An agent-based virtual power plant electricity-carbon-green certificate multi-market collaborative transaction decision method and system

Abstract

The application discloses a kind of virtual power plant electric-car-green certificate multi-market coordination transaction decision-making method and system based on agent, belong to virtual power plant operation optimization and energy market transaction technical field.The method first collects virtual power plant internal new energy generation, energy storage operation, load state and power market, carbon market and green certificate market transaction information, based on unified state object model constructs and includes the multi-market fusion state space of power state, carbon state and green attribute state, and establishes the coupling mapping relationship between power transaction value, carbon emission reduction value and green certificate environmental value;Through market prediction agent, the change trend of electricity price, carbon price and green certificate price is respectively predicted;Further construct the multi-market joint optimization model with power transaction income, carbon asset income and green certificate environmental value income comprehensive optimization as target, and consider operating constraint, carbon emission constraint and market risk constraint;Through transaction decision-making agent, power transaction strategy, carbon transaction strategy and green certificate transaction strategy are generated, and transaction execution is realized through standardized interface;At the same time, according to actual transaction result, model parameter is rolled up and updated.The application realizes the coordinated release of virtual power plant electric energy value, carbon asset value and green environmental value, and improves the intelligent decision-making level of virtual power plant participating in multi-market transaction.

Description

Technical Field

[0001] This invention belongs to the fields of intelligent optimization scheduling of power systems, virtual power plant operation and management, and artificial intelligence decision-making technology. Specifically, it relates to a multi-market collaborative trading decision-making method and system for virtual power plants based on intelligent agents, which is applicable to virtual power plant operation platforms, integrated energy service systems, zero-carbon park energy management systems, new energy aggregation operation systems, and power market trading auxiliary decision-making systems. Background Technology

[0002] With the large-scale integration of new energy sources, the continuous advancement of power market reforms, and the increasing requirements for carbon emission control, virtual power plants, as a new form of energy organization that can aggregate distributed new energy sources, energy storage, and adjustable loads, are gradually becoming important participants in power system operation and market transactions. Virtual power plants, by aggregating various internal resources, can participate in electricity markets, ancillary services markets, and demand response markets. Furthermore, with the development of carbon markets and green electricity certificate markets, virtual power plants also have the potential to participate in carbon emission trading and green certificate trading. However, current virtual power plant operations primarily revolve around a single electricity market, typically focusing on reducing electricity purchase costs or increasing market revenue, without fully considering the coupling relationship between electricity market revenue, carbon market revenue, and the environmental value of green certificates. Currently, the methods for making trading decisions in the virtual power plant market have the following main shortcomings: (1) Unified modeling of multi-market values ​​has not been achieved. Existing methods typically establish separate models for electricity trading, carbon trading, or green certificate trading: the electricity market focuses on electricity prices, the carbon market focuses on carbon emission costs, and the green certificate market focuses on the value of green attributes. The lack of a unified value evaluation mechanism among different markets makes it difficult for virtual power plants to maximize overall revenue. (2) Market price fluctuations affect the accuracy of trading decisions. Electricity prices, carbon prices, and green certificate prices all have high uncertainty. Existing technologies usually use historical prices or single forecast results to make trading decisions, which makes it difficult to describe the risks of future market changes and easily leads to fluctuations in the returns of trading strategies. (3) Lack of a coordinated optimization mechanism for electricity, carbon and green certificates. Existing virtual power plant trading strategies usually only consider the revenue from electricity trading, while ignoring the value of carbon emission reduction, the environmental value of green certificates and the opportunities for synergistic value of multiple markets. Therefore, they cannot fully explore the low-carbon operation value of virtual power plants. (4) Manually formulating trading strategies is inefficient. Currently, when virtual power plants participate in multi-market transactions, trading personnel need to manually formulate trading plans based on market prices, resource status, and policy requirements. Faced with complex scenarios involving multiple time scales, multiple market constraints, and multiple resource coordination, manual decision-making is difficult to meet the needs of real-time optimization. In recent years, the rapid development of artificial intelligence, big data analytics, and intelligent agent technology has provided new technological pathways for multi-market collaborative trading of virtual power plants. Existing virtual power plant resource sensing and carbon optimization technologies provide a certain foundation for multi-market trading, but there is still a lack of technical solutions for joint decision-making across multiple markets, including electricity, carbon, and green certificates. Therefore, there is an urgent need for a method and system that can integrate information from the electricity market, carbon market, and green certificate market, and realize multi-market collaborative trading of virtual power plants through intelligent forecasting, value assessment, and optimized decision-making. Summary of the Invention

