Robust optimization scheduling method and device for virtual power plant participating in electricity-carbon market based on uncertainty
Patent Information
- Application Number
- CN202611175773.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-04
- Publication Date
- 2026-10-27
AI Technical Summary
[0008]有鉴于此，针对现有技术中虚拟电厂参与电碳市场优化调度方法难以在系统经济性与运行可靠性之间实现灵活可控平衡的缺陷，有必要提出一种基于不确定度的虚拟电厂参与电碳市场的鲁棒优化调度方法和装置，实现经济性与安全性的优化权衡
[0080]所述目标函数最小化模块，配置为以虚拟电厂总运行成本最小化为目标构建目标函数；其中，所述总运行成本包括电力交易成本、燃气轮机运行成本和碳交易成本；
Smart Images
Figure CN122993052A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power system optimization scheduling technology, and in particular to a robust optimization scheduling method and apparatus for virtual power plants participating in the electricity carbon market based on uncertainty. Background Technology
[0002] With the deepening implementation of the "dual-carbon" strategy, the proportion of renewable energy power generation, represented by wind and solar power, is increasing. However, the inherent intermittency, volatility, and uncertainty of renewable energy output pose significant challenges to the safe and stable operation of the power system. Virtual Power Plants (VPPs), as an advanced energy management model capable of aggregating and managing distributed power sources (wind and solar), energy storage systems, and controllable loads, have become a key vehicle for promoting renewable energy consumption and participation in electricity market transactions. In the context of the coordinated operation of the electricity and carbon markets, VPPs need to develop optimized dispatch plans for the next 24 hours during the day-ahead phase to minimize total operating costs, including electricity trading costs, gas turbine operating costs, and carbon trading costs. Accurately predicting and addressing the uncertainties in wind and solar output is a core element in developing this plan and improving the operational efficiency of VPPs.
[0003] Currently, the existing technical solutions for optimizing the dispatch of virtual power plants participating in the electricity carbon market mainly include the following:
[0004] (1) Deterministic optimization method. This method formulates a dispatch plan based on the point-based predicted values of wind and solar power output (e.g., day-ahead forecast curves), treating the predicted values as actual values. It obtains the power output plan of each unit and the power grid interaction by solving deterministic mathematical programming problems (e.g., mixed integer linear programming). This method completely ignores the existence of prediction errors. When the actual wind and solar power output is lower than the predicted value, the virtual power plant will face a power shortage and be forced to urgently start high-cost backup power sources or purchase electricity from the electricity market at high prices, resulting in actual operating costs far exceeding planned costs and even threatening system security. Conversely, if the output is higher than the predicted value, it may lead to wind and solar curtailment, resulting in resource waste.
[0005] (2) Traditional robust optimization method (box uncertainty set). This method recognizes the uncertainty of wind and solar power output and describes it as a bounded interval (e.g., output fluctuates within ±20% of the predicted value). To ensure the system can operate safely under all possible uncertainties, it assumes that wind and solar power output reaches the most unfavorable boundary value (i.e., the lower limit of output) simultaneously in all time periods. This method's assumptions are too extreme and conservative, and are almost impossible to occur in reality. To cope with such an extremely harsh hypothetical scenario, the scheduling plan needs to reserve a large amount of expensive reserve capacity (such as high-load operation of gas turbines), resulting in extremely high day-ahead planning costs and severely impaired economic efficiency.
[0006] (3) Stochastic optimization method. This method requires assuming a precise probability distribution of wind and solar power output and simulating uncertainty by generating a large number of scenarios. In actual virtual power plant operation, the true probability distribution of wind and solar power output is difficult to obtain accurately. At the same time, this method has high computational complexity and is difficult to solve. The quality of its decision results is highly dependent on the assumed probability distribution. Once the distribution deviates significantly from the actual distribution, the optimization effect will be greatly reduced.
[0007] In summary, existing technologies suffer from either "overly optimistic (deterministic optimization)" or "overly conservative (traditional robust optimization)" shortcomings when dealing with the scheduling problem of virtual power plants participating in the carbon electricity market. There is a lack of a scheduling method that can achieve a flexible and controllable balance between system economy and operational reliability. Summary of the Invention
[0008] In view of this, in order to address the shortcomings of existing methods for optimizing the scheduling of virtual power plants participating in the carbon electricity market, which are difficult to achieve a flexible and controllable balance between system economy and operational reliability, it is necessary to propose a robust optimization scheduling method and device for virtual power plants participating in the carbon electricity market based on uncertainty, so as to achieve an optimal trade-off between economy and safety.
[0009] This invention provides a robust optimization scheduling method for virtual power plants participating in the electricity carbon market based on uncertainty, comprising:
[0010] Obtain day-ahead forecast data for virtual power plants;
[0011] An uncertain set of wind and solar power output is constructed based on the aforementioned day-ahead forecast data; wherein, the uncertain set adopts an improved uncertain set form based on uncertainty budget, with the forecast error of total wind and solar power output as the uncertainty parameter, and the uncertainty budget parameter limits the number of time periods in which the worst-case scenario is simultaneously obtained within the scheduling cycle;
[0012] An objective function is constructed with the goal of minimizing the total operating cost of the virtual power plant; wherein the total operating cost includes electricity trading costs, gas turbine operating costs, and carbon trading costs.
[0013] The constraints for constructing the virtual power plant include deterministic constraints, robust power balance constraints, and uncertainty budget constraints; the robust power balance constraints require that real-time power balance is maintained for all possible wind and solar power output scenarios within the uncertainty set.
