Virtual power plant scale trading method for e-car-greencertificate market

Abstract

This invention discloses a virtual power plant-scale trading method for the electricity-carbon-green certificate market, belonging to the field of electricity market trading. The method includes the following steps: S1, obtaining a standardized time-series dataset and a set of equipment constraint parameters; S2, using a VMD-Bi-MGLSTM multi-branch prediction network to complete ultra-short-term predictions and generate a set of typical uncertainty scenarios; S3, constructing a coupled pricing and trading model for the electricity-carbon-green certificate three markets; S4, establishing a multi-objective collaborative optimization model for virtual power plants; S5, performing hierarchical distributed solution on the multi-objective collaborative optimization model to output the optimal trading and operating strategies; S6, executing transactions and performing closed-loop verification via blockchain; S7, dynamic updating. By employing the above-mentioned virtual power plant-scale trading method for the electricity-carbon-green certificate market, the method achieves optimal synergy among multiple objectives of virtual power plant economics, low carbon emissions, and grid integration, completely solving the problems of fragmented trading, coarse predictions, lack of pricing, and insufficient security and reliability in traditional trading methods.

Description

Technical Field

[0001] This invention relates to the field of electricity market trading technology, and in particular to a virtual power plant-scale trading method for the electricity-carbon-green certificate market. Background Technology

[0002] Virtual power plants (VPPs), as the core carrier aggregating various resource types such as distributed photovoltaic, wind power, energy storage, flexible loads, carbon capture and storage (CCS), and power-to-gas (P2G), are key players participating in multi-market transactions under the new power system. Currently, VPP transactions suffer from problems such as fragmentation of the three markets, rudimentary handling of uncertainty, lack of pricing mechanisms, insufficient privacy protection, and weak low-carbon collaboration. A detailed analysis follows: 1. The electricity, carbon, and green certificate markets operate independently, without forming a price linkage and quota coordination mechanism, making it impossible to achieve a balance between economic and low-carbon goals; 2. The uncertainty of wind and solar power output and load is not accurately predicted and has high computational complexity when relying solely on single-scenario methods or robust optimization. 3. The pricing of multi-VPP transactions relies on a single model of master-slave game or Nash negotiation, without taking into account both the dominant power and the fairness of the returns. Summary of the Invention

[0003] The purpose of this invention is to provide a virtual power plant-scale trading method for the electricity-carbon-green certificate market, thereby solving the aforementioned technical problems.

[0004] To achieve the above objectives, this invention provides a virtual power plant-scale trading method for the electricity-carbon-green certificate market, comprising the following steps: S1, collecting multi-source operation data, market data, power grid operation data, and meteorological data of virtual power plants by category, and obtaining a standardized time-series dataset and a set of equipment constraint parameters after preprocessing; S2, based on the standardized time-series dataset output by S1, using a VMD-Bi-MGLSTM multi-branch prediction network to complete ultra-short-term predictions and generate a set of typical uncertainty scenarios; S3, based on the output results of S2, constructing a three-market coupled pricing and trading model for electricity, carbon, and green certificates; S4, combining the three-market coupled pricing and trading model constructed in S3, establishing a multi-objective collaborative optimization model for virtual power plants; S5, performing hierarchical distributed solution on the multi-objective collaborative optimization model output by S4, and outputting the optimal trading and operation strategies; S6, executing transactions and performing closed-loop verification through blockchain; S7, dynamically updating.

[0005] Therefore, the virtual power plant-scale trading method for the electricity-carbon-green certificate market described above has the following beneficial effects: 1. More efficient market coordination: Construct a price elasticity linkage matrix for electricity, carbon, and green certificates, break down the barriers to independent operation of the three markets, achieve price and quota coordination, and unify economic and low-carbon goals; 2. More accurate predictions and scenarios: The VMD-Bi-MGLSTM multi-branch network is used to improve the accuracy of ultra-short-term predictions, and the improved particle swarm optimization K-means scenario reduction is combined to efficiently characterize the uncertainty of wind and solar load. 3. Fairer pricing and returns: A hybrid pricing model combining master-slave game theory and Nash negotiation balances the dominance of virtual power plants with the fairness of returns for various distributed energy entities; 4. Optimized solution for greater efficiency and reliability: The three-layer distributed solution architecture balances robustness, global optimization, and computational efficiency, meeting the real-time and security requirements of engineering projects. 5. Comprehensive improvement in operational indicators: Simultaneously achieving the lowest operating costs, the least carbon emissions, and the highest renewable energy consumption rate, adapting to the low-carbon requirements of the new power system.

[0006] 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

[0007] Figure 1 This is a flowchart of the virtual power plant-scale trading method for the electricity-carbon-green certificate market described in this invention. Detailed Implementation

[0008] 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.

[0009] 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.

[0010] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0011] like Figure 1As shown, the virtual power plant-scale trading method for the electricity-carbon-green certificate market includes the following steps: S1. Collect multi-source operation data, market data, grid operation data, and meteorological data of virtual power plants by category, and obtain standardized time-series datasets and equipment constraint parameter sets after preprocessing; S2. Based on the standardized time-series dataset output by S1, use the VMD-Bi-MGLSTM multi-branch prediction network to complete ultra-short-term predictions and generate a set of typical uncertainty scenarios; S3. Based on the output results of S2, construct a three-market coupled pricing and trading model for electricity, carbon, and green certificates; S4. Combine the three-market coupled pricing and trading model constructed in S3 to establish a multi-objective collaborative optimization model for virtual power plants; S5. Perform hierarchical distributed solution on the multi-objective collaborative optimization model output by S4 to output the optimal trading and operation strategies; S6. Execute transactions and perform closed-loop verification through blockchain; S7. Dynamically update.

[0012] The virtual power plant multi-source operation data mentioned in step S1 includes distributed wind power output, distributed photovoltaic power output, energy storage state of charge (SOC), carbon capture equipment operation parameters, power-to-gas equipment operation parameters, carbon emissions, and load data; among which, the load data includes residential flexible adjustable load, commercial flexible adjustable load, and industrial basic rigid load. Market data includes time-of-use electricity pricing in the electricity market, carbon market allowances and prices, green certificate market allowances and prices, and trading rules; Power grid operation data includes node voltage, line capacity, network loss parameters, and power balance constraints; Meteorological data includes wind speed, wind direction, light intensity, temperature, and humidity.

[0013] The preprocessing described in step S1 involves sequentially performing outlier removal, linear interpolation to fill missing values, and Min-Max normalization on the subsets of virtual power plant multi-source operation data, market data, power grid operation data, and meteorological data.

