A virtual power plant cross-market collaborative optimization decision method and device based on a goose optimization algorithm and a storage medium

Abstract

The application provides a virtual power plant cross-market collaborative optimization decision method and device based on a goose optimization algorithm and a storage medium, relates to the technical field of power systems and power markets, and comprises the following steps: S1, acquiring virtual power plant resource data, prediction data and market data; S2, dividing resource sets and boundary evaluation according to response speed; S3, constructing a cross-market collaborative optimization model based on the capacity boundary; S4, iteratively solving the model by using the goose optimization algorithm; and S5, generating and issuing output control instructions of each resource in the virtual power plant. The application can more accurately reflect the actual adjustable capacity of the aggregated resources by constructing a virtual power plant dynamic adjustment capacity evaluation model and combining the collaborative optimization decision mechanism of the day-ahead spot energy market and the frequency modulation auxiliary service market, realizes the unified coordination of multi-market declaration and dispatching plans, and has the advantages of more accurate adjustment capacity evaluation, stronger market collaborative optimization capacity and higher executable optimization results.

Description

Technical Field

[0001] This invention relates to the field of power system and power market technology, and in particular to a cross-market collaborative optimization decision-making method, device and storage medium for virtual power plants based on the goose optimization algorithm. Background Technology

[0002] With the continuous increase in the proportion of renewable energy and the ongoing reform of the power market, virtual power plants, by aggregating various resources such as wind power, photovoltaics, energy storage, and adjustable loads to participate in the day-ahead spot electricity market and frequency regulation ancillary service market, have become an important way to improve the system's flexible regulation capabilities and promote the consumption of renewable energy. In multi-market operation scenarios, virtual power plants not only need to formulate reasonable market application strategies, but also need to balance returns, risks, and actual execution capabilities.

[0003] However, in existing technologies, the assessment of virtual power plant regulation capacity usually adopts static capacity superposition or simplified boundary estimation methods, which are difficult to accurately reflect the response speed of different resources, continuous regulation capacity, and the impact of wind power and photovoltaic prediction errors on the adjustable capacity. This can easily lead to overly aggressive or overly conservative applications, resulting in a lack of reliable constraints on subsequent market applications and dispatch results, and insufficient actual executability.

[0004] Meanwhile, existing cross-market optimization methods mostly model the spot market and the frequency regulation market separately or use weak coupling, which fails to effectively characterize the coordinated allocation relationship between fast and slow regulation capabilities in different markets. They also cannot simultaneously take into account factors such as segmented pricing, price fluctuation risk, deviation assessment, and regulation costs. Therefore, there is an urgent need for an integrated method that can realize the dynamic regulation capability assessment of virtual power plants and cross-market collaborative optimization decision-making. Summary of the Invention

[0005] The purpose of this invention is to provide a method, device, and storage medium for cross-market collaborative optimization decision-making of virtual power plants based on the goose optimization algorithm, which solves the problems in the background art of inaccurate assessment of the dynamic adjustment capability of virtual power plants, insufficient cross-market collaborative reporting capability, and poor actual executability of optimization results.

[0006] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides a cross-market collaborative optimization decision-making method for virtual power plants based on the goose optimization algorithm, executed by an electronic device, the method comprising: S1. Obtain the resource set of the virtual power plant and its operating constraint parameters, the day-ahead power forecast data and forecast error statistics of each resource, and the price forecast data and fluctuation statistics of the day-ahead spot electricity market and frequency regulation ancillary service market. S2. Divide the resource set into a fast-regulation resource set and a slow-regulation resource set according to the resource adjustment response speed. Under the condition of considering the day-ahead power prediction data and the prediction error statistics, solve the adjustment capability evaluation model to obtain the fast adjustment capability boundary and slow adjustment capability boundary for each scheduling period. S3. Based on the fast regulation capability boundary and slow regulation capability boundary obtained in step S2, construct a cross-market collaborative optimization model for the virtual power plant to participate in both the day-ahead spot electricity market and the frequency regulation ancillary service market. The objective function of the cross-market collaborative optimization model is to maximize the expected total revenue of the virtual power plant in the two markets, and the objective function includes regulation cost and risk penalty terms. The decision variables of the cross-market collaborative optimization model include at least: the frequency regulation declaration capacity and frequency regulation declaration price for each scheduling period, and the declaration electricity volume and declaration price for each segment corresponding to the multiple quotation curves in the spot market for each scheduling period. S4. Under the conditions of satisfying the boundary constraints of adjustment capacity, the constraints of spot multi-segment price and the constraints of market rules, the cross-market collaborative optimization model is solved iteratively by using the goose optimization algorithm to obtain the frequency adjustment declaration parameters, spot multi-segment price curves and resource scheduling plans for each scheduling period. The goose optimization algorithm uses a fitness function as an evaluation metric. This fitness function includes at least the objective function value of the cross-market collaborative optimization model and a penalty term for constraint violations. The penalty coefficient λ is dynamically updated according to the following rules: in, The initial penalty coefficient, This represents the current iteration number. The maximum number of iterations, The attenuation coefficient is... The standard deviation of group fitness The average fitness of the population; S5. Based on the solution results of step S4, generate and output frequency regulation declaration information and spot price information for market reporting, and generate and issue output control instructions for each resource within the virtual power plant.

[0007] Preferably, in step S2, for wind power resources and / or photovoltaic resources belonging to the slow-adjustment resource set, uncertainty correction is performed on their adjustable power upper limit and adjustable power upper limit based on the prediction error statistics to obtain the corrected slow-adjustment capability boundary. The greater the uncertainty represented by the prediction error statistics, the greater the shrinkage of the corresponding adjustable capability boundary.

[0008] Preferably, in step S2, a continuous adjustment capability limit is imposed on the slow adjustment resource so that the adjustable power of the slow adjustment resource does not exceed the preset maximum sustainable adjustment power in multiple consecutive scheduling periods, thereby forming a slow adjustment capability boundary that includes the continuous adjustment limit.

[0009] Preferably, the adjustment capability boundary constraints in step S3 include at least: The frequency regulation capacity requested during any scheduling period shall not exceed the rapid adjustment capacity boundary of that scheduling period; The total declared electricity volume in the spot market during any scheduling period shall not exceed the sum of the slow regulation capacity boundary and the remaining fast regulation capacity for that scheduling period, wherein the remaining fast regulation capacity is the portion remaining after deducting the frequency regulation declared capacity for that scheduling period from the fast regulation capacity boundary for that scheduling period.

[0010] Preferably, the goose optimization algorithm in step S4 further includes: Leader goose selection mechanism: An exponential selection probability is constructed based on the difference between the individual fitness and the group's optimal fitness, and the leader goose is determined according to the selection probability. V-shaped formation update strategy: The position of the leader goose is updated by a combination of linear decreasing factor and random factor, and a guiding term is introduced from the individual's historical best solution and the group's global best solution; the position of the non-leader goose is updated by a neighborhood individual difference term and a random perturbation term, wherein the random perturbation term includes a Gaussian random perturbation term.

[0011] Preferably, the multi-segment price curve of the spot market satisfies the following constraints: within the same scheduling period, the cumulative declared electricity volume of each price segment increases according to the price segment number, and the declared price of each price segment remains monotonically unchanged according to the price segment number, and the number of segments of the multi-segment price curve satisfies the preset segment number range constraint. Furthermore, when constructing the cross-market collaborative optimization model and calculating the expected return in the spot market, the predicted winning bid volume is determined based on the predicted clearing price and the multi-segment bidding curve. When the predicted clearing price falls into the bidding interval corresponding to the i-th bidding segment and the (i+1)-th bidding segment, the change in the winning bid volume within the bidding interval is determined based on the proportion of the price difference of the predicted clearing price within the bidding interval to the price difference of the adjacent bidding segments, and the predicted winning bid volume is obtained by combining the volume increment of the adjacent bidding segments.