[0014] The objective function and constraints are transformed into a mixed-integer linear programming problem and solved to obtain the optimal scheduling scheme for the virtual power plant in each time period.
[0015] Preferably, the construction of the uncertain set of wind and solar power output includes:
[0016] Define the total output of the landscape as Then its predicted value is The improved uncertain set can be represented as:
[0017]
[0018] in, For improved uncertain sets, Contribute to the overall scenery This is the predicted value for the total power output of wind and solar power. To contribute to wind power, This is the predicted value of wind power output. Contribute to photovoltaic power Forecast values of photovoltaic power output As a fluctuation indicator variable, For the range of uncertainty, The uncertainty budget parameter reflects the decision-maker's risk preference and takes values in the range [0, T0].
[0019] Preferably, the objective function is expressed as:
[0020]
[0021] in, Total operating cost, For electricity transaction costs, For the operating costs of gas turbines, This refers to the cost of carbon trading.
[0022] Preferably, the electricity transaction cost Represented as:
[0023]
[0024] Where T is the scheduling period. The electricity price for time period t. The power purchased during time period t. Let t represent the electricity sales volume during time period t;
[0025] The operating cost of the gas turbine Represented as:
[0026]
[0027] in, This is the fuel cost coefficient. For the gas turbine output during time period t, To fix operating costs, This refers to the operating status of the gas turbine. For start-up costs, This is a startup status indicator variable;
[0028] The carbon trading costs Represented as:
[0029]
[0030] in, For carbon price, Excess carbon emissions are expressed as:
[0031]
[0032] in, For carbon emission factors of gas turbines, Carbon emission factors from purchasing electricity from the power grid Free carbon credits.
[0033] Preferably, the deterministic constraints include:
[0034] The output of a gas turbine should be between its minimum and maximum output.
[0035]
[0036] Gas turbine start-up constraints:
[0037]
[0038] Energy storage charging and discharging power constraints:
[0039]
[0040]
[0041] Simultaneous charging and discharging are prohibited:
[0042]
[0043] Energy storage energy balance constraints:
[0044]
[0045] Energy storage energy range constraints:
[0046]
[0047] Initial and final energy equality constraint:
[0048]
[0049] Power purchase and sale constraints:
[0050]
[0051]
[0052] Simultaneous purchase and sale of electricity is prohibited:
[0053]
[0054] in, and These represent the minimum and maximum output values of the gas turbine. This is a charging state variable; 1 indicates charging. The energy storage charging power during time period t. For maximum charge / discharge power of energy storage, The energy storage discharge power during time period t. This is a discharge state variable, where 1 represents discharge. The state of charge of the stored energy at the end of time period t. For energy storage charging efficiency, For scheduling time intervals, For energy storage and discharge efficiency, For the minimum permissible state of charge of energy storage, For the maximum permissible state of charge of energy storage, This represents the initial state of charge of the energy storage. This is a state variable for electricity purchase; 1 indicates that electricity is purchased from the grid. For the maximum power to interact with the power grid, This is a power sales status variable; 1 indicates that power is sold to the grid.
[0055] Preferably, the robust power balance constraint includes:
[0056] For any permissible wind and solar power output scenario, the power balance constraint must hold:
[0057]
[0058]
[0059] uncertain sets Substituting the worst-case scenario into the minimum power output of wind and solar power, we introduce an integer variable to indicate whether the worst-case scenario has been achieved during this period. Then the robust power balance constraint is expressed as:
[0060]
[0061] Simplifying the above equation, we get:
[0062]
[0063] in, For load demand.
[0064] Preferably, the uncertainty budget constraint is expressed as:
[0065]
[0066] in, An integer variable representing whether the worst-case scenario is reached during this period. This is a budget parameter for uncertainty that reflects the risk preferences of decision-makers.
[0067] Preferably, the mixed-integer linear programming problem is expressed as:
[0068]
[0069] Deterministic constraints: gas turbine operation constraints, energy storage system constraints, and grid interaction constraints;
[0070]
[0071]
[0072]
[0073] in, It is a linear combination of all cost terms in the objective function. It includes all continuous decision variables, and the deterministic constraints include all equality and inequality constraints without uncertain parameters.
[0074] Preferably, solving the mixed-integer linear programming problem includes:
[0075] By setting different uncertainty budget parameters and solving them separately, a series of scheduling schemes with different protection levels are obtained;
[0076] Based on the comparison between theoretical operating costs and actual operating costs under worst-case scenarios, the optimal uncertainty budget parameter is selected as the robustness parameter of the scheduling scheme.
[0077] Secondly, the present invention provides a robust optimization scheduling device for virtual power plants participating in the carbon electricity market based on uncertainty, comprising: a data acquisition module, an uncertainty set construction module, an objective function minimization module, a constraint condition construction module, and a scheduling scheme solution module;
[0078] The data acquisition module is configured to acquire day-ahead forecast data of the virtual power plant;
[0079] The uncertainty set construction module is configured to construct an uncertain set of wind and solar power output based on the day-ahead forecast data; wherein, the uncertain set adopts an improved uncertain set form based on uncertainty budget, with the forecast error of total wind and solar power output as the uncertainty parameter, and the uncertainty budget parameter limits the number of time periods in which the worst case is obtained simultaneously within the scheduling cycle;
[0080] The objective function minimization module is configured to construct an objective function with the goal of minimizing the total operating cost of the virtual power plant; wherein the total operating cost includes electricity trading costs, gas turbine operating costs, and carbon trading costs;
[0081] The constraint construction module is configured to construct constraints for the virtual power plant. The constraints include deterministic constraints, robust power balance constraints, and uncertainty budget constraints. The robust power balance constraints require that real-time power balance is maintained for all possible wind and solar power output scenarios within the uncertainty set.