[0014] Step S2 specifically includes the following steps: S21. Perform mode decomposition on the standardized time series dataset output by S1 to obtain stationary intrinsic mode components; S211, Read wind power output sequence Photovoltaic power output sequence and load power sequence Variational mode decomposition is performed on each sequence to construct and solve constrained variational problems, thereby decomposing the nonstationary original sequence into... a stable intrinsic mode component and residual components: ; Constraints: ; In the formula, The first decomposition obtained A set of intrinsic modal components; For the first The set of center frequencies corresponding to each modal component; Indicates time The partial derivatives; The Dirac distribution function; Represents the convolution integral; The imaginary unit; For the first The center frequency corresponding to each modal component; The current standardized original sequence to be decomposed is taken in sequence. , , ; S212. Solve the constrained variational problem of S211 iteratively using the alternating direction multiplier method, and output the corresponding sequences. One set of stationary intrinsic mode components and one set of residual components; S22, Based on the output of step S21 A bidirectional Bi-MGLSTM network is constructed and trained using a set of stationary intrinsic mode components and a set of residual components. S221. Divide the stationary intrinsic mode components output by S21 into training, validation, and test sets in an 8:1:1 ratio, and set the input sequence length. ; S222. An improved bidirectional multi-gated LSTM unit is constructed by adding a double forget gate and a double input gate to the standard LSTM. The improved bidirectional multi-gated LSTM unit includes a forward multi-gated LSTM subnetwork, a backward multi-gated LSTM subnetwork, and a bidirectional feature fusion layer. Both the multi-gated LSTM subnetwork and the backward multi-gated LSTM subnetwork include a first forget gate, a second forget gate, a comprehensive forget gate, a first input gate, a second input gate, a comprehensive input gate, a candidate cell state, a cell state, an output gate, and a hidden state, arranged sequentially. The gating update calculation formula is as follows: ; In the formula, and These are the outputs of the first forget gate and the second forget gate, respectively. For the comprehensive forget gate output; and These are the outputs of the first input gate and the second input gate, respectively. For integrated input gate output; Output gate output; Candidate cell state; In cellular state; The current hidden layer state; is; , , , , , These are the weights of the first forget gate, the second forget gate, the first input gate, the second input gate, the output gate, and the candidate cell state, respectively. , , , , , For each gated bias term; This represents the hidden layer state from the previous time step. The model input at the current time is... or or ; Use the Sigmoid activation function; The activation function is a linear correction unit. It is the hyperbolic tangent activation function; S223, based on the improved bidirectional multi-gated LSTM unit of S222, constructs a bidirectional Bi-MGLSTM network, extracts temporal features from the forward and backward directions respectively, and obtains the final hidden layer output: ; ; ; In the formula, and These are the outputs of the forward MG-LSTM hidden layer and the backward MG-LSTM hidden layer at the current time, respectively. This is the final hidden layer output after bidirectional feature fusion, and it is considered as the fused feature. For forward multi-gated LSTM sub-cell functions; For backward multi-gated LSTM sub-unit functions; For forward multi-gated LSTM subnetworks in The hidden layer state at any given time; For backward multi-gated LSTM subnetworks in The hidden layer state at any given time; For feature fusion weights; For feature fusion bias term; S224, The fusion feature output by S223 By connecting to the fully connected layer, the predicted values ​​of the single-mode components are obtained. ; S225. The prediction results of all intrinsic mode components are linearly superimposed to obtain the ultra-short-term prediction values ​​of wind power, photovoltaic power, and load: ; In the formula, For the first Each sample corresponds to The ultra-short-term forecast values ​​of wind power, solar power, or load at any given moment; These are the predicted values ​​for the residual components; S226. Using the quantile loss function as the training objective, minimize the prediction error, and the formula for calculating the loss function is: ; in, ; ; In the formula, The quantile loss function value; The number of samples in the training set; The piecewise function of quantile loss; To predict residuals; quantiles; S227. Complete model training and testing, and output wind power prediction sequences. Photovoltaic prediction series and load forecast sequence ; S23, Wind power prediction sequence based on the output of S22 and photovoltaic prediction series Latin hypercube sampling is used to generate an initial uncertainty scenario set; S231. Using the predicted value output by S22 as the mean and the statistical standard deviation of historical data as the fluctuation range, construct the probability distribution of random variables for wind power, photovoltaics, and load: ; In the formula, , , These are the actual wind power output, actual photovoltaic power output, and actual load, respectively. , , They are respectively The square of the standard deviation of wind power output, solar power output, and load at any given moment; It is a normal distribution function; S232. Stratified sampling is performed on the above three types of random variables using Latin hypercube sampling to generate... The initial scenario is set, and the sampling expression is as follows: ; ; In the formula, For random variables in Uniformly stratified sampled values ​​within the interval; Let be the cumulative probability distribution function of the random variable; These are the generated scene sample values; It is the inverse function of the cumulative probability distribution function of the random variable; S233. Combine the sampled values ​​of each type of random variable at all times to form... A complete set of intraday operational scenarios is denoted as the initial scenario set. , For the first Complete intraday operational scenarios; S24. Based on the initial scene set output by S23, the improved particle swarm optimization algorithm is used to optimize K-means clustering to achieve scene reduction, and a typical uncertain scene set and probability weights are obtained.

[0015] Step S24 specifically includes the following steps: S241. Initialize K-means clustering parameters and set the number of typical scenarios. Euclidean distance is used as the similarity measure between scenes: ; In the formula, For the first Complete intraday operation scenario With the Complete intraday operation scenario The Euclidean distance between them; and They are respectively Moment Scene and scene The power value; This represents the total number of scheduling periods; S242. Construct an improved particle swarm optimization algorithm to adaptively optimize cluster centers with the objective of minimizing the total error of K-means clustering, and incorporate adaptive inertia weights. The calculation formula is: ; In the formula, and These are the upper and lower limits of the inertia weight, respectively; For the first The fitness value of each particle; This is the minimum fitness value for the population; This represents the average fitness value of the population. The improved velocity and position update expressions for the particle swarm optimization algorithm are as follows: ; ; In the formula, and The first The first particle Dimensional components in Time and The iteration speed at any given moment; and All are learning factors; and All A random number that is uniformly distributed within an interval; For the first The historical optimal position of each particle; The optimal position for the entire population; and The first The first particle Dimensional components in Time and The position at that moment; S243. Substitute the optimal initial cluster centers output by the improved particle swarm optimization algorithm into the K-means algorithm, and then process the output of S23. The initial scene is iteratively clustered until the cluster centers no longer change; S244. Calculate the proportion of scenes within each cluster and use it as the probability weight of the corresponding typical scene: , ; In the formula, For the first Probability weights for a typical scenario; For the first The number of scenarios contained within each cluster, with a one-to-one correspondence between the number of clusters and the number of typical scenarios; S245, Output 5 sets of typical uncertainty scenarios and the corresponding probability weights , Indicates the first The uncertain operating state vectors corresponding to each typical scenario, and , They represent the first In a typical scenario The optimal predicted output of wind power, the optimal predicted output of photovoltaic power, and the optimal load demand values, which are subject to constant uncertainties and fluctuations. This indicates the transpose operation.