[0012] Preferably, the adjustment cost item includes market price-perceived adjustment cost, which is used to determine the unit adjustment cost of the resource by a combination of fixed adjustment cost and price-sensitive item. The price-sensitive item is related to the deviation of the predicted electricity price from the benchmark price and is characterized by a response sensitivity coefficient.

[0013] Preferably, the risk penalty items include at least: Price volatility risk penalty determined by the statistical measures of price volatility predicted by the spot market and the statistical measures of price volatility predicted by the frequency modulation market; The deviation risk penalty is determined by the deviation assessment unit price, the prediction error statistic, and the indicator function, where the indicator function is either 0 or 1 and is used to characterize whether the deviation assessment triggering conditions are met.

[0014] Secondly, the present invention provides a virtual power plant cross-market collaborative optimization decision-making device based on the goose optimization algorithm, comprising: The data acquisition module is used to perform step S1 in the method as described in the first aspect of the present invention; A capability assessment module is used to perform step S2 in the method as described in the first aspect of the present invention; A collaborative modeling module is used to perform step S3 in the method as described in the first aspect of the present invention; The goose optimization solution module is used to execute step S4 in the method described in the first aspect of the present invention, and to iteratively solve according to the penalty coefficient dynamic update rule in the method described in the first aspect of the present invention; The output and control module is used to perform step S5 in the method as described in the first aspect of the present invention.

[0015] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it is used to implement the above-mentioned cross-market collaborative optimization decision-making method for virtual power plants based on the goose optimization algorithm.

[0016] Compared with the prior art, the beneficial effects of the present invention are: I. This invention divides the aggregated resources of a virtual power plant into fast-regulating resources and slow-regulating resources according to the resource regulation response speed, and constructs a dynamic regulation capability assessment model by combining day-ahead power prediction data, prediction error statistics, and continuous regulation capability constraints. This can more accurately obtain the fast regulation capability boundary and slow regulation capability boundary for each scheduling period, thereby improving the refinement, authenticity, and practical feasibility of the virtual power plant regulation capability assessment.

[0017] Second, this invention introduces the fast and slow adjustment capability boundaries into a cross-market collaborative optimization model of the day-ahead spot electricity market and the frequency regulation ancillary service market, and incorporates the adjustment cost, price fluctuation risk penalty, and deviation risk penalty into a unified objective function. This enables integrated collaborative decision-making of spot price quotations, frequency regulation applications, and resource scheduling plans, thereby improving the virtual power plant's revenue acquisition capability, risk control capability, and resource utilization efficiency in a multi-market environment.

[0018] Third, this invention employs a goose optimization algorithm that includes dynamic penalty coefficient updates, a leader goose selection mechanism, and a V-shaped formation update strategy to iteratively solve the cross-market collaborative optimization model. This effectively balances global search capability and local convergence capability under complex constraints, improving the feasibility, stability, and efficiency of the optimization solution and the generation of application schemes. Attached Figure Description

[0019] Figure 1 This is a flowchart of a cross-market collaborative optimization decision-making method for virtual power plants based on the goose optimization algorithm of the present invention. Detailed Implementation

[0020] The following is combined Figure 1 The technical solutions in the embodiments of the present invention will be clearly and completely described herein. It should be understood that the described embodiments are only a part of the embodiments of the present invention, and not all of them. Other implementation methods obtained by those skilled in the art based on the embodiments of the present invention without creative effort are all within the protection scope of the present invention.

[0021] This embodiment provides a cross-market collaborative optimization decision-making method for virtual power plants based on the Goose Optimization Algorithm, executed by an electronic device. The method is designed for application scenarios where virtual power plants participate in both the day-ahead spot electricity market and the frequency regulation ancillary service market. It first evaluates the dynamic adjustment capability of the aggregated resources of the virtual power plant, then uses the obtained fast adjustment capability boundary and slow adjustment capability boundary as constraint inputs to the cross-market collaborative optimization model, and finally outputs reportable market declaration parameters and executable resource scheduling plans.

[0022] In this embodiment, the scheduling period is discretized into multiple sub-times. For example, it is advisable The FM market is organized according to FM time periods, with This indicates the frequency modulation period. When the frequency modulation period differs from the time scale of the spot market sub-time, a time scale conversion factor can be used. In addition to the time-time mapping relationship, the frequency regulation declaration capacity is converted to the corresponding sub-time to ensure consistency with the capacity boundary constraints and spot price constraints at the sub-time scale.

[0023] Step S1: Obtain virtual power plant resource data, forecast data, and market data; Obtain the resource set and its operating constraint parameters of the virtual power plant, the day-ahead power forecast data and forecast error statistics of each resource, and the price forecast data and fluctuation statistics of the day-ahead spot electricity market and frequency regulation ancillary service market.

[0024] The resource set may include, but is not limited to, energy storage resources, interruptible loads, adjustable industrial loads, wind power resources, photovoltaic resources, distributed gas turbine units, and other adjustable resources. The operating constraint parameters may include the maximum operating power, minimum operating power, ramp rate limit, charge / discharge state constraint, continuous adjustment time constraint, state of charge constraint, start / stop constraint, or other parameters characterizing the operating boundary of the resource.

[0025] The day-ahead power forecast data is used to characterize the predicted output or load level of each resource in subsequent scheduling cycles. The forecast error statistics are used to characterize the degree of uncertainty between the predicted and actual values ​​of the corresponding resource. The price forecast data includes the day-ahead spot electricity market price forecast and the capacity compensation-related price, mileage price, or other settlement price forecast in the frequency regulation ancillary service market; the volatility statistics are used to characterize the degree of uncertainty of the corresponding market price, such as standard deviation, variance, quantile difference, or other statistics reflecting volatility.

[0026] In one implementation, the prediction error statistics include, but are not limited to, the sample standard deviation of the error between the predicted and measured values; the price fluctuation statistics include, but are not limited to, the sample standard deviation of the predicted price series. For different types of resources, the same or different statistical methods can be used to obtain the corresponding error statistics.

[0027] Step S2: Divide resources into fast and slow categories according to their response speed and solve the dynamic adjustment capability assessment model; After obtaining the data in step S1, the resource set is divided into fast-adjustment resource sets according to the resource adjustment response speed. With slow-adjustment resource set Furthermore, considering the day-ahead power prediction data and prediction error statistics, the regulation capacity assessment model is solved to obtain the fast regulation capacity boundary and slow regulation capacity boundary for each scheduling period.

[0028] In this embodiment, let the total resource set be denoted as . The resource set is denoted as (The rest of the text is incomplete and cannot be translated.) The set of resources for slow adjustment is denoted as Then we have:

[0029] in, This can include rapidly responding resources such as energy storage and interruptible loads. This can include resources such as wind power, solar power, and other slow-adjustable loads.

[0030] For ease of modeling, the following variables are defined: and These respectively represent rapid resource adjustment At midnight The adjustable power output and the adjustable power output; and These represent slow resource adjustment. At midnight The adjustable power output and the adjustable power output; , This represents the resource capacity weighting factor; This represents the average spot price of the previous day; , These represent the fast and slow resources at sub-times, respectively. Price perception unit adjustment cost; This represents the penalty factor for fluctuations in new energy sources; This indicates that slow, uncertain resources are at a specific time step. The prediction error statistic or the standard deviation of the output uncertainty; This represents a subset of wind / solar resources in the slow-regulating resource set; Indicates wind power / solar resources At midnight The predicted power; Indicates slow resource adjustment At midnight The power level involved in slow regulation or the actual sustainable regulation power.