[0082] The scheduling scheme solving module is configured to transform the objective function and the constraints into a mixed integer linear programming problem for solving, so as to obtain the optimal scheduling scheme for each time period of the virtual power plant.
[0083] As can be seen from the above technical solution, the robust optimization scheduling method for virtual power plants participating in the electricity carbon market based on uncertainty, provided in this embodiment of the invention, firstly acquires the day-ahead forecast data of the virtual power plant, such as wind power output forecast, photovoltaic power output forecast, load demand forecast, electricity price data, and carbon price data. Then, based on the day-ahead forecast data, an uncertain set of wind and solar power output is constructed. This uncertain set adopts an improved uncertain set form based on uncertainty budgeting, using the forecast error of total wind and solar power output as the uncertainty parameter. The uncertainty budget parameter limits the number of time periods within the scheduling cycle where the worst-case scenario is simultaneously obtained. Furthermore, an objective function is constructed with the goal of minimizing the total operating cost of the virtual power plant, including electricity trading costs, gas turbine operating costs, and carbon trading costs. Simultaneously, constraints on the virtual power plant are constructed, which may include deterministic constraints, robust power balance constraints, and uncertainty budget constraints. The robust power balance constraint requires that real-time power balance is maintained for all possible wind and solar power output scenarios within all uncertain sets. Finally, the objective function and constraints are transformed into a mixed-integer linear programming problem for solution, yielding the optimal scheduling scheme for each time period of the virtual power plant. Therefore, this scheme abandons the extreme assumption of all uncertainties worsening simultaneously in traditional methods, instead limiting the number of periods where adverse situations occur concurrently. It constructs a non-extreme, more realistic set of uncertainties through an uncertainty budget parameter. This parameter represents the maximum number of severe scenarios for which decision-makers are willing to reserve reserve capacity across all scheduling periods. This achieves precise control over the safety costs incurred by the scheduling scheme in dealing with uncertainty, thus realizing the optimal trade-off between economy and safety. Attached Figure Description
[0084] Figure 1 The flowchart illustrates a robust optimization scheduling method for virtual power plants participating in the electricity carbon market based on uncertainty, as provided in this embodiment of the invention.
[0085] Figure 2 This is a robustness-cost trade-off diagram.
[0086] Figure 3 This is a comparison chart of theoretical cost and worst-case actual cost. Detailed Implementation
[0087] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0088] The core objective of virtual power plant dispatch is to minimize the sum of costs associated with power trading strategies (purchasing / selling electricity), gas turbine start-up and shutdown, and carbon emission quota usage, while satisfying system power balance, equipment operating constraints, and power interaction limitations with the grid. Due to the uncertainty of wind and solar power output, virtual power plant dispatch decisions face risks. If actual wind and solar power output is lower than predicted, the virtual power plant needs to increase gas turbine output or purchase electricity from the grid to compensate for the power shortfall, increasing operating costs. If actual wind and solar power output is higher than predicted, the virtual power plant may need to curtail wind and solar power or sell electricity to the grid, impacting economic viability. Robust optimization methods consider the worst-case scenario for wind and solar power output, seeking dispatch schemes that still guarantee feasibility and economic viability even under the worst-case conditions.
[0089] In actual operation, virtual power plants face the severe challenge of uncertain wind and solar power output. Traditional deterministic optimization methods formulate scheduling plans based on predicted values. When actual wind and solar power output deviates from the predicted values, it may lead to power imbalance, increase operating costs, and even threaten system safety. While stochastic optimization methods can handle uncertainty, they require knowledge of the precise probability distribution of uncertain parameters, which is often difficult to obtain in practical applications. Robust optimization methods, as an effective means of handling uncertainty, do not require assumptions about the probability distribution of uncertain parameters; they only need to know their fluctuation range. By constructing an uncertainty set, they seek the optimal solution in the worst-case scenario, achieving optimal economic efficiency while ensuring system reliability. Specifically, for example... Figure 1 As shown, this embodiment of the invention provides a robust optimization scheduling method for virtual power plants participating in the electricity carbon market based on uncertainty. This method may include the following steps:
[0090] Step 101: Obtain the day-ahead forecast data for the virtual power plant;
[0091] Step 102: Construct an uncertain set of wind and solar power output based on the day-ahead forecast data; wherein, the uncertain set adopts an improved uncertain set form based on uncertainty budget, with the forecast error of total wind and solar power output as the uncertainty parameter, and the uncertainty budget parameter limits the number of time periods in which the worst case is obtained simultaneously within the scheduling cycle;
[0092] Step 103: Construct an objective function with the goal of minimizing the total operating cost of the virtual power plant; wherein the total operating cost includes electricity trading costs, gas turbine operating costs, and carbon trading costs;
[0093] Step 104: Construct constraints for the virtual power plant, including deterministic constraints, robust power balance constraints, and uncertainty budget constraints; the robust power balance constraints require that real-time power balance is maintained for all possible wind and solar power output scenarios within the uncertainty set.
[0094] Step 105: Transform the objective function and the constraints into a mixed-integer linear programming problem and solve it to obtain the optimal scheduling scheme for each time period of the virtual power plant.
[0095] The following provides a more detailed explanation of each of the above steps.
