Multi-dimensional target constraint-oriented aggregation subject market transaction matching method

Abstract

The invention provides an aggregation subject market transaction matching method for multi-dimensional target constraint, and relates to the technical field of resource allocation optimization, and the method comprises the steps: collecting transaction participation data of a plurality of aggregation subjects, carrying out the feature extraction of each transaction participation data, and obtaining an aggregation subject feature set; constructing a multi-dimensional target optimization model by taking cost, income and stability as multiple targets based on the aggregated main body feature set, and introducing output constraints, time sequence constraints and market rule constraints to form a target constraint matrix; and under the constraint of the target constraint matrix, carrying out iterative solution on the multi-dimensional target optimization model by adopting a multi-target optimization algorithm, and generating an optimal transaction matching scheme of each aggregation main body. Through the method and the device, the technical problem of poor feasibility of a transaction matching scheme caused by difficulty in processing complex multi-target conflicts due to adoption of a simplified static optimization model in the prior art is solved, and the feasibility of the aggregation subject market transaction matching scheme is improved through multi-dimensional target optimization.

Description

Technical Field

[0001] This application relates to the field of resource allocation optimization technology, specifically to a method for matching aggregated subject market transactions oriented towards multi-dimensional objective constraints. Background Technology

[0002] In existing electricity market aggregate trading, simplified static optimization models are typically used to schedule the output and trading strategies of each aggregate entity. While feasible solutions can be obtained when dealing with single or simple weighted objectives, the models are too simplified and cannot fully consider the multidimensional trading needs of aggregate entities. They also cannot accurately reflect the operational characteristics and diversified value demands of aggregate entities in the real market environment. As a result, the generated trading strategies are difficult to meet the differentiated needs of each entity in the actual market. This not only reduces the efficiency of trading matching but also makes the trading matching strategies less executable or unstable, making it difficult to achieve intelligent matching between aggregate entities and market transactions.

[0003] In summary, existing technologies suffer from the technical problem that the simplified static optimization model makes it difficult to handle complex multi-objective conflicts, resulting in poor feasibility of transaction matching schemes in actual operation. Summary of the Invention

[0004] The purpose of this application is to provide a market transaction matching method for aggregated entities oriented to multi-dimensional objective constraints, in order to solve the technical problem in the prior art that the use of simplified static optimization models makes it difficult to handle complex multi-objective conflicts, resulting in poor feasibility of transaction matching schemes in actual operation.

[0005] To achieve the above objectives, this application provides a method for market transaction matching of aggregated entities under multi-dimensional objective constraints. The method includes: collecting transaction participation data from multiple aggregated entities; extracting features from each transaction participation data to obtain an aggregated entity feature set; constructing a multi-dimensional objective optimization model based on the aggregated entity feature set, with cost, revenue, and stability as multiple objectives, and introducing output constraints, time series constraints, and market rule constraints to form an objective constraint matrix; and iteratively solving the multi-dimensional objective optimization model using a multi-objective optimization algorithm under the constraints of the objective constraint matrix to generate the optimal transaction matching scheme for each aggregated entity.

[0006] Optionally, historical transaction data and real-time transaction information of multiple aggregation entities are collected, including output, load, energy storage status, participation period and transaction frequency; the collected historical transaction data and real-time transaction information of multiple aggregation entities are preprocessed, including noise reduction, normalization and missing value compensation; the average output, peak power, load flexibility, revenue volatility and transaction execution deviation are calculated on the preprocessed historical transaction data and real-time transaction information of multiple aggregation entities to obtain characteristic indicators reflecting multiple aggregation entities, and the feature set of the aggregation entities is constructed.

[0007] Optionally, with cost, revenue, and stability as multi-dimensional objectives, an evaluation function relationship is established between the feature vectors in the feature set of the aggregated entities and the multi-dimensional objectives. Stability is used to quantify the stability of the trading strategies of multiple aggregated entities under market fluctuations and the consistency of matching strategies, including the rate of change of trading duration, trading volatility, and probability of default. Based on the evaluation function relationship, a multi-objective evaluation function is constructed with the objectives of minimizing cost, maximizing revenue, and maximizing stability. The multi-dimensional objective optimization model is constructed with the highest evaluation fitness of the multi-objective evaluation function as the optimization objective, and is used to generate the optimal trading matching scheme for multiple aggregated entities.

[0008] Optionally, based on the power generation capacity, energy storage capacity, and available output range of multiple aggregation entities, the minimum and maximum output of each entity in the current time period are set to determine the output constraints; based on the continuity requirements of the trading strategies of multiple aggregation entities and the output change rate limit, the upper limit of output change in continuous time periods is set to determine the time-series constraints; based on market trading rules, price upper and lower limits, and trading capacity limits, the maximum and minimum trading volumes that multiple aggregation entities can participate in in the market are set to determine the market rule constraints; the output constraints, time-series constraints, and market rule constraints are arranged according to the entity and constraint type, and represented in a matrix to obtain the target constraint matrix, which is used to constrain the solution of the multi-objective optimization model.

[0009] Optionally, the target constraint matrix is ​​used as a constraint condition, and the target is evaluated based on the multidimensional target optimization model. This includes randomly collecting empirical transaction matching schemes for target fitness evaluation, adjusting optimization variables based on the evaluation results to obtain optimization solution curves for each dimension; searching for variables in each dimension based on these optimization solution curves, updating the optimization solution curves for each dimension, and iteratively searching until the transaction matching scheme with the highest total evaluation value of the optimization solution curves for each dimension is found, thus obtaining the optimal transaction matching scheme.

[0010] Optionally, the direction of change of the optimization solution curves in each iteration is monitored, including the gradient or trend of cost, benefit, and stability objectives; strategies with consistent or similar curve change directions are grouped into the same partition, while strategies with significantly different change directions are grouped into other partitions; differentiated search strategies are adopted for different partitions, with increased search diversity for partitions with large fluctuations in change direction or unstable gradients, and accelerated convergence for partitions with stable change direction or consistent gradients; after each iteration, the partition division is readjusted according to the curve change trend, and the search strategy is continuously optimized. When the search objective or number of searches is reached, the transaction matching scheme with the highest global evaluation result is output, thus obtaining the optimal transaction matching scheme.

[0011] Optionally, real-time market transaction feedback information is collected to identify discrepancies in the transaction matching scheme; dynamic compensation and correction are performed based on the discrepancy identification results to achieve adaptive matching.