[0031] The aforementioned adjustment capacity assessment model employs a weighted optimization structure combining returns, risks, and adjustment costs. This ensures that the output adjustment capacity boundary not only reflects the physical adjustability space of resources but also embodies economic and uncertainty constraints. Its objective function can be expressed as:

[0032] in, Indicates the sub-time index; This indicates that the resource index can be adjusted quickly. This indicates a slow-tuning resource index; Indicates rapid adjustment of the resource set; This represents a set of resources that are adjusted slowly. This represents a subset of wind and / or solar resources in the slow-regulating resource set; Indicates rapid resource adjustment Ability weighting factor; Indicates slow resource adjustment Ability weighting factor; and These respectively represent rapid resource adjustment At midnight The adjustable power output and adjustable power output; and These represent slow resource adjustment. At midnight The adjustable power output and adjustable power output; Indicates rapid resource adjustment At midnight Price perception unit adjustment cost; Indicates slow resource adjustment At midnight Price perception unit adjustment cost; This represents the average spot price of the previous day; Indicates sub-time New energy fluctuation penalty factor; Indicates wind power and / or photovoltaic resources At midnight The prediction error statistic or the standard deviation of the output uncertainty; In another equivalent formulation, it can also be written as a comprehensive utility function. The maximization form is as follows: at each sub-time point, the weighted sum of the adjustable power of fast and slow resources is used as the capacity benefit term; the reduction in realizable capacity due to new energy prediction errors is used as the uncertainty penalty term; and the losses, depreciation, or opportunity costs of resources caused by adjustment actions are used as the price perception adjustment cost term.

[0033] in, This represents the comprehensive utility function value of the adjustment capacity assessment model; Indicates the sub-time index; This indicates that the resource index can be adjusted quickly. This indicates a slow-tuning resource index; This represents a resource index, indicating any resource in the fast-tuning resource set and the slow-tuning resource set; Indicates rapid adjustment of the resource set; This represents a set of resources that are adjusted slowly. This represents a collection of uncertain resources such as wind power and solar power. Indicates rapid resource adjustment Ability weighting factor; Indicates slow resource adjustment Ability weighting factor; and These respectively represent rapid resource adjustment At midnight The adjustable power output and adjustable power output; and These represent slow resource adjustment. At midnight The adjustable power output and adjustable power output; Indicates sub-time New energy fluctuation penalty factor; Representing uncertain resources At midnight The prediction error statistic or the standard deviation of the output uncertainty; Representing resources At midnight Price perception unit adjustment cost; This represents the average spot price of the previous day; and Representing resources At midnight The adjustable power and adjustable power. The above formula means that, at each sub-time point, the weighted sum of the adjustable power and adjustable power of fast and slow resources is taken as the capacity benefit term, the reduction in realizable capacity caused by the new energy prediction error is taken as the uncertainty penalty term, and the loss, depreciation or opportunity cost of resources caused by the adjustment action is taken as the price perception adjustment cost term.

[0034] For ease of implementation, the capability weighting factor , The virtual power plant operator can set the weighting factor based on factors such as resource response speed, ramp-up capability, and sustainable adjustment time. In a preferred embodiment, the weight of fast resources is no less than the weight of slow resources. The renewable energy fluctuation penalty factor... The settings can be configured by the operator based on historical backtesting results, risk appetite, and deviation assessment costs. The larger the value, the more conservative the competency assessment process.

[0035] (i) Capacity boundary constraints of slow-moving, uncertain resources such as wind power / solar power For new energy equipment such as wind power and photovoltaics, in scenarios where there is no wind or solar curtailment recovery, no backup capacity, or no available margin, its adjustable power capacity can be taken as zero. In scenarios where there is wind or solar curtailment recovery, backup capacity, or other available margin, its adjustable power upper limit and adjustable power upper limit can be determined based on the predicted power and the equipment operating boundary, and further uncertainty correction can be made by combining the prediction error statistics.

[0036] In a basic implementation, it can be written as:

[0037]

[0038] in, This refers to wind power and / or solar power resources in the slow-regulating resource set. At midnight Adjustable power; This refers to wind power and / or solar power resources in the slow-regulating resource set. At midnight Adjustable power; This represents the index of wind power and / or photovoltaic resources in the slow-adjustment resource set; Indicates the sub-time index; This represents a subset of wind and / or solar resources in the slow-regulating resource set; Indicates wind power and / or photovoltaic resources At midnight The predicted power; Indicates sub-time New energy fluctuation penalty factor; Indicates wind power and / or photovoltaic resources At midnight The prediction error statistic or output uncertainty standard deviation. The above formula means that for wind power and / or photovoltaic resources in the slow-adjustment resource set, in scenarios with no wind or solar curtailment recovery, no reserve capacity, or no available margin, the up-adjustable power is zero; the down-adjustable power upper limit is determined by the predicted power and the uncertainty correction term, and the larger the prediction error statistic, the smaller the down-adjustable power upper limit.

[0039] Furthermore, for slow-moving and uncertain resources such as wind power / solar power... It can be done at midnight. First, calculate the uncorrected upper and lower power limits, then base the calculation on the prediction error statistics. Perform shrinkage correction to make The larger the contraction, the greater the slow adjustment capability boundary.

[0040] Uncorrected upper and lower limits can be determined based on physical availability margins. For example, lowering the uncorrected upper limit can be defined as:

[0041] The upward adjustment of the uncorrected upper limit can be defined as:

[0042] in, Indicates wind power and / or photovoltaic resources At midnight The downward adjustment did not correct the upper limit; Indicates wind power and / or photovoltaic resources At midnight The upward adjustment did not correct the upper limit; Indicates wind power and / or photovoltaic resources At midnight The predicted power; and Representing resources Minimum and maximum permissible operating power; This represents an index of wind power and / or photovoltaic resources; This represents the sub-time index. For scenarios without power constraints and no backup margin, increasing the uncorrected upper limit is acceptable. For scenarios with power rationing recovery, reserve capacity, or other available margins, the uncorrected upper limit can be adjusted upwards by taking a positive value as described above.

[0043] Based on this, the boundary can be corrected using a linear shrinkage method, namely:

[0044]

[0045] in, Indicates wind power and / or photovoltaic resources At midnight The adjustable power limit after linear contraction correction; Indicates wind power and / or photovoltaic resources At midnight The adjustable power limit after linear contraction correction; Indicates wind power and / or photovoltaic resources At midnight The downward adjustment did not correct the upper limit; Indicates wind power and / or photovoltaic resources At midnight The upward adjustment did not correct the upper limit; and Representing resources The downward adjustment of the boundary contraction coefficient and the upward adjustment of the boundary contraction coefficient; Indicates wind power and / or photovoltaic resources At midnight The prediction error statistic or the standard deviation of the output uncertainty; This represents an index of wind power and / or photovoltaic resources; Indicates the sub-time index.

[0046] Alternatively, a correction can be made using a confidence interval shrinkage method, i.e.:

[0047]

[0048] in, Indicates wind power and / or photovoltaic resources At midnight The adjustable power limit is based on the confidence interval contraction correction; Indicates wind power and / or photovoltaic resources At midnight The adjustable power limit is based on the confidence interval contraction correction; Indicates wind power and / or photovoltaic resources At midnight The predicted power; Represents the confidence coefficient; Indicates wind power and / or photovoltaic resources At midnight The prediction error statistic or the standard deviation of the output uncertainty; and Representing resources Minimum and maximum permissible operating power; This represents an index of wind power and / or photovoltaic resources; This represents the sub-time index. Regardless of whether linear shrinkage or confidence interval shrinkage is used, the following condition is met: The larger the value, the smaller the resulting adjustable upper and lower limits, meaning the capacity boundary contraction monotonically increases with increasing uncertainty. Substituting the revised upper and lower power limits into subsequent capacity constraints allows the adjustable boundaries of wind / solar resources to adaptively change with uncertainty at each sub-time, thus avoiding nominally adjustable but actually unrealizable declaration results.

[0049] (ii) Capacity boundary constraints for other fast and slow resources For energy storage, interruptible loads, and other fast and slow adjustment resources that are not wind and solar fluctuating resources, their up-adjustment and down-adjustment capabilities are jointly constrained by the maximum operating power, minimum operating power, and predicted power of the resource.