[0096] For step 101, obtain the day-ahead forecast data of the virtual power plant.
[0097] In this step, the day-ahead forecast data may include wind power output forecast, solar power output forecast, load demand forecast, electricity price data, and carbon price data, etc.
[0098] For step 102, an uncertain set of wind and solar power output is constructed based on the day-ahead forecast data; wherein, the uncertain set adopts an improved uncertain set form based on uncertainty budget, with the forecast error of total wind and solar power output as the uncertainty parameter, and the number of time periods in the scheduling cycle that simultaneously obtain the worst case is limited by the uncertainty budget parameter.
[0099] In this step, the robust optimization method based on uncertainty budgeting proposed by Bertsimas and Sim is used to handle the uncertainty of wind and solar power output. This method introduces a robustness parameter to control the conservatism of the solution, achieving a flexible trade-off between conservatism and economy.
[0100] Specifically, the predicted output values for wind power and solar power are as follows: and The actual output can be expressed as the sum of the predicted value and the prediction error. Let the uncertainty magnitude be... Wind power prediction error And photovoltaic prediction error It typically fluctuates within a certain range, as shown below:
[0101]
[0102] This article takes This means that the power output of wind and solar power can fluctuate by ±20% based on the predicted value.
[0103] Furthermore, traditional robust optimization typically employs a box-type uncertainty set, assuming that the uncertain parameters in all time periods may simultaneously take the worst-case scenario:
[0104]
[0105] While the box-type uncertainty set is simple in form, it is overly conservative. It assumes that wind and solar power output reaches its worst-case scenario simultaneously in all time periods, which is virtually impossible in reality, leading to poor scheduling efficiency. To overcome the over-conservatism of the box-type set, this scheme introduces an uncertainty budget parameter to limit the number of time periods where the worst-case scenario occurs simultaneously. The total wind and solar power output is defined as... Then its predicted value is The improved uncertain set can be represented as:
[0106]
[0107] in, For improved uncertain sets, Contribute to the overall scenery This is the predicted value for the total power output of wind and solar power. To contribute to wind power, This is the predicted value of wind power output. Contribute to photovoltaic power Forecast values of photovoltaic power output As a fluctuation indicator variable, For the range of uncertainty, The uncertainty budget parameter reflects the decision-maker's risk preference and takes values in the range [0, T0]. This indicates that the solar energy output during that period reached the worst-case scenario (lower limit). This indicates that the output equals the predicted value. Uncertainty budget. It is an integer parameter representing the maximum number of time periods allowed to be taken in the worst-case scenario. The value reflects the decision-maker's risk preference: This corresponds to deterministic optimization, without considering any uncertainties; The corresponding conservative robust optimization is equivalent to a box-type uncertain set; This corresponds to a trade-off between conservatism and economy.
[0108] For step 103, an objective function is constructed with the goal of minimizing the total operating cost of the virtual power plant; wherein the total operating cost includes electricity trading costs, gas turbine operating costs, and carbon trading costs.
[0109] In this step, the objective function can be expressed as:
[0110]
[0111] in, Total operating cost, For electricity transaction costs, For the operating costs of gas turbines, This refers to the cost of carbon trading.
[0112] Specifically, in one embodiment, the electricity transaction cost It can be represented as:
[0113]
[0114] Where T is the scheduling period (24 hours). The electricity price for time period t. The power purchased during time period t. Let t represent the electricity sales volume during time period t.
[0115] In one embodiment, the gas turbine operating cost includes fuel costs, solid fuel operating costs, and aerodynamic costs. Specifically, the gas turbine operating cost... It can be represented as:
[0116]
[0117] in, This is the fuel cost coefficient. For the gas turbine output during time period t, To fix operating costs, This represents the operating state of the gas turbine (0-1 variables). For start-up costs, This is a startup status indicator variable.
[0118] In one embodiment, the carbon trading cost is levied on carbon emissions exceeding the free allowance, and can be expressed as:
[0119]
[0120] in, For carbon price, This refers to excess carbon emissions. Excess carbon emissions consist of two parts: emissions from gas turbines and emissions from electricity purchases, which can be specifically expressed as:
[0121]
[0122] in, For carbon emission factors of gas turbines, Carbon emission factors from purchasing electricity from the power grid Free carbon credits.
[0123] For step 104, construct the constraints of the virtual power plant. The constraints include deterministic constraints, robust power balance constraints, and uncertainty budget constraints. The robust power balance constraints require that real-time power balance is achieved for all possible wind and solar power output scenarios within the uncertainty set.
[0124] In this step, the deterministic constraints may include:
[0125] The output of a gas turbine should be between its minimum and maximum output.
[0126]
[0127] Gas turbine start-up constraints:
[0128]
[0129] Energy storage charging and discharging power constraints:
[0130]
[0131]
[0132] Simultaneous charging and discharging are prohibited:
[0133]
[0134] Energy storage energy balance constraints:
[0135]
[0136] Energy storage energy range constraints:
[0137]
[0138] Initial and final energy equality constraint:
[0139]
[0140] Power purchase and sale constraints:
[0141]
[0142]
[0143] Simultaneous purchase and sale of electricity is prohibited:
[0144]
[0145] in, and These represent the minimum and maximum output values of the gas turbine. This is a charging state variable; 1 indicates charging. The energy storage charging power during time period t. For maximum charge / discharge power of energy storage, The energy storage discharge power during time period t. This is a discharge state variable, where 1 represents discharge. The state of charge of the stored energy at the end of time period t. For energy storage charging efficiency, For scheduling time intervals, For energy storage and discharge efficiency, For the minimum permissible state of charge of energy storage, For the maximum permissible state of charge of energy storage, This represents the initial state of charge of the energy storage. This is a state variable for electricity purchase; 1 indicates that electricity is purchased from the grid. For the maximum power to interact with the power grid, This is a power sales status variable; 1 indicates that power is sold to the grid.
