A multi-micronet robust game optimization scheduling method and system based on a dynamic auction algorithm

Abstract

The application discloses a kind of multi-micronet robust game optimization scheduling method and system based on dynamic auction official algorithm, and relates to energy system scheduling and optimization technical field.The method constructs system physical model and double-layer robust optimization model based on Stackelberg master-slave game, upper layer maximizes system operator's profit to formulate price signal, lower layer maximizes the worst scenario income of multi-micronet alliance to optimize resource scheduling, combined with multiple constraint conditions, Stackelberg equilibrium is solved iteratively using dynamic auction official algorithm.The application converges to global optimal solution quickly through three-dimensional hybrid driving price updating mechanism, realizes system economic benefit maximization, operation robust and reliable, and has good explainability and practical application value.

Description

Technical Field

[0001] This invention relates to the field of energy system scheduling and optimization technology, and more specifically to a robust game-theoretic optimization scheduling method and system for multi-microgrids based on a dynamic auctioneer algorithm. Background Technology

[0002] To address the challenges of global climate change and energy security, building a new power system based on renewable energy has become a global consensus, driving a profound transformation of the energy system towards cleaner and lower-carbon directions. Against this backdrop, Distribution-Microgrid Integrated Energy Systems (DM-IES), as a key technology for achieving "horizontal multi-energy complementarity and vertical source-grid-load-storage coordination," have attracted widespread attention from the industry. DM-IES integrates various energy flows such as electricity, heat, cooling, and gas within a distribution network, and utilizes advanced energy conversion and storage technologies to achieve cascaded energy utilization and a significant improvement in overall efficiency. A typical DM-IES architecture is complex, usually comprising multiple heterogeneous microgrids with different functions, shared energy storage (SES) power stations, centralized renewable energy power stations, and other diverse entities, all operating within a single distribution area.

[0003] In actual operation, DM-IES faces three major technical challenges: difficulty in coordinating the conflicting interests of multiple stakeholders; uncertainty on both the source and load sides threatening energy supply security; and computational bottlenecks in solving the two-layer optimization model. These challenges severely restrict the full realization of its comprehensive effectiveness. The specific situations are as follows:

[0004] (1) Coordination among multiple stakeholders and conflicts of interest

[0005] The primary challenge to the efficient collaborative operation of DM-IES lies in the fact that the various components within the system—microgrids, shared energy storage operators, and renewable energy power plants—are all independent economic entities with differing and even conflicting operational objectives. Traditional centralized optimization scheduling methods assume the existence of a central controller capable of uniformly scheduling all resources, which is clearly impractical in a multi-entity, market-driven environment. Therefore, game theory, particularly the Stackelberg master-follower game model, which clearly characterizes the hierarchical decision-making relationship between leaders and followers in the market, is considered an effective tool for addressing such problems.

[0006] However, existing game theory models have significant limitations. Most studies focus solely on electricity trading, neglecting the complex coupling relationships between electricity, heat, and cold energy flows, as well as the enormous potential of Integrated Demand Response (IDR) in enhancing system flexibility. Furthermore, existing models fail to adequately depict the complex game dynamics involving emerging market players such as multiple heterogeneous microgrids and independent shared energy storage operators. This lack of effective coordination is not neutral; it inevitably leads to system-level economic inefficiency. For example, without guidance, microgrids may simultaneously purchase electricity from the main grid during peak consumption periods, exacerbating grid load and driving up purchase costs. Conversely, during periods of high renewable energy generation, disorderly low-price competition may occur, harming overall interests. Therefore, developing an advanced dispatch mechanism that can effectively coordinate the interests of multiple parties and guide their behavior towards global optimization is a pressing technical challenge.

[0007] (2) The ubiquitous source-load uncertainty

[0008] DM-IES's high dependence on renewable energy sources such as wind and solar power introduces significant uncertainty, randomness, and volatility into its operation. This uncertainty on the supply side, coupled with the random fluctuations in electricity, heat, and cooling load demand on the user side, constitutes a high degree of uncertainty on both the source and load sides, posing a serious threat to the system's economic operation and energy supply reliability.

[0009] Existing technologies primarily employ two types of methods to address this problem: First, stochastic optimization. This method relies on precise assumptions about the probability distribution of uncertain variables and approximates the uncertainty through scenario generation and reduction. However, the reliability of its calculation results heavily depends on the accuracy of the probability model, and it may lose information about low-probability, high-impact extreme events crucial to system safety during scenario reduction. Second, robust optimization. This method does not rely on probability distributions but instead constructs a set of uncertainties and seeks the optimal solution that still satisfies all constraints under all possible worst-case scenarios within that set.

[0010] Robust optimization offers the advantage of providing deterministic security guarantees for system operation. However, existing research rarely integrates robust optimization with multi-agent game theory frameworks to study the optimal robust strategy equilibrium problem for various stakeholders under uncertain environments. Any non-robust scheduling scheme, whether based on deterministic predictions or stochastic schemes based on expected value optimization, is inherently fragile. A tiny prediction deviation, such as a cloud drifting over a photovoltaic power station, can render the entire scheduling plan infeasible, leading to high emergency power purchase costs or forced load shedding. Therefore, the technical challenge lies not only in "handling" uncertainty but also in providing a scheduling strategy that can "guarantee" the system's safe and economical operation even in the worst-case scenario—that is, embedding risk management into the optimization model.