[0050] For rapid resource adjustment ,have:

[0051]

[0052] For slow-adjustment resources ,have:

[0053]

[0054] in, and These respectively represent rapid resource adjustment At midnight The adjustable power output and adjustable power output; and These represent slow resource adjustment. At midnight The adjustable power output and adjustable power output; and These respectively represent rapid resource adjustment Maximum operating power and minimum operating power; and These represent slow resource adjustment. Maximum operating power and minimum operating power; Indicates rapid resource adjustment At midnight The predicted power; Indicates slow resource adjustment At midnight The predicted power; Indicates rapid adjustment of the resource set; This represents a set of resources that are adjusted slowly. This indicates that the resource index can be adjusted quickly. This indicates a slow-tuning resource index; Indicates the sub-time index.

[0055] (iii) Constraints on the continuous adjustment capacity of slow resources Considering that slow resources often have limited continuous adjustment capabilities over multiple consecutive scheduling periods, a continuous adjustment capability constraint is imposed on slow adjustment resources:

[0056] in, Indicates slow resource adjustment At midnight The actual sustainable regulating power or the power level involved in slow regulation; Indicates slow resource adjustment Maximum sustainable adjustable power; This represents a set of resources that are adjusted slowly. This indicates a slow-tuning resource index; This represents the sub-time index. This constraint limits the intensity of continuous adjustment of slow resources over multiple consecutive scheduling periods, thus forming a slow adjustment capability boundary that includes continuous adjustment constraints.

[0057] (iv) Methods for obtaining prediction error statistics To obtain the statistics on the uncertainty of new energy output, it can be defined as the statistical standard deviation of the error between the predicted and measured values:

[0058]

[0059] in, Representing resources At midnight The prediction error statistic or the standard deviation of the output uncertainty; Representing resources At midnight The prediction error; Representing resources At midnight The measured power; Representing resources At midnight The predicted power; Represents a resource index; Indicates the sub-time index.

[0060] In one alternative implementation, This can be obtained using a rolling window statistical method. Specifically, for the same resource and the same sub-time point... , can be taken from the past Data from the past, or retrieve data from the previous day. Data from several days of the same type are used as a sample set. These days can be categorized based on weekdays, weekends, holidays, seasons, or weather types. An error sequence is constructed from the difference between predicted and measured power in the sample, and the sample standard deviation of this error sequence is calculated as the corresponding sub-time point. of To improve statistical robustness, outliers can be truncated before calculation, for example, by quantile truncation, or by using an exponentially weighted moving standard deviation to reflect recent volatility changes. These methods enable those skilled in the art to reproduce experiments and implement the capability assessment process of this invention, given historical forecast and measured data.

[0061] (v) Market price perception adjustment cost model To enable adjustment costs to adapt to market price signals, a market price-perceived adjustment cost model is introduced. This model is applicable to rapid resource... and slow resources Their unit adjustment costs can be expressed as follows:

[0062]

[0063] in, Indicates rapid resource adjustment At midnight Price perception unit adjustment cost; Indicates slow resource adjustment At midnight Price perception unit adjustment cost; , For fixed adjustment costs, it is used to characterize fixed costs such as adjustment losses, battery life depreciation, and equipment maintenance; , This is the response sensitivity coefficient, used to characterize the sensitivity of resource adjustment costs to deviations from market prices; , This is the benchmark price for the corresponding resource; For the moment Market-forecasted electricity prices; This indicates that the resource index can be adjusted quickly. This indicates a slow-tuning resource index; This represents the sub-time index. This model allows the unit adjustment cost of a resource to be determined jointly by fixed cost items and price-sensitive items, thereby enhancing the adaptability of adjustment capacity assessment to market price fluctuations.

[0064] (vi) Output of fast and slow adjustment capability boundary After solving the above-mentioned regulation capacity assessment model, the fast regulation capacity boundary and slow regulation capacity boundary of the virtual power plant at each sub-time point are output. In one implementation, the fast regulation capacity boundary and slow regulation capacity boundary respectively include an upward regulation boundary and a downward regulation boundary; in another implementation, the upward and downward regulation boundaries can also be aggregated into the total available capacity in the corresponding direction according to business needs.

[0065] For example, the total fast adjustment capability boundary and the total slow adjustment capability boundary can be output as follows:

[0066]

[0067] in, Indicates sub-time The overall rapid adjustment capability boundary; Indicates sub-time The overall slow-speed regulation capability boundary; and These respectively represent rapid resource adjustment At midnight The adjustable power output and adjustable power output; and These represent slow resource adjustment. At midnight The adjustable power output and adjustable power output; Indicates rapid adjustment of the resource set; This represents a set of resources that are adjusted slowly. This indicates that the resource index can be adjusted quickly. This indicates a slow-tuning resource index; Indicates the sub-time index.

[0068] In subsequent cross-market collaborative optimization, the capacity boundary in the corresponding direction can be selected as the model constraint input based on the direction in which the spot market and the frequency regulation market utilize the upward or downward adjustment capacity.

[0069] In one implementation, the aforementioned adjustment capability assessment model can be transformed into a linear programming problem or an equivalent convex optimization problem. Since the model is of moderate size, linear programming methods such as the simplex method or interior point method can be used, and a commercial or open-source solver can be invoked for solving.

[0070] Step S3: Construct a cross-market collaborative optimization model based on fast and slow capability boundaries Based on the fast and slow regulation capability boundaries obtained in step S2, a cross-market collaborative optimization model is constructed for the virtual power plant to participate in both the day-ahead spot electricity market and the frequency regulation ancillary service market. The objective function of the cross-market collaborative optimization model is to maximize the expected total revenue of the virtual power plant in both markets, and the objective function includes regulation cost terms and risk penalty terms.

[0071] In this embodiment, the decision variables of the cross-market collaborative optimization model include at least: Each frequency modulation period Frequency modulation declaration capacity With frequency modulation declaration price ; Each sub-time Number of price segments corresponding to multiple price curves in the spot market ; Cumulative declared electricity volume for each bidding segment ; Bidding prices for each quotation segment .

[0072] The declared electricity volume for each bidding segment can be obtained from the difference between adjacent cumulative declared electricity volumes. For example, the first... The increase in power consumption of a segment can be expressed as:

[0073] in, Indicates sub-time The The incremental electricity volume declared for each pricing segment; Indicates sub-time The The cumulative declared electricity volume corresponding to each bidding segment; Indicates sub-time The The cumulative declared electricity volume corresponding to each bidding segment.

[0074] Therefore, in this invention, the declared electricity volume of each segment and the cumulative declared electricity volume of each segment can be converted to each other. Further, in an optional embodiment, whether each bidding segment is activated in the spot market can be indicated by the number of bidding segments, whether the declared electricity volume of the corresponding bidding segment is zero, or by an explicitly set bidding segment activation variable. Thus, the electricity volume, price, and participation in clearing of each bidding segment can all be controlled by optimization variables.

[0075] The objective function of the cross-market collaborative optimization model can be expressed as:

[0076] in, This indicates the FM market during FM hours. Expected returns; This indicates that the spot market is at midnight. Expected returns; This represents the total adjustment cost of each resource and each time period during the cross-market collaborative optimization process. It can be obtained by accumulating the adjustment costs of each resource at each sub-time or each frequency adjustment period based on the market price perception adjustment cost model in step S2. This represents the cross-market risk penalty, used to characterize the impact of spot market price volatility risk, frequency regulation market price volatility risk, and deviation assessment risk on total returns; Indicates the frequency modulation period index; Indicates the sub-time index.

[0077] (a) Expected Return Model of the FM Market The expected revenue of the FM market can be modeled as "capacity compensation + mileage revenue". In one implementation, capacity compensation is only provided if there is a positive price difference. Therefore, the FM market will generate revenue during FM hours. The expected return can be expressed as:

[0078] in, This indicates the FM market during FM hours. Expected returns; Indicates frequency modulation period Frequency modulation application capacity; Indicates the probability of winning a bid in the FM market; This indicates the current daytime market price of electricity or the benchmark price for capacity compensation. This represents the approved variable cost constant; This indicates the unit's declared capacity during the frequency regulation period. Predicted FM mileage generated by the internal call; This indicates the predicted price for clearing out FM mileage; Indicates the performance indicators of the generating unit or aggregated resources; This indicates the frequency modulation period index.