[0003] Purpose of the invention The purpose of this invention is to provide a collaborative trading decision-making method and system for virtual power plants across multiple markets (electricity, carbon, and green certificates) based on intelligent agents. This addresses the problems in existing technologies, such as the independent operation of the electricity market, carbon market, and green certificate market, the inability to uniformly evaluate their value, high trading risks due to market price fluctuations, and the reliance on manual formulation and insufficient intelligence in virtual power plant multi-market trading strategies. By constructing a joint decision-making mechanism across these markets, the invention achieves synergistic optimization of the economic benefits, carbon reduction value, and green attribute value of virtual power plants.

[0004] Technical solution To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides a multi-market collaborative trading decision-making method for virtual power plants based on intelligent agents, comprising the following steps:

[0005] Step S1: Obtain virtual power plant operation status data and multi-market transaction data, and construct a multi-market integrated status space. Acquire internal resource operation data and external market environment data for the virtual power plant. Internal resource operation data includes: distributed renewable energy output data, energy storage system operation status data, flexible load operation data, controllable load regulation capacity data, and virtual power plant tradable capacity data. Multi-market transaction data includes: electricity market transaction prices, electricity market clearing results, carbon emission rights trading prices, carbon emission quota data, green certificate trading prices, and green certificate supply and demand data.

[0006] Furthermore, the acquired data undergoes time synchronization, data cleaning, anomaly detection, format conversion, and data fusion to form a unified state space for multi-market collaborative decision-making. This electricity-carbon-green certificate multi-market integrated state space includes an electricity state space, a carbon state space, and a green certificate state space, wherein: (1) The power state space is used to describe the power production, consumption, and trading states of a virtual power plant, and is represented as: X e={P_g, P_l, SOC, P_trade}, where P_g represents the output of new energy sources, P_l represents the load power, SOC represents the state of charge of energy storage, and P_trade represents the power traded.

[0007] (2) The carbon state space is used to describe the carbon emissions, carbon quotas and carbon asset trading status, and is represented as: X_c = {E,C_q, C_p}, where E represents carbon emissions, C_q represents carbon quotas and C_p represents carbon prices.

[0008] (3) The green certificate state space is used to describe the green attributes of new energy, the quantity of green certificates and the trading status of green certificates, and is represented as: X_g = {Q_g, P_g^c}, where Q_g represents the quantity of green certificates and P_g^c represents the price of green certificates.

[0009] Step S2: Establish a multi-market value coupling model for electricity, carbon, and green certificates.

[0010] Establish a coupled mapping relationship between the value of electricity trading, the value of carbon emission reduction, and the environmental value of green certificates. This coupled mapping relationship includes: (1) When the power generation of new energy increases, the revenue from electricity sales in the electricity market increases, while the cost of carbon emissions decreases, and the environmental value of green certificates is released simultaneously. (2) When the energy storage charging and discharging strategy is adjusted, the revenue from electricity purchase and sale in the electricity market, the cost of carbon quotas and the revenue from holding green certificates are coupled in time. (3) The three types of market values ​​are related in a unified state space. Changes in the decision of any one type of market will affect the profit space of the other two types of markets.

[0011] A comprehensive value model is established based on the economic benefits and environmental value generated by the virtual power plant's participation in different markets. The comprehensive benefit objective is expressed as: R = R e + R_c + R_g − C Where: R represents the comprehensive revenue of the virtual power plant, R e R_c represents revenue from electricity market transactions, R_g represents revenue from green certificate transactions, and C represents operating costs.