[0016] Step S3 specifically includes the following steps: S31, Uncertainty Scenario Set Based on S2 Output and the corresponding probability weights Construct a price elasticity linkage matrix for the electricity, carbon, and green certificate markets; S311. Extract the electricity market price under each uncertainty scenario set. Carbon market prices Green certificate market price ; S312, Calculate the price elasticity coefficient of electricity. Carbon price elasticity coefficient Green certificate self-price elasticity coefficient Electricity-carbon cross-price elasticity coefficient Electricity-Green Certificate Cross-Price Elasticity Coefficient Carbon-electric cross-price elasticity coefficient Carbon-Green Certificate Cross-Price Elasticity Coefficient Green Certificate-Electric Cross Price Elasticity Coefficient Green certificate-carbon cross-price elasticity coefficient : ; ; ; ; ; ; ; ; ; In the formula, For the first Changes in electrical load power over a period of time; For the first Original electrical load power for the time period; For the first Changes in electricity prices over time periods; For the first The original market price of electricity during that period; For the first Changes in carbon emissions over a period of time; For the first Original carbon emissions for a given period; For the first Changes in carbon market prices over a period of time; For the first The original carbon market price during that period; For the first Changes in demand for green certificates over a period of time; For the first The demand for original green certificates during the specified time period; For the first Changes in the market price of green certificates over a given period; For the first The original market price of the green certificate during the specified period; S313, Constructing a Price Elasticity Linkage Matrix for the Electricity-Carbon-Green Certificates Three-Market System : ; In the formula, Let be an electroelastomer matrix, where any element is . ; Let be the carbon emission self-elastic submatrix, where any element is ; Let be the self-elastic submatrix of the green certificate, where any element is ; Let be an electric-carbon cross-elastic submatrix, where any element is... ; Let be the cross-elastic submatrix of electricity-green certificates, where any element is ; Let be a carbon-electric cross-elastic submatrix, where any element is... ; Let be a carbon-green certificate cross-elasticity submatrix, where any element is... ; Let be the green certificate-electric cross-linked elastic submatrix, where any element is ; The green certificate-carbon cross-elastic submatrix is ​​given by the following formula: (The formula is not provided in the original text.) ; S32, Price Elasticity Linkage Matrix of the Electricity-Carbon-Green Certificate Three Markets Based on the Output of S31 Establish a master-slave game upper-level model with virtual power plants as the dominant players to optimize transaction pricing; Wherein, the objective function is: ; Constraints: ; In the formula, The expected total revenue of the virtual power plant in all typical scenarios; For virtual power plants Report electricity transaction prices at all times; For virtual power plants Report carbon trading prices regularly; For virtual power plants Report the green certificate transaction price at all times; For the first In a typical scenario Electricity transaction volume at any given moment; For the first In a typical scenario Carbon allowance trading volume at any given time; For the first In a typical scenario The trading volume of green certificates at any given moment; For the first In a typical scenario The operating cost of a virtual power plant at any given moment; and These are the electricity sales price from the power grid and the electricity purchase price from the power grid, respectively. and These are the lowest and highest transaction prices in the carbon market on that day; and These are the lowest and highest transaction prices in the green certificate market on that day; The improved particle swarm optimization algorithm described in step S2 is used to solve the objective function and satisfy the constraints to obtain the bid combination that maximizes the expected revenue of the virtual power plant. And calculate the change in virtual power plant transaction prices: ; In the formula, The relative rate of change of the best electricity price. This represents the absolute change in electricity price. This serves as the benchmark price for the electricity market. This represents the relative change rate of the optimal carbon price quote. This represents the absolute change in carbon price quotes; As the benchmark price for the carbon market; The relative change rate of the optimal bid price for green certificates. This represents the absolute change in the price of green certificates. This serves as the benchmark price for the green certificate market. S33, Quotation combination based on S32 output Establish a master-slave game lower-level model for the optimal response of each resource subject; S331. Construct the following normalized response objective function: ; In the formula, For the first The normalized optimization objective of each distributed energy entity; For the first The operating costs of a single distributed energy entity; For the first The operation and maintenance costs of energy storage charging and discharging for a single distributed energy entity; For the first Demand response compensation costs for individual distributed energy entities; For the first The carbon emission costs of a distributed energy entity; For the first Electricity trading volume of each distributed energy entity; For the first Carbon quota trading volume of each distributed energy entity; For the first The trading volume of green certificates of a distributed energy entity; S332. The distributed energy entities in the lower-level model of the master-slave game substitute the changes in the virtual power plant trading bids output by S32 into the price elasticity linkage matrix of the electricity-carbon-green certificate three markets constructed by S31, and calculate the load adjustment, carbon emission adjustment and green certificate demand adjustment under the linkage of the three markets: ; In the formula, , , These are respectively load adjustment, carbon emission adjustment, and green certificate demand adjustment; S333, will , , Substitute the core decision correction quantity into the normalized response objective function of each distributed energy entity, and add the following constraints: ; In the formula, and The first Distributed energy entity The lower and upper limits of electrical output or load at any given time; For the first Distributed energy entity The maximum amount of carbon allowances available at any given time; For the first Distributed energy entity The maximum number of green certificates available at any given time; For the first Distributed energy entity At any given time, the electrical reference power is... For the first Distributed energy entity Time-based carbon emissions; For the first Distributed energy entity Demand for time-based green certificates; S334. Solve step S333 using a convex optimization algorithm to obtain the optimal trading volume for each entity that satisfies the price linkage response rules of the electricity-carbon-green certificate three-markets. ; S34. Based on the equilibrium result of the master-slave game, Nash negotiation is introduced to achieve fair distribution of benefits among multiple parties and output the final collaborative transaction price.

[0017] In this embodiment, after step S34, step S35 is also included: based on the transaction price and transaction volume determined by the hybrid game, establish a blockchain relay chain cross-chain trusted transaction rule, and output the blockchain cross-chain transaction rule file, smart contract text, and transaction hash index. Step S35 specifically includes the following steps: S351. Construct a relay chain cross-chain architecture, which includes an electricity trading chain, a carbon compliance chain, a green certificate rights confirmation chain, and a relay regulatory chain. S352. Establish the following cross-chain interaction rules: Electricity trading: Power delivery and settlement are automatically triggered after a transaction is completed; Carbon trading: Carbon allowances are simultaneously cancelled upon completion of the transaction; Green certificate transactions: After the transaction is completed, the ownership of the green certificate will be transferred and the certificate will be cancelled. S353. Establish transaction verification rules: ; In the formula, This represents the maximum transmission capacity of the line. Available carbon quotas; This represents the total number of available green certificates. S354. Generate a cross-chain smart contract, write the transaction price, transaction volume, and scenario constraints into the contract to achieve automatic execution, traceability, and tamper-proof.

[0018] S355 outputs the blockchain cross-chain transaction rule file, smart contract text, and transaction hash index, which serve as the basis for S4 scheduling and S6 transaction execution.

[0019] Step S34 specifically includes the following steps: S341. Set the independent operating revenue of each distributed energy entity when it does not participate in collaborative transactions as the point of negotiation breakdown. : ; In the formula, for The benchmark price of electricity in the market at any given time; For the first When each distributed energy entity operates independently Real-time electricity transaction volume; for The benchmark price of the carbon market at any given time; For the first When each distributed energy entity operates independently Carbon trading volume at any given time; for Real-time green certificate trading volume; For the first When each distributed energy entity operates independently Real-time green certificate trading volume; For the first When each distributed energy entity operates independently Total operating cost at any given moment; S342. Constructing the logarithmic objective function for Nash negotiation: ; Constraints: ; in, ; In the formula, , , They are respectively the ones to be solved Distributed energy entities at all times With distributed energy entities The price of electricity, carbon, and green certificates is coordinated in the trading of these currencies. For the first After each distributed energy entity participates in collaborative transactions Total operating cost at any given time; where, To logarithmize the objective function value for the Nash negotiation; For the first The actual benefits of each distributed energy entity participating in collaboration; The total number of distributed energy entities participating in the collaboration; S343. Iteratively solve for the Nash negotiation equilibrium and output the final cooperative transaction price. , , .

