A master-slave game optimization method for energy storage leasing energy system suitable for wind power fluctuation scene

Abstract

The application discloses a kind of energy storage leasing energy system master-slave game optimization methods suitable for wind power fluctuation scene, belong to power grid dispatching technical field.The method steps include: S1, energy system is divided into upper distribution network operator, middle pumped storage operator and lower microgrid group alliance, and three-layer master-slave game optimization model is constructed;S2, the equilibrium solution is obtained by solving the master-slave game optimization model, and the energy system is dispatched based on the equilibrium solution.The application constructs the three-layer master-slave game model of distribution network operator-pumped storage operator-microgrid group alliance, and realizes the collaborative decision and benefit coordination among multiple subjects by taking time-of-use electricity purchasing and selling price and energy storage leasing price as unified price signal.

Description

Technical Field

[0001] This invention belongs to the field of power grid dispatching technology, specifically relating to a master-slave game optimization method for energy storage leasing energy systems applicable to wind power fluctuation scenarios. Background Technology

[0002] With increasing demand for fossil fuels, carbon emissions have risen significantly. Therefore, the efficient utilization of renewable energy is of significant research value, and traditional single-energy systems are evolving towards integrated energy systems. With the integration of renewable energy, the randomness and volatility of high-proportion wind power output exacerbate the difficulty of grid integration and amplify the uncertainty of distribution network operation, hindering the economic and safe operation of the system. Current research on microgrid group game optimization mainly considers carbon trading costs and network security boundaries, constructing an optimization scheduling model that takes into account wind power uncertainty. This model explores demand response strategies that combine energy storage leasing with adjustable load resources, addressing both the generation and load sides, and studies these strategies from the perspectives of price signals, capacity allocation, and risk constraints.

[0003] Existing game-theoretic optimization methods for microgrids with a high proportion of wind turbines mostly employ deterministic optimization, neglecting the impact of the volatility of wind power generation and the uncertainties in the complex relationships of multi-layered master-slave games on the system. Secondly, traditional battery energy storage models suffer from poor investment efficiency, low energy storage utilization during certain periods, and small installed capacity. While they can absorb renewable energy and achieve some peak shaving and valley filling, they cannot effectively reduce the investment and operating costs of microgrid clusters. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing game optimization methods that do not consider the uncertainty of wind power generation and traditional energy storage models, and to provide a master-slave game optimization method for energy storage leasing energy systems applicable to wind power fluctuation scenarios.

[0005] To achieve the above objectives, the present invention provides the following solution: a master-slave game optimization method for energy storage leasing energy systems applicable to wind power fluctuation scenarios, comprising the following steps: S1. Divide the energy system into an upper-level distribution network operator, a middle-level pumped storage operator, and a lower-level microgrid group alliance, and construct a three-level master-slave game optimization model. S2. Solve the master-slave game optimization model to obtain the equilibrium solution, and schedule the energy system based on the equilibrium solution.

[0006] More preferably, the master-slave game optimization model includes: a distribution network operator model, a pumped storage operator model, and a microgrid group alliance model; the distribution network operator model and the pumped storage operator model constitute an outer master-slave game; the pumped storage operator model and the microgrid group alliance model constitute an inner master-slave game.

[0007] More preferably, the distribution network operator model aims to minimize the total operating cost, and the objective function includes: ; In the formula, Indicates total operating cost; This represents the power interaction cost between the distribution network operator and the microgrid group alliance; This represents the electricity interaction cost between the distribution network operator and the pumped storage operator; This indicates the cost for distribution network operators to purchase electricity from the upstream power grid; in, ; In the formula, N Indicates the number of microgrids; T Indicates the index of the scheduling period; express t The electricity purchase price set by the power distribution network operator for a given time period; express t Electricity sales prices set by the time-of-use distribution network operator; microgrid n exist t Active power sold to distribution network operators during a given time period; microgrid n exist t Active power purchased from distribution network operators during a given time period; ; In the formula, This indicates that pumped storage operators are in t Active power sold to distribution network operators during a given time period; This indicates that pumped storage operators are in t Active power purchased from distribution network operators during a given time period; ; In the formula, This indicates the settlement price given by the superior power grid; Indicates that the distribution network operator is t Active power purchased from the upper-level power grid during a given time period; The power distribution network operator model uses electricity price constraints to guide pumped storage operators and microgrid group alliances in making power purchase and sale decisions: ; ; In the formula, , , These represent the sets of peak, flat, and trough periods, respectively. , , These represent the unified electricity sales price decision variables corresponding to peak, flat, and low periods, respectively, with the superscripts max and min indicating the upper and lower bounds, respectively. This indicates the average upper limit of electricity sales prices.