[0012] Furthermore, electricity market revenue is calculated based on the volume of electricity traded and the real-time electricity price: R e = P e × Q e , where P e For electricity price, Q eThe trading volume is the electricity. Carbon market revenue is calculated based on the carbon emission reductions of the virtual power plant and the carbon price: R_c = ΔE_c × P_c, where ΔE_c is the carbon emission reduction and P_c is the carbon trading price. Green certificate revenue is calculated based on the number of green certificates traded and the green certificate price: R_g = N_g × P_g, where N_g is the number of green certificates and P_g is the green certificate price. This model achieves a unified quantification of the value of electricity, carbon assets, and the green environment.

[0013] Step S3: Predict future market trends based on a multi-agent prediction model.

[0014] Building market forecasting agents includes: (1) Electricity price prediction agent: Input historical electricity prices, load prediction results, new energy output prediction results, and market supply and demand information, and output the electricity price prediction sequence within the future time window.

[0015] (2) Carbon price prediction agent: Input historical carbon price, energy price, emission data, and policy impact parameters, and output carbon price prediction sequence.

[0016] (3) Green certificate price prediction agent: Input new energy power generation, green certificate supply, enterprise green consumption demand, and market transaction data, and output green certificate price prediction sequence.

[0017] Furthermore, each predictive agent employs an artificial intelligence model, including but not limited to one or more of the following: machine learning model, deep learning model, Transformer model, temporal prediction model, reinforcement learning model, or large language model.

[0018] Step S4: Construct a joint optimization trading model for electricity, carbon, and green certificates.

[0019] Based on the forecast results, a multi-market joint optimization model is constructed. The optimization objective is to maximize the overall return and reflect the integration of the three market values. max F = α·R e + β·R_c + γ·R_g − C − λ·CVaR Where: α is the electricity revenue weight, β is the carbon revenue weight, γ is the green certificate value weight, λ is the risk weight, and CVaR is the conditional risk value.

[0020] Optimization constraints include: (1) Virtual power plant operation constraints: power balance constraints, energy storage SOC constraints, equipment capacity constraints, and load regulation constraints. (2) Electricity market trading constraints: maximum trading capacity, contract constraints, and output constraints. (3) Carbon market constraints: carbon emission quota constraints, carbon emission intensity constraints, and carbon trading capacity constraints. (4) Green certificate market constraints: Green certificate holding constraints, green certificate trading volume constraints, and green consumption ratio constraints.

[0021] Step S5: Generate multi-market trading strategies through the trading decision-making agent.

[0022] The trading decision-making intelligent agent includes a state awareness module, a strategy generation module, a risk assessment module, and a feedback learning module. Based on market forecast results, virtual power plant resource status, a comprehensive revenue model, and risk assessment results, it generates trading strategies. The trading strategies include: Electricity market strategy: electricity purchase, electricity sales, and trading hours.

[0023] Carbon market strategy: carbon allowance purchase volume, carbon allowance sale volume, and trading timing.

[0024] Green certificate market strategy: green certificate purchase volume, green certificate sale volume, and holding strategy.

[0025] Step S6: Transaction execution and feedback optimization.

[0026] The generated trading strategies are sent to power trading platforms, carbon trading platforms, and green certificate trading platforms. Simultaneously, actual trading results, market price changes, and resource execution status are collected and fed back to the optimization model for rolling updates to the trading strategies for the next cycle. This forms a closed-loop operating mechanism of "prediction—optimization—trading—feedback."

[0027] Secondly, the present invention provides a virtual power plant electricity-carbon-green certificate multi-market collaborative trading decision-making system based on intelligent agents, comprising: The multi-market data acquisition module is used to acquire electricity market data, carbon market data, green certificate market data, and virtual power plant resource status data. The multi-market value modeling module is used to build electricity revenue models, carbon asset revenue models, and green certificate value models. The market forecasting agent module includes an electricity price forecasting agent, a carbon price forecasting agent, and a green certificate price forecasting agent, which are used to predict future market trends. The joint optimization decision module is used to solve for the optimal trading strategy based on the comprehensive return objective and risk constraints. The trading decision-making intelligent agent module is used to generate electricity trading strategies, carbon trading strategies, and green certificate trading strategies. The execution feedback module is used to receive transaction results and update and optimize the model.

[0028] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method.