[0020] Step S4 specifically includes the following steps: S41. Modeling of coupled carbon capture equipment; S411. Calculate carbon capture capacity: ; and ; In the formula, For the first In a typical scenario Carbon capture rate at any given time; Carbon capture efficiency per unit operating power of carbon capture equipment; For the first In a typical scenario The operating power of the carbon capture equipment at all times; To fix the carbon capture power of the carbon capture equipment; The duration of the scheduling period; and They are respectively Minimum and maximum operating power of the carbon capture equipment at all times; S412. Calculate the operating cost of carbon capture equipment: ; In the formula, For the first In a typical scenario Operating costs of real-time carbon capture equipment; Cost per unit operating power of carbon capture equipment; The fixed operating costs of carbon capture equipment; S42, Modeling of coupled electro-pneumatic equipment; S421. Calculate the gas production of the electro-gas conversion equipment: ; and In the formula, For the first In a typical scenario Gas production rate of the electro-gas conversion equipment at any given time; Energy conversion efficiency of electro-gas conversion equipment; For the first In a typical scenario The operating power of the electro-pneumatic equipment at all times; and They are respectively The minimum and maximum operating power of the electro-pneumatic equipment at all times; S422. Calculate the carbon emissions and operating costs of the electricity-to-gas conversion equipment: ; ; In the formula, For the first In a typical scenario Carbon emissions from electro-gas conversion equipment; Carbon emission coefficient per unit gas production from natural gas combustion; For the first In a typical scenario Operating costs of the constant-time electro-gas conversion equipment; The unit power cost of the electro-gas conversion equipment; S43. Construct a demand response regulation model based on the price elasticity linkage matrix of the electricity-carbon-green certificate three markets; S431. Calculate the total adjustment of demand response under the price linkage of the electricity-carbon-green certificate three-market system: ; Constraints: ; In the formula, , , The first In a typical scenario The load demand response adjustment, carbon emission demand response adjustment, and green certificate demand response adjustment at any given time; , , The first In a typical scenario The difference between the price of electricity, carbon allowances, and green certificates in a coordinated transaction and the benchmark price; For the first In a typical scenario Always ensure user comfort when using electricity; For the first In a typical scenario Time-based load baseline value; This represents the lower limit of electrical comfort. S432. Calculate demand response cost: ; In the formula, For the first In a typical scenario Real-time demand response costs; Increase unit cost to load; Reduce unit cost to reduce load; S44. Construct a multi-objective collaborative optimization model for a virtual power plant: ; In the formula, The final single-objective optimization function value; , and All are target weights; in, Objective function for minimizing total operating cost: ; In the formula, The value of the economic objective function; For the first In a typical scenario Real-time virtual power plant and grid electricity purchase and sale costs; For the first In a typical scenario Operating costs of instantaneous energy storage devices; For the first In a typical scenario Profits from the coordinated trading of electricity, carbon, and green certificates; Objective function for minimizing total carbon emissions: ; In the formula, The objective function value is the low-carbon performance. For the first In a typical scenario Real-time carbon emissions from power sources within a virtual power plant; For the first In a typical scenario Net emissions from carbon quota trading at any given time; The objective function for maximizing the renewable energy absorption rate is: ; In the formula, The objective function value for renewable energy consumption; For the first In a typical scenario Actual wind power output at any given moment; For the first In a typical scenario Real-time photovoltaic output; For the first In a typical scenario Constantly abandoning wind and solar energy; Add the following constraints: Power balance constraints: ; In the formula, For the first In a typical scenario Distributed thermal power output at all times; and The first In a typical scenario Real-time energy storage discharge power and charging power; For the first In a typical scenario Real-time virtual power plant and grid power purchase and sale capacity; Constraints of energy storage devices: ; In the formula, and These are the minimum and maximum values ​​of the energy storage charging power, respectively. and These are the minimum and maximum values ​​of the energy storage discharge power, respectively. and The first In a typical scenario Time and The state of charge of the stored energy at any given moment; and These are energy storage charging efficiency and energy storage discharging efficiency, respectively. and These are the minimum and maximum values ​​of the energy storage state of charge, respectively; Carbon quotas and green certificate constraints: ; In the formula, For the first In a typical scenario The amount of carbon allowances traded by virtual power plants at any given time; For the first In a typical scenario The maximum amount of carbon allowances available at any given time; Power grid security constraints: .

[0021] Step S5 specifically includes the following steps: S51. Construct a high-level multi-stage robust optimization model to handle scenario uncertainties; S511. The typical uncertainty scenario set based on the output of S2 is divided into day-ahead scheduling, intraday scheduling and real-time scheduling according to the time periods 0:00-6:00, 6:00-18:00 and 18:00-24:00 respectively. S512, Define the set of robust uncertainties : ; S513. Constructing a multi-stage robust optimization objective function: ; In the formula, These are decision variables (including equipment operating power and transaction volume). S52. Using the improved particle swarm optimization algorithm described in step S2, solve for the robust optimization global optimal solution; S521. Determine the decision variables for solving the problem using the improved particle swarm optimization algorithm. : ; S522. Design of a fitness function for a mid-level improved particle swarm optimization algorithm: ; In the formula, To improve the fitness value of the particle swarm optimization algorithm; S523. Iterative solution, outputting the global optimal solution. ; S524, Find the global optimal solution Substitute the virtual power plant multi-objective collaborative optimization model described in step S4 to verify whether the constraints are met. If they are met, proceed to step S53; otherwise, return to step S523. S53. Construct a lower-level alternating direction multiplier method-column constraint generation algorithm to achieve distributed solution; S531. Divide the global optimization problem into the following 5 main subproblems and 1 global coordination subproblem: Subproblem 1: Optimization of carbon capture operation; Its decision variables ; Objective function: ; Constraints: ; In the formula, The objective function value for the carbon capture operation optimization subproblem; As a penalty factor; For power balance constrained Lagrange multipliers; Subproblem 2: Optimization of electro-gas conversion operation; Its decision variables ; Objective function: ; Constraints: ; In the formula, The objective function value for the sub-problem of optimizing the operation of the electric-to-gas conversion; Sub-problem 3: Energy storage equipment operation optimization sub-problem; Its decision variables ; Objective function: ; Constraints: ; In the formula, The objective function value for the sub-problem of optimizing the operation of energy storage devices; Subproblem 4: Distributed power source operation optimization subproblem; Its decision variables ; Objective function: ; Constraints: ; In the formula, , , All are secondary cost coefficients for power supplies; The objective function value for the subproblem of optimizing the operation of distributed power sources; Subproblem 5: Optimization of trading among multiple distributed energy entities; Its decision variables ; Objective function: ; Constraints: ; In the formula, The objective function value for the sub-problem of multi-distributed energy entity trading optimization; Subproblem 6: Global Coordination Subproblem; Its decision variables ; Objective function: ; Iterative update formula: ; In the formula, For the first The Lagrange multiplier corresponding to the power balance constraint in the next iteration; The objective function value for the global coordination subproblem; For the first In this iteration, the power interaction between the virtual power plant and the power grid after global coordination; For the first In the nth iteration, the 1st The virtual power plant and grid interaction power output from each main sub-problem; For the first The Lagrange multiplier corresponding to the power balance constraint in the next iteration; S532. Solve the subproblems divided in step S541 using the multiplier method-column constraint generation algorithm, and output the optimal solutions to each main subproblem. S54. The algorithm employs a multi-stage robust optimization layer to handle uncertainty, an improved particle swarm optimization algorithm in the middle layer to solve for the global optimum, and a multiplier method-column constraint generation algorithm in the lower layer to achieve distributed deployment. After iterative convergence, it outputs the optimal set of decision variables. , The first A distributed energy entity in The optimal electricity trading volume, carbon quota trading volume and green certificate trading volume at any given time; the optimal operating power and carbon capture volume of carbon capture equipment; the optimal operating power, gas production and equivalent carbon emissions of power-to-gas equipment; the optimal output of distributed power sources; and the optimal interaction power between virtual power plants and the power grid. S55. Verification: Calculate the economic objective function value, low-carbon objective function value, and new energy consumption objective function value of the optimal decision variable set, and determine whether they meet the multi-objective optimization requirements of S4. If so, output the optimal decision variable set as the optimal trading and operation strategy.