[0008] More preferably, the pumped storage operator model aims to minimize its own daily net operating cost, and the objective function includes: ; In the formula, The daily net operating cost for pumped storage operators; This indicates the net cost settled between pumped storage operators and distribution network operators; This represents the operation and maintenance costs for pumped storage operators; This refers to the rental income collected by pumped storage operators from the microgrid cluster alliance; in, ; ; In the formula, The maintenance cost coefficient per unit power; , They represent t The charging and discharging power of the pumped storage operator's own energy storage section during a given period; ; In the formula, , They represent t Periodic pumped storage operators are microgrids n The established unit price for electricity capacity and unit price for electricity power; , They represent t Periodic pumped storage operators are microgrids n The leased power capacity and power quota.

[0009] More preferably, the constraints of the pumped storage operator model include: mutual exclusion constraints on distribution network transactions, dynamic constraints on state of charge, power conservation constraints, and lease price constraints. The mutual exclusion constraints for transactions in the distribution network include:

[0010] In the formula, This indicates the rated power of the pumped storage operator's own energy storage section; The dynamic constraints on the state of charge include:

[0011] In the formula, express t Energy status of pumped storage units during specific time periods; Indicates the time step; , These represent the pumping efficiency and power generation efficiency of the pumped storage unit, respectively. , These represent the upper and lower bounds of the charged state of a pumped storage unit, respectively. The power conservation constraints include: ; The rental price constraints include: .

[0012] More preferably, the microgrid group alliance model aims to minimize daily operating costs, and the objective function includes: ; In the formula, , , , , , In order, the costs are: power purchase and sale interaction costs between the microgrid group alliance and the distribution operator, power generation costs of micro gas turbines, power generation costs of renewable energy, external pumped storage leasing fees, demand response compensation costs, and renewable energy curtailment penalty costs. in, ; In the formula, , These represent the power generation cost coefficients for photovoltaic and wind turbines, respectively. , They represent t Time-of-use microgrids n The output of photovoltaic and wind turbines;

[0013] In the formula, express t Time-of-use microgrid n The equivalent wind speed in the area; , , They represent microgrids n The cut-in velocity, rated velocity, and cut-out velocity of the wind turbine unit; microgrid n Rated power of the fan unit.

[0014] More preferably, the constraints of the microgrid cluster alliance model include: microgrid power balance constraints, renewable energy curtailment ratio constraints, transferable load constraints, load reduction constraints, and micro gas turbine output constraints;

[0015] ; ;

[0016] ; In the formula, express t Time period by microgrid q To microgrids n The power transmission capacity, express t Time period by microgrid n To microgrids q The power transmission capacity.

[0017] More preferably, in S2, the method for solving the master-slave game optimization model to obtain the equilibrium solution includes: S21. Given a set of upper-level price decisions in the outer master-slave game, solve the optimal response of the pumped storage operator and the robust optimal response of the microgrid group alliance in the inner master-slave game respectively. S22. Iteratively update the upper-level price decision based on the approximate gradient update strategy until convergence, and obtain the equilibrium solution.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention constructs a three-tiered master-slave game model involving a distribution network operator, pumped storage operator, and microgrid cluster alliance. It uses time-of-use electricity pricing and energy storage leasing prices as unified price signals to achieve collaborative decision-making and interest coordination among multiple stakeholders. This mechanism effectively characterizes the interaction relationships between stakeholders at different levels, avoiding system imbalances caused by single-stakeholder or single-layer optimization, and improving the overall economic efficiency and operational stability of the integrated energy system. Attached Figure Description

[0019] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a game structure diagram of an embodiment of the present invention; Figure 2 This is a schematic diagram of the master-slave game optimization method for energy storage leasing energy systems applicable to wind power fluctuation scenarios, provided in an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0023] Example 1: like Figure 1 , Figure 2 As shown in the figure, this embodiment provides a master-slave game optimization method for energy storage leasing energy systems applicable to wind power fluctuation scenarios. The method will be described in detail in conjunction with the following steps.

[0024] S1. Divide the energy system into an upper-level distribution network operator, a middle-level pumped storage operator, and a lower-level microgrid group alliance, and construct a three-level master-slave game optimization model.

[0025] In this embodiment, the master-slave game optimization model includes: a distribution network operator model, a pumped storage operator model, and a microgrid group alliance model. The distribution network operator model and the pumped storage operator model constitute the outer master-slave game; the pumped storage operator model and the microgrid group alliance model constitute the inner master-slave game. The pumped storage operator is treated as an independent operator, and under the guidance of time-of-use pricing, it sets energy storage leasing prices, responds to the net charging and discharging demand of the microgrid group alliance, coordinates its own charging and discharging with grid purchases and sales, and balances operation and maintenance costs with leasing revenue.

[0026] The distribution network operator model aims to minimize total operating costs, and its objective function includes: (1) In the formula, Indicates total operating cost; This represents the power interaction cost between the distribution network operator and the microgrid group alliance; This represents the electricity interaction cost between the distribution network operator and the pumped storage operator; This indicates the cost for distribution network operators to purchase electricity from the upstream power grid.