[0011] (3) Computational bottleneck of bi-level optimization

[0012] Combining master-slave game theory with robust optimization inevitably leads to a high-dimensional, non-convex, and nonlinear two-level mixed integer programming problem, which is computationally extremely challenging. Current techniques typically employ heuristic intelligent algorithms (such as Particle Swarm Optimization, PSO) to solve the upper-level leader pricing problem. However, standard heuristic algorithms often perform poorly when dealing with high-dimensional decision spaces exceeding one hundred dimensions, commonly exhibiting problems such as premature convergence, susceptibility to local optima, and unstable solution quality, making it difficult to guarantee the quality and global optimality of the final equilibrium solution found.

[0013] In summary, existing technologies have significant gaps in achieving collaborative, robust, and efficient scheduling of complex DM-IES. The market urgently needs a novel optimization scheduling method that must: 1) accurately simulate the strategic interactions among multiple independent stakeholders in complex DM-IES; 2) intrinsically guarantee at the mathematical model level that the scheduling scheme can withstand worst-case uncertainties on both the source and load sides, ensuring the system's operational safety and reliability; and 3) be equipped with efficient and reliable solution algorithms capable of rapidly solving the resulting complex bi-level optimization problem and converging to a high-quality global equilibrium solution. Summary of the Invention

[0014] In view of this, the present invention provides a robust game optimization scheduling method and system for multi-micronet based on dynamic auctioneer algorithm, which solves the problems existing in the background technology.

[0015] To achieve the above objectives, the present invention provides the following technical solution:

[0016] A robust game optimization scheduling method for multi-micronets based on a dynamic auctioneer algorithm includes the following steps:

[0017] S1. Construct a mathematical model for a multi-microgrid integrated energy system;

[0018] S2. Construct a two-layer robust optimization model consisting of an upper-level leader model and a lower-level follower alliance model. The upper-level leader model aims to maximize the total profit of the system operator, and the decision variable is the price signal vector. The lower-level follower alliance model is a robust optimization model that aims to maximize the alliance's profit under the worst-case uncertainty scenario, and the decision variable is the operating strategy of all controllable resources within the alliance.

[0019] S3. Establish constraints covering power balance, thermal power balance, cold power balance, power transmission between microgrids, power transmission between distribution network and microgrid, equipment output, SES operation, output of external wind farms, and reserve capacity.

[0020] S4. The dynamic auctioneer algorithm is used to iteratively solve the two-layer robust optimization model. By initializing the price, generating and submitting the tender, driving the price update in three dimensions and checking the convergence, the optimized scheduling strategy corresponding to the Stackelberg equilibrium solution is output.

[0021] Optionally, S1 specifically includes the following steps:

[0022] S11. Based on the system operator as the leader and the cooperative alliance of multiple heterogeneous microgrids, a shared energy storage power station and a centralized renewable energy power station as followers, establish an architecture model for a multi-microgrid integrated energy system.

[0023] S12. Take into account the user's IDR behavior and conduct modeling for demand response according to the three different load types: electrical load, thermal load, and cooling load.

[0024] S13. Considering the uncertainties of wind, solar and load, a model based on probability distribution and scenario generation is adopted for modeling, and finally typical scenarios of wind power, photovoltaic and electric heating cooling loads are obtained.

[0025] Optionally, the specific steps of S12 are as follows:

[0026] S121. Incorporate price-based electricity demand response into microgrids 1 and 3, and adjust the electricity load elasticity coefficient. The expression is:

[0027]

[0028] Electricity price elasticity matrix The expression is:

[0029]

[0030] Electricity load after demand response Represented as:

[0031]

[0032] In the formula: , These represent electricity consumption. and electricity price The relative increment; This indicates the electricity consumption before optimization. This indicates the change in electricity consumption after the controllable load power consumption status has been adjusted. Represents the elastic coefficient. and These represent the electricity price and its change, respectively.

[0033] S122. In microgrid 2, the incentive-based load demand response is adopted, and the calculation method is as follows:

[0034]

[0035]

[0036]

[0037]

[0038] In the formula: , , These represent the base load, the load that can be reduced, and the load that can be transferred, respectively. This represents the total load that can be transferred within the scheduling period; and These represent the maximum power values ​​for transferable and load-reducible loads, respectively. and Corresponding transfer compensation unit price and reduction compensation unit price; This indicates the compensation generated by an incentive-based demand response; This represents the load power under incentive-driven demand response. T Indicates the total number of scheduling periods;

[0039] The heat sources for S123, heat load 2, and heat load 3 are supplied by gas turbines and gas boilers, and an excitation-based heat load demand response mechanism is adopted. The calculation method for transferable heat load is as follows:

[0040]

[0041]

[0042]

[0043] In the formula: Indicates the heat load that can be transferred; and These represent the start and end times of the heat load transfer, respectively. Indicates the duration of continuous operation of the load; and These respectively represent the operating status and start / stop status of the heat load; express t Transferable heat load at any given time;

