon transaction costs; Expressed as a benchmark carbon price; is the growth rate of carbon trading price; It is expressed as the interval length of CEA transaction volume; according to the relationship between carbon emissions and energy supply of coal-fired and gas-fired units, the actual carbon emissions of each VPP are described as shown in formula (8): （8） Where: Carbon emissions expressed as VPPi; and They are respectively expressed as the carbon emissions of purchased electricity and purchased heat of VPPi; It is expressed as the carbon emission of VPPi in CHP operation; and They are respectively represented as the electricity and heat purchased by VPPi from the upper network at time t; , and They are respectively expressed as the carbon emission coefficient of coal-fired units; , and They are respectively expressed as carbon emission coefficients of gas turbine units; Expressed as GT power output in VPPi at time t.

4. A virtual power plant low-carbon bargaining optimization method based on differentiated multi-agent collaborative game as claimed in claim 3, characterized in that: The specific process of step S13 is as follows: In the MVPP-LA collaborative operation model, each VPP aims to maximize profits, and its objective function is shown in formulas (9) and (10). Among them, the operating income comes from selling energy to LA, while the operating costs include DER power generation, energy-carbon trading, energy transmission and DR instructions. Under the cooperation framework, LA represents the interests of all energy users in the MVPP power supply area. Through the DR mechanism incentive, LA reduces part of the power and thermal loads and negotiates with MVPP to achieve maximum benefits, which can be expressed as formula (11): （9） （10） （11） Where: It is expressed as the energy transaction cost between VPPi and LA; It is expressed as the energy transmission cost between VPPi and LA; Unit operating cost expressed as VPPi; It is expressed as the energy transmission cost between VPPi and other VPPs; It is expressed as the energy transaction cost between VPPi and the upper network; It is expressed as the carbon transaction cost between VPPi and the carbon market; It is expressed as the energy transaction cost between VPPi and other VPPs; DR compensation cost expressed as VPPi; It is expressed as the carbon trading cost between VPPi and other VPPs; and They are respectively expressed as the reduction of electrical load and thermal load of VPPi at time t; and They are respectively expressed as the average heat purchase and sale prices of the upper heat network; and They are respectively expressed as the price of electricity purchased and sold to the upper-level power grid; and They are respectively expressed as the compensation price for reducing electricity and heat load by VPPi; and They are respectively represented as the electricity and heat traded from VPPi to LA at time t; Expressed as the power transmission cost coefficient from VPPi to VPPj; As a combination of multiple energy supply units, the combined heat and power unit CHP converts and distributes electrical energy and thermal energy through a gas turbine (GT) and a waste heat recovery boiler (WHRB); GT is used to generate electricity, and WHRB provides thermal energy by recovering the flue gas emitted by GT; according to the operating characteristics of the gas turbine GT, the operating cost of CHP can be calculated by the relationship between electrical and thermal power output and fuel consumption; for ESS, the uncertainty and volatility of renewable energy output are stabilized by controlling its charging and discharging state, and it is assumed that its rental cost is linearly related to the stored energy; therefore, the operating cost formula of DER in VPP is shown in formula (12): （12） Where: Unit operating cost expressed as VPPi; It is represented by the GT power output in VPPi at time t; Expressed as ESS rental price; Expressed as natural gas price; Expressed as GT's operating efficiency; and They are respectively represented as the charging and discharging power in VPPi at time t.