[0012] Optionally, the real-time transaction feedback information includes market electricity price fluctuation data, system load change data, transaction execution status information of multiple aggregate entities, and market clearing data. The collected real-time transaction feedback information is synchronized and timestamp-calibrated to ensure time sequence consistency of each data source. The real-time transaction feedback information is aligned with the market-collected data corresponding to the optimal transaction matching scheme, and the fluctuation difference between each transaction data is calculated.

[0013] Optionally, based on the difference identification results, market volatility feature indicators are extracted; the feature set of the aggregation entity is updated based on the market volatility feature indicators to obtain the feature snapshot matrix for the current period; and the transaction matching scheme of the aggregation entity is solved iteratively based on the feature snapshot matrix to output an adaptively optimized transaction matching scheme.

[0014] Optionally, the market volatility characteristic indicators include: price deviation rate, load volatility coefficient, and main transaction deviation amount.

[0015] Optionally, the parameters in the target constraint matrix are adjusted based on the feature snapshot matrix; the multidimensional target optimization model is iteratively solved according to the adjusted and updated target constraint matrix to output the adaptive optimal transaction matching scheme corresponding to the current time.

[0016] Optionally, the feature snapshot matrix contains dynamic features of each aggregation entity, including cost features, revenue features, output capacity, and market fluctuation features.

[0017] Optionally, key indicators affecting the constraints are calculated from the feature snapshot matrix, including price fluctuation impact factor, load fluctuation impact factor, transaction execution deviation factor, and energy storage availability factor; the extracted key indicators are mapped to the corresponding parameters in the target constraint matrix, and the constraint parameters are adjusted and updated to obtain the adjusted and updated target constraint matrix.

[0018] Optionally, an influence transformation relationship between key indicators and constraint parameters is established; based on the influence transformation relationship, the extracted key indicators are transformed and calculated to obtain real-time constraint correction results; the mapping transformation is repeatedly performed on multiple aggregate subjects and constraint dimensions to form an adjusted and updated target constraint matrix.

[0019] Optionally, physical market rule verification is performed on the parameters in the adjusted and updated target constraint matrix. The physical market rule verification includes whether the upper and lower limits of output exceed the equipment capacity, whether the energy storage charging and discharging limits meet the SOC safety range, and whether the market transaction capacity exceeds the system's allowed upper limit. Boundary correction is performed on the constraint parameters that do not meet the physical market rule verification to determine the adjusted and updated target constraint matrix.

[0020] The technical solution provided in this application has at least the following technical effects or advantages: By collecting transaction participation data from multiple aggregate entities, feature extraction is performed on each transaction participation data to obtain an aggregate entity feature set. Based on the aggregate entity feature set, a multi-dimensional objective optimization model is constructed with cost, benefit, and stability as multiple objectives, and output constraints, time series constraints, and market rule constraints are introduced to form an objective constraint matrix. Under the constraints of the objective constraint matrix, a multi-objective optimization algorithm is used to iteratively solve the multi-dimensional objective optimization model to generate the optimal transaction matching scheme for each aggregate entity. In other words, by constructing a multi-dimensional objective optimization model that integrates cost, benefit, and stability, and introducing an objective constraint matrix that integrates output, time series, and market rules for iterative solving and dynamic feedback, market transaction matching of different aggregate entities is achieved, improving the feasibility of market transaction matching schemes.

[0021] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating the aggregated entity market transaction matching method for multidimensional objective constraints proposed in this application.

[0024] Figure 2 This is a flowchart illustrating the process of constructing a multi-dimensional objective optimization model in the aggregated subject market transaction matching method oriented towards multi-dimensional objective constraints in this application. Detailed Implementation

[0025] This application addresses the technical problem in existing technologies where simplified static optimization models struggle to handle complex multi-objective conflicts, leading to poor feasibility of matching schemes in practice. By providing a multi-dimensional objective constraint-based market transaction matching method, this application solves the problem of poor feasibility of existing matching schemes due to the difficulty in handling complex multi-objective conflicts. The method constructs a multi-dimensional objective optimization model that integrates cost, benefit, and stability, and introduces an integrated output, time series, and market rule objective constraint matrix for iterative solving and dynamic feedback. This enables market transaction matching of different aggregated entities, improving the feasibility of market transaction matching schemes.

[0026] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0027] For examples, please refer to the appendix. Figure 1 This application provides a method for matching aggregated entity market transactions under multi-dimensional objective constraints, wherein the method specifically includes the following steps: Collect transaction participation data from multiple aggregate entities, extract features from each transaction participation data, and obtain the feature set of the aggregate entity.

[0028] Furthermore, this application also includes the following steps: collecting historical transaction data and real-time transaction information of multiple aggregation entities, including output, load, energy storage status, participation period and transaction frequency; preprocessing the collected historical transaction data and real-time transaction information of multiple aggregation entities, including denoising, normalization and missing value compensation; calculating the average output, peak power, load flexibility, revenue volatility and transaction execution deviation of the preprocessed historical transaction data and real-time transaction information of multiple aggregation entities to obtain characteristic indicators reflecting multiple aggregation entities, and constructing the feature set of the aggregation entities.

[0029] Specifically, an aggregator is an entity participating in electricity market transactions. It aggregates multiple distributed energy resources into a single unit to participate in market transactions, and each aggregator can participate independently. Transaction participation data includes output data, load data, energy storage status data, and transaction data. Historical transaction data represents the entire process of transactions completed by the aggregator in past market cycles, used to characterize its long-term behavior, while real-time transaction information refers to the latest status within the current transaction cycle.

[0030] Because the raw data collection process contains noise, inconsistent dimensions, or missing values, preprocessing is necessary. Statistical methods are used to identify and remove abnormal data points caused by sensor malfunctions or communication interference; indicators of different dimensions and magnitudes are normalized, unifying indicators such as output, load, and energy storage status to the range [0,1]; and missing portions of data records are appropriately filled using interpolation and mean shifting to ensure the integrity of the dataset.