[0027] in, (2) In the formula, n =1,..., N : Indicates the microgrid quantity index; t =1,..., T : Represents the index of the scheduling period; express t The electricity purchase price set by the power distribution network operator for a given time period; express t Electricity sales prices set by the time-of-use distribution network operator; microgrid n exist t Active power sold to distribution network operators during a given time period; microgrid n exist t Active power purchased from the distribution network operator during a given period.

[0028] (3) In the formula, This indicates that pumped storage operators are in t Active power sold to distribution network operators during a given time period; This indicates that pumped storage operators are in t Active power purchased from the distribution network operator during a given period.

[0029] (4) In the formula, This indicates the settlement price given by the superior power grid; Indicates that the distribution network operator is t Active power purchased from the upper-level power grid during a given period.

[0030] In the three-level game, distribution network operators formulate time-of-use electricity purchase and sale prices based on load-side demand and energy supply-side output plans, and use price signals to guide pumped storage operators in the middle level and microgrid groups in the lower level to make electricity purchase and sale decisions.

[0031] Specifically, the distribution network operator model uses electricity price constraints to guide pumped storage operators and microgrid alliances in making electricity purchase and sale decisions: (5) (6) In the formula, , , These represent the sets of peak, flat, and trough periods, respectively. , , These represent the unified electricity sales price decision variables corresponding to peak, flat, and low periods, respectively, with the superscripts max and min indicating the upper and lower bounds, respectively. This indicates the average upper limit of electricity sales prices.

[0032] Formula (5) uses time-of-use pricing as a control signal at the upper level to guide lower-level entities to form differentiated electricity purchase and sale strategies at different times. Formula (6) imposes an upper limit on the average unit price of electricity sales throughout the entire dispatch cycle to prevent excessive price increases, thus balancing economic efficiency with the acceptability of lower-level followers.

[0033] Given time-of-use pricing and microgrid demand, pumped storage operators aim to minimize their daily net operating costs. Therefore, the objective function of the pumped storage operator model includes: (7) In the formula, The daily net operating cost for pumped storage operators; This indicates the net cost settled between pumped storage operators and distribution network operators; This represents the operation and maintenance costs for pumped storage operators; This refers to the rental income collected by pumped storage operators from the microgrid cluster alliance.

[0034] in, (8) Operation and maintenance costs of pumped storage Based on its power flux, its expression is: (9) In the formula, The maintenance cost coefficient per unit power; , They represent t The charging and discharging power of the pumped storage operator's own energy storage section during specific time periods.

[0035] (10) In the formula, , They represent t Periodic pumped storage operators are microgrids n The established unit price for electricity capacity and unit price for electricity power; , They represent t Periodic pumped storage operators are microgrids n The leased power capacity and power quota.

[0036] Pumped storage operators need to determine their operating strategies based on the needs of the microgrid. Therefore, the different behavioral strategies of the microgrid cluster alliance are aggregated into a single net power information, and the aggregated net charging and discharging power is as follows: (11) In the formula, express t The net charge and discharge power of the microgrid group during the time period (sum of all microgrids), a positive value indicates "net charging of the microgrid group", and a negative value indicates "net discharging of the microgrid group"; express t Total charging power of the microgrid group during the time period; Indicate t Total discharge power of the microgrid group during the time period.

[0037] Given the demand on the microgrid side, the free energy storage of pumped storage operators should be directional and unidirectional (i.e., charging and discharging cannot coexist at the same time), and there is a power limit, with the following constraints: (12) In the formula, express t Net charging and discharging demand of microgrid clusters during specific time periods; This indicates the rated power of the pumped storage operator's own energy storage section.

[0038] Based on the characteristics of pumped storage units and the features of operators, the constraints of the pumped storage operator model include: mutual exclusion constraints on distribution network transactions, dynamic constraints on state of charge, power conservation constraints, and lease price constraints.

[0039] The mutual exclusion constraints for transactions in the distribution network include: (13) Dynamic constraints on the state of charge include: (14) In the formula, express t Energy status of pumped storage units during specific time periods; Indicates the time step; , These represent the pumping efficiency and power generation efficiency of the pumped storage unit, respectively. , These represent the upper and lower bounds of the charged state of a pumped storage unit, respectively. This represents the stored energy at the initial moment of the pumped storage unit's cycle (time 0). This indicates the rated energy capacity of the pumped storage unit.

[0040] Power conservation constraints include: (15) Rental price constraints include: (16) In the formula, , These represent the upper and lower limits of the unit price for electricity capacity, respectively. , These represent the upper and lower limits of the unit price of electricity, respectively.

[0041] Microgrid clusters need to be optimized to minimize daily operating costs, given upper-level and middle-level electricity purchase / sale prices and pumped storage leasing prices, while considering the characteristics of each unit, the capacity of pumped storage units, and the multi-source coupling of curtailment penalties. (17) In the formula, , , , , , In order, the costs are: power purchase and sale interaction costs between the microgrid group alliance and the distribution operator, power generation costs of micro gas turbines, power generation costs of renewable energy, external pumped storage leasing fees, demand response compensation costs, and renewable energy curtailment penalty costs.