[0146] In one embodiment, the power balance constraint is the only constraint containing uncertain parameters. For any permissible wind and solar power output scenario, the power balance constraint must hold:
[0147]
[0148]
[0149] uncertain sets Substituting this into the equation, the worst-case scenario corresponds to the minimum output of wind and solar power, because in this case, the system needs more output from conventional units or to purchase electricity to meet the load, resulting in the highest operating costs. An integer variable is introduced to indicate whether the worst-case scenario has occurred during this period. Then the robust power balance constraint is expressed as:
[0150]
[0151] By rearranging the above equation and moving the definite terms to the right side, we get:
[0152]
[0153] in, For load demand.
[0154] In one embodiment, the uncertainty budget constraint is expressed as:
[0155]
[0156] in, An integer variable representing whether the worst-case scenario is reached during this period. This is a budget parameter for uncertainty that reflects the risk preferences of decision-makers.
[0157] For step 105, the objective function and the constraints are transformed into a mixed integer linear programming problem and solved to obtain the optimal scheduling scheme for each time period of the virtual power plant.
[0158] In this step, the robust optimization scheduling model for virtual power plants participating in the electricity carbon market can be transformed into the following mixed-integer linear programming problem:
[0159]
[0160] Deterministic constraints: gas turbine operation constraints, energy storage system constraints, and grid interaction constraints;
[0161]
[0162]
[0163]
[0164] in, It is a linear combination of all cost terms in the objective function. Includes all continuous decision variables ( , , , , , (etc.), the determination constraint includes all equality and inequality constraints that do not contain uncertain parameters.
[0165] In the specific solution process, different uncertainty budget parameters can be set to solve the problem separately, resulting in a series of scheduling schemes with different levels of protection. Based on the comparison between the theoretical operating cost and the actual operating cost under the worst-case scenario, the optimal uncertainty budget parameter is selected as the robustness parameter of the scheduling scheme.
[0166] As can be seen from the above, when a virtual power plant formulates its day-ahead dispatch plan, it is assumed that there are at most [number missing] [units missing]. The worst-case scenario where wind and solar power output reaches its lower limit during a given period is used as the basis for scheduling gas turbine output and power purchase and sale plans. This assumption essentially sets up an adjustable risk buffer for the uncertainty of wind and solar power output. The larger the capacity, the more backup capacity the system reserves, and the stronger its defense against extreme scenarios. However, the operating costs of the gas turbine, such as fuel costs and electricity purchase costs, will also increase accordingly. The smaller the value, the closer the dispatch plan is to a deterministic scenario, and the better the economic efficiency; however, the higher the risk of power shortages in actual operation. Adjustment The value of can flexibly control the conservatism of the scheduling scheme, achieving an optimal trade-off between system reliability and economy. In summary, this robust optimization model is a mixed-integer linear programming problem. Based on the above model, the intlinprog function in MATLAB R2016b is used to solve the mixed-integer linear programming problem. First, the solution parameters are set, and then, combined with the previously set equipment parameters and constraints, it is determined whether the convergence condition is met. The results are compared by budgeting under different uncertainties. The scheduling results are then analyzed, and finally, the model's economy, low carbon emissions, and robustness are assessed.
[0167] In addition, the present invention provides a robust optimization scheduling device for virtual power plants participating in the carbon electricity market based on uncertainty, comprising: a data acquisition module, an uncertainty set construction module, an objective function minimization module, a constraint condition construction module, and a scheduling scheme solution module;
[0168] The data acquisition module is configured to acquire day-ahead forecast data of the virtual power plant;
[0169] The uncertainty set construction module is configured to construct an uncertain set of wind and solar power output based on the day-ahead forecast data; wherein, the uncertain set adopts an improved uncertain set form based on uncertainty budget, with the forecast error of total wind and solar power output as the uncertainty parameter, and the uncertainty budget parameter limits the number of time periods in which the worst case is obtained simultaneously within the scheduling cycle;
[0170] The objective function minimization module is configured to construct an objective function with the goal of minimizing the total operating cost of the virtual power plant; wherein the total operating cost includes electricity trading costs, gas turbine operating costs, and carbon trading costs;
[0171] The constraint construction module is configured to construct constraints for the virtual power plant. The constraints include deterministic constraints, robust power balance constraints, and uncertainty budget constraints. The robust power balance constraints require that real-time power balance is maintained for all possible wind and solar power output scenarios within the uncertainty set.
[0172] The scheduling scheme solving module is configured to transform the objective function and the constraints into a mixed integer linear programming problem for solving, so as to obtain the optimal scheduling scheme for each time period of the virtual power plant.
[0173] The following specific examples will further illustrate this scheme.
[0174] For a 24-hour scheduling cycle, the value of Γ ranges from 0 to 24.
[0175] 1) When Γ=0: This means that no uncertainty is considered, the model degenerates into deterministic optimization, which is the most economical, but the safety margin is zero.
[0176] 2) When Γ=24: This means that the worst-case scenario needs to be defended for all time periods. The model degenerates into the traditional box-type robust optimization, which has the highest security but the worst economy.