[0029] Beneficial effects Compared with the prior art, the present invention has the following beneficial effects: (1) Achieve unified evaluation and coupled modeling of the three markets. This invention unifies the modeling of electricity market revenue, carbon market revenue and green certificate environmental value, and establishes a value coupling mapping relationship between the three markets, realizing the transformation from traditional single-market trading to multi-market collaborative trading. (2) Improve the operating revenue of virtual power plants. Improve the overall revenue level of virtual power plants by jointly optimizing electricity trading, carbon asset operation and green certificate trading strategies. (3) Reduce market trading risks. By introducing the CVaR risk assessment mechanism, the impact of market price fluctuations is fully considered, thereby improving the robustness of trading strategies. (4) Achieve intelligent trading decisions. Through multi-agent prediction and trading decision-making agents that include state perception, strategy generation, risk assessment, and feedback learning, automated strategy generation can be achieved in complex market environments. (5) Forming a complete business closed loop. This invention can collaborate with a unified state object model, data platform, and multi-agent scheduling system to achieve a complete closed loop of "resource perception → state modeling → price prediction → transaction decision → market execution → revenue feedback". Attached Figure Description

[0030] Figure 1 This is the overall flowchart of the multi-market collaborative trading of electricity, carbon, and green certificates in this invention. Figure 2 This is a diagram illustrating the state space construction architecture of the fusion of electricity, carbon, and green certificates in this invention. Figure 3 This is a diagram of the multi-market joint transaction optimization decision-making architecture of the present invention. Figure 4 This is a flowchart illustrating the closed-loop process for transaction execution and feedback optimization in this invention. Detailed Implementation

[0031] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. It should be noted that these embodiments are only used to illustrate the technical solutions of the present invention and do not limit the scope of protection of the present invention.

[0032] Example 1: Decision-making process for multi-market collaborative trading of electricity, carbon, and green certificates like Figure 1 As shown, this embodiment provides a multi-market collaborative trading decision-making method for virtual power plants based on intelligent agents, encompassing electricity, carbon, and green certificate markets. This embodiment uses the joint trading of a virtual power plant in a new energy park within the electricity, carbon, and green certificate markets as an application scenario. The virtual power plant includes: photovoltaic power generation units, wind power generation units, energy storage units, flexible load resources, and controllable energy equipment. By aggregating and managing these distributed resources, the overall market trading capabilities of the virtual power plant are enhanced.

[0033] I. Multi-market operation data collection First, the virtual power plant's operating status data and external market transaction data are acquired through the data acquisition module. The virtual power plant operation status data includes: (1) new energy power generation; (2) energy storage state of charge (SOC); (3) energy storage charging and discharging power; (4) flexible load operation status; and (5) adjustable capacity information. External market data includes: (1) day-ahead market electricity price; (2) real-time market electricity price; (3) carbon emission trading price; (4) green certificate trading price; (5) carbon emission quota information; and (6) green certificate supply and demand information. The data interface module connects to the power trading platform, carbon emission trading platform, and green certificate trading platform respectively to achieve multi-market data acquisition.

[0034] II. Construction of the State Space of Electricity-Carbon-Green Certificates The acquired data is input into the state fusion module, which constructs a multi-market fusion state space for electricity, carbon, and green certificates based on a unified state object model. The unified state object model provides a standardized data description framework, ensuring the consistency and traceability of multi-source heterogeneous data during the state fusion process. The fusion state space includes: (1) Power state space, used to describe generation state, power consumption state, energy storage state, and trading state. The power state space is represented as: X e = {P_g, P_l, SOC, P_trade}, where P_g represents the output of new energy sources, P_l represents the load power, SOC represents the state of charge of energy storage, and P_trade represents the power traded. (2) Carbon state space, used to describe carbon emissions, carbon emission reductions, and carbon quota status. The carbon state space is represented as: X_c = {E, C_q, C_p}, where E represents carbon emissions, C_q represents carbon quota, and C_p represents carbon price. (3) Green certificate state space, used to describe the green attributes of new energy, the quantity of green certificates, and the trading status of green certificates. The green certificate state space is represented as: X_g = {Q_g, P_g^c}, where Q_g represents the quantity of green certificates and P_g^c represents the price of green certificates. Through the above method, a unified description of the value of electricity, carbon value, and green environment value is achieved.