[0042] in, (18) Formula (18) and Formula (2) are mirror pairs, using the same set of trading power but considering different entities (distribution network / microgrid group), so as to maintain settlement consistency at the game and system aggregation levels.

[0043] (19) In the formula, express t Time-of-use microgrids n The output of the micro gas turbine; , They represent microgrids n Fuel cost coefficient of gas turbine; =1, ..., J Pollutant type index; express j Penalty coefficient per unit emission of pollutants.

[0044] (20) In the formula, , These represent the power generation cost coefficients for photovoltaic and wind turbines, respectively. , They represent t Time-of-use microgrids n The output of photovoltaic and wind turbines.

[0045] .(twenty one) ;(twenty two) In the formula, , These represent the compensation unit prices for transferable load and load that can be reduced, respectively. , They represent t Time-of-use microgrids n Transferable load and load reduction.

[0046] In this embodiment, formula (10) represents the revenue of pumped storage operators from leasing energy storage to microgrid groups. This revenue is paid by the microgrid groups as their cost. Therefore, formula (10) is consistent with the expression of formula (21).

[0047] ;(twenty three) In the formula, , These represent the curtailment penalty coefficients for wind turbines and solar power, respectively. , They represent t Time-of-use microgrids n Wind / solar curtailment rate of wind turbines and photovoltaic units.

[0048] The load and renewable energy consumption are coupled into an equivalent load to evaluate the power surplus / deficit status of the microgrid, as expressed below: ;(twenty four) In the formula, express t Time-of-use microgrids n The equivalent load (<0 indicates insufficient power, >0 indicates surplus power); express t Time-of-use microgrids n Local load.

[0049] Based on the characteristics of each unit in the microgrid and the features of the operator, the constraints include: microgrid power balance constraints, renewable energy curtailment ratio constraints, transferable load constraints, load reduction constraints, and micro gas turbine output constraints; specifically as follows: (25) (26) (27) (28) (29) In the formula, express t Time period by microgrid q To microgrids n The power transmission capacity, express t Time period by microgrid n To microgrids q The power transmission capacity.

[0050] Based on the above description of the system operation mechanism, both pumped storage operators and microgrid alliances make optimal decisions based on pricing signals set by distribution network operators. Their optimization results, in turn, influence the power plant's pricing strategy, thus allowing for the solution of this model. However, this game theory model is essentially a deterministic optimization, and the superiority of the resulting scheduling scheme largely depends on the accuracy of the predictions.

[0051] In most wind-solar hybrid microgrids or active distribution networks, the installed capacity and output of wind turbines are often higher than those of solar power, especially in coastal areas, high-wind-speed areas, or areas rich in wind resources, where wind power has become one of the most important renewable power sources for microgrids. In the application scenario targeted by this invention, the installed capacity and energy share of wind power in renewable energy are significantly higher than those of solar power. Therefore, from a system perspective, the main source of overall power balance and dispatch risk is the uncertainty of wind power, while solar power plays a more supplementary role.

[0052] In the optimized scheduling model of this invention, a piecewise power curve based on nameplate data is used to determine the wind turbine output. Based on equivalent modeling, the relationship between its active power output and wind speed is as follows: (30) In the formula, express t Time-of-use microgrid n The equivalent wind speed in the area; , , They represent microgrids n The cut-in velocity, rated velocity, and cut-out velocity of the wind turbine unit; microgrid n Rated power of the fan unit.

[0053] This model means: low wind speed area Wind energy is insufficient to overcome mechanical losses and starting resistance, so the wind turbine does not generate electricity and its output power is zero.

[0054] Wind speed zone on slope The wind turbine maximizes its wind energy capture efficiency under the control of power factor and pitch angle, and its output power increases approximately cubically with wind speed.

[0055] Rated wind speed zone To limit the mechanical and electrical stress on the unit, the wind turbine maintains its output power at the rated value through control methods such as pitch control. The wind speed in the vicinity no longer increases.

[0056] Extremely high wind speed shutdown area For safety reasons, the fan will automatically stop and output zero power when the wind speed exceeds the cut-out wind speed.

[0057] Based on the above models and statistical characteristics, wind speed is affected by multiple factors such as multi-scale atmospheric motion and topographic turbulence, and is widely regarded as a complex stochastic process. The relationship between wind speed and wind power is also nonlinear, which makes the wind power prediction error have obvious heteroscedasticity and autocorrelation, bringing more serious uncertainty impact to scheduling. In contrast, photovoltaic power generation is constrained by solar altitude angle and sunshine duration, and has a stable daily cycle and seasonal pattern. "Clear sky irradiance" can be given relatively accurately through physical models, with less fluctuation.

[0058] Due to the volatility and uncertainty of wind power generation, it is necessary to construct an uncertainty set for wind power generation fluctuations and incorporate the impact of this uncertainty into the model. Its expression is as follows: (31) In the formula, This represents the wind power uncertainty variable introduced considering wind power fluctuations; The space representing the set of uncertainty; N T The space representing the operating cycle; This indicates the maximum deviation in the output power of the wind turbine. This represents the uncertainty adjustment parameter. A larger value indicates that there is more wind power and the time period is closer to the lower bound, resulting in higher uncertainty. express t Forecast values ​​of wind power output for different time periods; express t Is the wind power output during that period the worst-case scenario?