[0044]

[0045]

[0046]

[0047] In the formula: and These represent the base heat load and the heat load after demand response adjustment, respectively. and These represent the prices for heat load transfer and reduction, respectively. This indicates the benefits derived from heat load demand response; This indicates the starting point at which the heat load can be reduced. This indicates the end point at which the heat load can be reduced. express t The heat load can be reduced at any time. This indicates that the total heat load can be reduced within the scheduling cycle. This represents the minimum power required to reduce the heat load. This indicates the maximum power that can reduce the heat load;

[0048] S124. The cooling load is equivalent to a cooling demand response with adjustment capabilities, expressed as follows:

[0049]

[0050]

[0051] In the formula: The voting value represents the feeling of hot or cold. This indicates the cooling load power after demand response. C represents the heat loss rate, and C represents the heat capacity. , They represent t Time period t Indoor temperature during the -1 hour period, express t Outdoor temperature during the time period; S Indicates the cooling area. Indicates the scheduling time interval.

[0052] Optionally, the specific steps of S13 are as follows:

[0053] Wind speed follows a Weibull distribution, and its probability density function is:

[0054]

[0055] In the formula: v For wind speed, , These are the parameters of the Weibull distribution;

[0056]

[0057] In the formula: It signifies that one has contributed merit and effort. Indicates the cut-in wind speed. Indicates the cut-out wind speed. Indicates the rated wind speed of the unit. Indicates the rated output power;

[0058] The output of photovoltaic power generation is related to the intensity of solar radiation, which follows a Beta distribution with the following probability density function:

[0059]

[0060] In the formula: r Indicates light intensity. This indicates the maximum light intensity. , The coefficient of the Beta distribution;

[0061] The prediction errors for electricity, heat, and cooling loads follow a normal distribution:

[0062]

[0063] In the formula: L This represents the actual value of the load. This represents the predicted load value. The standard deviation of the prediction error;

[0064] Taking into account the uncertainties of wind, solar, and load, Latin hypercube sampling technology is used to generate multiple scenarios, and clustering is performed through clustering algorithms to finally obtain typical scenarios of wind power, photovoltaic, and electric heating and cooling loads.

[0065] Optionally, S2 specifically includes the following steps:

[0066] S21. The objective function of the upper-level leader model is to maximize the interactive benefits of the system operators. :

[0067]

[0068] In the formula: This indicates that the leaders have formulated a micro-network. i Internal transaction price for electricity sales, microgrid i Electricity sold to the distribution network This indicates the cost price of electricity purchased from an external power grid. microgrid i Purchased power from the distribution network; and These represent the total compensation expenditure paid by the leader to incentivize followers to participate in electricity and heat demand response, respectively, with the unit price being a decision variable for the leader. N Indicates the number of microgrids;

[0069] The decision variables in the upper-level leader model are a set of price signal vectors. By issuing price signal vectors, the upper-level leader guides the behavior of the lower-level alliance in order to maximize its own profits.

[0070] S22. The objective function of the lower-level follower alliance model is to maximize the total revenue of the lower-level alliance. :

[0071]

[0072] In the formula: z Represents a vector of uncertain scenarios. Z Represents a set of uncertainties. This represents the total revenue of the alliance. Indicates the total cost of the alliance;

[0073] Specifically, it can be elaborated as follows:

[0074]

[0075] In the formula: , They represent microgrids respectively. i exist t The compensation benefits obtained from electricity and heat demand response at all times; Indicates power generation equipment g Unit power generation subsidy, Indicates power generation equipment g exist t Constant effort This indicates that the leaders have formulated a micro-network. i Internal transaction price for electricity sales, microgrid i From the power purchase capacity of the distribution network, Indicates the price of natural gas. microgrid i Gas purchase volume, Indicates power generation equipmentg The unit operation and maintenance cost This represents the cost per unit of standby capacity. This indicates that the system is providing an increased reserve capacity; PV Indicates photovoltaic power generation, CHP Indicates combined heat and power (CHP). WT Indicates wind power generation;

[0076] The decision variables in the lower-level follower alliance model are the operating strategies of all controllable resources within the alliance, which are specific responses to the leader's price signals.

[0077] Optionally, S4 specifically includes the following steps:

[0078] S41, System operator sets initial price vector and iteration parameters;

[0079] S42. The system operator broadcasts the initial price vector, solves for the optimal scheduling scheme and the dual variables corresponding to the power balance constraints, and submits the solution.

[0080] S43. The system operator uses a three-dimensional hybrid driving strategy to update prices;

[0081] S44. If the preset convergence condition is met, output the current price vector and the corresponding scheduling scheme as the Stackelberg equilibrium solution; otherwise, return to S42 and repeat the iteration.

[0082] Optionally, the specific steps of S42 are as follows:

[0083] S421, The system operator broadcasts the current number to followers. k Price vector of the wheel ;

[0084] S422. The follower uses the price vector as the known input parameter of the lower-level robust optimization model and transforms it into a single-level, equivalent mixed-integer linear programming problem using KKT conditions and strong duality theory.