[0177] 3) When Γ takes the median value (preferably 6 in this invention): This means that the model assumes that within 24 hours, there will be a maximum of 6 periods where the solar power output will reach the predicted lower limit. This is equivalent to purchasing a "safety insurance" policy for the system, using a pre-controlled increase in costs to avoid high penalty costs in extreme scenarios. This method allows decision-makers to flexibly adjust the value of Γ according to their own risk preferences.
[0178] The effectiveness of the robust optimization strategy in this scheme is further verified through simulation analysis. Two comparison scenarios are set up here:
[0179] Scenario 1 (Deterministic Scenario): In this scenario, the robustness parameter Γ is set to 0, and the uncertainty of new energy output is not taken into account, assuming that the predicted value is completely accurate.
[0180] Scenario 2 (Uncertainty Scenario): In this scenario, the robustness parameter Γ is set to 6, while considering that there is a prediction error of ±20% in the output of new energy sources.
[0181] For the power output structure:
[0182] When Γ=0: the utilization rate of wind and solar power is high, the output of gas turbine is low, the system is "wind and solar dependent", the economy is optimal but there are risks.
[0183] When Γ=6: To cope with uncertainty, the proportion of wind and solar power output decreases, gas turbine output increases, and energy storage shifts from "arbitrage" to "backup". The system transforms into a "multi-source coordinated stability type", with a slight decrease in economic efficiency but an improvement in safety.
[0184] For energy storage dispatch strategies:
[0185] When Γ=0: Energy storage mainly engages in peak-valley arbitrage, charging during valley hours and discharging during peak hours, resulting in drastic fluctuations in the SOC curve and no reserved backup capacity.
[0186] When Γ=6: the energy storage strategy switches to safety backup, completes charging ahead of schedule, accelerates the discharge pace, and reserves buffer space to cope with the fluctuations in new energy sources.
[0187] For grid interaction power:
[0188] When Γ=0: VPP is a "surplus generator", which mainly sells electricity to the grid to generate revenue.
[0189] When Γ=6: VPP’s electricity sales activities decrease significantly, and a large amount of electricity needs to be purchased during the evening peak to fill the power gap. The role of the power grid changes from “source of revenue” to “safe backup power source”.
[0190] For carbon emission analysis:
[0191] When Γ=0: carbon emissions are low, with an excess emission of 184,062.9 kg, but this is on the premise of ignoring prediction errors and has a high operational risk.
[0192] At Γ=6: Due to high-load operation of the gas turbine and increased electricity purchases, excess emissions rise to 221,710.4 kg, an increase of 20.4% compared to the deterministic scenario. This indicates that robust optimization, while improving safety, comes at the cost of increased carbon emissions.
[0193] like Figure 2 The diagram shows the robustness-cost trade-off curve of a virtual power plant, visually illustrating the impact of the robustness coefficient Γ on the total system cost. As Γ increases from 0 to 12, the total system cost increases approximately linearly: under deterministic scheduling, wind and solar utilization is highest, resulting in the lowest total system cost of -318,988 yuan; when Γ increases to 6, the total cost is -259,242 yuan, an increase of approximately 18.7% compared to the deterministic scenario; and when Γ=12, the total cost reaches -199,370 yuan, exhibiting the strongest robustness but the worst economic efficiency. This result demonstrates that the improved safety brought about by robustness optimization comes at the cost of economic efficiency. The linearity of the curve reflects that, under the current parameters, the marginal cost of robustness is relatively stable; in other words, for every unit increase in the robustness parameter, the system pays a clear "safety premium." Based on the comparative analysis of power structure, energy storage operation, grid interaction, and carbon emissions above, it can be found that as Γ increases, the system's operating strategy will shift from "economic priority" to "safety priority." The proportion of wind and solar power output will decrease, the backup role of gas turbines and grid power purchases will strengthen, energy storage will shift from an arbitrage tool to a backup buffer, and carbon emission levels will also rise accordingly.
[0194] In summary, the value of Γ=6 represents a good balance between economy and robustness in this example: compared to deterministic dispatch, its cost increase is acceptable (approximately 18.7%), while the system's ability to cope with fluctuations in renewable energy output is substantially improved, and the increase in carbon emissions is within an acceptable range. This conclusion provides a quantitative basis for selecting appropriate robustness parameters in the actual operation of virtual power plants.
[0195] By adjusting the robustness coefficient Γ, virtual power plant (VPP) optimization strategies with different robustness can be obtained, where Γ represents the ability of the decision scheme to withstand risks. It plays a crucial role in determining the VPP optimization strategy, helping decision-makers understand the impact of different values of Γ on the optimization strategy and formulate more reasonable optimization strategies based on their risk appetite and revenue objectives. Table 1 (negative costs indicate benefits) lists the changes in theoretical cost, worst-case actual cost, and power deficit of the virtual power plant under different values of Γ.