[0022] Step S532 specifically includes the following steps: S5321. Initialize Coordination Variables Lagrange multipliers Punishment factor ; S5322. Solve each main subproblem in parallel, and obtain the optimal solution for each subproblem based on the current coordination variable and Lagrange multipliers. ; S5323. Employ a column constraint generation algorithm to generate new constraint columns, supplement them to each subproblem, and correct the feasible region of the subproblems: ; In the formula, Generate thresholds for column constraints; S5324. Update the coordination variables and Lagrange multipliers to achieve coordinated convergence of the subproblems: ; In the formula, and The first The second iteration and the first Lagrange multipliers in the next iteration; S5325. Distributed Convergence Determination: If... If the distributed iteration fails, stop and output the optimal solutions to each main subproblem; otherwise, let... Then return to S5322. This represents the convergence threshold for distributed iteration.

[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 virtual power plant-scale trading method for the electricity-carbon-green certificate market, characterized by: Includes the following steps: S1. Collect multi-source operation data of virtual power plants, market data, power grid operation data, and meteorological data, and preprocess them to obtain a standardized time-series dataset and a set of equipment constraint parameters. S2. Based on the standardized time-series dataset output from S1, use a VMD-Bi-MGLSTM multi-branch prediction network to complete ultra-short-term predictions and generate a set of typical uncertainty scenarios. S3. Based on the output of S2, construct a three-market coupled pricing and trading model for electricity, carbon, and green certificates. S4. Combine the three-market coupled pricing and trading model constructed in S3 to establish a multi-objective collaborative optimization model for virtual power plants. S5. Perform hierarchical distributed solution on the multi-objective collaborative optimization model output from S4 to output the optimal trading and operation strategies. S6. Execute transactions and perform closed-loop verification through blockchain. S7. Dynamically update.

2. The virtual power plant-scale trading method for the electricity-carbon-green certificate market according to claim 1, characterized in that: The virtual power plant multi-source operation data mentioned in step S1 includes distributed wind power output, distributed photovoltaic power output, energy storage state of charge (SOC), carbon capture equipment operation parameters, power-to-gas equipment operation parameters, carbon emissions, and load data; among which, the load data includes residential flexible adjustable load, commercial flexible adjustable load, and industrial basic rigid load. Market data includes time-of-use electricity pricing in the electricity market, carbon market allowances and prices, green certificate market allowances and prices, and trading rules; Power grid operation data includes node voltage, line capacity, network loss parameters, and power balance constraints; Meteorological data includes wind speed, wind direction, light intensity, temperature, and humidity.

3. The virtual power plant-scale trading method for the electricity-carbon-green certificate market according to claim 1, characterized in that: The preprocessing described in step S1 involves sequentially performing outlier removal, linear interpolation to fill missing values, and Min-Max normalization on the subsets of virtual power plant multi-source operation data, market data, power grid operation data, and meteorological data.