[0085] S423. By solving the MILP problem, the follower obtains the optimal 24-hour cooperative scheduling scheme. and dual variables;

[0086] S424, Followers will contain the optimal scheduling scheme. and key dual variables The complete information package is submitted to the system operator.

[0087] Optionally, in S43, the three-dimensional hybrid driving strategy integrates three economic mechanisms: market supply and demand, profit orientation, and marginal cost alignment, expressed as:

[0088]

[0089] In the formula: and It was extracted from the tenders submitted by the followers. t Total demand and total supply information for the real-time system. for The learning rate; Is it the auctioneer's profit? t Price gradient at any time, for The learning rate; These are the dual variables of the power balance constraint obtained from the bidders. This indicates the price set by the system operator. for The learning rate.

[0090] Optionally, the preset convergence conditions in S44 include:

[0091] a) Continuous N In this iteration, the system operator's profit increase Less than a preset minimum threshold ;

[0092] b) Price vector update magnitude Less than the threshold;

[0093] c) Number of iterations k Reaching the preset maximum number of iterations If the termination condition is not met, then let k = k +1, return to S42, and begin a new round of auction.

[0094] A robust game optimization scheduling system for multi-micronets based on a dynamic auctioneer algorithm, which performs any of the methods described above, includes:

[0095] The system modeling module is used to construct mathematical models of multi-microgrid integrated energy systems.

[0096] The two-layer model construction module is used to construct a two-layer robust optimization model consisting of an upper-layer leader model and a lower-layer follower alliance model. The upper-layer leader model aims to maximize the total profit of the system operator, and the decision variable is the price signal vector. The lower-layer follower alliance model is a robust optimization model that aims to maximize the alliance's profit under the worst-case uncertainty scenario, and the decision variable is the operating strategy of all controllable resources within the alliance.

[0097] The constraint configuration module is used to set constraints covering power balance, thermal power balance, cold power balance, power transmission between microgrids, power transmission between distribution network and microgrid, equipment output, SES operation, output of external wind farms, and reserve capacity.

[0098] The equilibrium solution module is used to iteratively solve the two-layer robust optimization model using the dynamic auctioneer algorithm. It outputs the optimization scheduling strategy corresponding to the Stackelberg equilibrium solution by initializing the price, generating and submitting the tender, driving price updates in three dimensions and performing convergence checks.

[0099] As can be seen from the above technical solution, compared with the prior art, the present invention discloses a robust game optimization scheduling method and system for multi-micronets based on the dynamic auctioneer algorithm, which has the following beneficial effects:

[0100] (1) Achieve outstanding economic benefits and maximize system-level profits

[0101] This invention achieves system-wide economic benefits far exceeding those of traditional scheduling strategies. It efficiently solves the Stackelberg equilibrium using the Dynamic Auctioneer Algorithm (DAA). Through an intelligent price discovery process, the DAA enables the system operator, as the leader, to formulate optimal price signals, thereby precisely guiding the microgrid consortium of followers to adopt a collaborative scheduling strategy that not only benefits themselves but also maximizes the overall system profit. This method, through a market mechanism, successfully guides the dispersed interests of multiple independent entities towards global optimality.

[0102] (2) Provide deterministic operational robustness and power supply reliability

[0103] This invention generates feasible and safe scheduling schemes even under the most unfavorable fluctuations in renewable energy output and load demand, fundamentally improving the system's energy supply reliability—a feat unmatched by deterministic or stochastic optimization methods. This effect directly stems from the robust optimization paradigm with a max-min objective built into the lower-level follower model. This model does not rely on potentially biased probability predictions but directly optimizes for the worst-case scenario within a pre-defined set of uncertainties, thus intrinsically incorporating the necessary safety margin into the scheduling scheme.

[0104] (3) It possesses high computational performance and excellent solution quality.

[0105] Traditional algorithms like PSO often get stuck in local optima when dealing with high-dimensional problems, while DAA can converge to a higher-quality global equilibrium solution much faster. The core of this computational performance advantage stems from DAA's unique three-dimensional hybrid-driven price update mechanism, particularly the "marginal value alignment" component. In traditional methods, the leader's price adjustment process resembles a "blind men and an elephant" search; however, in this invention, by requiring followers to submit their internal energy balance constraints' dual variables (shadow prices) along with their "bids," the leader can glimpse the followers' true marginal costs and marginal valuations of energy. This provides the leader with economic insight. Price updates are no longer blind searches, but adjustments guided by economic logic that quickly bridge the gap between external pricing and internal value. This efficient feedback loop significantly reduces ineffective search space and avoids price oscillations in non-economic ranges, thus achieving a dual improvement in algorithm convergence speed and solution quality.

[0106] (4) Enhance the interpretability and practical application value of the model.