5. A virtual power plant low-carbon bargaining optimization method based on differentiated multi-agent collaborative game as claimed in claim 4, characterized in that: The specific process of step S14 is as follows: S141. Establish cogeneration equipment constraints: The operation of gas turbine GT and waste heat recovery boiler WHRB is subject to hourly power generation constraints. GT operation is limited by the ramp-down rate, which is described as shown in equation (13): （13） Where: It is expressed as the electrical output of GT in VPPi at time t; It is expressed as the maximum power output of GT in VPPi; and They are respectively expressed as the maximum rising and falling rates of GT in VPP; It is expressed as the WHRB heat output in VPPi at time t; It is expressed as the maximum heat output of WHRB in VPPi at time t; S142. Establishing energy storage system charging and discharging constraints: In order to alleviate the impact of energy consumption and renewable energy fluctuations on VPP scheduling, it is necessary to control the charging and discharging state of ESS; at any time t, the ESS energy storage depends on the energy storage level at the previous time t~1, and the energy storage should remain consistent at the beginning and end of the cycle; the charging and discharging amount, frequency and energy storage capacity are all limited, and the charging and discharging operations cannot be performed simultaneously; the specific constraint conditions are described as shown in formula (14): （14） Where: and They are represented as binary variables that determine the charge and discharge state of ESS at time t, 1 indicates the charge and discharge state, and 0 indicates the non-charge and discharge state; and They are respectively expressed as the charging and discharging efficiency of ESS; It is represented as the energy stored in VPPi by ESS at time t; It is expressed as the degradation coefficient of ESS in VPPi; and They represent the minimum and maximum capacities of ESS in VPPi, respectively; and They are respectively represented as the maximum charge and discharge capacity of ESS in VPPi; It is expressed as the maximum number of discharge and charge times of ESS in VPPi; and They are respectively represented as the charging and discharging powers in VPPi at time t; S143. Establish wind power and photovoltaic constraints: Since wind power, photovoltaic power and load have random fluctuations in different time periods, a random optimization method is used to quantify uncertainty. Based on the Latin hypercube sampling scenario generation method, 500 scenarios are generated for each uncertainty factor, and 5 typical scenarios and their corresponding probabilities are selected. The probability of each scenario is multiplied by its predicted value and then summed to obtain the random optimization result. The specific constraint description is shown in formula (15): （15） Where: and They are represented as the stochastically optimized wind power and PV power generation of VPPi at time t, respectively; and Respectively represent the time t in the scenario and PV and wind power forecasts under VPPi; and Respectively represented as VPPi in scenario and probability; S144. Establish demand response DR constraints: Based on the DR mechanism, the electrical load and thermal load need to satisfy the equation (16), which can be limited to the customer's energy comfort range to a certain extent; （16） Where: and They are respectively represented as the random optimized electrical load and scenario of VPPi at time t. Power forecast for VPPi; and are the random optimal heat load and scenario of VPPi at time t, respectively. Heat load prediction of VPPi; and They are expressed as the maximum power and heat load reduction percentages of VPPi, respectively; and They are respectively expressed as the reduction of electrical load and thermal load of VPPi at time t; S145. Establish energy-carbon trading volume constraints on VPP: In order to avoid network congestion caused by MVPP energy trading, an upper limit constraint is set on the energy trading volume, and the energy-carbon interaction between MVPPs must meet the balance constraint; the specific description is shown in formula (17): （17） Where: and They are respectively represented as the electricity and heat purchased by VPPi from the upper network at time t; and They represent the maximum power and heat purchased by VPP from the upper network respectively; and They are the maximum power and heat sold by VPP to the upper network respectively; and They are respectively expressed as the maximum power and heat exchanged between VPPs; and They are respectively represented as the electric power and heat traded by VPPi from VPPj at time t; It is expressed as the carbon trading volume from VPPi to VPPj; It is represented as the amount of electricity sold by VPPi to the upper network at time t; It is represented as VPPi selling heat power to the upper network at time t; S146. Establish VPP energy balance constraints: The VPP electrical and thermal balance, including the internal unit energy supply and the interaction with the LA, MVPP and upper network energy, is described as shown in Equation (18): （18） Where: It is represented as the amount of electricity traded from VPPi to LA at time t; It is represented by the amount of electricity purchased by VPPi from the upper network at time t; It is represented as the amount of electricity sold by VPPi to the upper network at time t; It is expressed as the heat traded from VPPi to LA at time t; It is expressed as the heat output of WHRB in VPPi at time t; It is represented as VPPi buying heat from the upper network at time t; It is represented as VPPi selling heat power to the upper network at time t; It is represented as the stochastic optimized wind power of VPPi at time t; Denoted as VPPi the electrical power traded from VPPj at time t.