[0079] The probability of winning a bid in the frequency modulation market can be determined based on the relationship between the predicted clearing results and the bid prices, for example, it can be expressed as:

[0080] in, This indicates the FM market during FM hours. The probability of winning the bid; This indicates the FM market during FM hours. Forecast value of FM mileage clearing price; This indicates the FM market during FM hours. The declared price; Indicates the probability of an event occurring; This indicates the frequency regulation period index. That is, when the predicted price for frequency regulation mileage is not lower than the declared price, the corresponding bid is considered more likely to be successful.

[0081] (II) Expected Return Model in the Spot Market The spot market at midnight The expected return can be determined based on the predicted probability of winning the bid, the predicted clearing price, and the predicted winning volume. In one implementation, the predicted winning volume can be expressed as:

[0082] in, Indicates sub-time The predicted winning bid volume; Indicates sub-time The The probability of winning a bid in the spot market for each price range; Indicates sub-time The The incremental electricity volume declared for each pricing segment; Indicates sub-time The number of price segments in the spot market; Indicates the sub-time index; This indicates the quote segment index.

[0083] Furthermore, the spot market at midnight The expected return can be expressed as:

[0084] in, This indicates that the spot market is at midnight. Expected returns; Indicates sub-time The predicted clearing price; Indicates sub-time The predicted winning bid volume; Indicates the sub-time index.

[0085] The probability of winning the bid can be expressed as:

[0086] in, Indicates sub-time The The probability of winning a bid in the spot market for each price range; Indicates the probability of an event occurring; Indicates sub-time The predicted clearing price; Indicates sub-time The The bid price for each quotation segment; Indicates the sub-time index; This indicates the price range index. That is, the predicted clearing price is not lower than the [number]th [quote range index]. When submitting a price quote, the quoted price segment has a chance of winning the bid.

[0087] When the predicted clearing price falls within a certain bidding range, the predicted winning volume can be estimated using interpolation within that range. If the predicted clearing price... Falling in Section and the Between segments, the following conditions must be met:

[0088] in, Indicates sub-time The The bid price for each quotation segment; Indicates sub-time The The bid price for each quotation segment; Indicates sub-time The predicted clearing price; Indicates the sub-time index; This indicates the quote segment index.

[0089] The predicted winning bid volume can then be expressed as:

[0090] in, Indicates sub-time The predicted winning bid volume; Indicates sub-time The The cumulative declared electricity volume corresponding to each bidding segment; Indicates sub-time The predicted clearing price; Indicates sub-time The The bid price for each quotation segment; Indicates sub-time The The bid price for each quotation segment; Indicates the first The first price segment is relative to the first The increase in electricity volume for each pricing segment; Indicates the sub-time index; This represents the price segment index. The fraction indicates the proportion of the predicted clearing price within adjacent price segments, thus allowing estimation of changes in the winning bid volume within that segment. This method provides a more detailed reflection of the impact of spot segmented pricing on the winning bid volume and revenue.

[0091] In the aforementioned expected return model for the frequency modulation market and the expected return model for the spot market... and All of them can be regarded as random variables, and their distributions can be obtained based on historical data and prediction models, thus providing a basis for subsequent expected return calculations and risk statistics.

[0092] (III) Feasible Calculation Methods for Expected Returns and Probability of Winning the Bid To ensure that the expected returns, probability terms, and risk statistics in the spot market and frequency regulation market have clear data sources and reproducible implementation processes, in one optional implementation method, random variables can be calculated using scenario generation or sampling estimation.

[0093] Specifically, historical spot clearing prices, frequency regulation mileage prices, deviation assessment prices, and forecast error sequences for wind power, photovoltaic power, or load can be collected first, and then the forecast mean for each period can be output by combining the day-ahead forecasting model. Subsequently, the spot clearing price and frequency regulation mileage price can be fitted with parametric distributions, such as normal distribution or log-normal distribution fitting; alternatively, kernel density estimation or bootstrap resampling can be used to obtain empirical distributions. For forecast errors, they can be grouped according to the same type of day to obtain the corresponding empirical distribution.

[0094] Then, generate A combined scenario. For each scenario The spot clearing price was obtained by sampling separately. FM mileage price Prediction error Random variables are used to determine the feasibility of the scenario and the conditions for triggering deviation assessments.

[0095] In each scenario, based on multiple price curves and the scenario clearing price Calculate and predict the winning bid amount The winning bid volume can still be determined using the aforementioned segmented interpolation method; simultaneously, based on the frequency regulation bid price and the frequency regulation market clearing rules, the winning bid volume and the mileage to be used will be determined, and the scenario revenue will be calculated. .

[0096] After calculating the revenue for all scenarios, the expected revenue can be estimated as follows:

[0097] in, This represents the expected value of the total revenue; Indicates the total number of scenes generated; Indicates the scene index; Indicates the first The revenue value in each scenario.

[0098] Price volatility statistics, such as standard deviation, can also be calculated from sample scenario values. This is relevant to the indicator function in the deviation risk penalty. The value can be set to 0 or 1 in each scenario based on whether the deviation exceeds the threshold or whether the assessment rule is triggered. In this way, without changing the overall modeling concept of the invention, the expected terms, probability terms, and risk terms all have a clear calculation process, which facilitates review and engineering implementation.

[0099] (iv) Boundary Coupling Constraints of Adjustment Capability To achieve coordinated allocation of the same aggregated resources between the spot market and the frequency regulation market, it is necessary to set boundary coupling constraints on regulation capacity to limit the occupation of fast and slow regulation capacity by the two types of markets.

[0100] In one implementation, when the FM market is categorized by FM time period Organizations and spot markets are organized by time interval. During organization, the frequency modulation (FM) declaration capacity can be converted into the rapid adjustment capacity occupancy at the corresponding sub-time point through time-segment mapping relationships. For any sub-time point after conversion... The requested frequency modulation capacity occupancy should not exceed the rapid adjustment capability boundary of that sub-time, that is:

[0101] in, This indicates that the time interval has been mapped and then converted to the sub-time interval. Frequency modulation application capacity occupancy; Indicates sub-time The overall rapid adjustment capability boundary; Indicates the sub-time index.

[0102] In this embodiment, the total declared electricity volume in the spot market This represents the total declared power equivalent value in the spot market after conversion using a unified time scale. When the original declared quantity is given in the form of electricity, it is converted into power equivalent value according to the duration of the corresponding sub-time point before participating in subsequent constraint calculations.

[0103] Furthermore, at any sub-time The total declared electricity volume in the spot market is constrained by both the slow regulation capacity boundary and the remaining fast regulation capacity, and can be written as:

[0104] in, Indicates sub-time The total declared volume of the spot market or the cumulative declared volume at the end of the period; Indicates sub-time The total slow-speed regulation capability boundary; Indicates sub-time The remaining rapid adjustment capability; Indicates the sub-time index.

[0105] In a preferred embodiment, to ensure that the spot price curve makes full use of available adjustment capacity, the total bid volume in the spot market can be directly set to the aforementioned upper limit, i.e.:

[0106] The portion within parentheses represents the remaining fast regulation capacity, indicating the capacity available for the spot market after deducting the frequency regulation market share. This coupling relationship prevents excessive frequency regulation applications from crowding out available spot market electricity, or excessive spot market applications from compressing frequency regulation reserve space.

[0107] (v) Constraints of multi-segment price curves in the spot market To meet the segmented pricing rules of the spot market, a multi-segment pricing curve constraint is set up so that the cumulative declared electricity volume of each pricing segment increases in sequence according to the segment number within the same scheduling period, the declared price of each pricing segment remains monotonically unchanged according to the segment number, and the range of the number of pricing segments is restricted.