[0035] III. Establishment of a Value Coupling Model for Electricity, Carbon, and Green Certificates Establish a coupled mapping relationship between the value of electricity trading, the value of carbon emission reduction, and the environmental value of green certificates. Taking a scenario of large-scale renewable energy generation as an example: when photovoltaic power generation increases, the revenue from electricity sales in the electricity market (α·R) increases. eAs carbon emissions increase, carbon emission costs decrease (β·R_c increases) due to the substitution of thermal power, and the environmental value of green certificates (γ·R_g) is released simultaneously. There is a positive synergy among the returns of the three types of markets. Taking changes in energy storage charging and discharging strategies as an example: when energy storage is charging, the cost of purchasing electricity in the electricity market increases, but the cost of carbon allowances decreases. The returns on holding green certificates may vary depending on the charging period, indicating a temporal coupling among the three types of markets.

[0036] The objective function for overall return is: max F = α·R e + β·R_c + γ·R_g − C − λ·CVaR.

[0037] IV. Multi-market price forecast The fused state space is input into the market forecasting agent. The market forecasting agent includes an electricity price forecasting agent, a carbon price forecasting agent, and a green certificate price forecasting agent.

[0038] The electricity price forecasting agent generates a future time window electricity price forecast sequence based on historical electricity price data, load forecast results, and renewable energy output forecast results, using artificial intelligence models (including but not limited to machine learning models, deep learning models, time series forecasting models, and large language models). For example, it can predict the electricity price P for the next 24 hours. e (t), t=1,2,...,24. The carbon price prediction agent generates a carbon price prediction sequence P_c(t) based on historical carbon prices, energy prices, and emission levels. The green certificate price prediction agent generates a green certificate price prediction sequence P_g(t) based on new energy power generation, green certificate supply and demand, and green consumption demand.

[0039] V. Construction of a Joint Optimization Model for Electricity-Carbon-Green Certificates Based on the prediction results, a multi-market joint trading optimization model is constructed. This embodiment aims to maximize the overall revenue of the virtual power plant. The objective function is: Max F = α·R e + β·R_c + γ·R_g − C − λ·CVaR Where: R e R_c represents the revenue from electricity market transactions, R_g represents the revenue from carbon market transactions, C represents the operating cost, CVaR represents the market risk cost, α is the weight of electricity revenue, β is the weight of carbon revenue, γ is the weight of green certificate value, and λ is the risk weight.

[0040] Furthermore, the revenue from electricity trading R e = Σ P e (t)·Q eCarbon trading revenue R_c = Σ P_c(t)·Q_c(t), green certificate trading revenue R_g = Σ P_g(t)·Q_g(t).

[0041] VI. Optimize Constraint Settings The joint optimization model includes the following constraints: (1) Power balance constraint: The output of new energy sources in the virtual power plant + the discharge of energy storage + the purchase of external power = load demand + energy storage charging. (2) Energy storage operation constraints: The energy storage SOC satisfies SOC_min ≤ SOC(t) ≤ SOC_max. (3) New energy operation constraints: The actual output of new energy satisfies 0 ≤ P_new(t) ≤ P_max(t). (4) Carbon emission constraints: The carbon emissions of the virtual power plant satisfy E(t) ≤ E_quota. (5) Green certificate trading constraints: The number of green certificates traded satisfies 0 ≤ Q_g(t) ≤ Q_green.

