Source network load storage double-layer optimization scheduling method and device and computer equipment

Abstract

The application relates to a source-grid-load-storage double-layer optimization scheduling method and device and a computer device. The method comprises the following steps: a double-layer optimization model containing an upper-layer power grid operator model and a lower-layer load aggregator model is constructed; the upper-layer power grid operator model is associated with the lower-layer load aggregator model through a power balance coupling constraint; based on a KKT condition of the lower-layer load aggregator model, binary auxiliary variables and positive numbers are introduced, each nonlinear complementary relaxation condition is equivalently converted into a linear inequality constraint condition, the double-layer optimization model is linearized, and a single-layer mixed integer linear programming model is converted; and the single-layer mixed integer linear programming model is used to output an optimal scheduling scheme of the power grid operator and the load aggregator. By adopting the method, multi-element resources of the source-grid-load-storage can be efficiently cooperated, economy and low carbon are considered, and solving efficiency and scheduling scheme completeness are improved.

Description

Technical Field

[0001] This application relates to the field of power grid dispatching technology, and in particular to a two-level optimized dispatching method, apparatus, computer equipment, computer-readable storage medium, and computer program product for source-grid-load-storage systems. Background Technology

[0002] With the advancement of the "dual carbon" target, the penetration rate of renewable energy in the power system is continuously increasing. To promote the consumption of new energy and ensure the safe and economical operation of the system, the coordinated and optimized scheduling among power sources, grids, loads, and storage has become a research hotspot in the field of power system dispatching technology.

[0003] Traditional power system dispatching primarily employs two approaches: First, traditional single-layer centralized dispatching treats grid operators and user-side resources as a single stakeholder, optimizing with the minimum total system cost as the sole objective. This approach ignores the differing interests of various stakeholders in a market environment. Second, basic two-layer optimization dispatching, while initially distinguishing between upper and lower layers, often fails to fully incorporate diverse resources such as thermal power, wind turbines, photovoltaics, energy storage, loads that can be reduced, and loads that can be transferred, and also fails to adequately account for the impact of carbon trading mechanisms on dispatching strategies. Furthermore, existing methods often rely on heuristic algorithms, resulting in low computational efficiency and a tendency to get trapped in local optima.

[0004] However, current scheduling methods or traditional approaches still have the following problems: First, the multi-stakeholder interest coordination mechanism is not sound, making it difficult to achieve coordinated resource optimization; second, the models are non-convex and highly nonlinear, making it difficult to balance solution efficiency and global optimality; third, the carbon trading mechanism is not closely integrated with the scheduling model, failing to fully leverage its incentive effect on clean energy; and fourth, the scheduling model does not cover all resource types, limiting the applicability of the optimization results in actual systems. Summary of the Invention

[0005] Based on this, it is necessary to provide a two-layer optimization scheduling method, device, computer equipment, computer-readable storage medium, and computer program product that can efficiently coordinate multiple resources of source, grid, load, and storage, take into account both economic efficiency and low carbon emissions, and improve the solution efficiency and scheduling scheme completeness, in order to address the above-mentioned technical problems.

[0006] Firstly, this application provides a two-tiered optimal scheduling method for source-grid-load-storage systems, including:

[0007] A two-tier optimization model is constructed, comprising an upper-level grid operator model and a lower-level load aggregator model. The upper-level grid operator model aims to minimize the total cost, including the operating cost of thermal power units and the cost of carbon trading. The lower-level load aggregator model aims to minimize the comprehensive cost, including the cost of renewable energy generation, the operating cost of energy storage, the cost of flexible load compensation, and the revenue from certified voluntary emission reductions. The upper-level grid operator model and the lower-level load aggregator model are linked by power balance coupling constraints.

[0008] Based on the Karush-Kuhn-Tucker conditions of the lower-level load aggregator model, binary auxiliary variables and positive numbers are introduced, and each nonlinear complementary relaxation condition is equivalently transformed into a linear inequality constraint condition, so as to linearize the two-level optimization model and transform it into a single-level mixed integer linear programming model.

[0009] Using the single-layer mixed integer linear programming model, the optimal scheduling scheme for the power grid operator and load aggregator is output.

[0010] In one embodiment, the scheduling objects in the lower-level load aggregator model include wind turbine generators, photovoltaic generators, electrochemical energy storage systems, loads that can be reduced, and loads that can be transferred.

[0011] The constraints of the lower-level load aggregator model include: power flow constraints and node voltage security constraints of the distribution network; state of charge constraints and charge / discharge mutual exclusion constraints of the electrochemical energy storage system; the constraint that the reduction amount of the load that can be reduced in any scheduling period does not exceed the maximum reduction ratio of the basic load demand in the scheduling period; and the balance constraint that the total transfer-in amount and the total transfer-out amount of the load that can be transferred out are equal in the scheduling cycle.

[0012] In one embodiment, the Karush-Kuhn-Tucker conditions based on the lower-level load aggregator model introduce binary auxiliary variables and positive numbers, and each nonlinear complementary relaxation condition is equivalently transformed into a linear inequality constraint condition to linearize the bi-level optimization model, including:

[0013] Construct a Lagrangian function for the optimization problem of the lower-level load aggregator model;

[0014] Based on the Lagrangian function, the Karush-Kuhn-Tucker optimality condition set, which includes stability conditions, primal feasibility conditions, dual feasibility conditions, and complementary relaxation conditions, is derived.

[0015] For the complementary relaxation conditions, binary auxiliary variables are introduced, and the Big M method is used to convert each nonlinear complementary relaxation condition into two linear inequality constraints.

[0016] The set of optimality conditions after linearization, together with all the constraints of the upper-level power grid operator model, constitutes the constraint set of the single-level mixed integer linear programming model.

[0017] In one embodiment, the carbon trading cost in the upper-level grid operator model is calculated based on the difference between the actual carbon emissions of thermal power units and the free carbon allowances allocated based on power generation, as well as the carbon trading price; the certified voluntary emission reduction revenue in the lower-level load aggregator model is calculated based on the predicted output of wind turbines and photovoltaic power units, the corresponding carbon dioxide emission reduction factors, and the certified voluntary emission reduction trading price.

[0018] In one embodiment, the constraints of the upper-level grid operator model include flexibility modification constraints and operational physical constraints for thermal power units; the flexibility modification constraints are used to reduce the minimum technical output limit of the thermal power units by introducing a modification ratio coefficient; the operational physical constraints include upper and lower limit constraints for unit output, ramp rate constraints, and minimum continuous start-stop time constraints.

[0019] In one embodiment, the charge-discharge mutual exclusion constraint of the electrochemical energy storage system is modeled by introducing binary variables; the unit reduction compensation cost of the load that can be reduced and the unit transfer compensation cost of the load that can be transferred are included as cost terms in the objective function of the lower-level load aggregator model.

[0020] Secondly, this application also provides a two-layer optimized scheduling device for source-grid-load-storage, comprising:

[0021] The model building module is used to construct a two-layer optimization model comprising an upper-layer grid operator model and a lower-layer load aggregator model. The upper-layer grid operator model aims to minimize the total cost, including thermal power unit operating costs and carbon trading costs. The lower-layer load aggregator model aims to minimize the comprehensive cost, including renewable energy generation costs, energy storage operating costs, flexible load compensation costs, and certified voluntary emission reduction benefits. The upper-layer grid operator model and the lower-layer load aggregator model are linked through power balance coupling constraints.

[0022] The linearization module is used to introduce binary auxiliary variables and positive numbers based on the Karush-Kuhn-Tucker conditions of the lower-level load aggregator model, and to convert each nonlinear complementary relaxation condition into a linear inequality constraint condition, so as to linearize the two-level optimization model and transform it into a single-level mixed integer linear programming model.

[0023] The calculation module is used to output the optimal scheduling scheme for the grid operator and load aggregator using the single-level mixed integer linear programming model.

