Multi-market coordinated multi-virtual power plant bidding method under low carbon perspective

Abstract

This invention discloses a multi-market collaborative bidding method for virtual power plants from a low-carbon perspective, belonging to the field of multi-market collaborative bidding for electricity. The method includes the following steps: S1, constructing a five-flow coupling model of electricity, carbon, heat, hydrogen, and green certificates, and classifying virtual power plant types to generate a triple uncertainty scenario; S2, building a three-order nested game bidding model; S3, constructing a joint risk management model, outputting risk costs and robust bidding feasible regions; S4, based on the electricity-carbon supply-demand ratio and robust feasible regions, realizing bidirectional matching and pairing transactions of electricity and carbon among multiple virtual power plants, generating initial bidding schemes for the five markets; S5, coupling the game bidding model, risk costs, and initial bidding schemes to obtain the optimal equilibrium bidding solution for the five markets; S6, feasibility verification and closed-loop optimization. Using the above-mentioned multi-market collaborative bidding method for virtual power plants from a low-carbon perspective, the economic benefits and long-term stable operation of the multi-virtual power plant alliance are maximized under the premise of low-carbon compliance.

Description

Technical Field

[0001] This invention relates to the field of multi-market collaborative bidding technology for electricity, and in particular to a multi-virtual power plant bidding method from a low-carbon perspective. Background Technology

[0002] With the deepening construction of new power systems, the high proportion of new energy sources (wind power and photovoltaics) connected to the grid has led to a significant increase in the volatility of power system supply and demand. Virtual power plants (VPPs), as the core carriers aggregating distributed power sources, energy storage, and controllable loads, have become an important support for smoothing system fluctuations and improving the absorption of new energy sources due to their flexible adjustment capabilities. At the same time, the gradual improvement of multiple market mechanisms, such as the national carbon market, the renewable energy green electricity certificate market, the hydrogen energy trading market, and the heat market, has provided a foundation for multiple VPPs to participate in multi-market collaborative bidding and achieve synergistic optimization of economic benefits and low-carbon emission reduction.

[0003] Currently, existing technologies for optimizing multi-VPP bidding have been researched, mainly focusing on bidding in a single electricity market or electricity-carbon linkage bidding. Relying solely on electricity and carbon markets for resource allocation cannot fully utilize the idle resources of heat, hydrogen, and renewable energy green certificates on the virtual power plant side, resulting in low comprehensive resource utilization. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-market collaborative bidding method for multiple virtual power plants from a low-carbon perspective, thereby solving the aforementioned technical problems.

[0005] To achieve the above objectives, this invention provides a multi-market collaborative bidding method for multiple virtual power plants from a low-carbon perspective, comprising the following steps: S1, collecting physical, multi-market, and uncertainty data of multiple virtual power plants, constructing a five-flow coupling model of electricity, carbon, heat, hydrogen, and green certificates, classifying virtual power plant types, and generating a triple uncertainty scenario; S2, building a third-order nested game bidding model based on the five-flow coupling model and the electricity-carbon supply-demand ratio; S3, integrating the triple uncertainty scenario and the third-order nested game bidding model to construct a joint risk management model, outputting risk costs and robust bidding feasible regions; S4, realizing bidirectional matching and pairing transactions of electricity and carbon among multiple virtual power plants based on the electricity-carbon supply-demand ratio and robust feasible regions, generating initial bidding schemes for the five markets; S5, coupling the game bidding model, risk costs, and initial bidding schemes to obtain the optimal equilibrium bidding solution for the five markets; S6, feasibility verification and closed-loop optimization.

[0006] Therefore, the multi-market collaborative multi-virtual power plant bidding method adopted by this invention from a low-carbon perspective has the following beneficial effects: 1. Two-way synergistic optimization of electricity and carbon, breaking through the limitations of single market bidding: Through dynamic matching of electricity and carbon supply and demand ratio and P2P two-way trading within the alliance, the supply and demand balance of electricity and carbon quotas is optimized simultaneously, solving the problem of "disconnect between electricity trading and carbon emissions" under the traditional single market bidding model, and significantly improving the overall resource allocation efficiency of multi-virtual power plant alliances; 2. Five-flow coupling + multi-market linkage to tap into the full-dimensional revenue potential: Based on the five-flow coupling model of electricity-carbon-heat-hydrogen-green certificates, it links five major markets: electricity, carbon trading, heat, hydrogen energy and green certificates. It breaks through the limitation of traditional solutions that only focus on the electricity market, fully taps into the multi-market revenue potential of multiple virtual power plants, and improves the overall economic benefits of the alliance. 3. Three-order nested game + evolutionary stable equilibrium to ensure long-term stability of the alliance: The three-order nested game framework of "cluster-led pricing - individual optimization operation - strategy evolution screening" is adopted. The fair distribution of benefits is achieved through asymmetric Nash negotiation, and the cooperative bidding strategy is screened by evolutionary stable equilibrium (ESS), which solves the problem of conflict of interest of virtual power plants in the alliance and ensures the long-term stable operation of the alliance. 4. Multi-source uncertainty management to improve the robustness and feasibility of the solution: The triple risk cost model is constructed by integrating CVaR (Conditional Value at Risk), IGDT (Information Gap Decision Theory), and DRO (Distributed Robust Optimization). The bidding solution is dynamically corrected through deep reinforcement learning, which effectively mitigates decision-making biases caused by multi-source uncertainties such as wind and solar power output, electricity price, and carbon price, and improves the robustness of the solution in complex market environments. 5. Closed-loop rolling optimization mechanism to adapt to dynamic operating scenarios: Construct a full-process closed-loop update mechanism based on low-carbon and economic indicators to dynamically correct equipment operating parameters, market transaction prices and uncertainty scenarios, realize rolling iterative optimization of bidding schemes, adapt to the dynamic operating environment in the actual scheduling process, and ensure the continuous effectiveness of the scheme.