[0108] In one implementation, the following constraints may be set:

[0109]

[0110]

[0111]

[0112]

[0113]

[0114]

[0115]

[0116] Among them, the definition ; Indicates sub-time The The cumulative declared electricity volume corresponding to each bidding segment; Indicates sub-time The The increase in electricity volume for each pricing segment; Indicates sub-time The The bid price for each quotation segment; Indicates sub-time Total reported electricity volume in the spot market; and Representing sub-times respectively Minimum and maximum allowable declared electricity volume; and These represent the minimum and maximum bid prices allowed by market rules, respectively. and Indicates the range of price segments; Indicates the sub-time index; This indicates the quote segment index.

[0117] In a directly implementable manner, for any given scheduling period, the spot market multi-segment price curve consists of the number of segments. Cumulative reported electricity volume for each segment Price declaration for each segment Composition, in which The range of the number of segments can be set by the rules of the trading center, for example, it can be limited to 1 to... The segment can also further limit the minimum electricity increment of a single bidding segment to avoid zero-length segments or extremely small segments affecting the stability of the clearing calculation.

[0118] The start and end points of the first segment can be consistent with the available declaration range for that time period. For example, the start point of the first segment can correspond to the minimum declared electricity volume, and the end point of the last segment cannot exceed the upper limit of available declarations on the spot market side for that time period. Preferably, the upper limit of available declarations on the spot market side is jointly determined by the slow adjustment capability boundary and the remaining fast adjustment capability for the corresponding time period, thereby ensuring that the multi-segment spot price curve is consistent with the aforementioned cross-market capability coupling constraints.

[0119] The above constraints ensure that the bidding curve exhibits a stepped increasing structure within the same time period, and that the number of bidding segments, the increasing relationship of electricity volume, and the monotonic relationship of price all meet the market bidding rules. This also facilitates the subsequent calculation and prediction of the winning bid electricity volume using interpolation within segments.

[0120] (vi) FM Market Application and Enforcement Constraints FM market applications must simultaneously meet market rule constraints and resource availability constraints. In one implementation, at least the following constraints may be set:

[0121]

[0122] in, Indicates frequency modulation period Frequency modulation application capacity; Indicates frequency modulation period The upper limit of frequency modulation application capacity; Indicates frequency modulation period The declared price for frequency modulation; Indicates frequency modulation period The upper limit of the frequency modulation declaration price; This indicates the frequency modulation period index. It can be determined by the rapid adjustment capability boundary of the corresponding frequency modulation period, or it can be determined in conjunction with the trading center's rules regarding the maximum application capacity; the aforementioned The price can be set at the upper limit stipulated by the trading center.

[0123] In one alternative implementation, FM market rule constraints may also include at least the following: Application capacity range constraints: The application capacity for frequency regulation should be within the allowable range and should not exceed the rapid adjustment capacity boundary of the corresponding time period or its conservative value. The conservative value can be the value of the rapid adjustment capacity boundary of the corresponding time period after being reduced by a safety factor, or the value after being corrected by combining the forecast error statistic, historical realization rate and safety margin. Price range and step size constraints for frequency modulation (FM) bids: The bid price must meet the lower and upper limits of the market price and the minimum bid step size requirement. Performance and assessment constraints: For the resource set participating in frequency regulation, additional constraints such as response speed, ramp rate, state of charge, and sustainable adjustment time are imposed to ensure that the declared capacity can be realized within the assessment period; when market rules require settlement based on performance coefficients, performance coefficients can be introduced into the revenue calculation. ; Time-based mapping constraints: When the frequency regulation market is organized by hour and the spot market is organized by sub-time, a mapping relationship can be used to distribute the frequency regulation application capacity into the occupancy at the sub-time level. For example, it can be evenly occupied within an hour or allocated according to the historical call ratio, so that the frequency regulation application constraints are consistent with the boundary constraints of the rapid adjustment capability at the sub-time scale.

[0124] By applying the above-mentioned rules and constraints, it can be ensured that the output frequency regulation application information not only meets the resource capacity boundary, but also complies with the transaction rules and assessment rules, thereby reducing the risk that the optimal solution of the model cannot be reported or is difficult to execute.

[0125] (vii) Cross-market risk penalty items To mitigate the risks associated with spot price volatility, frequency regulation price volatility, and deviation assessments, a cross-market risk penalty term is introduced into the objective function. An example of this can be expressed as:

[0126] in, Indicates cross-market risk penalties; This represents the risk preference coefficient, and ; Indicates sub-time The standard deviation of the spot market forecast price; Indicates sub-time Standard deviation of FM market forecast prices; Indicates sub-time The total declared volume in the spot market or the total declared volume after conversion to a uniform time scale; Indicates sub-time Frequency modulation application capacity occupancy; Indicates sub-time Deviation assessment unit price; Indicates sub-time Statistical measure of the error in new energy or load forecasting; Indicates sub-time Execution deviation; Indicates sub-time The preset deviation threshold; Indicates an indicator function; Indicates the sub-time index.

[0127] In one implementation, the indicator function takes the value of 1 when the absolute value of the actual deviation exceeds a preset deviation threshold, or when it meets the deviation assessment triggering rules stipulated by the power grid trading center; otherwise, it takes the value of 0. The preset deviation threshold can be set by market rules, assessment methods, or the risk control strategy of the virtual power plant operator.

[0128] Step S4: Solve the cross-market collaborative optimization model using the goose optimization algorithm. Under the conditions of satisfying the boundary constraints of adjustment capacity, the constraints of spot multi-segment pricing, and the constraints of market rules, the goose optimization algorithm is used to iteratively solve the cross-market collaborative optimization model constructed in step S3 to obtain the frequency regulation declaration parameters, spot multi-segment pricing curves, and resource scheduling plans for each scheduling period.

[0129] Since the cross-market collaborative optimization model includes conditional expectation, probability term, piecewise pricing variable, risk penalty term and multiple types of coupling constraints, the solution space has strong nonlinearity and nonconvexity, and there may be multiple local optima. Therefore, the goose optimization algorithm is adopted to approximate the global better solution through leader goose guidance, V-shaped formation collaboration and perturbation update.

[0130] In this embodiment, the decision variables are encoded as a position matrix of individual geese in the flock. To facilitate a unified solution, the variables represented by frequency modulation time periods in step S3 can be uniformly converted to the sub-time scale after time period mapping during the encoding stage. The corresponding position matrix can be written as:

[0131] in, A matrix representing the positions of individual geese in a flock; Indicates sub-time The corresponding frequency modulation application capacity; Indicates sub-time The corresponding frequency modulation (FM) application price; Indicates sub-time The The cumulative declared electricity volume corresponding to each bidding segment; Indicates sub-time The The bid price for each quotation segment; Indicates sub-time The number of price segments in the spot market; Indicates the sub-time index; Indicates the price segment index; This represents the total number of sub-times. When the number of price segments in each sub-time is inconsistent, unused price segments can be uniformly encoded by padding with zeros, fixing the maximum number of segments, or setting a variable to enable price segments.

[0132] Each individual goose in the flock must include at least the frequency regulation declaration capacity, frequency regulation declaration price, cumulative declared electricity volume for each segment of the spot multi-segment price curve, declaration price for each segment, and the number of quotation segments.

[0133] (a) Fitness function and constraint violation Based on the objective function in step S3, the fitness function of the goose optimization algorithm is constructed as follows:

[0134] in, Indicates individual geese in a flock The fitness function value; This indicates the FM market during FM hours. Expected returns; This indicates that the spot market is at midnight. Expected returns; This represents the total adjustment cost in the cross-market collaborative optimization process; Indicates cross-market risk penalties; Indicates the penalty coefficient; The constraint violation function is used to represent the quantity of constraint violations. Indicates the frequency modulation period index; Indicates the sub-time index.