6. A virtual power plant low-carbon bargaining optimization method based on differentiated multi-agent collaborative game as claimed in claim 5, characterized in that: The specific process of step S2 is as follows: Assuming that MVPP and LA are independent entities in the cooperation, each participant hopes to find a balanced transaction strategy in the negotiation to maximize its own interests; therefore, the reasonable determination of transaction price and transaction volume becomes the focus of the MVPP-LA cooperation alliance; the traditional Nash bargaining model, as shown in formula (19), can quantify the benefits of multilateral cooperation for each participant; in the Nash bargaining process, the transaction volume of a VPP and the transaction price with other VPPs and LA are regarded as decision variables; and the Nash bargaining breaking point is a fixed value, at which point each VPP only trades with its upper network and interacts with LA alone in the Stackelberg game; at the Nash bargaining breaking point, VPP, as a leader, sets energy prices according to user energy demand, while LA, as a follower, adjusts its energy use according to VPP price signals; VPP and LA play continuous games with the goal of maximizing their respective interests until an equilibrium solution is reached; The traditional bargaining model is described as shown in formula (19): （19） Where: and are the Nash bargaining breaking points, which represent the benefits of LA and VPPi before cooperation; and They represent the increased benefits of LA and VPPi through bargaining respectively; From equations (9) to (18), we can see that the established Nash bargaining model is a non-convex nonlinear optimization problem, where equation (7) indicates that the step-type carbon trading cost is a piecewise function, which can be obtained by using the big M method and introducing the convex transformation of Boolean variables; Equations (9) and (11) contain the product of two variables, transaction volume and transaction price, which are difficult to linearize using conventional methods. Therefore, in order to solve the non-convex model, it is decomposed into two sub-problems according to the overall strategy and individual interests of the cooperative game. These two sub-problems include the maximization and optimal allocation of the MVPP-LA alliance’s interests, which respectively determine the optimal interaction volume and interaction price within the alliance. By substituting the LA and MVPP objective functions and constraints, the overall cooperation benefit maximization subproblem can be obtained, as shown in formula (20): （20） Where: It is expressed as the energy transmission cost between VPPi and LA; Unit operating cost expressed as VPPi; It is expressed as the energy transmission cost between VPPi and other VPPs; It is expressed as the energy transaction cost between VPPi and the upper network; It is expressed as the carbon transaction cost between VPPi and the carbon market; It is expressed as the energy transfer cost between VPPi and other VPPs; Since formula (20) eliminates the transaction costs between entities, only the transaction volume of each entity can be obtained; substituting it into the original bargaining model (19), the optimal molecular problem of alliance benefits can be obtained, which is described as shown in formula (21): （21） Where: , , , , They represent the optimal solutions obtained by formula (20) respectively; the English character annotations in formula (21) are explained in formula (10); A nonlinear mapping method based on exponential function is proposed to quantify the energy interaction contribution of each VPP in the alliance, and to evaluate the role of each VPP in Nash bargaining. The contribution quantification function is described as shown in formula (22): （22） Where: and They represent the energy purchased and sold by VPPi within the alliance, respectively; It is expressed as the contribution factor of VPPi to MVPP cooperation; and They are respectively represented as the power and heat traded by VPPi from VPPj at time t; Based on the traditional bargaining formula (19), combined with formula (22), the contribution-based bargaining model of the benefit distribution sub-problem formula (21) is modified as shown in formula (23): （23） Where: and are the Nash bargaining breaking points, which represent the benefits of LA and VPPi before cooperation; and They represent the increased benefits of LA and VPPi through bargaining respectively; It is expressed as the contribution factor of VPPi to MVPP cooperation; To ensure that MVPP and LA have priority in bargaining transactions within the alliance, the energy bargaining price should be within the range of the energy purchase and sale price of the superior network, and the CEA bargaining between MVPPs should be constrained, that is, not lower than the carbon market sales price and not higher than the carbon market purchase price; the description is shown in formula (24): （24） Where: It is expressed as the electricity transaction price between VPPi and VPPj at time t; and They are respectively represented as the electricity purchase and electricity sales prices from the upper power grid at time t; and They are respectively expressed as the average heat purchase and sale price of the upper heat network; It is expressed as the heat transaction price between VPPi and VPPj at time t; It is expressed as the electricity transaction price of VPPi at time t; It is expressed as the heat transaction price between VPPi and LA at time t; Expressed as a benchmark carbon price; It is expressed as the CEA transaction price between VPPi and VPPj; Expressed as the carbon price corresponding to the trading range of carbon market trading volume.