[0196] Table 1
[0197] 0 -318987.60 -61087.60 257900.00 4 -279181.25 -107247.92 171933.33 6 -259241.97 -130291.97 128950.00 8 -239302.69 -153336.02 85966.67 12 -199369.84 -199369.84 0.00
[0198] like Figure 3 This graph compares the theoretical cost of a virtual power plant in the predicted scenario and the actual operating cost in the worst-case scenario under different robustness coefficients. The graph shows a significant difference between deterministic and robust dispatch when facing uncertainties in renewable energy. In the deterministic dispatch scenario, the system operates based on wind and solar power forecasts, resulting in the lowest theoretical cost and a significant economic advantage. However, in the worst-case scenario of insufficient wind and solar power output, the power deficit triggers a high penalty for power shortages, causing the actual operating cost to spike dramatically. The original theoretical economic advantage is completely lost, demonstrating strong operational vulnerability. As the robustness coefficient Γ gradually increases, the theoretical cost rises slightly, but the actual operating cost decreases significantly, and the difference between the two continuously narrows. When Γ=6, the actual cost of the system in the worst-case scenario is significantly lower than in the scenario where Γ=0. This is equivalent to buying "insurance" for the virtual power plant, paying a pre-controllable premium to reduce the high penalty costs caused by wind and solar power output deviations. This achieves an effective balance between theoretical cost and actual risk. When Γ=12, the theoretical cost and the actual cost almost overlap, indicating that the robust scheduling has completely eliminated the risk of power deficit, and the system can still operate stably even in the worst-case scenario. This comparison result perfectly verifies the core value of robust optimization: by moderately sacrificing the economics in the prediction scenario, the operational risks brought about by the uncertainty of new energy sources can be significantly reduced, thereby avoiding the catastrophic costs that deterministic scheduling may face. Considering both cost and risk, the value of Γ=6 is an optimal compromise that balances economy and safety in this example.
[0199] In summary, the robust optimization scheduling method and apparatus for virtual power plants participating in the electricity carbon market based on uncertainty provided by this invention has at least the following beneficial effects:
[0200] (1) This invention solves the problem of traditional methods being "either too risky or too conservative" by adjusting the robustness parameter Γ. Taking the preferred embodiment as an example, when Γ=6, the actual operating cost of the virtual power plant in the worst scenario is -130,291.97 yuan, which is much higher than the actual cost of deterministic optimization (Γ=0) in the worst scenario, which is -61,087.60 yuan (a negative cost indicates a benefit, and the larger the value, the higher the benefit). The system power deficit is reduced significantly from the deterministic 257,900 kWh to 128,950 kWh, and the operational safety is significantly improved.
[0201] (2) This invention incorporates risk management into the scheduling framework. Through the robustness-cost trade-off curve, virtual power plant operators can intuitively select an optimal Γ value based on their own risk tolerance (such as prediction deviation) and market expectations. Experiments show that as Γ increases from 0 to 12, the total system cost increases approximately linearly (from -318,988 yuan to -199,370 yuan), providing a clear calculation basis for quantifying the "safety premium".
[0202] (3) This invention demonstrates that moderately sacrificing the theoretical economics of the predicted scenario (when Γ=6, the theoretical cost is about 18.7% higher than that of the deterministic scenario) can significantly reduce the catastrophic cost risk caused by wind and solar power output deviations in actual operation. This strategy of “avoiding uncontrollable large risks with controllable small costs” effectively improves the reliability of the scheduling plan and the actual execution effect.
[0203] (4) The robust optimization model based on uncertainty budgeting proposed in this invention can ultimately be transformed into a standard mixed-integer linear programming problem, which can be solved efficiently using mature commercial solvers (such as the intlinprog function in MATLAB R2016b), and has good computational feasibility and engineering promotion value. Compared with stochastic optimization that requires the generation of complex scenarios, this method has higher computational efficiency and does not rely on probability distribution information that is difficult to obtain accurately.
[0204] This specification also provides a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to perform the methods in any of the embodiments of the specification.
[0205] This specification also provides a computing device, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, it implements the method in any of the embodiments of the specification.
[0206] Since the device embodiments and method embodiments of the present invention are based on the same inventive concept, detailed descriptions can be found in the method embodiments, and will not be repeated here.
[0207] The modules or units in the device of this invention can be merged, divided, and deleted according to actual needs. The above-disclosed embodiments are merely preferred embodiments of the present invention and should not be construed as limiting the scope of the invention. Those skilled in the art will understand that implementing all or part of the processes of the above embodiments and making equivalent changes according to the claims of this invention still fall within the scope of the invention.
Claims
1. A robust optimization scheduling method for virtual power plants participating in the electricity carbon market based on uncertainty, characterized in that, include: Obtain day-ahead forecast data for virtual power plants; An uncertain set of wind and solar power output is constructed based on the aforementioned day-ahead forecast data; wherein, the uncertain set adopts an improved uncertain set form based on uncertainty budget, with the forecast error of total wind and solar power output as the uncertainty parameter, and the uncertainty budget parameter limits the number of time periods in which the worst-case scenario is simultaneously obtained within the scheduling cycle; An objective function is constructed with the goal of minimizing the total operating cost of the virtual power plant; wherein the total operating cost includes electricity trading costs, gas turbine operating costs, and carbon trading costs. The constraints for constructing the virtual power plant include deterministic constraints, robust power balance constraints, and uncertainty budget constraints; the robust power balance constraints require that real-time power balance is maintained for all possible wind and solar power output scenarios within the uncertainty set. The objective function and constraints are transformed into a mixed-integer linear programming problem and solved to obtain the optimal scheduling scheme for the virtual power plant in each time period.
2. The robust optimization scheduling method for virtual power plants participating in the electricity carbon market based on uncertainty, as described in claim 1, is characterized in that... The construction of the uncertain set of wind and solar power output includes: Define the total output of the landscape as Then its predicted value is The improved uncertain set can be represented as: ； in, For improved uncertain sets, Contribute to the overall scenery This is the predicted value for the total power output of wind and solar power. To contribute to wind power, This is the predicted value of wind power output. Contribute to photovoltaic power This is the predicted value for photovoltaic power output. As a fluctuation indicator variable, For the range of uncertainty, The uncertainty budget parameter reflects the decision-maker's risk preference and takes values in the range [0, T0].