4. The virtual power plant-scale trading method for the electricity-carbon-green certificate market according to claim 2, characterized in that: Step S2 specifically includes the following steps: S21. Perform modal decomposition on the standardized time series dataset output by S1 to obtain stationary intrinsic modal components; S211, Read wind power output sequence Photovoltaic power output sequence and load power sequence Variational mode decomposition is performed on each sequence to construct and solve constrained variational problems, thereby decomposing the nonstationary original sequence into... a stable intrinsic mode component and residual components: ； Constraints: ; In the formula, The first decomposition obtained A set of intrinsic modal components; For the first The set of center frequencies corresponding to each modal component; Indicates time The partial derivatives; The Dirac distribution function; Represents the convolution integral; The imaginary unit; For the first The center frequency corresponding to each modal component; The current standardized original sequence to be decomposed is taken in sequence. , , ; S212. Solve the constrained variational problem of S211 iteratively using the alternating direction multiplier method, and output the corresponding sequences. One set of stationary intrinsic mode components and one set of residual components; S22, Based on the output of step S21 A bidirectional Bi-MGLSTM network is constructed and trained using a set of stationary intrinsic mode components and a set of residual components. S221. Divide the stationary intrinsic mode components output by S21 into training, validation, and test sets in an 8:1:1 ratio, and set the input sequence length. ; S222. An improved bidirectional multi-gated LSTM unit is constructed by adding a double forget gate and a double input gate to the standard LSTM. The improved bidirectional multi-gated LSTM unit includes a forward multi-gated LSTM subnetwork, a backward multi-gated LSTM subnetwork, and a bidirectional feature fusion layer. Both the multi-gated LSTM subnetwork and the backward multi-gated LSTM subnetwork include a first forget gate, a second forget gate, a comprehensive forget gate, a first input gate, a second input gate, a comprehensive input gate, a candidate cell state, a cell state, an output gate, and a hidden state, arranged sequentially. The gating update calculation formula is as follows: ； In the formula, and These are the outputs of the first forget gate and the second forget gate, respectively. For the comprehensive forget gate output; and These are the outputs of the first input gate and the second input gate, respectively. This is a composite input gate output; Output gate output; Candidate cell state; In cellular state; The current hidden layer state; is; , , , , , These are the weights of the first forget gate, the second forget gate, the first input gate, the second input gate, the output gate, and the candidate cell state, respectively. , , , , , For each gated bias term; This represents the hidden layer state from the previous time step. The model input at the current time is... or or ; Use the Sigmoid activation function; The activation function is a linear correction unit. It is the hyperbolic tangent activation function; S223, based on the improved bidirectional multi-gated LSTM unit of S222, constructs a bidirectional Bi-MGLSTM network, extracts temporal features from the forward and backward directions respectively, and obtains the final hidden layer output: ； ； ； In the formula, and These are the outputs of the forward MG-LSTM hidden layer and the backward MG-LSTM hidden layer at the current time, respectively. This is the final hidden layer output after bidirectional feature fusion, and it is considered as the fused feature. For forward multi-gated LSTM sub-cell functions; For backward multi-gated LSTM sub-unit functions; For forward multi-gated LSTM subnetworks in The hidden layer state at any given time; For backward multi-gated LSTM subnetworks in The hidden layer state at any given time; For feature fusion weights; For feature fusion bias term; S224, The fusion feature output by S223 By connecting to the fully connected layer, the predicted values ​​of the single-mode components are obtained. ; S225. The prediction results of all intrinsic mode components are linearly superimposed to obtain the ultra-short-term prediction values ​​of wind power, photovoltaic power, and load: ； In the formula, For the first Each sample corresponds to The ultra-short-term forecast values ​​of wind power, solar power, or load at any given moment; These are the predicted values ​​for the residual components; S226. Using the quantile loss function as the training objective, minimize the prediction error, and the formula for calculating the loss function is: ； in, ； ； In the formula, The quantile loss function value; The number of samples in the training set; The piecewise function of quantile loss; To predict residuals; Quantities; S227. Complete model training and testing, and output wind power prediction sequences. Photovoltaic prediction series and load forecast sequence ; S23, Wind power prediction sequence based on the output of S22 and photovoltaic prediction series Latin hypercube sampling is used to generate an initial uncertainty scenario set; S231. Using the predicted value output by S22 as the mean and the statistical standard deviation of historical data as the fluctuation range, construct the probability distribution of random variables for wind power, photovoltaics, and load: ； In the formula, , , These are the actual wind power output, actual photovoltaic power output, and actual load, respectively. , , They are respectively The square of the standard deviation of wind power output, solar power output, and load at any given moment; It is a normal distribution function; S232. Stratified sampling is performed on the above three types of random variables using Latin hypercube sampling to generate... The initial scenario is set, and the sampling expression is as follows: ； ； In the formula, For random variables in Uniformly stratified sampled values ​​within the interval; Let be the cumulative probability distribution function of the random variable; These are the generated scene sample values; It is the inverse function of the cumulative probability distribution function of the random variable; S233. Combine the sampled values ​​of each type of random variable at all times to form... A complete set of intraday operational scenarios is denoted as the initial scenario set. , For the first Complete intraday operational scenarios; S24. Based on the initial scene set output by S23, the improved particle swarm optimization algorithm is used to optimize K-means clustering to achieve scene reduction, and a typical uncertain scene set and probability weights are obtained.

5. The virtual power plant-scale trading method for the electricity-carbon-green certificate market according to claim 4, characterized in that: Step S24 specifically includes the following steps: S241. Initialize K-means clustering parameters and set the number of typical scenarios. Euclidean distance is used as the similarity measure between scenes: ； In the formula, For the first Complete intraday operation scenario With the Complete intraday operation scenario The Euclidean distance between them; and They are respectively Moment Scene and scene The power value; This represents the total number of scheduling periods; S242. Construct an improved particle swarm optimization algorithm to adaptively optimize cluster centers with the objective of minimizing the total error of K-means clustering, and incorporate adaptive inertia weights. The calculation formula is: ； In the formula, and These are the upper and lower limits of the inertia weight, respectively; For the first The fitness value of each particle; This is the minimum fitness value for the population; This represents the average fitness value of the population. The improved velocity and position update expressions for the particle swarm optimization algorithm are as follows: ； ； In the formula, and The first The first particle Dimensional components in Time and The iteration speed at any given moment; and All are learning factors; and All A random number that is uniformly distributed within an interval; For the first The historical optimal position of each particle; The optimal position for the entire population; and The first The first particle Dimensional components in Time and The position at that moment; S243. Substitute the optimal initial cluster centers output by the improved particle swarm optimization algorithm into the K-means algorithm, and then process the output of S23. The initial scene is iteratively clustered until the cluster centers no longer change; S244. Calculate the proportion of scenes within each cluster and use it as the probability weight of the corresponding typical scene: ， ； In the formula, For the first Probability weights for a typical scenario; For the first The number of scenarios contained within each cluster, with a one-to-one correspondence between the number of clusters and the number of typical scenarios; S245, Output 5 sets of typical uncertainty scenarios and the corresponding probability weights , Indicates the first The uncertain operating state vectors corresponding to each typical scenario, and , They represent the first In a typical scenario The optimal predicted output of wind power, the optimal predicted output of photovoltaic power, and the optimal load demand values, which are subject to constant uncertainties and fluctuations. This indicates the transpose operation.