[0135] The constraint violation amount can be composed of the violation amount of the adjustment capability boundary constraint, the violation amount of the price monotonicity constraint, the violation amount of the market rule constraint, and the violation amount of the price segment range constraint, that is:

[0136] in, Indicates the total number of constraint violations; This indicates the amount of violation of the boundary constraints of the adjustment capability; This indicates a violation of the monotonicity constraint on the quoted price. This indicates the amount of violations of market rules and regulations; This indicates the number of violations of the price range constraint.

[0137] in: Violation of boundary constraints of adjustment capability

[0138] This is used to characterize the extent to which the declared capacity occupancy of frequency regulation exceeds the fast regulation capacity boundary, or the equivalent value of the total declared power in the spot market after being converted to a uniform time scale exceeds the sum of the slow regulation capacity boundary and the remaining fast regulation capacity. It can be expressed as:

[0139] in, This indicates that the time interval has been mapped and then converted to the sub-time interval. Frequency modulation application capacity occupancy; Indicates sub-time The overall rapid adjustment capability boundary; This represents the total value of the spot market bid power after being converted to a unified time scale at sub-time t. Indicates sub-time The total slow-speed regulation capability boundary; Indicates the sub-time index.

[0140] Spot price monotonicity constraint violation quantity

[0141] This is used to characterize the extent to which spot price quotes at different price levels fail to meet the monotonicity requirement and the degree to which cumulative declared electricity volume fails to meet the incremental requirement. It can be expressed as:

[0142] in, Indicates sub-time The The bid price for each quotation segment; Indicates sub-time The The bid price for each quotation segment; Indicates sub-time The The cumulative declared electricity volume corresponding to each bidding segment; Indicates sub-time The The cumulative declared electricity volume corresponding to each bidding segment; Indicates sub-time The number of price segments in the spot market; Indicates the sub-time index; This indicates the quote segment index.

[0143] Market rule constraints violation volume

[0144] This is used to characterize the extent to which frequency modulation (FM) bid capacity, FM bid price, and other market rule variables exceed the permissible upper and lower limits. For example, it can be represented as:

[0145] in, Indicates sub-time Mapped frequency modulation reporting capacity variable; Indicates sub-time The upper limit of frequency modulation application capacity after mapping; Indicates sub-time Mapped frequency modulation bid price variable; Indicates sub-time The upper limit of the frequency modulation (FM) declaration price after mapping; This represents the sub-time index. When constraints such as minimum capacity and minimum bid step size exist, the corresponding violations can also be included in this item.

[0146] Price segment range constraint violation

[0147] This is used to characterize the extent to which the number of spot price quote segments exceeds the allowed range. It can be represented as:

[0148] in, and These represent the minimum and maximum number of bid segments allowed, respectively. Indicates sub-time The number of price segments in the spot market; Indicates the sub-time index.

[0149] The fitness function described above can unify the profit maximization objective and constraint feasibility into a single evaluation index, which can be used to guide the search direction of individual geese.

[0150] (ii) Dynamic penalty coefficient update mechanism To enhance the feasibility convergence capability during the search process, and to balance the early global search with the later local convergence, in this embodiment, the penalty coefficient is... It adopts a dynamic update method. Its update rules can be expressed as:

[0151] in, This represents the penalty coefficient for the current iteration; Indicates the initial penalty coefficient; Indicates the current iteration number; Indicates the maximum number of iterations; Indicates the attenuation coefficient; Indicates the standard deviation of group fitness; This represents the average fitness of the population.

[0152] Through this update mechanism, the penalty coefficient can be dynamically adjusted as the iteration progresses and the fitness distribution of the population changes, thereby maintaining a certain exploratory capability in the early stages of the search and enhancing the suppression of solutions that do not meet the constraints in the later stages.

[0153] (III) Population Initialization In the initialization phase, the population size, maximum number of iterations, learning factor, perturbation parameters, and upper and lower boundaries of each decision variable are first set. Then, within the allowed upper and lower boundaries of each decision variable, the positions of individuals in each flock are randomly initialized. Each initialized individual corresponds to a set of candidate frequency modulation reporting parameters, a multi-segment spot price curve, and the number of price segments.

[0154] (iv) Leader Goose Selection Mechanism In each iteration, an exponential selection probability is constructed based on the difference between the individual's fitness and the group's optimal fitness, and the leader goose is selected according to this probability. The selection probability can be expressed as:

[0155] in, Indicates the first The probability that an individual will be selected as the leader goose; Indicates the first The fitness function value of each individual; This represents the optimal fitness value in the current population; Indicates the size of the goose population; and Represents an individual index.

[0156] This mechanism allows for the adaptive selection of leader geese based on individual merits, thereby guiding the migration direction of the group.

[0157] (v) V-formation update strategy After selecting the leader goose, different update rules are applied to the leader goose and non-leader goose individuals.

[0158] To avoid confusion with the response sensitivity coefficient in step S2, in this embodiment, the update coefficients in the goose optimization algorithm are denoted as follows: , , , , .

[0159] For the leader goose, its position update can be represented as:

[0160] in, This indicates that the current leader goose is in the [number]th [year]. The position at the next iteration; This indicates that the current leader goose is in the [number]th [year]. The position at the next iteration; This indicates the historical best position of the leader goose. Indicates the globally optimal position of the group; and For learning factors; and for Random numbers within the interval; Indicates the current iteration number; This indicates the maximum number of iterations.

[0161] For a non-leader goose, its position update can be represented as:

[0162] in, Indicates the first The non-leader goose individual in the first The position at the next iteration; Indicates the first The non-leader goose individual in the first The position at the next iteration; This indicates that the current leader goose is in the [number]th [year]. The position at the next iteration; Indicates the first The set of neighboring individuals of a non-leader goose; This represents the number of individuals in the neighborhood set; , , To update the coefficients; and for Random numbers within the interval; This indicates that the mean is 0 and the variance is 0. The Gaussian random perturbation term; and Represents an individual index; This indicates the current iteration number.

[0163] To gradually reduce the perturbation strength as the iteration progresses, we can set:

[0164] in, Indicates the first Perturbation variance at the next iteration; Indicates the initial perturbation variance; Indicates the current iteration number; This indicates the maximum number of iterations.

[0165] Through the above update method, non-leader geese can migrate along the direction of the leader geese, maintain group diversity by utilizing neighborhood difference information, and improve the later convergence stability by decreasing Gaussian perturbation.

[0166] When the number of iterations reaches the maximum number of iterations Alternatively, if the preset convergence condition is met, the iteration ends and the optimal solution is output.

[0167] Step S5: Output market declaration results and generate resource control instructions. Based on the solution results of step S4, frequency regulation declaration information and spot price information for market reporting are generated and output, and output control instructions for various resources within the virtual power plant are generated and issued.

[0168] In one implementation, the frequency regulation declaration capacity, frequency regulation declaration price, and the declaration electricity and declaration price of each segment of the multi-segment quotation curve in the spot market obtained in step S4 can be used to generate a submission file or submission message in accordance with the data format required by the power grid trading center, so that the virtual power plant can complete the market declaration.

[0169] After the market declaration is completed and a scheduling plan is formed, the resource scheduling plan can be converted into control instructions for various resources within the virtual power plant, such as charging and discharging power instructions for energy storage resources, adjustment instructions for interruptible loads, and output adjustment instructions for other adjustable loads, and sent to the virtual power plant energy management system, aggregation control platform, or field control terminal for execution.

[0170] In an exemplary implementation scenario, the virtual power plant aggregates energy storage resources, interruptible loads, wind power resources, photovoltaic resources, and other adjustable load resources. First, the predicted power, upper and lower limits of operating power, continuous regulation capability parameters, and prediction error statistics of each resource are obtained for each day-ahead scheduling period or sub-time. At the same time, the price prediction data and fluctuation statistics of the day-ahead spot market and frequency regulation ancillary service market are obtained.