7. A virtual power plant low-carbon bargaining optimization method based on differentiated multi-agent collaborative game as claimed in claim 6, characterized in that: The specific process of step S3 is as follows: S31. Solve the sub-problem of maximizing alliance benefits: From constraints (16) and (18), we can see that the maximum coalition benefit subproblem contains energy and carbon interaction coupling variables: , , , , ; Introduce auxiliary variables , , , , , , Decouple atomic problems; when equation (25) is satisfied, it indicates that all participants have reached a consensus on the bargaining; （25） Where: and They are respectively represented as the electric power and heat purchased by LA from VPPi at time t; and They are respectively represented as the electrical power and heat output by VPPi to LA at time t; It is expressed as the carbon trading volume from VPPi to VPPj at time t; Based on the basic principles of distributed optimization algorithm, Lagrangian multipliers and penalty factors are introduced to create the following augmented Lagrangian functions for LA and MVPP, respectively, as shown in Equations (26) and (27): （26） （27） Where: , , , , denote the Lagrange multipliers respectively; is the penalty factor; the English character annotations of formulas (26) and (27) are explained in formulas (13) to (18); S32. Solve the optimal benefit allocation sub-problem: The optimal benefit allocation subproblem includes coupling variables , , , , introduce corresponding auxiliary variables , , , , , , Decoupling; The relationship between variables is described as shown in formula (28): （28） Where: and They are respectively represented as the price of electricity and heat purchased by LA from MVPP at time t; and They are respectively represented as the price at which MVPP sells power and heat to LA at time t; It is expressed as the CEA transaction price between VPPi and VPPj; Equation (23) is logarithmically transformed into a convex optimization problem, and the LA and MVPP augmented Lagrangian functions are established as shown in Equations (29) and (30): （29） （30） Where: , , , , It means that the optimal solution is derived from equations (26) and (27); , , , , represents the Lagrange multiplier; It is expressed as a penalty factor; the English character annotations of formulas (29) and (30) are explained in formula (24); S33, solution based on distributed adaptive iterative algorithm: In order to protect the data privacy of MVPP and LA during the bargaining process, an improved ADMM is used to solve two sub-problems in a distributed manner. Based on the traditional fixed penalty factor ADMM, an update mechanism is proposed to dynamically adjust the penalty factor during the iteration process. Let κ and σ be the scale factors, and the penalty factor The adaptive update method is described as shown in formula (31): （31） Where: and They are expressed as raw residual and double residual respectively; is the number of iterations; κ and σ are scale factors, Expressed as a penalty factor; The variable update equation and convergence condition of the "benefit maximization-optimal benefit allocation" two-level subproblem based on the distributed adaptive DP-ADMM iterative algorithm are described as Equation (32) and Equation (33), respectively. For the initial value of the variable, the energy transaction volume and the Lagrange multiplier are set to 0, and the penalty factor is set to 10. −2 , the iterative update description of coupling variables and auxiliary variables is shown in formula (32): （32） Where: , , , , They are respectively represented as the Lagrange multipliers iterated to the dth time; , , , , They are respectively represented as iterations to the (d+1)th Lagrange multiplier; is the penalty factor; and They are respectively represented as the power and heat purchased by LA from VPPi at the dth iteration at time t; and They are respectively represented as the power and heat output by VPPi to LA at the dth iteration at time t; It is expressed as the carbon trading volume from VPPi to VPPj at time t; , , They are , , Introduce auxiliary variables, and They are respectively the power and heat traded by VPPi from VPPj at time t, , , They are respectively represented by the introduction of auxiliary variables at the dth iteration at time t, see formula (25); They respectively represent the introduction of auxiliary variables at the (d+1)th iteration; It is represented as the first LA optimal iteration variable parameter set; and They are represented as the optimal iterative variable parameter sets of VPPi and VPPj respectively; and They are respectively represented as the power and heat traded by VPPi from VPPj at time t; The dual variable and penalty factor update formula and the convergence condition of the variables are described as shown in formula (33): （33） Where: , , , , They are respectively represented as the Lagrange multipliers iterated to the dth time; , , , , They are respectively represented as the iteration to the d+1th Lagrange multiplier; Represented as the penalty factor for iteration to the dth time; and They are respectively represented as the power and heat purchased by LA from VPPi at the dth iteration at time t; and They are respectively represented as the power and heat output by VPPi to LA at the dth iteration at time t; It is expressed as the carbon trading volume from VPPi to VPPj at time t; , , They respectively represent the introduction of auxiliary variables at the dth iteration at time t, and the comments are shown in formula (25).

Images