[0024] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0025] A two-tier optimization model is constructed, comprising an upper-level grid operator model and a lower-level load aggregator model. The upper-level grid operator model aims to minimize the total cost, including the operating cost of thermal power units and the cost of carbon trading. The lower-level load aggregator model aims to minimize the comprehensive cost, including the cost of renewable energy generation, the operating cost of energy storage, the cost of flexible load compensation, and the revenue from certified voluntary emission reductions. The upper-level grid operator model and the lower-level load aggregator model are linked by power balance coupling constraints.

[0026] Based on the Karush-Kuhn-Tucker conditions of the lower-level load aggregator model, binary auxiliary variables and positive numbers are introduced, and each nonlinear complementary relaxation condition is equivalently transformed into a linear inequality constraint condition, so as to linearize the two-level optimization model and transform it into a single-level mixed integer linear programming model.

[0027] Using the single-layer mixed integer linear programming model, the optimal scheduling scheme for the power grid operator and load aggregator is output.

[0028] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0029] A two-tier optimization model is constructed, comprising an upper-level grid operator model and a lower-level load aggregator model. The upper-level grid operator model aims to minimize the total cost, including the operating cost of thermal power units and the cost of carbon trading. The lower-level load aggregator model aims to minimize the comprehensive cost, including the cost of renewable energy generation, the operating cost of energy storage, the cost of flexible load compensation, and the revenue from certified voluntary emission reductions. The upper-level grid operator model and the lower-level load aggregator model are linked by power balance coupling constraints.

[0030] Based on the Karush-Kuhn-Tucker conditions of the lower-level load aggregator model, binary auxiliary variables and positive numbers are introduced, and each nonlinear complementary relaxation condition is equivalently transformed into a linear inequality constraint condition, so as to linearize the two-level optimization model and transform it into a single-level mixed integer linear programming model.

[0031] Using the single-layer mixed integer linear programming model, the optimal scheduling scheme for the power grid operator and load aggregator is output.

[0032] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0033] A two-tier optimization model is constructed, comprising an upper-level grid operator model and a lower-level load aggregator model. The upper-level grid operator model aims to minimize the total cost, including the operating cost of thermal power units and the cost of carbon trading. The lower-level load aggregator model aims to minimize the comprehensive cost, including the cost of renewable energy generation, the operating cost of energy storage, the cost of flexible load compensation, and the revenue from certified voluntary emission reductions. The upper-level grid operator model and the lower-level load aggregator model are linked by power balance coupling constraints.

[0034] Based on the Karush-Kuhn-Tucker conditions of the lower-level load aggregator model, binary auxiliary variables and positive numbers are introduced, and each nonlinear complementary relaxation condition is equivalently transformed into a linear inequality constraint condition, so as to linearize the two-level optimization model and transform it into a single-level mixed integer linear programming model.

[0035] Using the single-layer mixed integer linear programming model, the optimal scheduling scheme for the power grid operator and load aggregator is output.

[0036] The aforementioned two-tiered optimization scheduling method, device, computer equipment, computer-readable storage medium, and computer program product for power generation, grid, load, and storage effectively coordinates the interests of different market participants by constructing a two-tiered game model with upper-level grid operators and lower-level load aggregators as the main players, and setting differentiated objective functions that take into account carbon trading costs and voluntary emission reduction benefits. This makes the scheduling scheme more in line with actual market operations. By adopting KKT conditions combined with auxiliary variable linearization, the non-convex two-tiered model is transformed into a standard mixed-integer linear programming problem. A mature solver is used to achieve efficient global optimization, solving the problems of low computational efficiency and susceptibility to local optima in traditional methods. By fully incorporating all resource elements such as thermal power, new energy, energy storage, and flexible loads into the model, and establishing precise physical and economic constraints, the completeness and executability of the scheduling scheme are significantly improved. Ultimately, the goal of synergistically optimizing system economic costs and carbon emission levels is achieved while ensuring the safe operation of the power grid. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 This is an application environment diagram of a two-layer optimization scheduling method for source-grid-load-storage in one embodiment;

[0039] Figure 2 This is a flowchart illustrating a two-layer optimization scheduling method for source-grid-load-storage in one embodiment;

[0040] Figure 3 This is a flowchart illustrating the two-layer optimization scheduling method for source-grid-load-storage in another embodiment;

[0041] Figure 4 This is a flowchart illustrating the two-layer optimization scheduling method of source-grid-load-storage in the most detailed embodiment of this application;

[0042] Figure 5 This is the network topology diagram in the most detailed embodiment of this application;

[0043] Figure 6 This is a diagram showing the scheduling results in the most detailed embodiment of this application;

[0044] Figure 7(a) is a cost change diagram of the game process of the upper and lower layer models in the most detailed embodiment of this application at different time periods;

[0045] Figure 7(b) is a cost breakdown diagram of the upper-level power grid operator model in the most detailed embodiment of this application;

[0046] Figure 7(c) is a cost breakdown diagram of the lower-level load aggregator model in the most detailed embodiment of this application;

[0047] Figure 8(a) is a cost convergence diagram of the upper-level power grid operator model in the most detailed embodiment of this application;

[0048] Figure 8(b) is a cost convergence diagram of the lower-level load aggregator model in the most detailed embodiment of this application;

[0049] Figure 9(a) is a diagram of the unit output of node 6 in the most detailed embodiment of this application;

[0050] Figure 9(b) is a graph showing the change in the energy storage SOC of node 6 in the most detailed embodiment of this application;

[0051] Figure 10 This is a structural block diagram of a two-layer optimized scheduling device for source-grid-load-storage in one embodiment;

[0052] Figure 11 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0054] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0055] The two-layer optimization scheduling method for source-grid-load-storage provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104, or it can be located in the cloud or on another network server.

[0056] The terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle systems, and projection devices. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted displays. Head-mounted displays can be virtual reality (VR) devices, augmented reality (AR) devices, and smart glasses. The server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0057] In one exemplary embodiment, such as Figure 2 As shown, a two-layer optimal scheduling method for source-grid-load-storage is provided, which is then applied to... Figure 1 Taking the server in the example, the explanation includes the following steps S202 to S206. Wherein:

[0058] Step S202: Construct a two-layer optimization model that includes an upper-layer grid operator model and a lower-layer load aggregator model. The upper-layer grid operator model aims to minimize the total cost, including the operating cost of thermal power units and the cost of carbon trading. The lower-layer load aggregator model aims to minimize the comprehensive cost, including the cost of renewable energy generation, the operating cost of energy storage, the cost of flexible load compensation, and the benefits of certified voluntary emission reductions. The upper-layer grid operator model and the lower-layer load aggregator model are linked by power balance coupling constraints.

[0059] Specifically, the overall architecture of the model is established as a master-slave game framework comprising an upper-level decision-maker and a lower-level decision-maker. The upper-level decision-maker corresponds to the grid operator, and the lower-level decision-maker corresponds to the load aggregator. As the leader, the grid operator's decision-making involves determining the active power output plan of the traditional thermal power units under its jurisdiction during each dispatch period; while the load aggregator, as the follower, makes decisions encompassing the output of distributed wind and solar power generation, the charging and discharging power of energy storage systems, and the adjustment of flexible loads on its behalf.

[0060] Secondly, objective functions are defined for the upper and lower optimization models respectively. For the upper-level model, the objective is to minimize the total cost borne by the grid operator. This total cost consists of two parts: the operating cost of thermal power generation and the carbon trading cost caused by carbon emissions from thermal power generation. For the lower-level load aggregator model, the objective is to minimize the load aggregator's overall operating cost. This overall cost is a net cost concept, which aggregates the generation cost of distributed renewable energy, the operating cost of using energy storage systems, the flexible load compensation cost paid to incentivize user participation in regulation, and deducts the certified voluntary emission reduction revenue that can be obtained through the market for providing clean electricity. Finally, to achieve the feasibility of the physical system and the linkage between upper and lower-level decisions, a key coupling constraint must be established to link the two models, namely the power balance constraint. This constraint requires that, in each scheduling period, the sum of the total thermal power output provided by the upper-level grid operator and the net exchange power of all resources on the lower-level load aggregator side (including distributed generation, energy storage, and net load after flexible load regulation) must match the total load demand of the system at that time. By establishing this core physical relationship, the upper-level decision-makers, while pursuing the minimization of their own costs, must also consider the impact of their decisions on the economically optimal response behavior of the lower-level aggregators, and vice versa. This constitutes a complete two-level optimization problem that reflects the interaction of market participants.