[0107] Unlike many optimization algorithms that operate in a "black box" manner, this invention cleverly reconstructs a complex bi-level mathematical programming problem into a dynamic economic auction process, enhancing the transparency and intuitive interpretability of the entire optimization process. System operators, dispatchers, and other actual users do not need to be optimization algorithm experts to intuitively understand the operational logic of the "auctioneer" setting prices, "bidders" responding to bids, and the final market clearing. This economic framework not only enhances users' trust in the optimization results but also facilitates model debugging, verification, and deployment, thereby significantly lowering the barrier to implementing this advanced technology in real-world energy management systems and possessing practical promotional value. Attached Figure Description

[0108] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0109] Figure 1 A flowchart of the robust game optimization scheduling method for multi-micronets based on dynamic auctioneer algorithm provided by the present invention;

[0110] Figure 2 A schematic diagram of the DM-IES operating structure provided by this invention;

[0111] Figure 3 The flowchart for solving DAA provided by this invention. Detailed Implementation

[0112] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0113] To address the challenge of coordinating complex energy systems involving multiple independent stakeholders under conditions of uncertainty in renewable energy generation and load demand, this invention discloses a robust game-theoretic optimization scheduling method for multi-microgrids based on a dynamic auctioneer algorithm, such as... Figure 1 As shown, it includes the following steps:

[0114] S1. Construct a mathematical model for a multi-microgrid integrated energy system;

[0115] S2. Construct a two-layer robust optimization model consisting of an upper-level leader model and a lower-level follower alliance model. The upper-level leader model aims to maximize the total profit of the system operator, and the decision variable is the price signal vector. The lower-level follower alliance model is a robust optimization model that aims to maximize the alliance's profit under the worst-case uncertainty scenario, and the decision variable is the operating strategy of all controllable resources within the alliance.

[0116] S3. Establish constraints covering power balance, thermal power balance, cold power balance, power transmission between microgrids, power transmission between distribution network and microgrid, equipment output, SES operation, output of external wind farms, and reserve capacity.

[0117] S4. The dynamic auctioneer algorithm is used to iteratively solve the two-layer robust optimization model. By initializing the price, generating and submitting the tender, driving the price update in three dimensions and checking the convergence, the optimized scheduling strategy corresponding to the Stackelberg equilibrium solution is output.

[0118] Next, for Figure 1 The process shown is described in detail to further understand the technical solution to be protected by this invention.

[0119] I. Physical Modeling of Multi-Microgrid Integrated Energy System

[0120] 1.1 Establishing a system architecture model

[0121] The specific technical environment used in this embodiment is a DM-IES, whose detailed physical and mathematical models are described as follows: The multi-microgrid integrated energy system includes a system operator as the leader and a cooperative alliance of followers consisting of multiple heterogeneous microgrids, a shared energy storage power station and a centralized renewable energy power station.

[0122] System Operator (SO): As the central coordinating entity of the system and the leader in the game model, it is responsible for formulating system-level pricing strategies;

[0123] Microgrid1 (MG1): An office-type microgrid equipped with a combined cooling and power (CCP) system. Internal equipment includes a gas turbine, an electric chiller, a waste heat recovery system, distributed photovoltaics, and a lithium bromide absorption chiller.

[0124] Microgrid2 (MG2): An industrial microgrid equipped with a combined heat and power (CHP) system, whose core equipment includes a CHP unit, a gas-fired boiler, and distributed photovoltaics;

[0125] Microgrid3 (MG3): A commercial microgrid equipped with a combined cooling, heating and power (CCHP) system, integrating a cogeneration unit, a gas boiler, an electric refrigeration and air conditioning system, and a photovoltaic system;

[0126] Shared Energy Storage (SES): As an independent market entity, it provides charging and discharging services to various microgrids and generates revenue through peak shaving and valley filling and market arbitrage;

[0127] Centralized wind farms: acting as shared renewable energy suppliers, selling electricity to various entities within the system;

[0128] External power distribution network: serves as the source of power exchange and backup support for the system.

[0129] See Figure 2 The diagram showing the DM-IES operating structure defines the energy interaction paths between the various entities within the system as follows:

[0130] 1) Each microgrid can engage in bidirectional electricity trading with the distribution network and bidirectional charging and discharging interactions with shared energy storage power stations. 2) Microgrids can achieve bidirectional electricity sharing and trading through point-to-point connections. 3) Wind farms, as centralized renewable energy suppliers, can sell electricity unidirectionally to microgrids, shared energy storage power stations, and the distribution network. 4) In addition to serving microgrids, shared energy storage power stations can purchase electricity from the distribution network during off-peak electricity prices or utilize excess wind power for charging, achieving peak shaving and valley filling, and arbitrage through buying low and selling high. Since energy storage participates in microgrid power supply, reverse power transmission from energy storage power stations to the grid is not considered due to grid connection policies.

[0131] 1.2 Establishing a User Model

[0132] The user's IDR behavior is taken into account, and demand response is modeled separately according to three different load types: electrical load, thermal load, and cooling load.

[0133] 1.2.1 Price-based demand response modeling

[0134] Modeling the response of user electricity consumption behavior to electricity prices mainly involves four methods, among which the electricity price elasticity matrix method is the most widely used. For microgrid 1 and microgrid 3, considering that urban residential users are relatively more sensitive to electricity price fluctuations, price-based load demand response is mainly included in these two microgrids.