[0031] After data preprocessing, characteristic indicators of the aggregation entity are generated through statistical calculations. Average output is obtained by averaging the output over each time period; peak power is obtained by taking the maximum output over each time period; load flexibility is obtained by calculating the difference between maximum and minimum load; and revenue volatility is obtained by calculating the standard deviation of daily or time-period revenue, with a larger value indicating greater revenue instability and higher risk. The transaction execution deviation rate is calculated as |(actual total output - total awarded electricity)| / total awarded electricity * 100%, where actual total output is the total electricity actually delivered to the grid by the aggregation entity, and the total awarded electricity is the amount of electricity allocated to an aggregation entity in the electricity market transaction that must be executed. The difference reflects the degree of deviation from the contract. These indicators are organized into aggregation entity feature sets, with each entity's feature vector containing multiple dimensional indicators. For example, suppose data from three aggregation entities over the past 7 days is collected, including virtual power plant A, load aggregator B, and energy storage power station C. It is found that the output data of virtual power plant A at a certain time point is suddenly recorded as -100MW due to communication interference, which is obviously wrong. The noise reduction process identifies it as an outlier and removes it. The load data of load aggregator B is recorded in kW, while the other entities are recorded in MW. Before normalization, they are all converted to MW. The SOC data of energy storage power station C is missing at 3:00 am. Based on the SOC of 80% at 2:55 am and 79% at 3:05 am, the SOC at 3:00 am should be 79.5% by linear interpolation, thus completing the missing value compensation. For virtual power plant A, the average output is 2.8MW, the peak power is the historical maximum of 4.9MW, the load flexibility is ±0.6MW, the standard deviation of its 7 daily revenue values ​​is 12500, and the transaction execution deviation rate is 2.5%. For load aggregator B, the average output is 2.2MW, the peak power is 3.1MW, the load flexibility is the maximum reducible power, i.e., -1.0MW, the standard deviation of its 7 daily revenue values ​​is 8200, and the transaction execution deviation rate is 1.8%. For energy storage power station C, which is basically in charge-discharge balance, the average output is 0MW, the peak power is 10MW, the load flexibility is ±10MW, the standard deviation of its 7 daily revenue values ​​is 15000, and the transaction execution deviation rate is 0.5%. The aggregation entity feature set comprehensively reflects the aggregation entity's output level, flexibility, revenue stability, and execution accuracy.

[0032] Based on the aggregated subject feature set, a multi-dimensional objective optimization model is constructed with cost, benefit, and stability as multiple objectives, and output constraints, time series constraints, and market rule constraints are introduced to form an objective constraint matrix.

[0033] Further details are attached. Figure 2As shown, this application further includes the following steps: establishing an evaluation function relationship between the feature vectors in the feature set of the aggregated entities and the multi-dimensional objectives, with cost, revenue, and stability as multi-dimensional objectives, wherein stability is used to quantify the stability of the trading strategies of multiple aggregated entities under market fluctuations and the consistency of matching strategies, including the rate of change of trading duration, trading volatility, and probability of default loss; based on the evaluation function relationship, constructing a multi-objective evaluation function with the objectives of minimizing cost, maximizing revenue, and maximizing stability; and constructing the multi-dimensional objective optimization model with the highest evaluation fitness of the multi-objective evaluation function as the optimization objective, for generating the optimal trading matching scheme for multiple aggregated entities.

[0034] Furthermore, this application also includes the following steps: Based on the power generation capacity, energy storage capacity, and available output range of multiple aggregation entities, setting the minimum and maximum output of each entity in the current time period to determine the output constraints; based on the continuity requirements of the trading strategies of multiple aggregation entities and the output change rate limit, setting the upper limit of output change over a continuous time period to determine the time-series constraints; based on market trading rules, price limits, and trading capacity limits, setting the maximum and minimum trading volumes that multiple aggregation entities can participate in in the market to determine the market rule constraints; arranging the output constraints, time-series constraints, and market rule constraints according to the entity and constraint type, performing matrix representation to obtain the target constraint matrix, which is used to constrain the solution of the multi-objective optimization model.

[0035] Specifically, costs are the expenditures of the aggregator in market transactions, including power generation or energy storage operation costs, electricity purchase costs, dispatch fees, etc.; revenues are the benefits obtained by the aggregator through participating in market transactions, including electricity sales revenue, incentive subsidies, or peak-valley arbitrage revenue; stability is used to quantify the smoothness of the aggregator's trading strategy under market fluctuations and the consistency of matching strategies, including the rate of change of trading duration, trading volatility, and probability of default. The rate of change of trading duration is the ratio of the magnitude of changes in output or load during continuous trading periods, trading volatility is the standard deviation of the fluctuations in trading revenue or output, and probability of default is the probability that the execution of the transaction deviates from the commitment or defaults.

[0036] For each aggregate entity's feature vector, calculate its cost, revenue, and stability indices under different trading strategies. The cost function can be constructed based on power output, electricity price, energy storage charging and discharging losses, and dispatching costs. The revenue function can be constructed based on power output sales revenue and market prices. The stability function integrates the rate of change of trading duration, trading volatility, and probability of default, such as stability score = w1*(1-rate of change of trading duration) + w2*(1-trading volatility) + w3*(1-probability of default), where w1, w2, and w3 are weighting coefficients that can be set according to the strategy priorities. The evaluation function is a functional relationship that maps the aggregate entity's feature vector to cost, revenue, and stability objectives, used to evaluate the optimization effect under different feature combinations.

[0037] A multi-objective evaluation function is constructed based on minimizing cost, maximizing revenue, and maximizing stability. This function is defined as: Evaluation Fitness = a * Revenue - b * Cost + c * Stability Score, where a, b, and c are weighting coefficients reflecting preferences for different objectives. The highest evaluation fitness of the multi-objective evaluation function is used as the optimization objective to construct a multi-dimensional objective optimization model. This model generates optimal transaction matching schemes for multiple aggregated entities. High fitness indicates that the transaction strategy performs better in terms of cost, revenue, and stability. The multi-dimensional objective optimization model takes the feature set of the aggregated entities as input, combines the multi-objective evaluation function and constraints, and generates the optimal transaction matching scheme through an optimization algorithm.

[0038] For example, assuming that the cost of Option 1 is 500, the revenue is 12000, and the stability of the transaction's rate of change, transaction volatility, and probability of default are 0.05, 0.08, and 0.02 respectively, then the stability is 1 / 3*(1-0.05)+1 / 3*(1-0.08)+1 / 3*(1-0.02)=0.95, and the evaluation fitness is 0.4*12000-0.2*500+0.4*0.95=4700.38.

[0039] Output constraints are calculated based on the real-time status of each aggregation entity. Based on the aggregation entity's power generation capacity and energy storage capacity, combined with the current energy storage status and available output range, the minimum and maximum output of each entity in the current time period are determined. The available output range is the power range that each entity can actually output in the current time period, determined by both power generation capacity and energy storage status constraints, and includes both minimum and maximum output. Output constraints ensure that the actual output of an aggregation entity within a certain time period does not exceed the maximum available power and does not fall below the minimum output, guaranteeing equipment safety and compliance with market rules.