[0061] Step S204: Based on the Karush-Kuhn-Tucker conditions of the lower-level load aggregator model, binary auxiliary variables and positive numbers are introduced, and each nonlinear complementary relaxation condition is equivalently transformed into a linear inequality constraint condition, so as to linearize the two-level optimization model and transform it into a single-level mixed integer linear programming model.

[0062] Specifically, firstly, a corresponding Lagrangian function needs to be constructed based on the objective function and all constraints of the lower-level load aggregator model. Then, based on this Lagrangian function, a set of mathematical expressions constituting the KKT optimality conditions (i.e., the Karush-Kuhn-Tucker conditions) is derived. This set of conditions collectively defines the necessary and sufficient conditions that the optimal solution to the original lower-level optimization problem must satisfy. These include: a stability condition, i.e., the first-order partial derivatives of the Lagrangian function with respect to all lower-level decision variables are zero; a primal feasibility condition, i.e., the solution must satisfy all the original constraints of the lower-level load aggregator model; a dual feasibility condition, i.e., the Lagrange multipliers corresponding to inequality constraints must be non-negative; and a complementary relaxation condition, i.e., the product of the Lagrange multiplier corresponding to each inequality constraint and the relaxation amount of the constraint itself must be zero. Among these, the complementary relaxation condition manifests as a series of nonlinear equations and is the key to the overall non-convexity and nonlinearity of the model. To eliminate this nonlinearity, linearization is required: for each complementary relaxation condition, a unique binary auxiliary variable and a sufficiently large positive number (usually denoted as M) are introduced. Using this binary variable and the positive number M, a nonlinear equation of the form "the product of the multiplier and the relaxation is zero" can be equivalently transformed into two linear inequality constraints. The value of the binary variable (0 or 1) is used to select and activate the original inequality constraint or its corresponding nonnegative multiplier constraint, while the positive number M ensures that, if not activated, the other constraint becomes a relaxed and ineffective constraint. After performing the above operations on all complementary relaxation conditions, the original lower-level optimization problem is equivalently replaced by its complete and linearized set of KKT conditions. At this point, this series of linearized KKT conditions, as new constraints, is merged with all the original constraints of the upper-level grid operator model to form a completely new and unified constraint system. Simultaneously, the objective function of the upper-level model is retained as the objective function of this new model, while the set of decision variables is expanded to include all original upper-level variables, original lower-level variables, the introduced KKT multipliers, and the newly introduced binary auxiliary variables. Thus, the original two-level optimization model has been completely transformed into a mathematical programming problem in which all constraints and objective functions are linear and contain both continuous and integer (binary) variables, i.e., a standard single-level mixed integer linear programming model.

[0063] Step S206: Using a single-level mixed integer linear programming model, output the optimal scheduling scheme for the power grid operator and the load aggregator.

[0064] Specifically, a single-level mixed-integer linear programming model is input into a mathematical programming solver capable of handling mixed-integer linear programming problems. This solver, based on mature algorithms such as branch and bound and cutting planes, automatically solves the model. Its objective is to minimize the original upper-level objective function (i.e., the total cost of the grid operator) while satisfying all integration constraints (including the original upper-level constraints and the lower-level optimality conditions after equivalent transformation and linearization). After completing the calculation, the solver outputs a set of optimal numerical solutions corresponding to all decision variables in the model. The implementer needs to identify and separate subsets of variables representing the decisions of the upper-level grid operator and the lower-level load aggregator from this global solution.

[0065] Specifically, the optimal dispatch scheme for grid operators is reflected in the active power output setpoints of each thermal power unit in the model for each dispatch period. This scheme directly provides the start-up, shutdown, and power commands for traditional power generation units. The optimal dispatch scheme for load aggregators is reflected in the planned output values ​​of each distributed wind turbine and photovoltaic unit, the charging and discharging power values ​​of each energy storage system in each period (positive values ​​represent discharging, negative values ​​represent charging), the reduction amount of each load that can be reduced, and the spatiotemporal transfer amount of each load that can be transferred. These separated decision variables collectively constitute the optimal strategy for both parties in a game equilibrium state. This can simultaneously guide the grid dispatch center to issue dispatch commands to thermal power units and the load aggregator to coordinate and control the distributed resources under its jurisdiction, thereby achieving coordinated and optimized operation of power generation, grid, load, and storage.

[0066] The aforementioned two-tiered optimization scheduling method for power generation, grid, load, and storage effectively coordinates the interests of different market participants by constructing a two-tiered game model with upper-level grid operators and lower-level load aggregators as the main players, and setting differentiated objective functions that take into account carbon trading costs and voluntary emission reduction benefits. This makes the scheduling scheme more in line with actual market operations. By adopting KKT conditions combined with auxiliary variable linearization, the non-convex two-tiered model is transformed into a standard mixed-integer linear programming problem. A mature solver is used to achieve efficient global optimization, solving the problems of low computational efficiency and susceptibility to local optima in traditional methods. By fully incorporating all resource elements such as thermal power, new energy, energy storage, and flexible loads into the model, and establishing precise physical and economic constraints, the completeness and executability of the scheduling scheme are significantly improved. Ultimately, the goal of synergistically optimizing system economic costs and carbon emission levels is achieved while ensuring the safe operation of the power grid.

[0067] In one exemplary embodiment, the scheduling objects in the lower-level load aggregator model include wind turbine generators, photovoltaic generators, electrochemical energy storage systems, loads that can be reduced, and loads that can be transferred.

[0068] The constraints of the lower-level load aggregator model include: power flow constraints and node voltage security constraints of the distribution network, state of charge constraints and charge / discharge mutual exclusion constraints of the electrochemical energy storage system, the maximum reduction ratio constraint that the reduction amount of the load that can be reduced in any scheduling period does not exceed the basic load demand of the scheduling period, and the balance constraint that the total transfer in and total transfer out of the load that can be transferred out are equal during the scheduling cycle.

[0069] Specifically, firstly, the specific types of the dispatchable objects need to be clearly defined, which includes wind turbine generators, photovoltaic generators, electrochemical energy storage systems, loads that can be reduced, and loads that can be transferred. In the mathematical model, these objects are characterized by a series of continuous or integer decision variables to represent their states in future dispatch periods. Examples include the planned active power output of wind turbine generators and photovoltaic generators, the charging or discharging power of energy storage systems, the actual reduction power of loads that can be reduced, and the transfer power of loads that can be transferred between periods. Power flow constraints and node voltage safety constraints in the distribution network are used to ensure that the optimized dispatch scheme does not lead to distribution line overload or node voltage exceeding limits. Linearized branch power flow models or second-order cone relaxation models are typically used to describe the coupling relationship between active and reactive power and node voltage amplitude. The state-of-charge constraints of the electrochemical energy storage system describe the dynamic process of its internal energy storage, ensuring that its charging and discharging processes comply with energy conservation, and that the remaining charge at any given time remains within the rated capacity range. Its charging and discharging mutual exclusion constraints, by introducing auxiliary binary variables, ensure that the energy storage system cannot be in a charging and discharging state simultaneously during any dispatch period. The constraint on load reduction requires that the actual power reduction during any scheduling period must not exceed a pre-agreed maximum proportion of the original base load demand for that period, as defined in the user contract. The constraint on load transfer requires that the total load transferred in must equal the total load transferred out throughout the entire scheduling cycle, thereby ensuring that the user's total electricity demand is met, with optimization adjustments made only to the electricity usage time.