[0007] In summary, compared with the prior art, the present invention improves carbon emission reduction rate by 43.08%, bidding revenue by 44.86%, solution convergence speed by 15.8%, uncertainty risk by 23.6%, and renewable energy consumption rate by 30.21%.

[0008] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0009] Figure 1 This is a flowchart of the multi-market collaborative multi-virtual power plant bidding method from a low-carbon perspective, as described in this invention. Detailed Implementation

[0010] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the embodiments of the present invention and are not intended to limit the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of this application. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout.

[0011] It should be noted that the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, such as a process, method, system, product, or server that includes a series of steps or units, not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such process, method, product, or device.

[0012] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0013] like Figure 1 As shown, the multi-market collaborative bidding method for multiple virtual power plants from a low-carbon perspective includes the following steps: S1, collecting physical, multi-market, and uncertainty data of multiple virtual power plants, constructing a five-flow coupling model of electricity, carbon, heat, hydrogen, and green certificates, classifying virtual power plant types, and generating a triple uncertainty scenario; S2, building a third-order nested game bidding model based on the five-flow coupling model and the electricity-carbon supply-demand ratio; S3, integrating the triple uncertainty scenario and the third-order nested game bidding model to construct a joint risk management model, outputting risk costs and robust bidding feasible regions; S4, realizing bidirectional matching and pairing transactions of electricity and carbon among multiple virtual power plants based on the electricity-carbon supply-demand ratio and robust feasible regions, generating initial bidding schemes for the five markets; S5, coupling the game bidding model, risk costs, and initial bidding schemes to obtain the optimal equilibrium bidding solution for the five markets; S6, feasibility verification and closed-loop optimization.