[0042] VII. Trading Decision-Making Agent Generates Trading Strategies The joint optimization model is input into the trading decision-making agent. This agent employs a multi-agent collaborative mechanism, including a state-aware agent, a strategy-generating agent, a risk-assessing agent, and a feedback-learning agent. Wherein: State-aware agents are used to perceive the resource status of virtual power plants, changes in the market environment, and the status of transaction execution in real time. The strategy generation agent is used to generate electricity trading strategies, carbon trading strategies, and green certificate trading strategies based on state perception results and optimization model output. The risk assessment agent is used to assess the market risk of the current trading strategy and calculate risk indicators such as CVaR; The feedback learning agent is used to receive transaction execution results and update and optimize model parameters and strategy generation rules. Each intelligent agent achieves information sharing and joint decision-making through a collaborative mechanism, generating the following based on electricity price forecasts, carbon price forecasts, green certificate price forecasts, and virtual power plant resource status: (1) Electricity trading strategy: including trading time, trading volume, and buying and selling direction. (2) Carbon trading strategy: including carbon quota purchase volume, carbon quota sale volume, and optimal trading timing. (3) Green certificate trading strategy: including green certificate holdings, green certificate purchases, and green certificate sales. VIII. Optimization of Transaction Execution and Feedback After the trading strategy is generated, it is output to the power trading platform, carbon trading platform, and green certificate trading platform through standardized interfaces. After the transaction is completed, the trading execution module collects the actual trading results, including the actual transaction price, the transaction volume, changes in carbon assets, and changes in green certificate assets, and feeds the execution results back to the status fusion module. Through a rolling update method, the system can optimize decisions for the next trading cycle.

[0043] Example 2: Multi-market collaborative trading in the context of large-scale new energy development When the virtual power plant detects a rapid increase in renewable energy output, the state fusion module identifies a surplus of renewable energy. In the electricity-carbon-green certificate value coupling model, the increase in renewable energy generation triggers a three-market linkage effect: electricity market sales revenue (α·R) e As carbon emission costs (β·R_c) increase, the environmental value of green certificates (γ·R_g) decreases, and the environmental value of green certificates is released simultaneously. The transaction decision-making agent executes: (1) increasing the proportion of green certificate sales; (2) reducing the risk of new energy curtailment; (3) adjusting the energy storage charging plan; and (4) optimizing the electricity market trading strategy. Ultimately, this achieves increased new energy consumption, increased electricity trading revenue, and release of green asset value.

[0044] Example 3: Trading Optimization under the Scenario of Rising Carbon Prices When carbon market prices continue to rise, the carbon price prediction agent anticipates the trend of carbon cost changes in advance. Rising carbon prices lead to increased carbon emission costs, which in turn affect electricity purchase and sale strategies and green certificate holding decisions in the electricity market. The trading decision agent executes the following: (1) purchasing carbon allowances in advance; (2) increasing the proportion of low-carbon energy operation; (3) adjusting energy storage operation strategies; and (4) reducing electricity purchases during high-carbon periods. This reduces the carbon trading risk of virtual power plants.

[0045] Example 4: Multi-market collaborative optimization scenario When the virtual power plant detects a surge in photovoltaic power generation, a drop in electricity prices, and a low carbon price, the trading decision-making agent executes a multi-market arbitrage strategy: (1) increasing energy storage charging capacity to utilize low-priced electricity for charging; (2) increasing the proportion of green certificate sales to realize the value of the green environment; (3) purchasing carbon allowances in advance to lock in low-carbon costs; and (4) discharging energy storage during periods of rising electricity prices to maximize overall returns. In this scenario, electricity revenue, carbon revenue, and green certificate revenue are released simultaneously, achieving coordinated arbitrage across the three markets.

[0046] Example 5: Collaborative Application with Energy Agent OS System In this embodiment, the present invention can be combined with existing virtual power plant resource digital profiling modules, electricity-carbon collaborative optimization scheduling modules, and multi-agent collaborative decision-making modules. Specifically, the resource digital profiling module provides resource adjustability data, the prediction module provides market price prediction data, and the agent decision-making module generates multi-market trading strategies. This ultimately forms a closed loop for intelligent virtual power plant trading: "resource perception → state modeling → price prediction → trading decision → market execution → revenue feedback."

[0047] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. For those skilled in the art, various modifications and variations can be made to the present invention without departing from its spirit and principles. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-market collaborative trading decision-making method for virtual power plants based on intelligent agents, encompassing electricity, carbon, and green certificates, characterized in that... Includes the following steps: S1: Acquire internal resource operation data and external market transaction data of virtual power plants, and construct a multi-market integrated state space of electricity, carbon and green certificates; S2: Based on the multi-market fusion state space, the market prediction agent predicts the electricity market price, carbon market price, and green certificate market price respectively, generating a multi-market price prediction sequence; S3: Based on the multi-market price forecast sequence and the operating status of the virtual power plant, construct an optimized model for the joint trading of electricity, carbon, and green certificates; S4: Generate a collaborative trading strategy for virtual power plants to participate in the electricity market, carbon market, and green certificate market based on the aforementioned joint trading optimization model; S5: Output the collaborative trading strategy to the corresponding trading execution platform, and update the multi-market fusion state space according to the trading execution results to achieve rolling optimization decision-making.