3. The robust optimization scheduling method for virtual power plants participating in the electricity carbon market based on uncertainty, as described in claim 2, is characterized in that... The objective function is expressed as: ； in, Total operating cost, For electricity transaction costs, For the operating costs of gas turbines, This refers to the cost of carbon trading.
4. The robust optimization scheduling method for virtual power plants participating in the electricity carbon market based on uncertainty, as described in claim 3, is characterized in that... The electricity transaction costs Represented as: ； Where T is the scheduling period. The electricity price for time period t. The power purchased during time period t. Let t represent the electricity sales volume during time period t; The operating cost of the gas turbine Represented as: ； in, This is the fuel cost coefficient. For the gas turbine output during time period t, To fix operating costs, This refers to the operating status of the gas turbine. For start-up costs, This is a startup status indicator variable; The carbon trading costs Represented as: ； in, For carbon price, Excess carbon emissions are expressed as: ； in, For carbon emission factors of gas turbines, Carbon emission factors from purchasing electricity from the power grid Free carbon credits.
5. The robust optimization scheduling method for virtual power plants participating in the electricity carbon market based on uncertainty, as described in claim 4, is characterized in that... The deterministic constraints include: The output of a gas turbine should be between its minimum and maximum output. ； Gas turbine start-up constraints: ； Energy storage charging and discharging power constraints: ； ； Simultaneous charging and discharging are prohibited: ； Energy storage energy balance constraints: ； Energy storage energy range constraints: ； Initial and final energy equality constraint: ； Power purchase and sale constraints: ； ； Simultaneous purchase and sale of electricity is prohibited: ； in, and These represent the minimum and maximum output values of the gas turbine. This is a charging state variable; 1 indicates charging. The energy storage charging power during time period t. For maximum charge / discharge power of energy storage, The energy storage discharge power during time period t. This is a discharge state variable, where 1 represents discharge. The state of charge of the stored energy at the end of time period t. For energy storage charging efficiency, For scheduling time intervals, For energy storage and discharge efficiency, For the minimum permissible state of charge of energy storage, For the maximum permissible state of charge of energy storage, This represents the initial state of charge of the energy storage. This is a state variable for electricity purchase; 1 indicates that electricity is purchased from the grid. For the maximum power to interact with the power grid, This is a power sales status variable; 1 indicates that power is sold to the grid.
6. The robust optimization scheduling method for virtual power plants participating in the electricity carbon market based on uncertainty, as described in claim 5, is characterized in that... The robust power balance constraint includes: For any permissible wind and solar power output scenario, the power balance constraint must hold: ； ； uncertain sets Substituting the worst-case scenario into the minimum power output of wind and solar power, we introduce an integer variable to indicate whether the worst-case scenario has been achieved during this period. Then the robust power balance constraint is expressed as: ； Simplifying the above equation, we get: ； in, For load demand.
7. The robust optimization scheduling method for virtual power plants participating in the electricity carbon market based on uncertainty, as described in claim 6, is characterized in that... The uncertainty budget constraint is expressed as: ； in, An integer variable representing whether the worst-case scenario is reached during this period. This is a budget parameter for uncertainty that reflects the risk preferences of decision-makers.
8. The robust optimization scheduling method for virtual power plants participating in the electricity carbon market based on uncertainty, as described in claim 7, is characterized in that... The mixed-integer linear programming problem is expressed as: ； Deterministic constraints: gas turbine operation constraints, energy storage system constraints, and grid interaction constraints; ； ； ； in, It is a linear combination of all cost terms in the objective function. It includes all continuous decision variables, and the deterministic constraints include all equality and inequality constraints without uncertain parameters.
9. The robust optimization scheduling method for virtual power plants participating in the electricity carbon market based on uncertainty, as described in claim 8, is characterized in that... Solving the mixed-integer linear programming problem includes: By setting different uncertainty budget parameters and solving them separately, a series of scheduling schemes with different protection levels are obtained; Based on the comparison between theoretical operating costs and actual operating costs under worst-case scenarios, the optimal uncertainty budget parameter is selected as the robustness parameter of the scheduling scheme.
10. A robust optimization scheduling device for virtual power plants participating in the electricity carbon market based on uncertainty, characterized in that, include: The module includes a data acquisition module, an uncertain set construction module, an objective function minimization module, a constraint construction module, and a scheduling scheme solution module. The data acquisition module is configured to acquire day-ahead forecast data of the virtual power plant; The uncertainty set construction module is configured to construct an uncertain set of wind and solar power output based on the day-ahead forecast data; wherein, the uncertain set adopts an improved uncertain set form based on uncertainty budget, with the forecast error of total wind and solar power output as the uncertainty parameter, and the uncertainty budget parameter limits the number of time periods in which the worst case is obtained simultaneously within the scheduling cycle; The objective function minimization module is configured to construct an objective function with the goal of minimizing the total operating cost of the virtual power plant; wherein, the total operating cost includes electricity trading costs, gas turbine operating costs, and carbon trading costs; The constraint construction module is configured to construct constraints for the virtual power plant. The constraints include deterministic constraints, robust power balance constraints, and uncertainty budget constraints. The robust power balance constraints require that real-time power balance is maintained for all possible wind and solar power output scenarios within the uncertainty set. The scheduling scheme solving module is configured to transform the objective function and the constraints into a mixed integer linear programming problem for solving, so as to obtain the optimal scheduling scheme for each time period of the virtual power plant.