[0014] Step S1 specifically includes the following steps: S11. Synchronously collect physical, market, and uncertainty data from multiple virtual power plants to obtain raw operational data. The physical data for multiple virtual power plants includes the installed capacity, operating efficiency, ramp rate, SOC upper and lower limits, and heat-to-power ratio constraints of wind turbines, photovoltaics, gas turbines, battery energy storage, cogeneration units, boilers (gas-fired boilers or heating boilers), electrolyzers, hydrogen storage tanks, thermal storage devices, and controllable loads. The market data includes electricity market data (corresponding to electricity flows: including day-ahead and real-time electricity prices, ancillary service prices (including transaction prices for frequency regulation, reserve, peak shaving, and other ancillary services), and electricity market transaction upper and lower limits). The data includes: carbon trading market data (corresponding to carbon flows: including tiered carbon trading benchmark price, tiered carbon price growth rate, and upper and lower limits for carbon quota market trading); heat market data (corresponding to heat flows: including heat trading time-of-use price and upper and lower limits for heat market trading); hydrogen energy market data (corresponding to hydrogen flows: including hydrogen production and sales prices and upper and lower limits for hydrogen market trading); and green certificate market data (corresponding to green certificate flows: including green certificate trading price and upper and lower limits for green certificate market trading). Uncertainty data includes wind and solar power output forecasts, load forecasts, electricity price fluctuation ranges, carbon price fluctuation ranges, hydrogen price fluctuation ranges, and user demand response behavior data. S12. Perform data cleaning and standardization, outlier removal, and linear interpolation to fill in missing values ​​on the original running data in sequence, and unify the time scale to 15 minutes to form a standardized original dataset. S13. Construct a five-flow coupling model of electricity, carbon, heat, hydrogen, and green certificates; Power flow parameter modeling: ; In the formula, For virtual power plants Time period Net power; For virtual power plants Time period Wind power output; For virtual power plants Time period Photovoltaic power output; For virtual power plants Time period The output of the gas turbine; For virtual power plants Time period The power generation capacity of the combined heat and power unit; and Virtual power plants Time period The charging and discharging power of the energy storage device; For virtual power plants Time period Electricity purchased from the power grid; For virtual power plants Time period The electrical power sold to the power grid; For virtual power plants Time period User electrical load power; For virtual power plants Time period The power consumption of the electrolytic cell; Carbon flow parameter modeling: ; In the formula, For virtual power plants Time period Net carbon emissions; , , These are the carbon emission coefficients for gas turbines, combined heat and power units, and electricity purchased from the grid. For virtual power plants Time period Free carbon credits; For virtual power plants Time period The amount of carbon allowances shared by the alliance; This represents the carbon emission reduction factor corresponding to a single green certificate. For virtual power plants Time period The number of green certificates held; Heat flux parameter modeling: ; in, ; In the formula, For virtual power plants Time period Net heat power; and Virtual power plants Time period The heating capacity of cogeneration units and boilers; and These are the heat charging power and heat dissipation power of the thermal storage device, respectively. This refers to the heat load power. The heat-to-power ratio of a combined heat and power (CHP) unit; Hydrogen flow parameter modeling: ; In the formula, For virtual power plants Time period Net hydrogen content; The efficiency of hydrogen production in the electrolyzer; For virtual power plants Time period The power consumption of the electrolytic cell; and Virtual power plants Time period The hydrogen charging and discharging capacity of the hydrogen storage device; Hydrogen loading; Green certificate flow parameter modeling: ; In the formula, For virtual power plants Time period The number of green certificates issued; The conversion factor for renewable energy green certificates; For virtual power plants Time period The output of renewable energy, and ; S14. Based on the five-flow coupling model of electricity-carbon-heat-hydrogen-green certificate, establish a comprehensive pricing model for five markets, a tiered carbon trading cost model, and an equipment operation and maintenance cost model to obtain a low-carbon parameter library. Five-market integrated pricing model: ; In the formula, For virtual power plants Time period The comprehensive market electricity price; , , , , These are the time-sharing prices for electricity, carbon allowances, heat, hydrogen, and green certificates, respectively. Tiered carbon trading cost model: ; in, ; In the formula, For virtual power plants Time period The cost of carbon trading; It serves as the benchmark price for carbon trading; For carbon trading tiered intervals; The carbon price gradient growth rate; For virtual power plants Time period The difference between net carbon emissions and free carbon allowances; Equipment maintenance cost model: ; In the formula, For virtual power plants Time period The total cost of equipment operation and maintenance; , , , , , The unit power operation and maintenance costs are respectively for wind turbines, photovoltaics, gas turbines, combined heat and power units, energy storage devices, and electrolyzers; S15. For the three types of uncertainty—source, load, and price—generate a triple scenario combining randomness, interval fuzziness, and distribution fuzziness: ; In the formula, For the total set of uncertain scenarios; These are respectively: a set of random uncertainties (which is a set of random scenarios of wind and solar power and electricity prices generated by Latin hypercube sampling), a set of interval uncertainties (which is a set of interval uncertainties of carbon price and hydrogen price constructed by information gap decision theory), and a set of distributed fuzzy uncertainties (which is a set of distributed Bruker scenarios constructed by Wasserstein distance). For the scene The probability of occurrence; S16. Calculate the electricity-carbon supply-demand ratio based on the five-current coupling model of electricity-carbon-heat-hydrogen-green certificates and classify virtual power plants: Electricity and carbon sales type: ; Electricity-carbon dual balance type: ; Purchase electricity or carbon fiber: ; in, ; In the formula, For virtual power plants Time period The supply and demand ratio of carbon electricity; For virtual power plants Time period The electricity supply-demand ratio; For virtual power plants Time period The carbon quota supply-demand ratio; For virtual power plants Time period Total power supply; For virtual power plants Time period Total electricity demand; For virtual power plants Time period Total carbon allowance supply; For virtual power plants Time period The total demand for carbon allowances.

[0015] The three-level nested game bidding model framework described in step S2 is as follows: The first layer is dominated by the multi-virtual power plant cluster operator, which sets dynamic pricing for the five markets of electricity, carbon, heat, hydrogen, and green certificates; each virtual power plant is a follower at the lower level, responding to pricing and optimizing its own operating strategy. The second layer calculates bargaining power based on the five-flow transaction contribution of each VPP and uses asymmetric Nash negotiation to fairly distribute the benefits of alliance cooperation. The third layer iteratively optimizes the bidding strategy of the virtual power plant based on market dynamics and selects a stable equilibrium solution.

[0016] In the first layer of the three-order nested game bidding model, the objective function of the upper-level multi-virtual power plant cluster operator is as follows: ; The objective function for the lower-level virtual power plant is as follows: ; in, ; ; ; ; In the formula, To maximize the total revenue of the multi-virtual power plant cluster; The number of virtual power plants; For scheduling periods; , , , , These are the trading revenues from the electricity market, carbon trading market, green certificate market, heat market, and hydrogen energy market, respectively. For virtual power plants Time period The total operating cost, and ; To minimize the cost of low-carbon operation, For virtual power plants Time period Energy purchase cost, For virtual power plants Time period Demand response compensation costs; For virtual power plants Time period The triple uncertainty risk cost; , , All are weighting coefficients; , , These are, respectively, the conditional value of risk based on random scenarios, the robust cost based on interval scenarios, and the sub-Bruker optimization cost based on distributed fuzzy scenarios; Value at risk; Confidence level; For the scene Virtual power plant exist Bidding losses during a specific time period; Let be the operating cost function of a virtual power plant that fluctuates with uncertainty variables under the scenario of interval uncertainty; For distributed fuzzy uncertainty sets Take the supremacy within; To follow a probability distribution The mathematical expectation.