[0040] Based on the continuity requirements and output change rate limits of multiple aggregated trading strategies, an upper limit is set for output changes over consecutive time periods, resulting in time-series constraints. Time-series constraints limit output changes within consecutive time periods, including output continuity requirements and upper limits for output change rates. These constraints limit the output change rates between adjacent time periods, preventing short-term sharp fluctuations and ensuring the executability of trading strategies. Market trading rules are the participation, settlement, and compliance regulations established by the market operator, including minimum matching units, minimum / maximum order sizes, matching frequency, quotas, and priorities. Price limits are the lowest / highest allowed price ranges in the market; for example, prices cannot fall below the minimum floor price or exceed the maximum limit, usually expressed in currency / energy units. Trading capacity limits are the upper limit of energy / power allowed to be traded within a single or cumulative time period. Extract the hard rules on participation and pricing from the market rules text, including the minimum transaction volume per order, the maximum tradable volume in a single time period, the cumulative upper limit for each entity in a certain time period or day, the lower and upper limits of pricing, and possible tiered or segmented rules. Determine the maximum and minimum transaction volumes that multiple aggregate entities can participate in in the market, that is, the minimum and maximum transaction power that the market allows each aggregate entity to submit or be accepted in a certain time period or scheduling cycle, and obtain the market rule constraints.

[0041] Output constraints, timing constraints, and market rule constraints are arranged according to subject and constraint type, and written in standard linear form to obtain the objective constraint matrix. This matrix is ​​used to solve the multi-objective optimization model. After each optimization, it is checked whether the candidate solution satisfies all the conditions in the matrix, thus ensuring that any generated transaction matching scheme is physically feasible, temporally consistent, and legally compliant. Output constraints and timing constraints ensure that the generated strategy meets the requirements of equipment capacity and continuity. Market rule constraints ensure that the transaction scheme is within the scope of legal pricing, trading volume, and capacity, avoiding the risk of violations.

[0042] Under the constraints of the objective constraint matrix, a multi-objective optimization algorithm is used to iteratively solve the multi-dimensional objective optimization model to generate the optimal transaction matching scheme for each aggregation entity.

[0043] Furthermore, this application also includes the following steps: using the target constraint matrix as a constraint condition, evaluating the target based on the multidimensional target optimization model, wherein, randomly collecting empirical transaction matching schemes for target fitness evaluation, adjusting the optimization variables based on the evaluation results, and obtaining optimization solution curves for each dimension; searching for variables in each dimension based on the optimization solution curves for each dimension, updating the optimization solution curves for each dimension, and iteratively searching until the transaction matching scheme with the highest total evaluation value of the optimization solution curves for each dimension is found, thereby obtaining the optimal transaction matching scheme.

[0044] Furthermore, this application also includes the following steps: monitoring the direction of change of the optimization solution curves of each dimension in each iteration, including the gradient or trend of cost, benefit and stability objectives; grouping strategies with consistent or similar curve change directions into the same partition, and grouping strategies with significantly different change directions into other partitions; adopting differentiated search strategies for different partitions, wherein increasing search diversity for partitions with large fluctuations in change direction or unstable gradients, and accelerating convergence for partitions with stable change direction or consistent gradients; after each iteration, readjusting the partition division according to the curve change trend, continuously optimizing the search strategy, and when the search objective or number of searches is reached, outputting the transaction matching scheme with the highest global evaluation result to obtain the optimal transaction matching scheme.

[0045] Specifically, the objective constraint matrix is ​​used as the constraint condition, and a multi-dimensional objective optimization model is used as the solution object. Before the solution begins, several empirical trading matching schemes are randomly selected from historical market data and strategy patterns as initial solutions to improve the initial quality of the search process. In each iteration, the objective fitness of each initial solution is calculated sequentially, that is, the comprehensive fitness value is calculated based on the three objective functions of cost, benefit, and stability. Based on the evaluation results, i.e., the objective fitness evaluation value, optimization variables such as output per time period, trading volume, and pricing strategy are adaptively adjusted, including selection, crossover, and mutation. Schemes with high fitness are preferentially selected as parents. The output arrangements of different schemes in a specific time period are exchanged through crossover operations, and the output value of a certain subject is randomly fine-tuned through mutation operations, thereby generating a new generation of offspring population.

[0046] Solutions with higher fitness are retained and further participate in the next round of search, while solutions with lower fitness are replaced or optimized, thus gradually forming solution curves in various dimensions that reflect the optimization trend. These solution curves are used to record and visualize the search progress during the operation of the multi-objective optimization algorithm.

[0047] Based on the optimization curves of each dimension, a search for variables in each dimension is conducted. Through hundreds or thousands of iterative searches, the solution space is continuously explored and expanded, causing the optimization curves of each dimension to gradually converge. When the optimization curves of each dimension no longer significantly improve or reach the maximum number of iterations, the iterative search is stopped, and the transaction matching scheme with the highest overall evaluation value of the optimization curves of each dimension is selected as the optimal transaction matching scheme.

[0048] For example, assume an initial population of 100 candidate schemes is randomly generated. Virtual power plant A's initial randomized output plans are 1.5MW, 3.0MW, 5.0MW, and 2.0MW, while load aggregator B's initial randomized output plans are 5.0MW, -3.0MW, 8.0MW, and 6.0MW, with negative values ​​representing charging. Check whether the output for each time period falls within the output constraints, timing constraints, and market rule constraints specified by the target constraint matrix. Assume the scheme passes the check. Calculate the scheme's revenue (8200), cost (300), and stability score (7.5), resulting in a fitness score of 3223. From the 100 schemes, select the 20 with the highest fitness as elite parents. Randomly select two parent schemes for crossover operations to generate offspring. Apply a small perturbation to one of the offspring schemes. The newly generated offspring schemes also need to undergo feasibility verification using the target constraint matrix. After 500 iterations, the optimization curve shows that the fitness steadily increased from the initial 3200 to 4250 and then plateaued, indicating that the algorithm has converged. The optimal trading matching scheme selected from the final population is as follows: the output plan of virtual power plant A is 2.0MW, 4.0MW, 4.2MW, and 5.0MW, and the output plan of load aggregator B is -2.0MW, 5.0MW, 8.0MW, and 5.0MW. This scheme has a revenue of 10900, a cost of 480, and a stability score of 8.8, so the evaluation fitness is 4267.52.