[0070] The output of wind turbines is greatly affected by the cut-in and cut-out wind speeds, and generally the output of wind turbines is greater at night than during the day. The output of photovoltaic power is mainly affected by the intensity of solar radiation.

[0071] The wind speed probability distribution function is:

[0072] ;

[0073] In the formula, for Wind speed at any moment For shape parameters, This is the scale parameter.

[0074] Wind power Represented as:

[0075] ;

[0076] In the formula, For rated capacity, These are the cut-in wind speed, cut-out wind speed, and rated wind speed, respectively.

[0077] Photovoltaic power for:

[0078] ;

[0079] In the formula, The conversion factor is... For the area of ​​the photovoltaic array, Light intensity.

[0080] Thermal power units:

[0081] thermal power unit output power for:

[0082] ;

[0083] In the formula, For diesel generator efficiency, This represents the fuel consumption rate.

[0084] Electrochemical energy storage:

[0085] Energy storage state of charge :

[0086] ;

[0087] In the formula, To store the remaining electricity, Configure capacity.

[0088] Carbon quotas:

[0089] ;

[0090] In the formula, For the first One thermal power unit Constant effort This refers to the carbon quota coefficient for thermal power units. It is a carbon emission factor.

[0091] In this embodiment, by fully incorporating wind turbines, photovoltaics, energy storage, and flexible loads into the lower-level load aggregator model and precisely setting their operation and interaction constraints, this method achieves refined management and control of distributed resources. Specifically, power flow and voltage constraints in the distribution network ensure the safety and feasibility of the dispatching scheme within the local power grid; dynamic energy constraints and charge / discharge mutual exclusion constraints for energy storage ensure the sustainability of its regulation capabilities and the rationality of its actions; and the quantitative proportions and total balance constraints for loads that can be reduced or transferred fully tap into demand-side potential while strictly adhering to user power supply guarantees and contractual boundaries.

[0092] In one exemplary embodiment, such as Figure 3 As shown, based on the Karush-Kuhn-Tucker conditions of the lower-level load aggregator model, binary auxiliary variables and positive numbers are introduced to equivalently transform each nonlinear complementary relaxation condition into a linear inequality constraint, thereby linearizing the bi-level optimization model, including:

[0093] Step S302: Construct a Lagrangian function for the optimization problem of the lower-level load aggregator model;

[0094] Step S304: Based on the Lagrangian function, derive the Karush-Kuhn-Tucker optimality condition set, which includes stability conditions, primal feasibility conditions, dual feasibility conditions, and complementary relaxation conditions.

[0095] Step S306: For the complementary relaxation conditions, a binary auxiliary variable is introduced, and the Big M method is used to convert each nonlinear complementary relaxation condition into two linear inequality constraints.

[0096] Step S308: The set of optimality conditions after linearization is combined with all the constraints of the upper-level power grid operator model to form the constraint set of the single-level mixed integer linear programming model.

[0097] Specifically, the Lagrangian function constructed for the optimization problem of the lower-level load aggregator model consists of the weighted sum of the lower-level objective function and all lower-level constraints (including equality constraints and inequality constraints):

[0098] ;

[0099] In the formula, For the lower-level objective function, For the set of all decision variables in the lower level, For the first The part of the equality constraint. The dual variable corresponding to the equation. For the first The part constrained by inequalities. is the dual variable corresponding to the inequality.

[0100] The optimality conditions of the lower-level load aggregator model optimization problem can be completely replaced by the following four sets of KKT conditions:

[0101] (1) Stability conditions:

[0102] Lagrange function For all lower-level decision variables The partial derivative is 0.

[0103] ;

[0104] In the formula, For Lagrange functions For all lower-level decision variables, The partial derivatives of the equations indicate that, at the optimal solution, the objective function cannot be further improved by fine-tuning any decision variables.

[0105] (2) Original feasibility conditions:

[0106] That is, all the original constraints must be satisfied.

[0107] (3) Conditions for duality:

[0108] All dual variables corresponding to inequality constraints must be non-negative.

[0109] ;

[0110] In the formula is the dual variable corresponding to the inequality.

[0111] (4) Complementary relaxation conditions:

[0112] For each inequality constraint, the product of its corresponding dual variable and the constraint relaxation must be zero.

[0113] ;

[0114] This condition implies that if an inequality constraint does not take the boundary (i.e., Then its shadow price It must be 0; otherwise, if a constrained shadow price Then the constraint must be a tight constraint. ).

[0115] Among them, the complementary relaxation condition in the KKT conditions This is a nonlinear product term, making the transformed model nonconvex and unsolvable directly. To preserve the linearity of the model, it must be linearized. Therefore, the Big M method is used for any complementary relaxation pair. Introduce a binary auxiliary variable and a sufficiently large positive number The original nonlinear constraints are replaced by the following two equivalent linear inequality constraints:

[0116] ;

[0117] ;

[0118] In the formula, These two constraints are penalized by a sufficiently large positive number, expressed as a penalty coefficient, and are expressed by binary variables. The values ​​of achieve a complementary relationship:

[0119] when At that time, the first constraint is restored to its original state. The second constraint is mandatory. Combining duality feasibility , can be obtained .

[0120] when At that time, the first constraint This becomes a redundant constraint, while the second constraint allows... Take a positive value. To satisfy the original complementarity condition, the solver will drive... It tends towards 0.

[0121] By performing this linearization process on the complementary relaxation conditions of all inequality constraints, the linearity of the entire model is preserved.

[0122] Through the above steps, the original two-layer model was successfully transformed into a large-scale single-layer MILP model. The equivalent model is structured as follows:

[0123] Objective function: The objective function of the optimization problem using the original upper-level power grid operator model. .

[0124] Decision variables: Includes all decision variables from the original upper and lower levels, as well as all KKT dual variables introduced for the transformation. and all the big M-method binary auxiliary variables used for linearization .

[0125] Constraints: All constraints of the optimization problem of the original upper-level power grid operator model, the original feasibility part of the KKT conditions, the stability conditions of the KKT conditions, the dual feasibility of the KKT conditions, and the KKT complementary relaxation conditions after linearization by the Big M method.

[0126] Although the final MILP model has a significantly increased number of variables and constraints, its structure is standard and can be efficiently solved using Gurobi, thus obtaining the Stackelberg equilibrium solution of the original two-layer master-slave game problem.

[0127] In this embodiment, the optimization problem of the lower-level load aggregator model is equivalent to its KKT optimality condition, and binary auxiliary variables and positive numbers M are introduced to linearize the complementary relaxation conditions. The complex non-convex two-level game model is strictly transformed into a standard single-level mixed integer linear programming model. This avoids the non-convex and nonlinear problem caused by master-slave nesting in the original problem, and allows for direct use of mature and efficient commercial mathematical programming solvers for solving the problem.

[0128] In an exemplary embodiment, the carbon trading cost in the upper-level grid operator model is calculated based on the difference between the actual carbon emissions of thermal power units and the free carbon allowances allocated based on power generation, as well as the carbon trading price; the certified voluntary emission reduction revenue in the lower-level load aggregator model is calculated based on the predicted output of wind turbines and photovoltaic power units, the corresponding carbon dioxide emission reduction factors, and the certified voluntary emission reduction trading price.

[0129] Specifically, the core of calculating carbon trading costs in the upper-level grid operator model lies in internalizing the economic signals of the carbon market into direct economic costs affecting thermal power dispatch decisions. Specifically, it first requires calculating the actual total carbon emissions of each thermal power unit during the dispatch cycle, which is obtained by multiplying the unit's active power output in each time period by its carbon emission factor per unit of electricity generated. Simultaneously, based on allocation rules or the baseline method, the total amount of free carbon allowances that the unit can obtain based on its historical data or electricity generation is determined. The carbon trading cost is the difference between the actual carbon emissions and the free carbon allowances, multiplied by the real-time unit price or predicted price in the carbon trading market. If the actual emissions exceed the allowance, the cost is positive, representing the need to pay for additional allowances; if the actual emissions are lower than the allowance, the cost is negative, representing the ability to generate revenue by selling excess allowances, which will offset the total cost in the objective function. For the certified voluntary emission reduction revenue in the lower-level load aggregator model, its calculation aims to quantify and monetize the environmental value of distributed clean generation, thereby creating a positive incentive for aggregators to actively absorb new energy sources.