[0017] In the second layer of the third-order nested game bidding model, the objective function for Nash negotiation payout allocation is as follows: ; in, ; In the formula, For virtual power plants The benefits of cooperation; For virtual power plants Independent operating revenue; For virtual power plants Bargaining power; , , , , The contribution weights for electricity, carbon allowances, heat, hydrogen energy, and green certificate trading are respectively. For virtual power plants Electricity trading volume; For virtual power plants Carbon quota trading volume; For virtual power plants Heat trading volume; For virtual power plants Hydrogen energy trading volume; For virtual power plants The volume of green certificate transactions.

[0018] In the third layer of the three-order nested game bidding model, the criteria for selecting a stable equilibrium solution are as follows: Simultaneously satisfy and The bidding strategy is a stable equilibrium solution; among which, Copying dynamic functions for bidding strategies The first derivative; Copying dynamic functions for bidding strategies The second derivative; The dynamic equation for replicating the bidding strategy is as follows: ; In the formula, For virtual power plants employing a cooperative bidding strategy Proportion; For virtual power plants using a non-cooperative bidding strategy Proportion; For virtual power plants Net revenue from five-market synergy using a bidding strategy; The average return of the bidding strategy for the virtual power plant group.

[0019] Step S4 specifically includes the following steps: S41. Based on the classification of virtual power plants in step S16, the following dynamic matching rule is set: Virtual power plants that sell electricity and carbon dioxide are preferentially matched with virtual power plants that purchase electricity and carbon dioxide, and the matching degree is calculated: ; In the formula, For virtual power plants With virtual power plants During the period The degree of matching; For virtual power plants During the period The supply and demand ratio of carbon electricity; S42. Sort the matching scores in descending order and filter before... Each virtual power plant is considered a matched virtual power plant, and the total amount of electricity carbon trading for each matched virtual power plant pair is calculated. Electricity trading volume: ; Carbon quota trading volume: ; In the formula, for Time-based virtual power plant To virtual power plants Electricity sales volume; and Virtual power plants Electricity surplus and virtual power plants The power shortage; for Time-based virtual power plant To virtual power plants Carbon quota sales; and Virtual power plants Carbon allowance surplus and virtual power plants Carbon quota deficit; S43. Calculate the carbon trading price for virtual power plant pairs; Electricity trading prices: ; Carbon trading prices: ; In the formula, For electricity trading prices; and Time periods The benchmark purchase price and sales price of electricity in the power grid; For carbon trading prices; and Time periods The benchmark buying and selling prices in the carbon allowance market; S44. Initial bidding scheme for the five markets of electricity-carbon-thermal-hydrogen-green certificates; ; ; ; ; ; In the formula, , , , , Virtual power plants exist The bidding volume for electricity, carbon, green certificates, heat, and hydrogen during the time period; For virtual power plants During the period The number of green certificates required to meet the renewable energy quota system requirements.

[0020] Step S5 specifically includes the following steps: S51. Construct a two-dimensional adaptive ADMM hybrid solution framework, coupling the game bidding model, joint risk cost, and five-market initial bidding scheme into a unified convex optimization problem, constructing the augmented Lagrangian function and completing the algorithm initialization: ; In the formula, For the augmented Lagrangian function under adaptive penalty factor; The total net revenue of five-market collaboration for multiple virtual power plant clusters during the scheduling period; For virtual power plants Total cost of low-carbon operation during the scheduling cycle; This is the transpose of the Lagrange multiplier vector; , All are constraint coefficient matrices; The constraint constant vector; For the set of decision variables in virtual power plant bidding; For the set of transaction control variables of multiple virtual power plant cluster operators; Iteration penalty factor; S52. Perform a two-dimensional adaptive penalty factor update combining residual balancing and bargaining weights to dynamically adjust the iterative penalty factor and improve the solution convergence speed: ; in, ; ; In the formula, For the first Sub-iteration fusion penalty factor; The penalty factor fusion weight coefficient; and These are adaptive penalty factors for the residual dimension and the bargaining dimension, respectively; The penalty factor scaling factor; For the first Sub-iteration fusion penalty factor; and The first The original residual and dual residual of the next iteration; and All are residual ratio thresholds; S53. Based on the adaptive penalty factor, perform ADMM distributed alternating iterative solution, updating decision variables and Lagrange multipliers until the iteration termination threshold condition is met, and output the ADMM hybrid solution result. , These are the original bid quantities for electricity, carbon, heat, hydrogen, and green certificates obtained through ADMM hybrid solution. S54. Construct a deep reinforcement learning DRL rolling correction model to dynamically correct the ADMM hybrid solution and output the optimal equilibrium bidding solution for the five markets.

[0021] In step S54, the bidding adjustment process is modeled as a Markov decision process, and state variables are set. Action variables reward function ;in, These are deep reinforcement learning models for virtual power plants. During the period The adjustment amounts for the bidding volumes of electricity, carbon, heat, hydrogen, and green certificates; For virtual power plants Net income from five market synergies; For virtual power plants Total operating costs; For virtual power plants The combined risk cost of triple uncertainty; The weighting factor for low-carbon benefits; Iterative optimization using the DQN value function: ; In the formula, The optimal action value function; Discount factor; For state variables Time period Next, from the next moment Among all feasible bid adjustment actions, select the one that allows the optimal action value function to be achieved. The action that reaches the maximum value, and the maximum overall reward value in that state; For expectation operators; The optimal bidding correction strategy is as follows: ; In the formula, , , , , These are the corrected optimal bid quantities for electricity, carbon, heat, hydrogen, and green certificates, respectively. After iterative convergence, the set of optimal equilibrium bidding solutions for the five markets is obtained. ;in, Virtual power plants To virtual power plants During the period The optimal electricity trading volume and carbon quota trading volume; Virtual power plants To virtual power plants During the period The optimal electricity trading price and carbon quota trading price.