[0049] In each iteration, the changing trends of all candidate strategies in the three dimensions of cost, benefit, and stability are monitored. Strategies with similar improvement directions are grouped into the same partition, while those with very different improvement directions are grouped into different partitions.

[0050] Differentiated search strategies are implemented for different partitions. For partitions with large fluctuations in the direction of change and unstable gradients, the search diversity is increased by adding random perturbations or expanding the search range during the optimization process to avoid getting trapped in local optima, such as increasing the mutation rate to explore more possibilities. For partitions with stable directions of change and consistent gradients, convergence is accelerated by reducing the search range and refining the search to quickly approach the optimal value, such as using a more greedy selection and a smaller variation length for refined development. After each iteration, the partitions are readjusted according to the new trends, forming a continuously self-optimizing search loop. When the termination condition is met, i.e., the search objective is reached or the maximum number of searches is reached, the solution with the highest global evaluation result is selected from all partitions as the final optimal transaction matching solution. By partitioning and dynamically adjusting the search strategy, optimization avoids getting trapped in local optima, effectively distinguishes between steady-state evolution strategies and fluctuating strategies, and improves optimization convergence efficiency.

[0051] Furthermore, this application also includes the following steps: collecting real-time market transaction feedback information and identifying the difference in the transaction matching scheme; performing dynamic compensation and correction based on the difference identification results, and performing adaptive matching.

[0052] Furthermore, this application also includes the following steps: the real-time transaction feedback information includes market electricity price fluctuation data, system load change data, transaction execution status information of multiple aggregate entities, and market clearing data; the collected real-time transaction feedback information is synchronized and timestamp calibrated to ensure time sequence consistency of each data source; the real-time transaction feedback information is aligned with the market collection data corresponding to the optimal transaction matching scheme, and the fluctuation difference between each transaction data is calculated.

[0053] Specifically, during the real-time operation phase of the electricity market, real-time market transaction feedback information is collected, including market electricity price fluctuation data, system load change data, transaction execution status information of multiple aggregate entities, and market clearing data. Real-time market transaction feedback information is a dynamic data stream obtained in real-time from the market environment during the transaction execution phase, reflecting the difference between planned and actual operating conditions. Market electricity price fluctuation data is the curve of market electricity price changes over different time periods, reflecting instantaneous changes in supply and demand; system load change data is the change in total electricity consumption of the actual total load of the power grid over different time periods; aggregate entity transaction execution status information describes whether each entity is supplying / purchasing electricity according to plan, including the actual output / consumption power and energy storage SOC of each aggregate entity, used to verify whether they are fulfilling their contractual obligations; market clearing data is official settlement information released by the market operator, including the final cleared volume, cleared price, and possible deviation assessment fees.

[0054] Because these data originate from different hardware and software systems, their timestamps may have millisecond-level deviations or different sampling frequencies, thus requiring rigorous data synchronization and timestamp calibration. All real-time transaction feedback data undergoes unified timestamp calibration and synchronization processing to ensure consistency across different sources at the same point in time. The real-time transaction feedback information is aligned with the market data corresponding to the aforementioned optimal transaction matching scheme, comparing the differences between the two. Specifically, at each identical moment, what actually happened is compared to what was initially expected, and volatility differences are calculated for cost, revenue, and stability-related indicators to reflect the degree of deviation between the actual execution and the planned matching scheme. The volatility difference is the difference between the actual value and the predicted value calculated after the alignment operation. For example, suppose an optimal scheme for virtual power plant A to output 4.0MW during the 14:00-15:00 period is generated, based on a predicted electricity price of 450 yuan / MWh and a predicted load of 2500MW. The actual electricity price at 14:05 was 480 yuan / MWh (timestamp 14:05:00.123); the actual load at 14:05 was 2550MW (timestamp 14:05:00.456); and the actual output of virtual power plant A was 3.7MW at 14:05 (timestamp 14:05:01.001). All these timestamps were truncated to the minute-level time of 14:05:00 to complete data synchronization and ensure time-series consistency. Forecast data for 14:05 was retrieved from historical records: predicted electricity price was 450 yuan / MWh, predicted load was 2500MW, and planned output of virtual power plant A was 4.0MW. After alignment, the price deviation rate is (480-450) / 450=+6.67%, the alignment deviation rate is (2550-2500) / 2500=+2.0%, and the execution deviation of the virtual power plant A transaction is (3.7-4.0) / 4.0=-7.5%.

[0055] By aligning and comparing real-time data with the original plan, it is possible to effectively distinguish whether the deviation stems from fluctuations in the external market environment or from insufficient execution capabilities within the aggregation entity. Based on this, the transaction matching plan can be adjusted to obtain the optimal transaction matching plan.

[0056] Furthermore, this application also includes the following steps: extracting market volatility feature indicators based on the difference identification results; updating the feature set of the aggregation subject based on the market volatility feature indicators to obtain the feature snapshot matrix for the current period; and performing rolling iterative solution for the transaction matching scheme of the aggregation subject based on the feature snapshot matrix to output an adaptively optimized transaction matching scheme.

[0057] Furthermore, this application also includes the following steps: the market volatility characteristic indicators include: price deviation rate, load volatility coefficient and main transaction deviation amount.

[0058] Furthermore, this application also includes the following steps: adjusting the parameters in the target constraint matrix based on the feature snapshot matrix; iteratively solving the multidimensional target optimization model according to the adjusted and updated target constraint matrix, and outputting the adaptive optimal transaction matching scheme corresponding to the current time.

[0059] Furthermore, this application also includes the following steps: the feature snapshot matrix contains the dynamic features of each aggregation entity, the dynamic features including cost features, revenue features, output capacity, and market fluctuation features.

[0060] Specifically, the difference identification result is the fluctuation difference calculated above, which is the quantitative deviation between the actual value and the predicted value. Market fluctuation characteristic indicators are extracted from it. Market fluctuation characteristic indicators include price deviation rate, load fluctuation coefficient, and entity transaction deviation. Price deviation rate is the ratio of the difference between the real-time market electricity price and the planned electricity price, reflecting unexpected changes in market price trends; load fluctuation coefficient is the ratio of the system load change to the average load, reflecting unexpected changes in the overall supply and demand balance of the system; entity transaction deviation is the absolute or relative difference between the actual output and the planned output of each aggregate entity, reflecting its own performance reliability. Key market fluctuation characteristic indicators are extracted from the difference identification result. These indicators are then used to update the static aggregate entity feature set in real time. For example, the latest price deviation rate is used to correct future revenue forecasts, and the actual output deviation of entities is used to correct their reliability scores, thereby generating a feature snapshot matrix that highly reflects the current situation.