[0059] The purpose of constructing the uncertainty set is to find the uncertain variables. u In the uncertain set U The most economically efficient scheduling scheme when the internal parameters change towards the worst-case scenario has the following form: (32) By internalizing wind power fluctuations, we can ensure that the strategy is optimized while maintaining feasibility and economy under budgetary conservatism, and unify price signals with physical constraints.

[0060] S2. Solve the master-slave game optimization model to obtain the equilibrium solution, and schedule the energy system based on the equilibrium solution.

[0061] Based on step S1, this invention establishes a multi-layer master-slave game model. The existence and uniqueness of the solution to this model can be proven using a multi-step inverse induction method. For ease of demonstration, this embodiment transforms the decision variables of each subject into vector form, denoted as the decision vector of the upper-level distribution network operator. , which is: (33) Let the decision vector of the mid-level pumped storage operator be . This includes the power purchased / sold from the distribution network operator at each time period, as well as the unit price for leased capacity and unit price for power to the microgrid. The corresponding decision-feasibility region is denoted as: (34) Its linear constraints are derived from the power unidirectionality of pumped storage units, the upper limit of rated power, the dynamic equation of state of charge, and the upper and lower limits of SOC.

[0062] The uncertainty set constructed above : (35) Let the decision vector of the lower-level microgrid group alliance be . z This includes the power purchased and sold by each microgrid, the output of micro gas turbines, transferable / reducible loads, wind and solar power curtailment rates, and power exchanged between microgrids, with the corresponding feasible region being: (36) Among them, power balance, curtailment ratio, transferable / reducible load constraints, and inter-microgrid power transfer constraints are all linear constraints.

[0063] Based on the definitions of the objective functions of each entity mentioned above, the costs of the upper-level distribution network operator, the middle-level pumped storage operator, and the lower-level microgrid cluster alliance operator are respectively calculated as follows: , , .

[0064] When this model reaches the game equilibrium point, its solution... Should meet: (37) That is, near this point, any one of the three entities cannot reduce its operating costs by unilaterally changing its own decision variables.

[0065] The robust optimization model of the three-layer master-slave game of the present invention can be transformed into the following representation: (38) In the formula, For distribution network operators, this is the decision-feasibility domain; This represents the feasible domain for mid-level pumped storage operators under a given upper-level purchase / sale electricity price. This represents the feasible domain for the lower-level microgrid cluster alliance under given purchase / sale electricity prices, pumped storage leasing strategies, and wind power uncertainties.

[0066] This embodiment makes the following assumptions: Assumption 1 (Non-empty, bounded, and convex nature of the feasible region) 1. Based on the physical output and rated capacity limitations of distribution networks, energy storage, and microgrid equipment, decision variables at the upper, middle, and lower levels. Both have upper and lower bounds and satisfy linear power balance and safe operation constraints, therefore , , All are sets of non-empty, bounded, closed and convex polyhedra.

[0067] 2. Uncertainty set of wind power It is composed of deviation interval constraints and uncertainty budget constraints, and is a set of non-empty, bounded, closed and convex polyhedra.

[0068] Assumption 2 (Continuity and strict convexity of the objective function) Objective functions at each level , , All of them consist of linear cost items and quadratic cost items (such as gas turbine fuel costs, demand response penalties, wind and solar curtailment penalties, etc.), and are continuous in their respective decision variables.

[0069] Assumption 3 (Constraining the linear independence of gradients) The equality constraint matrices of the three-layer game subjects satisfy the linear independence constraint on their respective feasible regions, that is, the constraint gradient vector sets are linearly independent. Thus, the Karush–Kuhn–Tucker (KKT) conditions are necessary and sufficient conditions for each layer of convex optimization problems.

[0070] Under the above assumptions, we will now prove that the solutions to the lower and middle level problems exist and are unique, and further prove that the equilibrium solution of the entire three-level master-slave game model exists and is unique.

[0071] Given upper and middle level decisions and uncertainty set Under these conditions, the lower-level microgrid group consortium solves the robust two-stage optimization problem in the following form: (39) For fixed The inner problem is: .because It is a non-empty, compact convex set, and exist Continuous (specifically manifested in wind power output and equivalent load containing) (linear terms), according to the extremum existence theorem, the above problem has at least one optimal solution. .

[0072] The inner maximum value function is denoted as: (40) The robustness problem can then be written as: According to robust optimization theory, if the original objective function... about To strictly convex, regarding If it is an affine or convex function, then about It still maintains convexity; in this invention, If only linear terms such as wind power output and wind curtailment penalty are entered into the objective function, the above conditions are satisfied. Therefore, It is a convex function; further considering assumption 2... The strict convexity of the property indicates that... about It is a strictly convex function. On the other hand, Defined by linear power balance, upper and lower bounds of output, curtailment ratio, and transferable load constraints, it is a non-empty compact convex set. Since a strictly convex function must have a unique minimum point on a non-empty compact convex set, formula (40) has one and only one optimal solution, denoted as: (41) Where, mapping It is the unique optimal response mapping of the lower-level microgrid group to the upper and middle-level strategies.