[0171] Secondly, the aggregated resources are divided into a fast-adjustment resource set and a slow-adjustment resource set according to the resource response speed. Based on the aforementioned objective function and constraints, the dynamic adjustment capability evaluation model is solved to obtain the fast adjustment capability boundary and slow adjustment capability boundary corresponding to each scheduling period.

[0172] Then, based on the fast adjustment capability boundary and the slow adjustment capability boundary, a cross-market collaborative optimization model is constructed to jointly optimize the frequency regulation declaration capacity, frequency regulation declaration price, the declaration electricity volume and declaration price of each segment of the multi-segment quotation curve in the spot market, and simultaneously consider the adjustment cost term and the risk penalty term in the objective function.

[0173] Next, the goose optimization algorithm described in step S4 is used for iterative solution. Through fitness function evaluation, dynamic penalty coefficient update, leader goose selection and V-formation update, a better solution to the cross-market collaborative optimization problem is obtained, and the frequency adjustment declaration parameters, spot price curves and resource scheduling plans for each scheduling period are output.

[0174] Finally, in step S5, the solution results are converted into market reporting information and / or into control commands for various resources within the virtual power plant for execution. Through this process, the integrated implementation of dynamic adjustment capability assessment and cross-market collaborative optimization decision-making for the virtual power plant can be achieved.

[0175] In another embodiment, the present invention also provides a virtual power plant cross-market collaborative optimization decision-making device based on the goose optimization algorithm, including a data acquisition module, a capacity assessment module, a collaborative modeling module, a goose optimization solution module, and an output and control module, which are respectively used to perform the above steps S1 to S5.

[0176] In another embodiment, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is used to implement the methods described in any of the above embodiments.

[0177] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A cross-market collaborative optimization decision-making method for virtual power plants based on the goose optimization algorithm, characterized in that, Performed by an electronic device, the method includes: S1. Obtain the resource set of the virtual power plant and its operating constraint parameters, the day-ahead power forecast data and forecast error statistics of each resource, and the price forecast data and fluctuation statistics of the day-ahead spot electricity market and frequency regulation ancillary service market. S2. Divide the resource set into a fast-regulation resource set and a slow-regulation resource set according to the resource adjustment response speed. Under the condition of considering the day-ahead power prediction data and the prediction error statistics, solve the adjustment capability evaluation model to obtain the fast adjustment capability boundary and slow adjustment capability boundary for each scheduling period. S3. Based on the fast regulation capability boundary and slow regulation capability boundary obtained in step S2, construct a cross-market collaborative optimization model for the virtual power plant to participate in both the day-ahead spot electricity market and the frequency regulation ancillary service market. The objective function of the cross-market collaborative optimization model is to maximize the expected total revenue of the virtual power plant in the two markets, and the objective function includes regulation cost and risk penalty terms. The decision variables of the cross-market collaborative optimization model include at least: the frequency regulation declaration capacity and frequency regulation declaration price for each scheduling period, and the declaration electricity volume and declaration price for each segment corresponding to the multiple quotation curves in the spot market for each scheduling period. S4. Under the conditions of satisfying the boundary constraints of adjustment capacity, the constraints of spot multi-segment price and the constraints of market rules, the cross-market collaborative optimization model is solved iteratively by using the goose optimization algorithm to obtain the frequency adjustment declaration parameters, spot multi-segment price curves and resource scheduling plans for each scheduling period. The goose optimization algorithm uses a fitness function as an evaluation metric. This fitness function includes at least the objective function value of the cross-market collaborative optimization model and a penalty term for constraint violations. The penalty coefficient λ is dynamically updated according to the following rules: in, The initial penalty coefficient, This represents the current iteration number. The maximum number of iterations, The attenuation coefficient is... The standard deviation of group fitness The average fitness of the population; S5. Based on the solution results of step S4, generate and output frequency regulation declaration information and spot price information for market reporting, and generate and issue output control instructions for each resource within the virtual power plant.

2. The method according to claim 1, characterized in that, In step S2, for wind power resources and / or photovoltaic resources belonging to the slow-adjustment resource set, uncertainty correction is performed on their adjustable power upper limit and adjustable power upper limit based on the prediction error statistics to obtain the corrected slow-adjustment capability boundary. The greater the uncertainty represented by the prediction error statistics, the greater the shrinkage of the corresponding adjustable capability boundary.

3. The method according to claim 1, characterized in that, In step S2, a continuous adjustment capability limit is imposed on the slow adjustment resource so that the adjustable power of the slow adjustment resource does not exceed the preset maximum sustainable adjustment power in multiple consecutive scheduling periods, thereby forming a slow adjustment capability boundary that includes the continuous adjustment limit.

4. The method according to claim 1, characterized in that, The adjustment capability boundary constraints in step S3 include at least: The frequency regulation capacity requested during any scheduling period shall not exceed the rapid adjustment capacity boundary of that scheduling period; The total declared electricity volume in the spot market during any scheduling period shall not exceed the sum of the slow regulation capacity boundary and the remaining fast regulation capacity for that scheduling period, wherein the remaining fast regulation capacity is the portion remaining after deducting the frequency regulation declared capacity for that scheduling period from the fast regulation capacity boundary for that scheduling period.

5. The method according to claim 1, characterized in that, The goose optimization algorithm in step S4 also includes: Leader goose selection mechanism: An exponential selection probability is constructed based on the difference between the individual fitness and the group's optimal fitness, and the leader goose is determined according to the selection probability. V-shaped formation update strategy: The position of the leader goose is updated by a combination of linear decreasing factor and random factor, and a guiding term is introduced from the individual's historical best solution and the group's global best solution; the position of the non-leader goose is updated by a neighborhood individual difference term and a random perturbation term, wherein the random perturbation term includes a Gaussian random perturbation term.

6. The method according to claim 1, characterized in that, The multi-segment price curves in the spot market satisfy the following constraints: within the same scheduling period, the cumulative declared electricity volume of each price segment increases according to the price segment number, and the declared price of each price segment remains monotonically unchanged according to the price segment number; and the number of segments of the multi-segment price curves satisfies the preset segment number range constraint. Furthermore, when constructing the cross-market collaborative optimization model and calculating the expected return in the spot market, the predicted winning bid volume is determined based on the predicted clearing price and the multi-segment bidding curve. When the predicted clearing price falls into the bidding interval corresponding to the i-th bidding segment and the (i+1)-th bidding segment, the change in the winning bid volume within the bidding interval is determined based on the proportion of the price difference of the predicted clearing price within the bidding interval to the price difference of the adjacent bidding segments, and the predicted winning bid volume is obtained by combining the volume increment of the adjacent bidding segments.

7. The method according to claim 1, characterized in that, The adjustment cost item includes market price-perceived adjustment costs, which are used to determine the unit adjustment cost of resources by a combination of fixed adjustment costs and price-sensitive items. The price-sensitive items are related to the deviation of the predicted electricity price from the benchmark price and are characterized by a response sensitivity coefficient.

8. The method according to claim 1, characterized in that, The risk penalty items include at least: Price volatility risk penalty determined by the statistical measures of price volatility forecasts from the spot market and the statistical measures of price volatility forecasts from the frequency modulation market; The deviation risk penalty is determined by the deviation assessment unit price, the prediction error statistic, and the indicator function, where the indicator function is either 0 or 1 and is used to characterize whether the deviation assessment triggering conditions are met.

9. A virtual power plant cross-market collaborative optimization decision-making device based on the goose optimization algorithm, characterized in that, include: The data acquisition module is used to execute step S1 in the method of claim 1; Capability assessment module, used to perform step S2 in the method of claim 1; A collaborative modeling module is used to perform step S3 in the method of claim 1; The goose optimization solution module is used to execute step S4 in the method of claim 1 and iteratively solve according to the penalty coefficient dynamic update rule in the method of claim 1; The output and control module is used to execute step S5 in the method of claim 1.

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

Images