2. The method according to claim 1, characterized in that, The virtual power plant's internal resource operation data includes new energy power generation, energy storage status of charge, energy storage charging and discharging power, flexible load operation status, and adjustable capacity information; the external market transaction data includes electricity market transaction prices, carbon emission transaction prices, green certificate transaction prices, carbon emission quota information, and green certificate supply and demand information.

3. The method according to claim 1, characterized in that, The multi-market integrated state space of electricity, carbon, and green certificates is established based on a unified state object model. The multi-market integrated state space includes an electricity state space, a carbon state space, and a green certificate state space. The electricity state space is used to describe the electricity production, consumption, and trading status of virtual power plants; the carbon state space is used to describe the carbon emissions, carbon quotas, and carbon asset trading status; and the green certificate state space is used to describe the green attributes of new energy sources, the number of green certificates, and the trading status of green certificates.

4. The method according to claim 1, characterized in that, The multi-market price forecast includes electricity price forecast, carbon price forecast, and green certificate price forecast; the forecasting method adopts an artificial intelligence model, including but not limited to one or more of machine learning models, deep learning models, time series forecasting models, reinforcement learning models, or large language models.

5. The method according to claim 1, characterized in that, The joint trading optimization model aims to maximize the comprehensive benefits of the virtual power plant; the comprehensive benefits are the weighted sum of electricity trading revenue, carbon asset trading revenue, and green certificate environmental value revenue, minus operating costs, energy storage degradation costs, and market risk costs.

6. The method according to claim 1, characterized in that, The joint trading optimization model includes one or more of the following constraints: virtual power plant power balance constraint, new energy output constraint, energy storage state of charge constraint, energy storage charging and discharging power constraint, carbon emission quota constraint (i.e., the carbon emissions of virtual power plants do not exceed the quota), green certificate trading quantity constraint (i.e., the trading volume of green certificates does not exceed the holding volume), and market trading capacity constraint (i.e., the trading volume of each market does not exceed the upper limit of the corresponding market capacity).

7. The method according to claim 1, characterized in that, The coordinated trading strategies include: electricity market trading strategies, carbon market trading strategies, and green certificate market trading strategies; wherein, the electricity market trading strategy includes trading time, trading direction, and trading volume; the carbon market trading strategy includes carbon quota purchase volume, sales volume, and trading timing; and the green certificate market trading strategy includes green certificate holding volume, purchase volume, and sales volume.

8. The method according to claim 1, characterized in that, The collaborative trading strategy is generated by a trading decision-making agent; the trading decision-making agent adopts a multi-agent collaborative mechanism, including a state-aware agent, a strategy-generating agent, a risk-assessing agent, and a feedback-learning agent. The trading strategy is dynamically adjusted based on the virtual power plant resource status, multi-market price forecasts, historical trading results, and profit and risk indicators.

9. A multi-market collaborative trading decision-making system for virtual power plants based on intelligent agents, comprising electricity, carbon, and green certificates, for implementing the method described in any one of claims 1 to 8, characterized in that, include: The data acquisition module is used to acquire virtual power plant operation data as well as electricity market, carbon market and green certificate market transaction data; The state fusion module is used to construct a multi-market integrated state space for electricity, carbon, and green certificates based on a unified state object model; The market forecasting module is used to generate forecasts of electricity prices, carbon prices, and green certificate prices through a market forecasting agent. The joint optimization module is used to construct and solve multi-market joint trading optimization models; The trading decision module is used to generate multi-market collaborative trading strategies through a trading decision-making intelligent agent; The trade execution module is used to output the trading strategy to the corresponding trading platform; The feedback optimization module is used to update and optimize the model based on the transaction execution results.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method described in any one of claims 1 to 8.

Images