[0073] In the given With the unique optimal response of the lower layer Under these conditions, the daily operation optimization problem for a mid-level pumped storage operator can be written as: ;in It consists of unidirectional constraints on pumped storage power, upper and lower bounds on charging and discharging power, and dynamic equations of charge state, and is a non-empty compact convex set.

[0074] Assuming 2, For about Since the function is strictly convex, the intermediate-level optimization problem is a standard strictly convex programming problem. Similarly, by the strict convexity and the LICQ condition, the intermediate-level problem has one and only one optimal solution: ;in This is the unique optimal response mapping for pumped storage operators to the upper-level electricity purchase / sale pricing strategy.

[0075] The optimal responses of the middle and lower layers are explicitly written as and Then the equivalent optimization problem for the upper-level distribution network operator is: (42) Based on hypothesis 1, Non-empty compact convex set; and It is the unique optimal response obtained on the convex programming problem, and its relation to Continuous change (obtained from the continuity theorem for solution mappings in parametric convex programming). Therefore, the composite function... about It remains a strictly convex function.

[0076] Based on hypothesis 2, the original objective about It is a strictly convex function; because , For the unique optimal response mapping, the corresponding It can be seen as in The weighted summation of the unique optimal execution results from the middle and lower layers, thus about It remains a strictly convex function. Strictly convex functions exist on convex sets. Since the minimum point on the upper level is unique, the upper-level problem has a unique optimal solution. .

[0077] Therefore, in a given At that time, the optimal responses of the middle and lower layers are uniquely determined as follows: (43) Therefore, the unique optimal solution to this game theory model is: (44) satisfy: In the given hour, This reduces the total cost for upper-level distribution network operators. Minimum.

[0078] In the given hour, Pumped storage operators are in the feasible field The unique minimum-cost solution.

[0079] In the given hour, The problem of lower-level robust microgrid scheduling in Ω M ( , , U The unique minimum cost solution on the ), corresponding to the worst-case wind power scenario. ∈ U There is at least one.

[0080] therefore, The unique equilibrium solution constitutes the three-layer master-slave game optimization model proposed in this invention, thus proving that the solution of the model of this invention exists and is unique.

[0081] After proving the existence and uniqueness of the optimal solution to the model, this invention proposes a distributed nested solution method to solve the model. The overall idea is as follows: a set of upper-level price decisions is given in the outer layer. The optimal response of the pumped storage operator in the middle layer is solved separately in the inner layer. Robust optimal response of the underlying microgrid group Then, by utilizing the convexity of the upper-level objective function, the price vector... Perform iterative updates until convergence.

[0082] Given a set of price vectors in the outer layer (No. k After the outermost iteration, the joint solution of the inner layers includes two steps: 1. Solving for the optimal response of the mid-level pumped storage operator.

[0083] In the given Under these conditions, the middle-level problem is: (45) In the formula, This represents the decision vector for mid-level pumped storage operators; Indicates at a given price The next middle feasible region, The mid-level cost function can be solved using existing convex optimization solvers.

[0084] 2. Solving for the optimal response of robust scheduling of the lower-level microgrid group.

[0085] upper layer With mid-level response As a parameter, the robustness problem of the lower two segments is: (46) In the formula, This represents the decision vector of the lower-level microgrid group alliance; This represents the lower-level feasible region given the upper and middle-level decision conditions; This represents the uncertainty vector of wind power, belonging to the uncertainty set. U ; This is the lower-level cost function.

[0086] Regarding the " "Robust structure: This invention employs the column and constraint generation (C&CG) method to construct the lower-level solution module, that is, to gradually approximate the uncertainty set with a 'finite set of scenarios'." U This module solves the lower-level convex optimization master problem in a finite scenario, then searches for new worst-case wind power scenarios through subproblems and adds them to the scenario set. This process is iterated until the worst-case cost converges within the tolerance range. The output of this module is the current price. Robust optimal scheduling solution .

[0087] Through the two inner steps described above, in each outer iteration In the middle, the inner solution module is called to obtain a unique optimal response pair: (47) In the formula, , They represent the first The optimal solutions for the middle and lower layers in the next outer layer iteration.

[0088] To obtain the optimal response of the inner layer , Then, the upper-level distribution network operator will adjust the price based on the current price. Based on the electricity purchase and sale situation, the price vector is updated to gradually converge to the unique optimal solution in order to reduce operating costs.

[0089] To ensure practicality and ease of implementation, this invention adopts a simple iterative rule based on the monotonic relationship between "power imbalance and price": defining each time period t Power imbalance: (48) In the formula, In the price vector Down t Net power purchase and sale in the power distribution network during the time period; For price vector The total active power purchased by the distribution network under the inner layer response; The total active power sold to the microgrid cluster alliance and the upper-level power grid.