[0130] In practice, the total power generation of wind and solar power generators during the dispatch cycle needs to be calculated based on their predicted output curves. Then, this clean power generation is multiplied by a carbon dioxide emission reduction factor, which represents the amount of carbon emissions avoided per kilowatt-hour of clean electricity relative to the grid average emission level or a specific baseline, to obtain the theoretical emission reduction. Finally, this certified emission reduction is multiplied by the price in the certified voluntary emission reduction trading market to obtain the expected revenue. This revenue is included as a negative cost in the lower-level objective function, directly incentivizing aggregators to maximize the utilization of wind and solar power.

[0131] In this embodiment, by embedding a carbon trading cost mechanism based on quotas and the market in the upper-level model, grid operators are driven to actively optimize thermal power output to reduce carbon emissions. At the same time, by embedding a voluntary emission reduction benefit mechanism based on actual emission reductions in the lower-level load aggregator model, load aggregators are incentivized to maximize the consumption of clean energy. This makes carbon emission costs and emission reduction benefits the core economic signals influencing the decisions of both parties, thereby spontaneously coordinating the behavior of both parties in the game process. This guides the system as a whole to significantly reduce fossil energy dependence and total carbon emissions while meeting electricity demand, achieving a deep integration and unification of economic dispatch and low-carbon goals.

[0132] In an exemplary embodiment, the constraints of the upper-level grid operator model include flexibility modification constraints and operational physical constraints for thermal power units; the flexibility modification constraints are used to reduce the minimum technical output limit of thermal power units by introducing a modification ratio coefficient; the operational physical constraints include upper and lower limit constraints for unit output, ramp rate constraints, and minimum continuous start-stop time constraints.

[0133] Specifically, the upper-level objective function is to minimize the total value of the grid operator's consideration of thermal power costs and carbon emissions trading, i.e.:

[0134] ;

[0135] In the formula, For the cost of generating electricity from thermal power units, These are the carbon emission trading costs for thermal power units.

[0136] The corresponding formulas for calculating thermal power costs and carbon emission costs are as follows:

[0137] ;

[0138] ;

[0139] In the formula, For the output of thermal power units, These are the cost coefficients for thermal power units. For carbon trading prices, For carbon emission allowances.

[0140] The corresponding constraints on thermal power units include ramp rate constraints, thermal power output constraints, start-up and shutdown constraints, and carbon quota constraints. Furthermore, to improve the flexibility of thermal power units, the minimum output value has been modified. The specific constraints are as follows:

[0141] ;

[0142] ;

[0143] ;

[0144] ;

[0145] In the formula, These represent the lower limits of output before and after the modification. This is the upper limit of the output of thermal power units. This refers to the conversion ratio coefficient for thermal power units. For the climbing speed limit, thermal power units Minimum power-on time and minimum power-off time, thermal power units arrive The duration of continuous power-on and power-off. thermal power units exist and The start / stop status at any given time. Carbon emissions purchased from the carbon market.

[0146] In this embodiment, by introducing flexibility modification constraints, the minimum technical output limit of the thermal power unit is significantly reduced, thereby greatly expanding its adjustable output range and enhancing its deep peak-shaving capability, enabling it to better adapt to fluctuations in renewable energy output. Simultaneously, by strictly setting upper and lower output limits, ramp-up rates, and minimum continuous start-stop times, operational safety and equipment lifespan are ensured during wide-range adjustments, avoiding equipment damage and operational risks caused by frequent or drastic adjustments.

[0147] In one exemplary embodiment, the charge-discharge mutual exclusion constraint of the electrochemical energy storage system is used for modeling by introducing binary variables; the unit reduction compensation cost of the load that can be reduced and the unit transfer compensation cost of the load that can be transferred are included as cost terms in the objective function of the lower load aggregator model.

[0148] Specifically, the objective function of the lower-level load aggregator model is to minimize the overall cost of the load aggregator, which includes the cost of renewable energy generation, the charging and discharging cost of energy storage, the cost of adjustable load, and the certified voluntary emission reduction benefits generated by renewable energy generation.

[0149] ;

[0150] In the formula, The costs of wind turbines and photovoltaic power generation are respectively. The cost of charging and discharging energy storage, In order to reduce load costs, For transferable load costs, Carbon trading price for distributed wind and solar power generation.

[0151] ;

[0152] ;

[0153] ;

[0154] ;

[0155] ;

[0156] ;

[0157] In the formula, These are the cost coefficients for wind turbines and photovoltaic power generation, respectively. The output of wind turbines and solar power, respectively. These are the charging and discharging costs of energy storage, The charging and discharging power of energy storage, To adjust prices per unit of load that can be reduced, The amount of load reduction that can be achieved. Adjustable price per unit of transferable load, This refers to the reduction in transferable load. The carbon trading prices for certified voluntary emission reductions from wind turbines and solar power are respectively. These represent the carbon dioxide emissions per unit of electricity generated in the distribution network. They are respectively The output of wind turbines and photovoltaics at all times.

[0158] The constraints of the lower-level load aggregator model are:

[0159] The power flow constraints, node voltage safety constraints, and line capacity constraints of the distribution network must not exceed their limits:

[0160] ;

[0161] ;

[0162] ;

[0163] In the formula, They are respectively The square term of the voltage between two points, These are the resistance and reactance of the line, respectively. These are the active power and reactive power flowing through the line, respectively. These are the upper and lower limits of the system operating voltage. for Time Node voltage, For the line The limit of transmission capacity, For the line exist The trends of the moment.

[0164] Energy storage constraints include state of charge constraints, charge / discharge power constraints, and charge / discharge mutual exclusion constraints:

[0165] ;

[0166] ;

[0167] ;

[0168] ;

[0169] In the formula, These represent the maximum energy storage capacity and the energy storage capacity value. For energy storage charging and discharging power and limits, To improve the charging and discharging efficiency of energy storage;

[0170] Regarding load reduction, a certain maximum reduction ratio should be maintained for reducible load, and the load transferable load should ensure that the transfer-in and transfer-out load values ​​remain equal.

[0171] ;

[0172] ;

[0173] In the formula, Let t be the amount of load reduction that can be achieved at time t. The load demand that can be reduced at time t. The maximum reduction coefficient, Let t be the amount of load that can be transferred at time t.

[0174] The output of a scenic viewpoint cannot exceed its maximum output value.

[0175] ;

[0176] ;

[0177] In the formula, The wind and light exert their respective forces at time t. These represent the maximum output values ​​of wind and solar power, respectively.

[0178] The upper and lower coupling constraints are the corresponding active power balance constraints:

[0179] ;

[0180] In the formula, for The load demand at any time, for Thermal power output at all times for Energy storage and output at all times for Solar power output at all times for The wind turbine output at all times, for The load can be reduced at any time. for Transferable load at any given time.

[0181] A binary variable needs to be introduced for the energy storage system in each scheduling period. This variable represents the working state of the energy storage (e.g., 1 represents the discharging state, and 0 represents the charging or idle state). By associating this binary variable with a continuous decision variable representing the charging and discharging power, and using a sufficiently large positive number (i.e., the Big M method), a set of linear constraints can be constructed, thereby forcing that in any given period, one and only one of the charging and discharging power can take a non-zero value, or both can be zero.