[0022] Step S6 specifically includes the following steps: S61. Set up a set of bidding strategies for virtual power plants. , The cooperative bidding strategy for participating in S4's matching transactions. This is a non-cooperative bidding strategy (independently participating in the market, without participating in alliance collaboration); S62. Calculate the average revenue of a cluster of multiple virtual power plants: ; In the formula, For virtual power plants The average return of the bidding strategy; For virtual power plants Net synergistic gains from five markets using a cooperative bidding strategy ; For virtual power plants Net income from independent operation using a non-cooperative bidding strategy; S63. Substituting the average revenue of the bidding strategy obtained in S62 into the dynamic equation for replicating the bidding strategy, we obtain the dynamic evolution formula for the proportion of the virtual power plant cooperation strategy: ; In the formula, The dynamic function for replicating the bidding strategy of virtual power plants characterizes the rate of change of the proportion of cooperative strategies. S64. Using the Euler method to iteratively solve for the proportion of cooperative bidding strategies at different times. Continue until convergence, then output the steady-state policy ratio. ; S65, First, filter those that meet the requirements. The steady-state point, then the selected... Perform the first derivative, if it satisfies Then the If a point is determined to be an asymptotically stable equilibrium point, only the bidding strategy corresponding to this asymptotically stable equilibrium point will be considered as the candidate optimal strategy. S66. Based on the candidate optimal strategy, sequentially verify the energy balance constraints and operational constraints of power flow, carbon flow, heat flow, hydrogen flow, and green certificate flow, and eliminate invalid strategies that do not meet the constraints. S67. Calculate the economic accounting indicators and low-carbon accounting indicators that have passed the feasibility verification; The economic accounting indicators are expressed as follows: ; In the formula, The total operating cost of a multi-virtual power plant cluster; , , , , These include the equipment operation and maintenance costs of the virtual power plant, energy purchase costs, carbon trading costs, demand response compensation costs, and the combined risk costs of three uncertainties. Total revenue across five markets for multi-virtual power plant clusters; , , , , These represent the trading revenues of virtual power plants in the electricity market, carbon trading market, green certificate market, heat market, and hydrogen market, respectively. The expression for low-carbon accounting indicators is as follows: ; In the formula, , , These are carbon emission reduction rate, renewable energy integration rate, and green certificate completion rate, respectively. Carbon emissions for independent operation; The actual carbon emissions generated after implementing the candidate optimal strategy after eliminating invalid strategies; The total renewable energy power actually consumed by the multi-virtual power plant cluster during the dispatch cycle; The total renewable energy generation capacity of a multi-virtual power plant cluster during the scheduling period; The number of valid green certificates actually held by the virtual power plant during the dispatch cycle; The number of green certificates required for a virtual power plant; S68. Update step S1 based on the accounting results.

[0023] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A multi-market collaborative bidding method for multiple virtual power plants from a low-carbon perspective, characterized by: Includes the following steps: S1. Collect physical, market, and uncertainty data from multiple virtual power plants, construct a five-flow coupling model of electricity, carbon, heat, hydrogen, and green certificates, classify virtual power plant types, and generate a triple uncertainty scenario; S2. Build a third-order nested game bidding model based on the five-flow coupling model and the electricity-carbon supply-demand ratio; S3. Integrate the triple uncertainty scenario and the third-order nested game bidding model to construct a joint risk management model, and output the risk cost and robust bidding feasible region; S4. Based on the electricity carbon supply-demand ratio and robust feasible region, realize bidirectional matching and pairing trading of electricity carbon from multiple virtual power plants, and generate initial bidding schemes for five markets; S5. Coupled game bidding model, risk cost and initial bidding scheme, to obtain the optimal equilibrium bidding solution for the five markets; S6. Feasibility verification and closed-loop optimization.