[0061] The feature snapshot matrix is ​​a matrix representing the dynamic characteristics of each entity in the current time period, and it is updated over time. The feature snapshot matrix dynamically displays the dynamic characteristics of each aggregated entity, including cost characteristics, revenue characteristics, output capacity, and market volatility characteristics. Cost characteristics are the electricity purchase / operating costs updated based on real-time electricity prices; revenue characteristics are the potential electricity sales revenue updated based on real-time electricity prices; output capacity is the real-time available generating power updated based on actual weather, equipment failures, or energy storage SOC; market volatility characteristics include the aforementioned price deviation rate, reflecting the uncertainty of the environment in which each aggregated entity operates.

[0062] The parameters in the objective constraint matrix are adjusted based on the feature snapshot matrix, and the updated constraints are used to iteratively solve the multidimensional objective optimization model again, thereby generating the adaptive optimal transaction matching scheme for the current period.

[0063] Furthermore, this application also includes the following steps: calculating key indicators affecting constraints from the feature snapshot matrix, including price fluctuation impact factor, load fluctuation impact factor, transaction execution deviation factor, and energy storage availability factor; mapping the extracted key indicators to the corresponding parameters in the target constraint matrix, adjusting and updating the constraint parameters to obtain the adjusted and updated target constraint matrix.

[0064] Furthermore, this application also includes the following steps: establishing the influence transformation relationship between key indicators and constraint parameters; performing transformation calculations on the extracted key indicators based on the influence transformation relationship to obtain real-time constraint correction results; and repeatedly performing mapping transformations on multiple aggregation subjects and constraint dimensions to form an adjusted and updated target constraint matrix.

[0065] Specifically, key indicators influencing constraints are calculated from the real-time updated feature snapshot matrix, including the price volatility impact factor, load volatility impact factor, trade execution deviation factor, and energy storage availability factor. The price volatility impact factor aims to capture the degree to which market price deviations from expectations stimulate the willingness and ability of entities to exert force. It quantifies the impact of abnormal market price fluctuations on the output decisions of aggregate entities. Large positive price fluctuations indicate that aggregate entities should be allowed to increase their maximum output to obtain more returns. If prices are consistently and significantly higher than predicted and fluctuate wildly, the price volatility impact factor will increase significantly, indicating that the market is strongly incentivizing aggregate entities to increase their output cap. The load volatility impact factor quantifies the impact of abnormal fluctuations in the total system load on market trading rules. Large positive load fluctuations suggest that the system may require more resource support, thus allowing for a more relaxed trading capacity limit to attract more resources to participate in balancing. The trade execution deviation factor quantifies the impact of the aggregate entity's own performance reliability on trading continuity constraints. A larger trade execution deviation factor indicates that the entity's behavior is more unpredictable, and the risk of its submitted trading strategies is higher. Therefore, it is necessary to limit the flexibility of its behavior and force its strategies to be smoother to protect the system and other market participants. The energy storage availability factor is used to quantify the impact of the current state of the energy storage system on its charging and discharging capabilities. When the SOC is in the middle ideal range, it means that the energy storage can be charged and discharged at full power; when the SOC is extremely high, the charging capability should be limited; when the SOC is extremely low, the discharging capability should be limited.

[0066] The system invokes preset influence transformation relationships, uses extracted key indicators as input, performs transformation calculations, and directly outputs real-time correction results for specific parameters in the target constraint matrix. Minimum output is adjusted to the baseline minimum output multiplied by (1 - price fluctuation impact factor), and maximum output is adjusted to the baseline maximum output multiplied by (1 + price fluctuation impact factor). When market price signals are positive, it not only encourages entities to generate more power but also allows them to reduce output more significantly. Minimum energy storage is adjusted to the baseline minimum energy storage multiplied by (1 - energy storage availability factor), and maximum energy storage is adjusted to the baseline maximum energy storage multiplied by (1 + energy storage availability factor). In practical applications, the energy storage availability factor is selectively applied based on the SOC level.

[0067] The baseline trading continuity constraint is multiplied by (1 - trading execution deviation factor) to obtain a new constraint. Trading continuity represents the ramp-up rate limit or the maximum change in output between adjacent periods. For entities with poor reliability, tightening their ramp-up rate means that their output is not allowed to change drastically in a short period of time, forcing their trading strategy to become smooth and stable. This reduces the risk of shock to the system caused by their sudden exit or large fluctuations, and is a risk control mechanism.

[0068] Market rules constrain the adjustment of trading ceilings / floors based on load fluctuation factors. When the system is under heavy load, temporarily raising the trading ceiling can incentivize more spare capacity to enter the market, serving as a safety valve mechanism to deal with emergencies.

[0069] The mapping transformation is repeatedly performed on multiple aggregate entities and constraint dimensions, and new trading volume limits are calculated for all entities. Once all constraint parameters are updated, an adjusted and updated target constraint matrix is ​​formed. The calculated price volatility impact factor, load volatility impact factor, trade execution deviation factor, and energy storage availability factor are used as raw materials and input into the influence transformation relationship to obtain real-time correction results for all constraint parameters. This process is performed synchronously on all aggregate entities and all relevant constraint dimensions. Each entity receives personalized constraint adjustments based on its unique trade execution deviation factor and energy storage availability factor, while all entities are simultaneously affected by global factors, namely the price volatility impact factor and trade execution deviation factor. All corrected parameters are reassembled to form an adjusted and updated target constraint matrix, accurately defining what can and cannot be done at the current moment. The solver reruns the optimization algorithm based on the adjusted and updated target constraint matrix. Because the search space has been intelligently corrected and guided, the solver can find the best solution suitable for the current rapidly changing environment more quickly and accurately—that is, the adaptive optimal trading matching solution.

[0070] By incorporating market volatility and the reliability of entity execution into the constraint layer as quantitative factors, the feasible domain of the optimization problem can reflect the real risks in real time, thereby generating a safer, more feasible, and reasonable matching solution under the current market conditions.

[0071] Furthermore, this application also includes the following steps: verifying the parameters in the adjusted and updated target constraint matrix using physical market rules, wherein the physical market rule verification includes whether the upper and lower limits of output exceed the equipment capacity, whether the energy storage charging and discharging limits comply with the SOC safety range, and whether the market transaction capacity exceeds the system's allowed upper limit; and performing boundary correction on the constraint parameters that do not meet the physical market rule verification to determine the adjusted and updated target constraint matrix.