[0182] For loads that can be reduced or transferred, their role in the model lies not only in providing adjustment potential but also in the real economic costs associated with such adjustments. The objective function of the lower-level load aggregator model needs to explicitly account for the compensation costs incurred in utilizing these resources. Specifically, the cost of a load that can be reduced is the product of its actual electricity reduction and a pre-agreed unit reduction compensation price; the cost of a load that can be transferred is the product of its transferred electricity and a unit transfer compensation price. Incorporating these costs as linear terms into the objective function forces the load aggregator's optimization process to economically weigh the compensation costs paid to users against the costs that could be saved through other means (such as using energy storage or obtaining CCER revenue).

[0183] In this embodiment, by introducing binary variables to model the mutual exclusion of energy storage charging and discharging, the physical feasibility of the energy storage system's operating state is ensured, avoiding invalid instructions from the model to charge and discharge simultaneously. This ensures the executability and safety of the scheduling scheme at the device level. Simultaneously, the adjustment and compensation costs of flexible loads are explicitly included in the objective function, prompting load aggregators to automatically weigh compensation expenditures against other operational benefits during the optimization process. This approach taps into demand-side potential while ensuring user participation and the feasibility of the solution.

[0184] The most detailed embodiment of this application is as follows:

[0185] like Figure 4 As shown, the first step is system initialization and data input. All parameters and boundary conditions required within the scheduling cycle are input, forming the baseline for the optimization model. Specifically, this includes: grid topology (e.g., the IEEE 33-node system), line parameters (resistance, reactance, capacity limits); source-side parameters, such as the cost coefficient, carbon emission factor, carbon quota coefficient, retrofit ratio coefficient, output upper and lower limits, ramp rate, and minimum start-up / shutdown time of thermal power units, as well as the cut-in / cut-out wind speed, rated capacity, solar intensity prediction curve, and power characteristic parameters of wind and solar power units; load-side parameters, namely the base load prediction curve for each node, the maximum reduction ratio and unit compensation price for loads that can be reduced, and the unit compensation price for loads that can be transferred; storage-side parameters, namely the configured capacity, rated power, charge / discharge efficiency, initial state of charge (SOC), and SOC upper and lower limits of the energy storage system; and market parameters, namely carbon trading prices and certified voluntary emission reduction trading prices.

[0186] Secondly, a two-tiered optimization scheduling model for source-grid-load-storage coordination is constructed. This model clearly distinguishes the stakeholders and decision-making scope of the upper-level decision-makers (grid operators) and the lower-level decision-makers (load aggregators). The upper-level model aims to minimize the total cost of the grid operator, and its objective function is the sum of the operating cost of thermal power units and the carbon trading cost. The operating cost of thermal power is modeled using a quadratic function, while the carbon trading cost is calculated by multiplying the difference between the actual carbon emissions of each thermal power unit and the free carbon allowance obtained based on its power generation by the carbon trading price. The decision variables of the upper-level model are the output plans of each thermal power unit in each time period, and its constraints include: upper and lower limits of output after flexibility upgrades (by introducing a modification ratio coefficient to reduce the original minimum technical output), ramp-up constraints, and minimum continuous start-up and shutdown time constraints. The lower-level model aims to minimize the comprehensive cost of the load aggregator, and its objective function is the sum of the cost of renewable energy generation, the operating cost of energy storage, and the cost of flexible load compensation, minus the certified voluntary emission reduction revenue generated by renewable energy generation. The decision variables of the lower-level model are the output of distributed wind and solar power, the charging and discharging power of energy storage, the amount of load reduction that can be reduced, and the amount of load transfer that can be transferred within its jurisdiction. Its constraints are a complete set, including: power flow constraints of the distribution network (using linearized models such as DistFlow), node voltage security constraints, and line transmission capacity constraints; dynamic update constraints of the energy storage system's state of charge (SOC), upper and lower limit constraints of charging and discharging power, and mutual exclusion constraints of charging and discharging states implemented by introducing binary variables; constraints that the amount of load reduction that can be reduced does not exceed a certain proportion of its base load demand for that period; constraints that the total transfer in and total transfer out of transferable load remain balanced throughout the entire dispatch cycle; and constraints that the output of wind and solar turbines does not exceed their predicted maximum output. The core connecting the upper and lower-level models is a global power balance coupling constraint. This constraint requires that, in each dispatch period, the sum of the total output of upper-level thermal power, the total output of lower-level wind and solar power, and the net discharge power of energy storage must equal the total system load demand minus the net load value adjusted for transferable load, thereby ensuring that the decisions of both sides are physically consistent.

[0187] Subsequently, the model is transformed and solved. Because the constructed two-layer model has a nested structure, the lower-layer decision variables depend on the upper-layer decisions, resulting in a non-convex and non-linear model, making direct solution difficult. This embodiment uses a method based on Karush-Kuhn-Tucker optimality conditions for equivalent transformation. The specific steps are: constructing a Lagrangian function for the lower-layer optimization problem; deriving its KKT optimality condition set, including stability conditions, primal feasibility conditions, dual feasibility conditions, and complementary relaxation conditions. The complementary relaxation conditions are non-linear equations. To handle this non-linearity, for each complementary relaxation condition, a binary auxiliary variable and a sufficiently large positive number M are introduced, and the big M method is used to equivalently transform it into two linear inequality constraints. Through this step, the lower-layer optimization problem is completely equivalently replaced by its linearized KKT conditions. These conditions are then merged with all constraints of the upper-layer model, transforming the original two-layer model into a single-layer mixed-integer linear programming model where the decision variables include all primal variables, KKT multipliers, and the newly introduced binary variable. This model is in standard form and can be efficiently solved using mature commercial mathematical programming solvers (such as Gurobi). The solver is invoked to compute this MILP model, aiming to minimize the original upper-level objective function while satisfying all constraints.

[0188] Finally, the optimal scheduling scheme is analyzed and output. After the solution is completed, the values ​​of key decision variables are extracted from the global optimal solution output by the solver. This includes: the time-sharing output plans of each thermal power unit, which constitute the scheduling instructions on the grid operator side; the time-sharing planned output of each distributed wind turbine and photovoltaic unit, the time-sharing charging and discharging plans of each energy storage system, the time-sharing reduction amount of each load that can be reduced, and the time-sharing transfer amount of each load that can be transferred, which together constitute the coordinated scheduling scheme on the load aggregator side. At the same time, key performance indicators such as total system cost, carbon emissions, and renewable energy absorption rate can be output. This scheme achieves the comprehensive optimization of system economy and low carbon emissions by driving the coordinated interaction of resources from power generation, grid, load, and storage through economic incentives and carbon market signals, under the premise of strictly meeting grid safety and equipment operation constraints.

[0189] This simulation example uses the IEEE 33-node distribution system as the simulation object. The system's base voltage is set to 12.66kV, the base capacity to be 10MVA, the dispatch cycle to be 24 hours, and the time step to be 1 hour. Specific parameters are configured as follows: a thermal power unit with a maximum output of 8000kW is configured at node 1; wind turbines with a total capacity of 2400kW are configured at nodes 8, 18, and 30; and photovoltaic units with a total capacity of 2100kW are configured at nodes 4, 12, and 25. Two energy storage systems are configured at nodes 6 and 21, each with a power and capacity of 600kW / 2400kWh and a charge / discharge efficiency of 95%. Loads can be reduced at nodes 24 and 32, with a maximum reduction ratio of 20%; loads can be transferred at nodes 15 and 28, with a maximum transfer ratio of 15%. The system has a certain amount of free carbon credits; any excess requires the purchase of carbon emission rights. Additionally, renewable energy generation can earn CCER (China Certified Emission Reduction) revenue. The specific network topology and node layout are shown below. Figure 5 As shown.