[0135] Electricity load elasticity coefficient The expression is:

[0136]

[0137] Electricity price elasticity matrix The expression is:

[0138]

[0139] Electricity load after demand response Represented as:

[0140]

[0141] In the formula: , These represent electricity consumption. and electricity price The relative increment; This indicates the electricity consumption before optimization. This indicates the change in electricity consumption after the controllable load power consumption status has been adjusted. Represents the elastic coefficient. and These represent the electricity price and its change, respectively.

[0142] 1.2.2 Modeling of Incentive-Driven Electric Load Demand Response

[0143] In Microgrid 2, since industrial users are more sensitive to direct economic incentives, incentive-based load demand response is adopted.

[0144]

[0145]

[0146]

[0147]

[0148] In the formula: , , These represent the base load, the load that can be reduced, and the load that can be transferred, respectively. This indicates the total amount of load that can be transferred within the scheduling period; and These represent the maximum power values ​​for transferable and load-reducible loads, respectively. and Corresponding transfer compensation unit price and reduction compensation unit price; This indicates the compensation generated by an incentive-based demand response; This represents the load power under incentive-driven demand response. T This represents the total number of scheduling periods (usually 24, corresponding to a full day of 24 hours, with a time step of 1 hour).

[0149] 1.2.3 Heat Load Demand Response Modeling

[0150] Heat loads 2 and 3 are supplied by gas turbines and gas boilers. Real-time transfer and reduction of heat loads are possible on the load side; therefore, an incentive-based heat load demand response mechanism is adopted. The transferable heat loads are as follows:

[0151]

[0152]

[0153]

[0154] In the formula: Indicates the heat load that can be transferred; and These represent the start and end times of the heat load transfer, respectively. Indicates the duration of continuous operation of the load; and These respectively represent the operating status and start / stop status of the heat load; express t Transferable heat load at any given time.

[0155]

[0156]

[0157]

[0158] In the formula: and These represent the base heat load and the heat load after demand response adjustment, respectively. and These represent the prices for heat load transfer and reduction, respectively. This indicates the benefits derived from heat load demand response; This indicates the starting point at which the heat load can be reduced. This indicates the end point at which the heat load can be reduced. express t The heat load can be reduced at any time. This indicates that the total heat load can be reduced within the scheduling cycle. This represents the minimum power required to reduce the heat load. This indicates the maximum power that can reduce the heat load.

[0159] 1.2.4 Modeling of Cooling Load Demand Response

[0160] Human perception of hot and cold comfort is influenced by numerous factors, and therefore can be simplified. In this embodiment, the cooling load primarily originates from air conditioning. Considering the characteristics of air conditioning usage, the cooling load can be equivalent to a cooling demand response with adjustable capabilities, expressed as follows:

[0161]

[0162]

[0163] In the formula: The voting value represents the feeling of hot or cold. This indicates the cooling load power after demand response. C represents the heat loss rate, and C represents the heat capacity. , They represent t Time period t Indoor temperature during the -1 hour period, express t Outdoor temperature during the time period; S Indicates the cooling area. Indicates the scheduling time interval.

[0164] 1.3 Uncertainty Modeling of Wind, Solar and Load

[0165] The output of wind and solar power, as well as the electricity, heat and cooling loads of users, have strong randomness and volatility. In order to describe the impact of these uncertainties on the system operation, this embodiment adopts a modeling method based on probability distribution model and scenario generation.

[0166] Numerous studies show that wind speed approximately follows a Weibull distribution, with the following probability density function:

[0167]

[0168] In the formula: v For wind speed, , These are the parameters of the Weibull distribution;

[0169]

[0170] In the formula: It signifies that one has contributed merit and effort. Indicates the cut-in wind speed. Indicates the cut-out wind speed. Indicates the rated wind speed of the unit. This indicates the rated output power.

[0171] The output of photovoltaic power generation is related to the intensity of solar radiation, which follows a Beta distribution with the following probability density function:

[0172]

[0173] In the formula: r Indicates light intensity. This indicates the maximum light intensity. , is the coefficient of the Beta distribution.

[0174] The prediction errors for electricity, heat, and cooling loads generally follow a normal distribution:

[0175]

[0176] In the formula: L This represents the actual value of the load. This represents the predicted load value. This represents the standard deviation of the prediction error.

[0177] Taking into account the uncertainties of wind, solar, and load, Latin hypercube sampling technology is used to generate multiple scenarios, and clustering is performed through clustering algorithms to finally obtain typical scenarios of wind power, photovoltaic, and electric heating and cooling loads.

[0178] II. A Two-Layer Robust Optimization Model Based on Master-Slave Game Theory

[0179] The optimization problem solved in this embodiment is constructed as a two-level mathematical programming model, whose structure clearly reflects the hierarchical decision-making relationship between leaders and followers. This step elaborates on the economic implications and mathematical expression of this model.

[0180] 2.1 Constructing an Upper-Level Leader Model

[0181] The goal of the upper-level leader model is to maximize the total profit of the system operator (leader), which consists of energy transactions with lower-level alliances, interactions with the external power grid, and demand response compensation payments to lower-level alliances.