[0072] Specifically, the parameters in the adjusted and updated target constraint matrix are verified using physical market rules. This includes verifying whether the upper and lower limits of output exceed the equipment capacity, whether the energy storage charging and discharging limits comply with the SOC safety range, and whether the market trading capacity exceeds the system's allowed upper limit. Verifying whether the upper and lower limits of output exceed the equipment capacity verifies the physical hard limit of the aggregation entity. No matter how strong the price-based incentives are, the planned output of any entity cannot exceed the total nameplate capacity of all its power generation equipment, and the planned output reduction cannot be lower than its minimum technical output. For example, a gas turbine unit with a maximum capacity of 10MW and an absolute minimum technical output of 2MW, after dynamic adjustment, has a maximum output of 10.5MW, exceeding the 10MW upper limit and violating the physical equipment capacity.

[0073] Whether the energy storage charge and discharge limits comply with the SOC safety range ensures that dispatch commands for energy storage do not cause its state of charge to exceed the manufacturer's specified safety window, thereby protecting battery health and extending its lifespan. The charge and discharge power commands in the simulation matrix are executed to predict the SOC value at the end of the dispatch cycle. For example, if the SOC safety range is 20% lower and 90% upper, and the matrix commands the energy storage to discharge at its maximum power of 8MW within the next hour, executing this command will predict that the SOC will decrease from the current 25% to 18%. Since 18% < 20%, the SOC will enter the danger zone.

[0074] Whether the market trading capacity exceeds the system's permissible limit refers to the fact that even if all participants have a strong willingness to trade, their total trading volume cannot exceed the transmission capacity limit of the power grid at a specific node or region; otherwise, transmission congestion will occur. The planned trading volumes of all participating entities at a critical node or region are summed to obtain the total trading power, which is then compared with the system's permissible limit given by the power grid operator. For example, if the system's permissible limit for a certain region is 100MW, and the total trading volume of all participants in that region, after adjustment for load fluctuation factors, reaches 105MW, exceeding the system's permissible limit of 100MW, it may cause line overload.

[0075] When the verification process detects parameters that do not meet the rules, a precise and automated boundary correction procedure is initiated. For any parameter exceeding the verification rules, the system ignores its previous dynamic adjustments and directly corrects its value to the closest legal boundary value. After verification and precise correction, a new target constraint matrix that is both dynamically intelligent and absolutely safe and compliant is determined. This updated target constraint matrix is ​​used to generate an adaptive optimal transaction matching scheme.

[0076] The revised constraints are strictly within the rated capacity of the equipment and the SOC of the energy storage, avoiding overpower or overcharge / over-discharge events during actual execution. Transaction volume is strictly controlled within the allowable limits for a single entity and the system, preventing transaction settlement failures or market penalties. The constraint matrix, validated by physical market rules, ensures that the multidimensional objective optimization solver will not generate infeasible solutions, improving convergence efficiency.

[0077] In summary, the aggregated entity market transaction matching method oriented towards multi-dimensional objective constraints provided in this application has the following technical effects: By collecting transaction participation data from multiple aggregate entities, feature extraction is performed on each transaction participation data to obtain an aggregate entity feature set. Based on the aggregate entity feature set, a multi-dimensional objective optimization model is constructed with cost, benefit, and stability as multiple objectives, and output constraints, time series constraints, and market rule constraints are introduced to form an objective constraint matrix. Under the constraints of the objective constraint matrix, a multi-objective optimization algorithm is used to iteratively solve the multi-dimensional objective optimization model to generate the optimal transaction matching scheme for each aggregate entity. In other words, by constructing a multi-dimensional objective optimization model that integrates cost, benefit, and stability, and introducing an objective constraint matrix that integrates output, time series, and market rules for iterative solving and dynamic feedback, market transaction matching of different aggregate entities is achieved, improving the feasibility of market transaction matching schemes.

[0078] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0079] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for matching aggregated entities in a market transaction based on multi-dimensional objective constraints, characterized in that: include: Collect transaction participation data from multiple aggregate entities, extract features from each transaction participation data, and obtain the feature set of the aggregate entity; Based on the aforementioned aggregated subject feature set, a multi-dimensional objective optimization model is constructed with cost, benefit, and stability as multiple objectives, and output constraints, time series constraints, and market rule constraints are introduced to form an objective constraint matrix; Under the constraints of the objective constraint matrix, a multi-objective optimization algorithm is used to iteratively solve the multi-dimensional objective optimization model to generate the optimal transaction matching scheme for each aggregation entity.

2. The aggregation entity market transaction matching method oriented towards multi-dimensional objective constraints according to claim 1, characterized in that, Also includes: Collect real-time market transaction feedback information and perform differential identification on the transaction matching scheme; Dynamic compensation and correction are performed based on the difference recognition results, and adaptive matching is then carried out.

3. The aggregation entity market transaction matching method oriented towards multi-dimensional objective constraints according to claim 2, characterized in that, Dynamic compensation and correction are performed based on the difference recognition results, and adaptive matching is carried out, including: Based on the differential identification results, market volatility characteristic indicators are extracted; The aggregated subject feature set is updated based on the market volatility characteristic indicators to obtain the feature snapshot matrix for the current period; Based on the feature snapshot matrix, the transaction matching scheme of the aggregation subject is solved by rolling iteration, and an adaptively optimized transaction matching scheme is output.

4. The aggregation entity market transaction matching method oriented towards multi-dimensional objective constraints according to claim 3, characterized in that, The market volatility characteristic indicators include: price deviation rate, load volatility coefficient, and main transaction deviation amount.

5. The aggregation entity market transaction matching method oriented towards multi-dimensional objective constraints according to claim 3, characterized in that, Collect real-time market transaction feedback information and perform differential identification on the transaction matching scheme, including: The real-time transaction feedback information includes market electricity price fluctuation data, system load change data, transaction execution status information of multiple aggregated entities, and market clearing data. The collected real-time transaction feedback information is synchronized and timestamp calibrated to ensure the time sequence consistency of each data source. The real-time transaction feedback information is aligned with the market data collected for the optimal transaction matching scheme, and the volatility difference between each transaction data is calculated.