[0190] After model optimization, the total operating cost of the system within 24 hours was determined to be 34,276.38 yuan. This cost is jointly determined by the game between the upper-level grid operator and the lower-level aggregator. This includes the upper-level cost, namely the grid operator's operating cost of 38,827.96 yuan, consisting of thermal power fuel costs of 30,836.37 yuan and carbon trading costs of 7,991.59 yuan. The lower-level source-load-storage aggregator's operating cost is -4,551.58 yuan, meaning it achieves a profit of 4,551.58 yuan. Although the aggregator bears the costs of new energy, energy storage, and load regulation (1,839.89 yuan), the certified voluntary emission reduction revenue of 6,391.47 yuan obtained by providing green electricity to the system far exceeds its operating cost, verifying the effectiveness of the incentive mechanism set in the model. This cost distribution clearly reflects the economic relationship of the two-level master-slave game: the upper-level grid, through payment incentives, guides the lower-level aggregator to actively optimize its resources to support the grid, ultimately achieving cost reduction and efficiency improvement at the system level. The final optimized scheduling results show the output of each unit as follows: Figure 6 As shown:

[0191] according to Figure 6 It visually displays the output of each part over a 24-hour period.

[0192] On the power generation side, thermal power units, as the main supporting power source of the system, generated a total of 67,893.64 kWh, accounting for 62.2%. Their output curve basically followed the trend of the system's net load, providing stable power support during nighttime and peak load periods when renewable energy output was insufficient. For renewable energy, wind power and solar power, as zero-marginal-cost sources, were prioritized for dispatch, with a total daily output of 40,710 kWh, representing a renewable energy penetration rate of 37.3%. Solar power output was concentrated between 7:00 and 18:00, peaking at 12:00; wind power had better output at night and in the early morning. The optimized dispatching model effectively promoted the local consumption of a high proportion of renewable energy. Figure 6 As can be seen, during the evening hours when photovoltaic output weakens, wind power output gradually increases, sharing the evening peak load with thermal power and energy storage discharge, forming a good "wind, solar, thermal, and energy storage" multi-energy complementary pattern, effectively mitigating the volatility of a single new energy source.

[0193] In terms of energy storage and demand-side response, both played a crucial role in smoothing load fluctuations and achieving peak shaving and valley filling. The energy storage system implemented an efficient charging and discharging strategy throughout the day, charging during the off-peak hours of 1:00-7:00 AM and the peak solar power generation and lower electricity prices of 12:00-3:00 PM, effectively absorbing excess green electricity. Discharging occurred during the two peak electricity consumption periods of 9:00-11:00 AM and 6:00-9:00 PM, reducing reliance on thermal power units during peak periods and lowering the system's peak output cost. On the demand side, the load that could be reduced was mainly called upon during the evening peak of 8:00 PM, reducing load by 1.2% and directly lowering the system's peak load. The load that could be transferred achieved a spatial and temporal transfer of load, shifting some non-rigid load from peak hours to the lower load periods in the early morning, achieving a positive interaction between supply and demand.

[0194] The system operating cost exhibits significant time-varying qualitative characteristics within the scheduling cycle. The collaborative interaction characteristics of resources on the source, grid, load, and storage sides are fully reflected in the time series. The cost change diagrams for each period of the game are shown in Figure 7(a), (b), and (c).

[0195] As shown in Figures 7(a), (b), and (c), the total system cost curve fluctuates with load demand and renewable energy output, reaching its lowest point of 470.78 yuan in period 2 and its highest point of 3592.21 yuan in period 20. In terms of cost composition, the upper-level grid bears the majority of operating costs, and its trend is highly positively correlated with the system's net load curve, especially during nighttime off-peak and evening peak periods, where thermal power output and carbon trading costs dominate. However, the CCER environmental benefits generated by renewable energy generation on the lower-level operating side strongly offset the positive system costs. Specifically, during the midday peak of photovoltaic power generation and the nighttime peak of wind power output, lower-level resources not only reduce the demand for upper-level thermal power output by providing clean energy, but also offset their own operation and maintenance costs and energy storage losses through high certified voluntary emission reduction benefits, resulting in a "net benefit" state for the lower-level total cost for most periods. In addition, the energy storage system and demand response mechanism played a key role in peak shaving during peak load periods. Although a small amount of cost was generated to compensate for the load that could be reduced and transferred, it effectively alleviated the pressure of thermal power plant ramping up and prevented further deterioration of the system's marginal cost. This verified the rationality and economy of the two-layer model in scheduling multiple types of resources in the time-series dimension.

[0196] This example uses a single-layer transformation method based on KKT conditions to convert the originally complex bi-layer nonlinear programming problem into a mixed-integer linear programming problem for solution. Its convergence performance and computational efficiency were verified in simulations. The iterative convergence process of the solver is shown in Figures 8(a) and (b). After approximately 50 iterations, the model quickly reaches the Nash equilibrium point, and the objective function value exhibits good convergence characteristics. Specifically, the upper-layer cost curve gradually decreases from its initial high level and stabilizes at 38827.96 yuan, indicating that the grid side effectively controls fuel and carbon emission expenditures by optimizing thermal power output plans. Meanwhile, the lower-layer cost curve dynamically adjusts during the game process, eventually converging to -4551.58 yuan, realizing a shift from simply pursuing low costs to a strategy that also considers environmental benefits. The smooth convergence trajectories of the two curves show that introducing the Big-M method to handle complementary relaxation conditions did not cause numerical oscillations or non-convergence. In terms of computational efficiency, the Gurobi solver successfully searched for the global optimum within a finite time, and the MIPGap was kept within a very small range, demonstrating the robustness of the model in handling IEEE 33-node power flow constraints and various types of discrete variables. This efficient convergence performance means that the algorithm is not only suitable for offline scheduling analysis, but also has the potential to be applied in actual daily or real-time scheduling scenarios in distribution networks.

[0197] As a key node equipped with an electrochemical energy storage system, Node 6's operating characteristics fully demonstrate the system's flexibility in responding to load fluctuations and price signals. The output and SOC changes of Node 6 are shown in Figure 9(a) and (b).

[0198] As shown in Figures 9(a) and 9(b), the energy storage system strictly follows the "low storage, high discharge" arbitrage strategy during the scheduling cycle, exhibiting a clear "two-charge, two-discharge" dual-cycle characteristic. Specifically, during the early morning off-peak period and midday off-peak period, driven by both low electricity prices and the demand for renewable energy consumption, the energy storage system performs high-power charging, rapidly increasing its state of charge (SOC) to the upper limit of 90%, effectively absorbing redundant system power and filling load troughs. During the morning and evening peak periods, the energy storage system switches to discharge mode, with the SOC rapidly decreasing to the lower limit of 10%. At this time, the energy storage output and thermal power jointly support the node load, significantly reducing the power supply pressure and carbon emission costs during peak periods. Furthermore, the SOC curve remains within the upper and lower limit constraints, and the final SOC of the scheduling cycle returns to the initial level, ensuring sustainability for the next scheduling cycle. This precise timing not only achieves peak shaving and valley filling at the node level but also verifies the superior performance of the two-layer game model in tapping the potential of distributed resource regulation and balancing economic efficiency and system stability by smoothing net load fluctuations.

[0199] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0200] Based on the same inventive concept, this application also provides a two-layer optimization scheduling device for source-grid-load-storage to implement the aforementioned two-layer optimization scheduling method. The solution provided by this device is similar to the implementation described in the above method. Therefore, the specific limitations of one or more embodiments of the two-layer optimization scheduling device for source-grid-load-storage provided below can be found in the limitations of the two-layer optimization scheduling method for source-grid-load-storage described above, and will not be repeated here.

[0201] In one exemplary embodiment, such as Figure 10 As shown, a two-layer optimized scheduling device for source-grid-load-storage is provided, comprising:

[0202] Model building module 1002 is used to construct a two-layer optimization model that includes an upper-layer grid operator model and a lower-layer load aggregator model. The upper-layer grid operator model aims to minimize the total cost, including the operating cost of thermal power units and the cost of carbon trading. The lower-layer load aggregator model aims to minimize the comprehensive cost, including the cost of renewable energy generation, the operating cost of energy storage, the cost of flexible load compensation, and the benefits of certified voluntary emission reductions. The upper-layer grid operator model and the lower-layer load aggregator model are linked by power balance coupling constraints.