2. The multi-market collaborative multi-virtual power plant bidding method from a low-carbon perspective as described in claim 1, characterized in that: Step S1 specifically includes the following steps: S11. Simultaneously collect physical, market, and uncertainty data from multiple virtual power plants to obtain raw operational data. The physical data from multiple virtual power plants includes the installed capacity, operating efficiency, ramp rate, SOC upper and lower limits, and heat-to-power ratio constraints of wind turbines, photovoltaics, gas turbines, battery energy storage, cogeneration units, boilers, electrolyzers, hydrogen storage tanks, thermal storage devices, and controllable loads. The market data includes electricity market data, carbon trading market data, heat market data, hydrogen energy market data, and green certificate market data. The uncertainty data includes wind and solar power output forecasts, load forecasts, electricity price fluctuation ranges, carbon price fluctuation ranges, hydrogen price fluctuation ranges, and user demand response behavior data. S12. Perform data cleaning and standardization, outlier removal, and linear interpolation to fill in missing values ​​on the original running data in sequence, and unify the time scale to 15 minutes to form a standardized original dataset. S13. Construct a five-flow coupling model of electricity, carbon, heat, hydrogen, and green certificates; Power flow parameter modeling: ； In the formula, For virtual power plants Time period Net power; For virtual power plants Time period Wind power output; For virtual power plants Time period Photovoltaic power output; For virtual power plants Time period The output of the gas turbine; For virtual power plants Time period The power generation capacity of the combined heat and power unit; and Virtual power plants Time period The charging and discharging power of the energy storage device; For virtual power plants Time period Electricity purchased from the power grid; For virtual power plants Time period The electrical power sold to the power grid; For virtual power plants Time period User electrical load power; For virtual power plants Time period The power consumption of the electrolytic cell; Carbon flow parameter modeling: ； In the formula, For virtual power plants Time period Net carbon emissions; , , These are the carbon emission coefficients for gas turbines, combined heat and power units, and electricity purchased from the grid. For virtual power plants Time period Free carbon credits; For virtual power plants Time period The amount of carbon allowances shared by the alliance; This represents the carbon emission reduction factor corresponding to a single green certificate. For virtual power plants Time period The number of green certificates held; Heat flux parameter modeling: ； in, ； In the formula, For virtual power plants Time period Net heat power; and Virtual power plants Time period The heating capacity of cogeneration units and boilers; and These are the heat charging power and heat dissipation power of the thermal storage device, respectively. This refers to the heat load power; The heat-to-power ratio of a combined heat and power (CHP) unit; Hydrogen flow parameter modeling: ； In the formula, For virtual power plants Time period Net hydrogen content; The efficiency of hydrogen production in the electrolyzer; For virtual power plants Time period The power consumption of the electrolytic cell; and Virtual power plants Time period The hydrogen charging and discharging capacity of the hydrogen storage device; Hydrogen loading; Green certificate flow parameter modeling: ； In the formula, For virtual power plants Time period The number of green certificates issued; The conversion factor for renewable energy green certificates; For virtual power plants Time period The output of renewable energy, and ; S14. Based on the five-flow coupling model of electricity-carbon-heat-hydrogen-green certificate, establish a comprehensive pricing model for five markets, a tiered carbon trading cost model, and an equipment operation and maintenance cost model to obtain a low-carbon parameter library. Five-market integrated pricing model: ； In the formula, For virtual power plants Time period The comprehensive market electricity price; , , , , These are the time-sharing prices for electricity, carbon allowances, heat, hydrogen, and green certificates, respectively. Tiered carbon trading cost model: ； in, ； In the formula, For virtual power plants Time period The cost of carbon trading; It serves as the benchmark price for carbon trading; For carbon trading tiered intervals; The carbon price gradient growth rate; For virtual power plants Time period The difference between net carbon emissions and free carbon allowances; Equipment maintenance cost model: ； In the formula, For virtual power plants Time period The total cost of equipment operation and maintenance; , , , , , The unit power operation and maintenance costs are respectively for wind turbines, photovoltaics, gas turbines, combined heat and power units, energy storage devices, and electrolyzers; S15. For the three types of uncertainty—source, load, and price—generate a triple scenario combining randomness, interval fuzziness, and distribution fuzziness: ； In the formula, For the total set of uncertain scenarios; These are respectively the set of random uncertainty, the set of interval uncertainty, and the set of distributed fuzzy uncertainty; For the scene The probability of occurrence; S16. Calculate the electricity-carbon supply-demand ratio based on the five-current coupling model of electricity-carbon-heat-hydrogen-green certificates and classify virtual power plants: Electricity and carbon sale type: ; Electro-carbon double balanced type: ; Purchase electricity or carbon fiber: ; in, ； In the formula, For virtual power plants Time period The supply and demand ratio of carbon electricity; For virtual power plants Time period The electricity supply-demand ratio; For virtual power plants Time period The carbon quota supply-demand ratio; For virtual power plants Time period Total power supply; For virtual power plants Time period Total electricity demand; For virtual power plants Time period Total carbon allowance supply; For virtual power plants Time period The total demand for carbon allowances.

3. The method of claim 2, wherein the low-carbon multi-market synergy multi-virtual power plant bidding method is characterized by: The three-level nested game bidding model framework described in step S2 is as follows: The first layer is dominated by the multi-virtual power plant cluster operator, which sets dynamic pricing for the five markets of electricity, carbon, heat, hydrogen, and green certificates; each virtual power plant is a follower at the lower level, responding to pricing and optimizing its own operating strategy. The second layer calculates bargaining power based on the five-flow transaction contribution of each VPP and uses asymmetric Nash negotiation to fairly distribute the benefits of alliance cooperation. The third layer iteratively optimizes the bidding strategy of the virtual power plant based on market dynamics and selects a stable equilibrium solution.