[0090] Since both the middle and lower-level problems are convex optimizations, and the electricity purchase and sale prices enter the objective function in a linear form, increasing the electricity sale price for a certain period will decrease the electricity purchase demand of microgrids and energy storage, while increasing the willingness to sell electricity. Since the price changes monotonically, this invention employs an approximate gradient update strategy.

[0091] The power imbalance is treated as an "approximate gradient direction" of the upper-level objective function, and the price vector is updated stepwise as a whole: (49) In the formula, Indicates the feasible region The projection operator is used to ensure that the updated price still satisfies the upper and lower bounds and the mean constraint. Indicates the first The step size of the next outer iteration can be a fixed small step size or adaptively adjusted according to the iteration. This represents the price adjustment direction vector.

[0092] If the length of the price range and the change in the price vector ( If both the change in the upper-level target and the change in the upper-level target are less than a preset threshold, then the outer-level iteration converges and outputs... As the only optimal solution of the three-layer master-slave game robust optimization scheduling model of this invention, the corresponding worst-case wind power scenario is also given. .

[0093] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A master-slave game optimization method for energy storage leasing systems applicable to wind power fluctuation scenarios, characterized in that, Includes the following steps: S1. Divide the energy system into an upper-level distribution network operator, a middle-level pumped storage operator, and a lower-level microgrid group alliance, and construct a three-level master-slave game optimization model. S2. Solve the master-slave game optimization model to obtain the equilibrium solution, and schedule the energy system based on the equilibrium solution; The master-slave game optimization model includes: a distribution network operator model, a pumped storage operator model, and a microgrid group alliance model; the distribution network operator model and the pumped storage operator model constitute the outer master-slave game; the pumped storage operator model and the microgrid group alliance model constitute the inner master-slave game. The microgrid group alliance model aims to minimize daily operating costs, and its objective function includes: ； In the formula, , , , , , In order, the costs are: power purchase and sale interaction costs between the microgrid group alliance and the distribution operator, power generation costs of micro gas turbines, power generation costs of renewable energy, external pumped storage leasing fees, demand response compensation costs, and renewable energy curtailment penalty costs. in, ； ； In the formula, express t Time-of-use microgrids n The output of the micro gas turbine; , They represent microgrids n Fuel cost coefficient of gas turbines; =1, ..., J Pollutant type index; express j Penalty coefficient per unit emission of pollutants; express t The electricity purchase price set by the power distribution network operator during a specific time period; express t Electricity sales prices set by the time-of-use distribution network operator; microgrid n exist t Active power sold to distribution network operators during a given time period; microgrid n exist t Active power purchased from distribution network operators during a given time period; ； In the formula, , These represent the power generation cost coefficients for photovoltaic and wind turbines, respectively. , They represent t Time-of-use microgrids n The output of photovoltaic and wind turbines; N Indicates the number of microgrids; T Indicates the index of the scheduling period; In the formula, express t Time-of-use microgrid n The equivalent wind speed in the area; , , They represent microgrids n The cut-in velocity, rated velocity, and cut-out velocity of the wind turbine unit; microgrid n Rated power of the fan unit; The power output of the wind turbine is assessed using a piecewise power curve based on nameplate data. Equivalent modeling is performed, and the relationship between active power output and wind speed is as follows: In the formula, microgrid n Rated power of the fan unit; ； ； In the formula, , These represent the compensation unit prices for transferable load and load that can be reduced, respectively. , They represent t Time-of-use microgrids n Transferable load and load reduction; , They represent t Periodic pumped storage operators are microgrids n The established unit price for electricity capacity and unit price for electricity power; , They represent t Periodic pumped storage operators are microgrids n The leased power capacity and power allowance; , These represent the compensation unit prices for transferable load and load that can be reduced, respectively. , They represent t Time-of-use microgrids n Transferable load and load reduction; The constraints of the microgrid group alliance model include: microgrid power balance constraints, renewable energy curtailment ratio constraints, transferable load constraints, load reduction constraints, and micro gas turbine output constraints. ； ； ； In the formula, express t Time period by microgrid q To microgrids n The power transmission capacity, express t Time period by microgrid n To microgrids q The power transmission capacity; , These represent the curtailment penalty coefficients for wind turbines and solar power, respectively. This indicates the power surplus or shortage status of a microgrid; Constructing the uncertainty set of wind power generation fluctuations: ； In the formula, This represents the wind power uncertainty variable introduced considering wind power fluctuations; The space representing the set of uncertainty; N T The space representing the operating cycle; This indicates the maximum deviation in the output power of the wind turbine. This represents the uncertainty adjustment parameter. A larger value indicates that there is more wind power and the time period is closer to the lower bound, resulting in higher uncertainty. express t Forecast values ​​of wind power output for different time periods; express t Is the wind power output during that period the worst-case scenario? Based on the constructed uncertainty set, the most economically efficient scheduling scheme is obtained when the uncertain variables change towards the worst-case scenario within the uncertainty set: ； In S2, the method for solving the master-slave game optimization model to obtain the equilibrium solution includes: S21. Given a set of upper-level price decisions in the outer master-slave game, solve the optimal response of the pumped storage operator and the robust optimal response of the microgrid group alliance in the inner master-slave game respectively. The process of finding the optimal response for pumped storage operators includes: Given a set of price vectors Under these conditions, the middle-level problem is: ； In the formula, This represents the decision vector for mid-level pumped storage operators; Indicates at a given price The next middle feasible region, Represents the middle-level cost function; The process of finding the robust optimal vector for a microgrid cluster alliance includes: upper layer With mid-level response As a parameter, the robustness problem of the lower two segments is: ； In the formula, This represents the decision vector of the lower-level microgrid group alliance; This represents the lower-level feasible region given the upper and middle-level decision conditions; This represents the uncertainty vector of wind power, belonging to the uncertainty set. U ; For the lower-level cost function; In each outer layer iteration In the middle, the inner solution module is called to obtain a unique optimal response pair: ； In the formula, , They represent the first The optimal solutions for the middle and lower layers in the next outer layer iteration; To obtain the optimal response of the inner layer , Then, the upper-level distribution network operator will adjust the price based on the current price. The electricity purchase and sale situation is used to update the price vector, gradually converging to the unique optimal solution; Define each time period t Power imbalance: ； In the formula, In the price vector Down t Net power purchase and sale in the power distribution network during the time period; For price vector The total active power purchased by the distribution network under the inner layer response; The total active power sold to the microgrid cluster alliance and the upper-level power grid; S22. Iteratively update the upper-level price decision based on the approximate gradient update strategy until convergence, to obtain the equilibrium solution; By treating the power imbalance as an approximate gradient direction of the upper-level objective function, the overall price vector is updated stepwise: ； In the formula, Indicates the feasible region The projection operator; Indicates the first The step size of the next outermost iteration; This represents the price adjustment direction vector; Price range length, price vector change When the changes in the upper-level target are all less than a preset threshold, the outer-level iteration converges, and the output is... As the sole optimal solution of the three-layer master-slave game robust optimization scheduling model, the corresponding worst-case wind power scenario is also given. .