[0203] The linearization module 1004 is used to introduce binary auxiliary variables and positive numbers into the Karush-Kuhn-Tucker conditions based on the lower-level load aggregation quotient model. It transforms each nonlinear complementary relaxation condition into a linear inequality constraint condition, thereby linearizing the two-level optimization model into a single-level mixed integer linear programming model.

[0204] The calculation module 1006 is used to output the optimal scheduling scheme for grid operators and load aggregators using a single-level mixed integer linear programming model.

[0205] In an exemplary embodiment, the scheduling objects in the lower-level load aggregator model include wind turbine generators, photovoltaic generators, electrochemical energy storage systems, loads that can be reduced, and loads that can be transferred. The constraints of the lower-level load aggregator model include: power flow constraints and node voltage security constraints of the distribution network, state of charge constraints and charge / discharge mutual exclusion constraints of the electrochemical energy storage system, the constraint that the reduction amount of the load that can be reduced in any scheduling period does not exceed the maximum reduction ratio of the basic load demand of the scheduling period, and the balance constraint that the total transfer-in amount and the total transfer-out amount of the load that can be transferred are equal during the scheduling cycle.

[0206] In an exemplary embodiment, the linearization module 1004 is specifically used to construct a Lagrangian function for the optimization problem of the lower-level load aggregator model; based on the Lagrangian function, derive the Karush-Kuhn-Tucker optimality condition set, which includes stability conditions, primal feasibility conditions, dual feasibility conditions, and complementary relaxation conditions; for the complementary relaxation conditions, introduce binary auxiliary variables and use the Big M method to equivalently transform each nonlinear complementary relaxation condition into two linear inequality constraints; and combine the linearized optimality condition set with all the constraints of the upper-level grid operator model to form the constraint set of a single-level mixed-integer linear programming model.

[0207] In an exemplary embodiment, the carbon trading cost in the upper-level grid operator model is calculated based on the difference between the actual carbon emissions of thermal power units and the free carbon allowances allocated based on power generation, as well as the carbon trading price; the certified voluntary emission reduction revenue in the lower-level load aggregator model is calculated based on the predicted output of wind turbines and photovoltaic power units, the corresponding carbon dioxide emission reduction factors, and the certified voluntary emission reduction trading price.

[0208] In an exemplary embodiment, the constraints of the upper-level grid operator model include flexibility modification constraints and operational physical constraints for thermal power units; the flexibility modification constraints are used to reduce the minimum technical output limit of thermal power units by introducing a modification ratio coefficient; the operational physical constraints include upper and lower limit constraints for unit output, ramp rate constraints, and minimum continuous start-stop time constraints.

[0209] In one exemplary embodiment, the charge-discharge mutual exclusion constraint of the electrochemical energy storage system is used for modeling by introducing binary variables; the unit reduction compensation cost of the load that can be reduced and the unit transfer compensation cost of the load that can be transferred are included as cost terms in the objective function of the lower load aggregator model.

[0210] Each module in the aforementioned two-layer optimized scheduling device for source-grid-load-storage can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0211] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 11 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a two-tier optimized scheduling method for source-network-load-storage systems.

[0212] Those skilled in the art will understand that Figure 11 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0213] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method described above.

[0214] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0215] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the method described above.

[0216] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0217] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0218] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0219] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A two-layer optimal scheduling method for source-grid-load-storage, characterized in that, The method includes: A two-tier optimization model is constructed, comprising an upper-level grid operator model and a lower-level load aggregator model. The upper-level grid operator model aims to minimize the total cost, including the operating cost of thermal power units and the cost of carbon trading. The lower-level load aggregator model aims to minimize the comprehensive cost, including the cost of renewable energy generation, the operating cost of energy storage, the cost of flexible load compensation, and the revenue from certified voluntary emission reductions. The upper-level grid operator model and the lower-level load aggregator model are linked by power balance coupling constraints. Based on the Karush-Kuhn-Tucker conditions of the lower-level load aggregator model, binary auxiliary variables and positive numbers are introduced, and each nonlinear complementary relaxation condition is equivalently transformed into a linear inequality constraint condition, so as to linearize the two-level optimization model and transform it into a single-level mixed integer linear programming model. Using the single-layer mixed integer linear programming model, the optimal scheduling scheme for the power grid operator and load aggregator is output.

2. The method according to claim 1, characterized in that, The scheduling objects in the lower-level load aggregator model include wind turbine generators, photovoltaic generators, electrochemical energy storage systems, loads that can be reduced, and loads that can be transferred. The constraints of the lower-level load aggregator model include: power flow constraints and node voltage security constraints of the distribution network; state of charge constraints and charge / discharge mutual exclusion constraints of the electrochemical energy storage system; the constraint that the reduction amount of the load that can be reduced in any scheduling period does not exceed the maximum reduction ratio of the basic load demand in the scheduling period; and the balance constraint that the total transfer-in amount and the total transfer-out amount of the load that can be transferred out are equal in the scheduling cycle.

3. The method according to claim 1, characterized in that, The Karush-Kuhn-Tucker conditions based on the lower-level load aggregator model introduce binary auxiliary variables and positive numbers, and transform each nonlinear complementary relaxation condition into a linear inequality constraint condition to linearize the bi-level optimization model, including: Construct a Lagrangian function for the optimization problem of the lower-level load aggregator model; Based on the Lagrangian function, the Karush-Kuhn-Tucker optimality condition set, which includes stability conditions, primal feasibility conditions, dual feasibility conditions, and complementary relaxation conditions, is derived. For the complementary relaxation conditions, binary auxiliary variables are introduced, and the Big M method is used to convert each nonlinear complementary relaxation condition into two linear inequality constraints. The set of optimality conditions after linearization, together with all the constraints of the upper-level power grid operator model, constitutes the constraint set of the single-level mixed integer linear programming model.

4. The method according to claim 1, characterized in that, The carbon trading cost in the upper-level grid operator model is calculated based on the difference between the actual carbon emissions of thermal power units and the free carbon allowances allocated based on power generation, as well as the carbon trading price; the certified voluntary emission reduction revenue in the lower-level load aggregator model is calculated based on the output of wind turbines and photovoltaic generators, the corresponding carbon dioxide emission reduction factors, and the certified voluntary emission reduction trading price.

5. The method according to claim 1, characterized in that, The constraints of the upper-level power grid operator model include flexibility modification constraints and operational physical constraints for thermal power units. The flexibility modification constraints are used to reduce the minimum technical output limit of the thermal power units by introducing a modification ratio coefficient. The operational physical constraints include upper and lower limits of unit output, ramp rate constraints, and minimum continuous start-stop time constraints.

6. The method according to claim 2, characterized in that, The charge-discharge mutual exclusion constraint of the electrochemical energy storage system is used for modeling by introducing binary variables; the unit reduction compensation cost of the load that can be reduced and the unit transfer compensation cost of the load that can be transferred are included as cost terms in the objective function of the lower-level load aggregator model.

7. A two-layer optimized scheduling device for source-grid-load-storage, characterized in that, The device includes: The model building module is used to construct a two-layer optimization model comprising an upper-layer grid operator model and a lower-layer load aggregator model. The upper-layer grid operator model aims to minimize the total cost, including thermal power unit operating costs and carbon trading costs. The lower-layer load aggregator model aims to minimize the comprehensive cost, including renewable energy generation costs, energy storage operating costs, flexible load compensation costs, and certified voluntary emission reduction benefits. The upper-layer grid operator model and the lower-layer load aggregator model are linked through power balance coupling constraints. The linearization module is used to introduce binary auxiliary variables and positive numbers based on the Karush-Kuhn-Tucker conditions of the lower-level load aggregator model, and to convert each nonlinear complementary relaxation condition into a linear inequality constraint condition, so as to linearize the two-level optimization model and transform it into a single-level mixed integer linear programming model. The calculation module is used to output the optimal scheduling scheme for the grid operator and load aggregator using the single-level mixed integer linear programming model.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images