4. The method of claim 3, wherein the low-carbon multi-market synergy multi-virtual power plant bidding method is characterized by: In the first layer of the three-order nested game bidding model, the objective function of the upper-level multi-virtual power plant cluster operator is as follows: ； The objective function for the lower-level virtual power plant is as follows: ； in, ； ； ； ； In the formula, To maximize the total revenue of the multi-virtual power plant cluster; The number of virtual power plants; For scheduling periods; , , , , These are the trading revenues from the electricity market, carbon trading market, green certificate market, heat market, and hydrogen energy market, respectively. For virtual power plants Time period The total operating cost, and ; To minimize the cost of low-carbon operation, For virtual power plants Time period Energy purchase cost, For virtual power plants Time period Demand response compensation costs; For virtual power plants Time period The triple uncertainty risk cost; , , All are weighting coefficients; , , These are, respectively, the conditional value of risk based on random scenarios, the robust cost based on interval scenarios, and the sub-Bruker optimization cost based on distributed fuzzy scenarios; Value at risk; Confidence level; For the scene Virtual power plant exist Bidding losses during a specific time period; Let be the operating cost function of a virtual power plant that fluctuates with uncertainty variables under the scenario of interval uncertainty; For distributed fuzzy uncertainty sets Take the supremacy within; To follow a probability distribution The mathematical expectation.

5. The method of claim 4, wherein the low-carbon multi-market synergy multi-virtual power plant bidding method is characterized by: In the second layer of the third-order nested game bidding model, the objective function for Nash negotiation payout allocation is as follows: ； in, ； In the formula, For virtual power plants The benefits of cooperation; For virtual power plants Independent operating revenue; For virtual power plants Bargaining power; , , , , The contribution weights for electricity, carbon allowances, heat, hydrogen energy, and green certificate trading are respectively. For virtual power plants Electricity trading volume; For virtual power plants Carbon quota trading volume; For virtual power plants Heat trading volume; For virtual power plants Hydrogen energy trading volume; For virtual power plants The volume of green certificate transactions.

6. The multi-market collaborative multi-virtual power plant bidding method from a low-carbon perspective as described in claim 5, characterized in that: In the third layer of the three-order nested game bidding model, the criteria for selecting a stable equilibrium solution are as follows: Simultaneously satisfy and The bidding strategy is a stable equilibrium solution; among which, Copying dynamic functions for bidding strategies The first derivative; Copying dynamic functions for bidding strategies The second derivative; The dynamic equation for replicating the bidding strategy is as follows: ； In the formula, For virtual power plants employing a cooperative bidding strategy Proportion; For virtual power plants using a non-cooperative bidding strategy Proportion; For virtual power plants Net revenue from five-market synergy using a bidding strategy; The average return of the bidding strategy for the virtual power plant group.

7. The multi-market collaborative multi-virtual power plant bidding method from a low-carbon perspective as described in claim 6, characterized in that: Step S4 specifically includes the following steps: S41. Based on the classification of virtual power plants in step S16, the following dynamic matching rule is set: Virtual power plants that sell electricity and carbon dioxide are preferentially matched with virtual power plants that purchase electricity and carbon dioxide, and the matching degree is calculated: ； In the formula, For virtual power plants With virtual power plants During the period The degree of matching; For virtual power plants During the period The supply and demand ratio of electricity carbon; S42, arrange the matching degrees in descending order, and screen the front The virtual power plant is taken as a completed virtual power plant, and the total amount of electricity-carbon transactions of the completed virtual power plant pair is calculated. Electricity transaction volume: ; Carbon quota trade volume: ; In the formula, for Time-based virtual power plant To virtual power plants Electricity sales volume; and Virtual power plants Electricity surplus and virtual power plants The power shortage; for Time-based virtual power plant To virtual power plants Carbon quota sales; and Virtual power plants Carbon allowance surplus and virtual power plants Carbon quota deficit; S43. Calculate the carbon trading price for virtual power plant pairs; Electricity trading prices: ; Carbon trading price: ; In the formula, For electricity trading prices; and Time periods The benchmark purchase price and sales price of electricity in the power grid; For carbon trading prices; and Time periods The benchmark buying and selling prices in the carbon allowance market; S44. Initial bidding scheme for the five markets of electricity-carbon-thermal-hydrogen-green certificates; ； ； ； ； ； In the formula, , , , , Virtual power plants exist The bidding volume for electricity, carbon, green certificates, heat, and hydrogen during the time period; For virtual power plants During the period The number of green certificates required to meet the renewable energy quota system requirements.