6. The virtual power plant-scale trading method for the electricity-carbon-green certificate market according to claim 5, characterized in that: Step S3 specifically includes the following steps: S31, Uncertainty Scenario Set Based on S2 Output and the corresponding probability weights Construct a price elasticity linkage matrix for the electricity, carbon, and green certificate markets; S311. Extract the electricity market price under each uncertainty scenario set. Carbon market prices Green certificate market price ; S312, Calculate the price elasticity coefficient of electricity. Carbon price elasticity coefficient Green certificate self-price elasticity coefficient Electricity-carbon cross-price elasticity coefficient Electricity-Green Certificate Cross-Price Elasticity Coefficient Carbon-electricity cross-price elasticity coefficient Carbon-Green Certificate Cross-Price Elasticity Coefficient Green Certificate-Electric Cross Price Elasticity Coefficient Green certificate-carbon cross-price elasticity coefficient : ； ； ； ； ； ； ； ； ； In the formula, For the first Changes in electrical load power over a period of time; For the first Original electrical load power for the time period; For the first Changes in electricity prices over time periods; For the first The original market price of electricity during that period; For the first Changes in carbon emissions over a period of time; For the first Original carbon emissions for a given period; For the first Changes in carbon market prices over a period of time; For the first The original carbon market price during that period; For the first Changes in demand for green certificates over a period of time; For the first The demand for original green certificates during the specified time period; For the first Changes in the market price of green certificates over a given period; For the first The original market price of the green certificate during the specified period; S313, Constructing a Price Elasticity Linkage Matrix for the Electricity-Carbon-Green Certificates Three-Market System : ； In the formula, Let be an electroelastomer matrix, where any element is . ; Let be the carbon emission self-elastic submatrix, where any element is ; Let be the self-elastic submatrix of the green certificate, where any element is ; Let be an electric-carbon cross-elastic submatrix, where any element is... ; Let be the cross-elastic submatrix of electricity-green certificates, where any element is ; Let be a carbon-electric cross-elastic submatrix, where any element is... ; Let be a carbon-green certificate cross-elasticity submatrix, where any element is... ; Let be the green certificate-electric cross-linked elastic submatrix, where any element is ; The green certificate-carbon cross-elastic submatrix is ​​given by the following formula: (The formula is not provided in the original text.) ; S32, Price Elasticity Linkage Matrix of the Electricity-Carbon-Green Certificate Three Markets Based on the Output of S31 Establish a master-slave game upper-level model with virtual power plants as the dominant players to optimize transaction pricing; Wherein, the objective function is: ； Constraints: ； In the formula, The expected total revenue of the virtual power plant in all typical scenarios; For virtual power plants Report electricity transaction prices at all times; For virtual power plants Report carbon trading prices regularly; For virtual power plants Report the green certificate transaction price at all times; For the first In a typical scenario Electricity transaction volume at any given moment; For the first In a typical scenario Carbon allowance trading volume at any given time; For the first In a typical scenario The trading volume of green certificates at any given moment; For the first In a typical scenario The operating cost of a virtual power plant at any given moment; and These are the electricity sales price from the power grid and the electricity purchase price from the power grid, respectively. and These are the lowest and highest transaction prices in the carbon market on that day; and These are the lowest and highest transaction prices in the green certificate market on that day; The improved particle swarm optimization algorithm described in step S2 is used to solve the objective function and satisfy the constraints to obtain the bid combination that maximizes the expected revenue of the virtual power plant. And calculate the change in virtual power plant transaction prices: ； In the formula, The relative rate of change of the optimal electricity price. This represents the absolute change in electricity price. This serves as the benchmark price for the electricity market. This represents the relative change rate of the optimal carbon price quote. This represents the absolute change in carbon price quotes; As the benchmark price for the carbon market; The relative change rate of the optimal bid price for green certificates. This represents the absolute change in the price of green certificates. This serves as the benchmark price for the green certificate market. S33, Quotation combination based on S32 output Establish a master-slave game lower-level model for the optimal response of each resource subject; S331. Construct the following normalized response objective function: ； In the formula, For the first The normalized optimization objective of each distributed energy entity; For the first The operating costs of a single distributed energy entity; For the first The operation and maintenance costs of energy storage charging and discharging for a single distributed energy entity; For the first Demand response compensation costs for individual distributed energy entities; For the first The carbon emission costs of a distributed energy entity; For the first Electricity trading volume of each distributed energy entity; For the first Carbon quota trading volume of each distributed energy entity; For the first The trading volume of green certificates for each distributed energy entity; S332. The distributed energy entities in the lower-level model of the master-slave game substitute the changes in the virtual power plant trading bids output by S32 into the price elasticity linkage matrix of the electricity-carbon-green certificate three markets constructed by S31, and calculate the load adjustment, carbon emission adjustment and green certificate demand adjustment under the linkage of the three markets: ； In the formula, , , These are respectively load adjustment, carbon emission adjustment, and green certificate demand adjustment; S333, will , , Substitute the core decision correction quantity into the normalized response objective function of each distributed energy entity, and add the following constraints: ； In the formula, and The first Distributed energy entity The lower and upper limits of electrical output or load at any given time; For the first Distributed energy entity The maximum amount of carbon allowances available at any given time; For the first Distributed energy entity The maximum number of green certificates available at any given time; For the first Distributed energy entity At any given time, the electrical reference power is... For the first Distributed energy entity Time-based carbon emissions; For the first Distributed energy entity Demand for time-based green certificates; S334. Solve step S333 using a convex optimization algorithm to obtain the optimal trading volume for each entity that satisfies the price linkage response rules of the electricity-carbon-green certificate three-markets. ; S34. Based on the equilibrium result of the master-slave game, Nash negotiation is introduced to achieve fair distribution of benefits among multiple parties and output the final collaborative transaction price.

7. The virtual power plant-scale trading method for the electricity-carbon-green certificate market according to claim 6, characterized in that: Step S34 specifically includes the following steps: S341. Set the independent operating revenue of each distributed energy entity when it does not participate in collaborative transactions as the point of negotiation breakdown. : ； In the formula, for The benchmark price of electricity in the current market; For the first When each distributed energy entity operates independently Real-time electricity transaction volume; for The benchmark price of the carbon market at any given time; For the first When each distributed energy entity operates independently Carbon trading volume at any given time; for Real-time green certificate trading volume; For the first When each distributed energy entity operates independently Real-time green certificate trading volume; For the first When each distributed energy entity operates independently Total operating cost at any given moment; S342. Constructing the logarithmic objective function for Nash negotiation: ； Constraints: ; in, ； In the formula, , , They are respectively the ones to be solved Distributed energy entities at all times With distributed energy entities The price of electricity, carbon, and green certificates is coordinated in the trading of these currencies. For the first After each distributed energy entity participates in collaborative transactions Total operating cost at any given time; where, To logarithmize the objective function value for the Nash negotiation; For the first The actual benefits of each distributed energy entity participating in collaboration; The total number of distributed energy entities participating in the collaboration; S343. Iteratively solve for the Nash negotiation equilibrium and output the final cooperative transaction price. , , .