[0182] Objective function: Maximize the interactive benefits for the system operator. :

[0183]

[0184] In the formula: This indicates that the leaders have formulated a micro-network. i Internal transaction price for electricity sales, microgrid i Electricity sold to the distribution network This indicates the cost price of electricity purchased from an external power grid. microgrid i Purchased power from the distribution network; and These represent the total compensation expenditure paid by the leader to incentivize followers to participate in electricity and heat demand response, respectively, with the unit price being a decision variable for the leader. N This indicates the number of microgrids.

[0185] Decision variables: The leader's decision variables are a set of price signal vectors. By issuing these price signal vectors, the leader guides the behavior of the lower-level alliance to maximize its own profits. These include: the time-of-use electricity price for each microgrid. Compensation prices for load shifting and reduction , Compensation price for heat load transfer and reduction , .

[0186] 2.2 Constructing a Lower-Level Follower Alliance Model

[0187] The goal of the lower-level model is to maximize the overall operational benefits of the consortium (including microgrids, shared energy storage, and wind farms) through coordinated scheduling of internal resources after receiving price signals from the upper level. This is a robust optimization problem, aiming to find the optimal solution under the worst-case uncertainty scenarios to ensure the economic efficiency and security of operation.

[0188] Objective function: Maximize the total revenue of the lower-level alliances. :

[0189]

[0190] In the formula: z Represents a vector of uncertain scenarios. Z Represents a set of uncertainties. This represents the total revenue of the alliance. This indicates the total cost of the alliance.

[0191] The objective function is a max-min structure. The outer max indicates that the alliance seeks the optimal scheduling strategy, while the inner min indicates that the strategy needs to maximize revenue under all possible worst-case scenarios z (such as the lowest photovoltaic output or the highest load) within the uncertainty set Z.

[0192] Specifically, it can be elaborated as follows:

[0193]

[0194] In the formula: , They represent microgrids respectively. i exist t The compensation benefits obtained from electricity and heat demand response at all times; Indicates power generation equipment g Unit power generation subsidy, Indicates power generation equipment g exist t Constant effort This indicates that the leaders have formulated a micro-network. i Internal transaction price for electricity sales, microgrid i From the power purchase capacity of the distribution network, Indicates the price of natural gas. microgrid i Gas purchase volume, Indicates power generation equipment g The unit operation and maintenance cost This represents the cost per unit of standby capacity. This indicates that the system is providing an increased reserve capacity; PV Indicates photovoltaic power generation, CHP Indicates combined heat and power (CHP). WT It refers to wind power generation.

[0195] Decision variables: The follower's decision variables are the operating strategies of all controllable resources within the alliance, which are specific responses to the leader's price signals, including: the output of each power generation unit. Shared energy storage charging and discharging plan Scheduling volume for various demand responses The system's required backup capacity .

[0196] III. Establishing Constraints Based on Actual Circumstances

[0197] 3.1 Power Balance Constraints

[0198]

[0199] In the formula: For micro-network jA collection of local power generation equipment; For generator sets g exist t Efforts made at all times; and These represent the power purchased / sold to the distribution network, respectively. and These are respectively shared energy storage to microgrids j Discharge / charge power; To share wind farms to microgrids j Electricity sales capacity; and These refer to the power purchased / sold between microgrids; This refers to the power consumption of the air conditioner. This refers to the net electrical load after demand response.

[0200] 3.2 Thermal power balance constraint

[0201]

[0202] In the formula: and They represent microgrids respectively. j Thermal power of medium-sized generating units and gas-fired boilers; This indicates the heat load after the demand response.

[0203] 3.3 Cooling power balance constraint

[0204]

[0205] In the formula: , and These represent the cooling output of the CCHP unit in microgrid 3, the output of the lithium bromide absorption chiller in microgrid 1, and the cooling power of the electric air conditioning in each microgrid, respectively. This indicates the cooling load demand.

[0206] 3.4 Power Transmission Constraints Between Microgrids

[0207]

[0208] In the formula: , , , , and They represent microgrids respectively. j To Micro Network i Purchase and sale power, maximum purchase and sale power, and purchase and sale power variables.

[0209] 3.5 Power Constraints for Distribution Network and Microgrid Transmission

[0210]

[0211] In the formula: , , and They represent microgrids respectively. j Maximum power purchased and sold from the distribution network and power purchase and sale variables.

[0212] 3.6 Equipment Output Constraints

[0213]

[0214]

[0215] In the formula: Indicates the output of the equipment; and These represent the minimum and maximum output values ​​of the equipment, respectively. Represents the running status variable; COP c For various coefficients; Indicates the calorific value of natural gas. G Indicates gas consumption; This indicates the output power of the gas turbine in microgrid 1. This represents the thermal power of the waste heat boiler in microgrid 1.

[0216] 3.7 SES Operational Constraints

[0217]

[0218] In the formula: , and These represent the stored charge state, the upper limit of the charged state, and the upper limit of the charged state, respectively; and These represent the charging and discharging power of the energy storage power station, respectively. and These respectively represent the energy storage power station to the microgrid j Charge and discharge variables; and These represent the charge and discharge efficiencies, respectively. and These represent the maximum charging and discharging power, respectively. This indicates the state of charge of the energy storage power station at the end of the scheduling cycle. This indicates the state of charge of the energy storage power station at the beginning of the scheduling cycle.