8. The multi-market collaborative multi-virtual power plant bidding method from a low-carbon perspective as described in claim 7, characterized in that: Step S5 specifically includes the following steps: S51. Construct a two-dimensional adaptive ADMM hybrid solution framework, coupling the game bidding model, joint risk cost, and five-market initial bidding scheme into a unified convex optimization problem, constructing the augmented Lagrangian function and completing the algorithm initialization: ； In the formula, For the augmented Lagrangian function under adaptive penalty factor; The total net revenue of five-market collaboration for multiple virtual power plant clusters during the scheduling period; For virtual power plants Total cost of low-carbon operation during the scheduling cycle; This is the transpose of the Lagrange multiplier vector; , All are constraint coefficient matrices; The constraint constant vector; For the set of decision variables in virtual power plant bidding; For the set of transaction control variables of multiple virtual power plant cluster operators; Iteration penalty factor; S52. Perform a two-dimensional adaptive penalty factor update combining residual balancing and bargaining weights, dynamically adjusting the iterative penalty factor: ； in, ； ； In the formula, For the first Sub-iteration fusion penalty factor; The penalty factor fusion weight coefficient; and These are adaptive penalty factors for the residual dimension and the bargaining dimension, respectively; The penalty factor scaling factor; For the first Sub-iteration fusion penalty factor; and The first The original residual and dual residual of the next iteration; and All are residual ratio thresholds; S53. Based on the adaptive penalty factor, perform ADMM distributed alternating iterative solution, updating decision variables and Lagrange multipliers until the iteration termination threshold condition is met, and output the ADMM hybrid solution result. , These are the original bid quantities for electricity, carbon, heat, hydrogen, and green certificates obtained through ADMM hybrid solution. S54. Construct a deep reinforcement learning DRL rolling correction model to dynamically correct the ADMM hybrid solution and output the optimal equilibrium bidding solution for the five markets.

9. The multi-market collaborative multi-virtual power plant bidding method from a low-carbon perspective as described in claim 8, characterized in that: In step S54, the bidding adjustment process is modeled as a Markov decision process, and state variables are set. ; Action variables reward function ;in, These are deep reinforcement learning models for virtual power plants. During the period The adjustment amounts for the bidding volumes of electricity, carbon, heat, hydrogen, and green certificates; For virtual power plants Net income from five market synergies; For virtual power plants Total operating costs; For virtual power plants The combined risk cost of triple uncertainty; The weighting factor for low-carbon benefits; Iterative optimization using the DQN value function: ； In the formula, The optimal action value function; Discount factor; For state variables Time period Next, from the next moment Among all feasible bid adjustment actions, select the one that allows the optimal action value function to be achieved. The action that reaches the maximum value, and the maximum overall reward value in that state; For expectation operators; The optimal bidding correction strategy is as follows: ； In the formula, , , , , These are the corrected optimal bid quantities for electricity, carbon, heat, hydrogen, and green certificates, respectively. After iterative convergence, the set of optimal equilibrium bidding solutions for the five markets is obtained. ;in, Virtual power plants To virtual power plants During the period The optimal electricity trading volume and carbon quota trading volume; Virtual power plants To virtual power plants During the period The optimal electricity trading price and carbon quota trading price.

10. The multi-market collaborative multi-virtual power plant bidding method from a low-carbon perspective as described in claim 9, characterized in that: Step S6 specifically includes the following steps: S61. Set up a set of bidding strategies for virtual power plants. , The cooperative bidding strategy for participating in S4's matching transactions. This is a non-cooperative bidding strategy; S62. Calculate the average revenue of a cluster of multiple virtual power plants: ； In the formula, For virtual power plants The average return of the bidding strategy; For virtual power plants Net synergistic gains from five markets using a cooperative bidding strategy ; For virtual power plants Net income from independent operation using a non-cooperative bidding strategy; S63. Substituting the average revenue of the bidding strategy obtained in S62 into the dynamic equation for replicating the bidding strategy, we obtain the dynamic evolution formula for the proportion of the virtual power plant cooperation strategy: ； In the formula, A dynamic function is used to replicate the bidding strategy for virtual power plants. S64. Using the Euler method to iteratively solve for the proportion of cooperative bidding strategies at different times. Continue until convergence, then output the steady-state policy ratio. ; S65, First, filter those that meet the requirements. The steady-state point, then the selected... Perform the first derivative, if it satisfies Then the If a point is determined to be an asymptotically stable equilibrium point, only the bidding strategy corresponding to this asymptotically stable equilibrium point will be considered as the candidate optimal strategy. S66. Based on the candidate optimal strategy, sequentially verify the energy balance constraints and operational constraints of power flow, carbon flow, heat flow, hydrogen flow, and green certificate flow, and eliminate invalid strategies that do not meet the constraints. S67. Calculate the economic accounting indicators and low-carbon accounting indicators that have passed the feasibility verification; The economic accounting indicators are expressed as follows: ； In the formula, The total operating cost of a multi-virtual power plant cluster; , , , , These include the equipment operation and maintenance costs of the virtual power plant, energy purchase costs, carbon trading costs, demand response compensation costs, and the combined risk costs of three uncertainties. Total revenue across five markets for multi-virtual power plant clusters; , , , , These represent the trading revenues of virtual power plants in the electricity market, carbon trading market, green certificate market, heat market, and hydrogen market, respectively. The expression for low-carbon accounting indicators is as follows: ； In the formula, , , These are carbon emission reduction rate, renewable energy integration rate, and green certificate completion rate, respectively. Carbon emissions for independent operation; The actual carbon emissions generated after implementing the candidate optimal strategy after eliminating invalid strategies; The total renewable energy power actually consumed by the multi-virtual power plant cluster during the dispatch cycle; The total renewable energy generation capacity of a multi-virtual power plant cluster during the scheduling period; The number of valid green certificates actually held by the virtual power plant during the dispatch cycle; The number of green certificates required for a virtual power plant; S68. Update step S1 based on the accounting results.

Images