6. The aggregation entity market transaction matching method oriented towards multi-dimensional objective constraints according to claim 3, characterized in that, Based on the aforementioned feature snapshot matrix, the transaction matching scheme for the aggregation entity is solved iteratively and adaptively, resulting in an optimized transaction matching scheme, including: The parameters in the target constraint matrix are adjusted based on the feature snapshot matrix; The multidimensional objective optimization model is iteratively solved based on the adjusted and updated objective constraint matrix, and the adaptive optimal transaction matching scheme corresponding to the current time is output.

7. The aggregation entity market transaction matching method oriented towards multi-dimensional objective constraints according to claim 6, characterized in that, The feature snapshot matrix contains the dynamic features of each aggregation entity, including cost features, revenue features, output capacity, and market fluctuation features.

8. The aggregation entity market transaction matching method oriented towards multi-dimensional objective constraints according to claim 7, characterized in that, Adjusting the parameters in the target constraint matrix based on the feature snapshot matrix includes: Calculate the key indicators affecting constraints from the feature snapshot matrix, including price fluctuation impact factor, load fluctuation impact factor, transaction execution deviation factor, and energy storage availability factor; The extracted key indicators are mapped to the corresponding parameters in the target constraint matrix, and the constraint parameters are adjusted and updated to obtain the adjusted and updated target constraint matrix.

9. The aggregation entity market transaction matching method oriented towards multi-dimensional objective constraints according to claim 8, characterized in that, Mapping the extracted key indicators to the corresponding parameters in the target constraint matrix includes: Establish the impact transformation relationship between key indicators and constraint parameters; Based on the aforementioned influence transformation relationship, the extracted key indicators are transformed and calculated to obtain real-time constraint correction results; The mapping transformation is repeated on multiple aggregation subjects and constraint dimensions to form an adjusted and updated target constraint matrix.

10. The aggregation entity market transaction matching method oriented towards multi-dimensional objective constraints according to claim 9, characterized in that, After generating the adjusted and updated target constraint matrix, the following steps are also included: The parameters in the adjusted and updated target constraint matrix are verified using physical market rules. The physical market rule verification includes whether the upper and lower limits of output exceed the equipment capacity, whether the energy storage charging and discharging limits meet the SOC safety range, and whether the market transaction capacity exceeds the system's allowed upper limit. Boundary corrections are performed on constraint parameters that do not meet the physical market rule verification, and the adjusted and updated target constraint matrix is ​​determined.

11. The aggregation entity market transaction matching method oriented towards multi-dimensional objective constraints according to claim 1, characterized in that, The process of obtaining the aggregated subject feature set includes: Collect historical transaction data and real-time transaction information from multiple aggregate entities, including output, load, energy storage status, participation period, and transaction frequency; The historical transaction data and real-time transaction information collected from multiple aggregate entities are preprocessed, including noise reduction, normalization, and missing value compensation. The average output, peak power, load flexibility, revenue volatility, and transaction execution deviation are calculated from the historical transaction data and real-time transaction information of multiple aggregate entities after preprocessing, so as to obtain characteristic indicators reflecting multiple aggregate entities and construct the feature set of the aggregate entities.

12. The aggregation entity market transaction matching method oriented towards multi-dimensional objective constraints according to claim 1, characterized in that, Based on the aforementioned aggregated subject feature set, a multi-dimensional objective optimization model is constructed with cost, benefit, and stability as multiple objectives, including: Using cost, revenue, and stability as multi-dimensional objectives, an evaluation function relationship is established between the feature vectors in the feature set of the aggregated entities and the multi-dimensional objectives. Stability is used to quantify the stability of the trading strategies of multiple aggregated entities under market fluctuations and the consistency of matching strategies, including the rate of change of trading, trading volatility, and probability of loss default. Based on the aforementioned evaluation function relationship, a multi-objective evaluation function is constructed with the objectives of minimizing cost, maximizing revenue, and maximizing stability. The multi-dimensional objective optimization model is constructed with the highest evaluation fitness of the multi-objective evaluation function as the optimization objective, and is used to generate the optimal transaction matching scheme for multiple aggregate entities.

13. The aggregation entity market transaction matching method oriented towards multi-dimensional objective constraints according to claim 1, characterized in that, A target constraint matrix is ​​formed by introducing output constraints, timing constraints, and market rule constraints, including: Based on the power generation capacity, energy storage capacity, and available output range of multiple aggregate entities, the minimum and maximum output of each entity in the current time period are set to determine the output constraints; Based on the continuity requirements of the trading strategies of multiple aggregate entities and the limit on the rate of change of output, an upper limit for the change of output over a continuous period is set to determine the time-series constraint. Based on market trading rules, price limits, and trading capacity restrictions, the maximum and minimum trading volumes that multiple aggregate entities can participate in in the market are set to determine the market rule constraints. The output constraints, timing constraints, and market rule constraints are arranged according to the subject and constraint type, and represented in a matrix to obtain the target constraint matrix, which is used to constrain the solution of the multi-objective optimization model.

14. The aggregation subject market transaction matching method oriented towards multi-dimensional objective constraints according to claim 12, characterized in that, Under the constraints of the objective constraint matrix, a multi-objective optimization algorithm is used to iteratively solve the multi-dimensional objective optimization model to generate the optimal transaction matching scheme for each aggregation entity, including: Using the target constraint matrix as a constraint condition, the target is evaluated based on the multidimensional target optimization model. In this process, empirical transaction matching schemes are randomly collected to evaluate the target fitness. Based on the evaluation results, the optimization variables are adjusted to obtain the optimization solution curves for each dimension. Based on the optimization solution curves of each dimension, the variables of each dimension are searched, and the optimization solution curves of each dimension are updated. Through iterative search, the transaction matching scheme with the highest total evaluation value of the optimization solution curves of each dimension is searched, and the optimal transaction matching scheme is obtained.

15. The aggregation entity market transaction matching method oriented towards multi-dimensional objective constraints according to claim 14, characterized in that, include: Monitor the direction of change of the optimization solution curves in each iteration, including the gradient or trend of cost, benefit and stability objectives; Strategies with similar or identical curve changes are grouped into the same partition, while strategies with significantly different curve changes are grouped into other partitions. Differentiated search strategies are adopted for different partitions. For partitions with large fluctuations in the direction of change or unstable gradients, the search diversity is increased, while for partitions with stable directions of change or consistent gradients, convergence is accelerated. After each iteration, the partition division is readjusted according to the curve change trend, and the search strategy is continuously optimized. When the search target or number of searches is reached, the transaction matching scheme with the highest global evaluation result is output to obtain the optimal transaction matching scheme.

Images