8. The virtual power plant-scale trading method for the electricity-carbon-green certificate market according to claim 7, characterized in that: Step S4 specifically includes the following steps: S41. Modeling of coupled carbon capture equipment; S411. Calculate carbon capture capacity: ； and ; In the formula, For the first In a typical scenario Carbon capture rate at any given time; Carbon capture efficiency per unit operating power of carbon capture equipment; For the first In a typical scenario The operating power of the carbon capture equipment at all times; To fix the carbon capture power of the carbon capture equipment; The duration of the scheduling period; and They are respectively Minimum and maximum operating power of the carbon capture equipment at all times; S412. Calculate the operating cost of carbon capture equipment: ； In the formula, For the first In a typical scenario Operating costs of real-time carbon capture equipment; Cost per unit operating power of carbon capture equipment; The fixed operating costs of carbon capture equipment; S42, Modeling of coupled electro-pneumatic equipment; S421. Calculate the gas production of the electro-gas conversion equipment: ； and In the formula, For the first In a typical scenario Gas production rate of the electro-gas conversion equipment at any given time; Energy conversion efficiency of electro-gas conversion equipment; For the first In a typical scenario The operating power of the electro-pneumatic equipment at all times; and They are respectively The minimum and maximum operating power of the electro-pneumatic equipment at all times; S422. Calculate the carbon emissions and operating costs of the electricity-to-gas conversion equipment: ； ； In the formula, For the first In a typical scenario Carbon emissions from electro-gas conversion equipment; Carbon emission coefficient per unit gas production from natural gas combustion; For the first In a typical scenario Operating costs of the constant-time electro-gas conversion equipment; The unit power cost of the electro-gas conversion equipment; S43. Construct a demand response regulation model based on the price elasticity linkage matrix of the electricity-carbon-green certificate three markets; S431. Calculate the total adjustment of demand response under the price linkage of the electricity-carbon-green certificate three-market system: ； Constraints: ; In the formula, , , The first In a typical scenario The load demand response adjustment, carbon emission demand response adjustment, and green certificate demand response adjustment at any given time; , , The first In a typical scenario The difference between the price of electricity, carbon allowances, and green certificates in a coordinated transaction and the benchmark price; For the first In a typical scenario Always ensure user comfort when using electricity; For the first In a typical scenario Time-based load baseline value; This represents the lower limit of electrical comfort. S432. Calculate demand response cost: ； In the formula, For the first In a typical scenario Real-time demand response costs; Increase unit cost to load; Reduce unit cost to reduce load; S44. Construct a multi-objective collaborative optimization model for a virtual power plant: ； In the formula, The final single-objective optimization function value; , and All are target weights; in, The objective function for minimizing total operating cost is: ； In the formula, The value of the economic objective function; For the first In a typical scenario Real-time virtual power plant and grid electricity purchase and sale costs; For the first In a typical scenario Operating costs of instantaneous energy storage devices; For the first In a typical scenario Profits from the coordinated trading of electricity, carbon, and green certificates; Objective function for minimizing total carbon emissions: ； In the formula, The objective function value is the low-carbon performance. For the first In a typical scenario Real-time carbon emissions from power sources within a virtual power plant; For the first In a typical scenario Net emissions from carbon quota trading at any given time; The objective function for maximizing the renewable energy absorption rate is: ； In the formula, The objective function value for renewable energy consumption; For the first In a typical scenario Actual wind power output at any given moment; For the first In a typical scenario Real-time photovoltaic output; For the first In a typical scenario Constantly abandoning wind and solar energy; Add the following constraints: Power balance constraints: ； In the formula, For the first In a typical scenario Distributed thermal power output at all times; and The first In a typical scenario Real-time energy storage discharge power and charging power; For the first In a typical scenario Real-time virtual power plant and grid power purchase and sale capacity; Constraints of energy storage devices: ； In the formula, and These are the minimum and maximum values ​​of the energy storage charging power, respectively. and These are the minimum and maximum values ​​of the energy storage discharge power, respectively. and The first In a typical scenario Time and The state of charge of the stored energy at any given moment; and These are energy storage charging efficiency and energy storage discharging efficiency, respectively. and These are the minimum and maximum values ​​of the energy storage state of charge, respectively; Carbon quotas and green certificate constraints: ； In the formula, For the first In a typical scenario The amount of carbon allowances traded by virtual power plants at any given time; For the first In a typical scenario The maximum amount of carbon allowances available at any given time; Power grid security constraints: 。 9. The virtual power plant-scale trading method for the electricity-carbon-green certificate market according to claim 8, characterized in that: Step S5 specifically includes the following steps: S51. Construct a high-level multi-stage robust optimization model to handle scenario uncertainties; S511, based on the typical uncertainty scenario set output by S2, is divided into day-ahead scheduling, intraday scheduling and real-time scheduling according to the time periods 0:00-6:00, 6:00-18:00 and 18:00-24:00 respectively; S512, Define the set of robust uncertainties : ； S513. Constructing a multi-stage robust optimization objective function: ； In the formula, For decision variables; S52. Using the improved particle swarm optimization algorithm described in step S2, solve for the robust optimization global optimal solution; S521. Determine the decision variables for solving the problem using the improved particle swarm optimization algorithm. : ； S522. Design of a fitness function for a mid-level improved particle swarm optimization algorithm: ； In the formula, To improve the fitness value of the particle swarm optimization algorithm; S523. Iterative solution, outputting the global optimal solution. ; S524, Find the global optimal solution Substitute the virtual power plant multi-objective collaborative optimization model described in step S4 to verify whether the constraints are met. If they are met, proceed to step S53; otherwise, return to step S523. S53. Construct a lower-level alternating direction multiplier method-column constraint generation algorithm to achieve distributed solution; S531. Divide the global optimization problem into the following 5 main subproblems and 1 global coordination subproblem: Subproblem 1: Optimization of carbon capture operation; Its decision variables ; Objective function: ； Constraints: ； In the formula, The objective function value for the carbon capture operation optimization subproblem; As a penalty factor; For power balance constrained Lagrange multipliers; Subproblem 2: Optimization of electro-gas conversion operation; Its decision variables ; Objective function: ； Constraints: ； In the formula, The objective function value for the sub-problem of optimizing the operation of the electric-to-gas conversion; Sub-problem 3: Energy storage equipment operation optimization sub-problem; Its decision variables ; Objective function: ； Constraints: ； In the formula, The objective function value for the sub-problem of optimizing the operation of energy storage devices; Subproblem 4: Distributed power source operation optimization subproblem; Its decision variables ; Objective function: ； Constraints: ； In the formula, , , All are secondary cost coefficients for power supplies; The objective function value for the subproblem of optimizing the operation of distributed power sources; Subproblem 5: Optimization of trading among multiple distributed energy entities; Its decision variables ; Objective function: ； Constraints: ； In the formula, The objective function value for the sub-problem of multi-distributed energy entity trading optimization; Subproblem 6: Global Coordination Subproblem; Its decision variables ; Objective function: ； Iterative update formula: ； In the formula, For the first The Lagrange multiplier corresponding to the power balance constraint in the next iteration; The objective function value for the global coordination subproblem; For the first In this iteration, the power interaction between the virtual power plant and the power grid after global coordination; For the first In the nth iteration, the 1st The virtual power plant and grid interaction power output from each main sub-problem; For the first The Lagrange multiplier corresponding to the power balance constraint in the next iteration; S532. Solve the subproblems divided in step S541 using the multiplier method-column constraint generation algorithm, and output the optimal solutions to each main subproblem. S54. The algorithm employs a multi-stage robust optimization layer to handle uncertainty, an improved particle swarm optimization algorithm in the middle layer to solve for the global optimum, and a multiplier method-column constraint generation algorithm in the lower layer to achieve distributed deployment. After iterative convergence, it outputs the optimal set of decision variables. , The first A distributed energy entity in The optimal electricity trading volume, carbon quota trading volume and green certificate trading volume at any given time; the optimal operating power and carbon capture volume of carbon capture equipment; the optimal operating power, gas production and equivalent carbon emissions of power-to-gas equipment; the optimal output of distributed power sources; and the optimal interaction power between virtual power plants and the power grid. S55. Verification: Calculate the economic objective function value, low-carbon objective function value, and new energy consumption objective function value of the optimal decision variable set, and determine whether they meet the multi-objective optimization requirements of S4. If so, output the optimal decision variable set as the optimal trading and operation strategy.

10. The virtual power plant-scale trading method for the electricity-carbon-green certificate market according to claim 9, characterized in that: Step S532 specifically includes the following steps: S5321. Initialize Coordination Variables Lagrange multipliers Punishment factor ; S5322. Solve each main subproblem in parallel, and obtain the optimal solution for each subproblem based on the current coordination variable and Lagrange multipliers. ; S5323. Employ a column constraint generation algorithm to generate new constraint columns, supplement them to each subproblem, and correct the feasible region of the subproblems: ； In the formula, Generate thresholds for column constraints; S5324. Update the coordination variables and Lagrange multipliers to achieve coordinated convergence of subproblems: ； In the formula, and The first The second iteration and the first Lagrange multipliers in the next iteration; S5325. Distributed Convergence Determination: If... If the distributed iteration fails, stop and output the optimal solutions to each main subproblem; otherwise, let... Then return to S5322. This represents the convergence threshold for distributed iteration.

Images