2. The master-slave game optimization method for energy storage leasing systems applicable to wind power fluctuation scenarios as described in claim 1, characterized in that, The distribution network operator model aims to minimize total operating cost, and its objective function includes: ； In the formula, Indicates total operating cost; This represents the power interaction cost between the distribution network operator and the microgrid group alliance; This represents the electricity interaction cost between the distribution network operator and the pumped storage operator; This indicates the cost for distribution network operators to purchase electricity from the upstream power grid; in, ； ； In the formula, This indicates that pumped storage operators are in t Active power sold to distribution network operators during a given time period; This indicates that pumped storage operators are in t Active power purchased from distribution network operators during a given time period; ； In the formula, This indicates the settlement price given by the superior power grid; Indicates that the distribution network operator is t Active power purchased from the upper-level power grid during a given time period; The power distribution network operator model uses electricity price constraints to guide pumped storage operators and microgrid group alliances in making power purchase and sale decisions: ； ； In the formula, , , These represent the sets of peak, flat, and trough periods, respectively. , , These represent the unified electricity sales price decision variables corresponding to peak, flat, and low periods, respectively, with the superscripts max and min indicating the upper and lower bounds, respectively. This indicates the average upper limit of electricity sales prices.

3. The master-slave game optimization method for energy storage leasing systems applicable to wind power fluctuation scenarios as described in claim 2, characterized in that, The pumped storage operator model aims to minimize its own daily net operating cost, and the objective function includes: ； In the formula, The daily net operating cost for pumped storage operators; This indicates the net cost settled between pumped storage operators and distribution network operators; This represents the operation and maintenance costs for pumped storage operators; This refers to the rental income collected by pumped storage operators from the microgrid cluster alliance; in, ； ； In the formula, The maintenance cost coefficient per unit power; , They represent t The charging and discharging power of the pumped storage operator's own energy storage section during a given period; 。 4. The master-slave game optimization method for energy storage leasing systems applicable to wind power fluctuation scenarios as described in claim 3, characterized in that, The constraints of the pumped storage operator model include: mutual exclusion constraints for distribution network transactions, dynamic constraints for state of charge, power conservation constraints, and lease price constraints. The mutual exclusion constraints for transactions in the distribution network include: In the formula, This indicates the rated power of the pumped storage operator's own energy storage section; The dynamic constraints on the state of charge include: In the formula, express t Energy status of pumped storage units during specific time periods; Indicates the time step; , These represent the pumping efficiency and power generation efficiency of the pumped storage unit, respectively. , These represent the upper and lower bounds of the charged state of a pumped storage unit, respectively. The power conservation constraints include: ； The rental price constraints include: 。

Images