[0219] 3.8 External Wind Farm Output Constraints

[0220]

[0221] In the formula: Indicates wind farm to microgrid j Electricity sales capacity, This indicates the charging power of the wind farm to the shared energy storage. This indicates the power output of the wind farm sold to the main power distribution network. express t The total available generating capacity of the wind farm at any given time.

[0222] 3.9 Reserve Capacity Constraints

[0223]

[0224]

[0225] In the formula: Indicates the minimum required reserve capacity. This indicates the load reduction provided by each microgrid. This indicates the percentage of backup coverage required. This indicates the proportion of the load that can be reduced to reserve. microgrid j exist t The maximum load that can be reduced at any given time; This represents the maximum fluctuation in the predicted electricity load. This represents the maximum fluctuation in distributed photovoltaic power output. This indicates the maximum fluctuation in the output of centralized wind power.

[0226] IV. Solving the Two-Level Model Based on the Dynamic Auctioneer Algorithm (DAA)

[0227] This step is the core innovation of this invention, proposing a Dynamic Auctioneer Algorithm (DAA) to iteratively solve the aforementioned two-layer master-slave game model. This algorithm is the core innovation of this invention; it cleverly transforms the leader's price optimization process into a dynamic market auction process, simulating the multi-round dynamic game between the "auctioneer" and "bidders" to achieve market equilibrium. This method is not only computationally efficient, but more importantly, it provides an intuitive economic explanation for the complex game-solving process, enhancing the interpretability and practicality of the model.

[0228] The algorithm mapping relationship is as follows:

[0229] To apply DAA, it is first necessary to establish a mapping relationship between the game model and the auction mechanism, assigning the mathematical entities in the optimization problem to the roles of market participants:

[0230] Auctioneer: This role is held by the upper-level system operator (leader). Their responsibilities include designing and publishing auction rules (i.e., energy prices and compensation standards), guiding the market through multiple rounds of bidding, and ultimately maximizing their own profits (i.e., the system's total profit).

[0231] Bidders are formed by a coalition of lower-level microgrid partners (followers). They are considered a rational, single entity whose role is to calculate and submit a "bid" that maximizes their own benefit (minimizes costs) based on the auctioneer's currently published price.

[0232] Auctioned Items: The right to consume or supply electricity at different times, and the capacity to provide demand response services. These are goods available for trading in the market.

[0233] Bid: A complete 24-hour optimal dispatch plan submitted by the bidder (follower). This plan represents their optimal response to current price signals, including the output plans of all equipment, energy interaction plans, and demand response dispatch quantities. This "bid" implicitly reveals the bidder's valuation and willingness to pay for various energy services at current prices, and is core information reflecting market feedback.

[0234] Auction Process: A multi-round iterative process between the auctioneer and bidders. In each round, the auctioneer announces the price, and bidders submit their bids. This process repeats until the market reaches equilibrium, where the auctioneer can no longer increase their profit by adjusting the price. The price and scheduling at this point are the Stackelberg equilibrium solution.

[0235] The execution process of DAA simulates a complete market clearing process, which can be described in detail through the following steps:

[0236] 4.1 Initialization

[0237] Auctioneer (leader) initializes the iteration counter k =1, and set a set of initial price vectors. This includes electricity prices for different time periods and demand response compensation prices. The initial price can be based on historical data, external market reference prices, or a preset benchmark value.

[0238] 4.2 Generate and submit tender documents

[0239] (1) Price broadcast: The auctioneer broadcasts the current price to the bidders (followers). k Price vector of the wheel .

[0240] (2) Solving the robust optimization problem: The bidder uses this price vector as the known input parameter of its lower-level robust optimization model (step 2.2). Since the model is a max-min problem, in order to solve this problem, the KKT (Karush-Kuhn-Tucker) conditions and strong duality theory are used to transform it into a single-level, equivalent mixed integer linear programming problem.

[0241] (3) Obtaining key information: By solving this MILP problem, bidders not only obtain their optimal 24-hour collaborative scheduling scheme, but also... (i.e., the "tender documents" for this round), and at the same time, a set of key economic information can be obtained—dual variables (i.e., shadow prices). Among them, the dual variables related to the power balance constraint... Its economic significance is: under the current scheduling scheme, at t The marginal benefit of increasing the supply of electricity by one unit at a time accurately reflects the internal marginal value of electricity to bidders.

[0242] (4) Tender submission: Bidders will include the optimal scheduling scheme. and key dual variables The complete information package was submitted to the auctioneer.

[0243] 4.3 Evaluate the tender documents and update the price

[0244] After receiving the bids, the auctioneer executes a three-dimensional hybrid-driven price update strategy to generate the price for the next round. This strategy is the core of the algorithm in this invention. It goes beyond simple gradient adjustment and integrates three economic mechanisms: market supply and demand, profit orientation, and marginal cost alignment, ensuring that the algorithm can converge to the Stackelberg equilibrium point quickly and stably.

[0245]

[0246] Component 1: Adjusting the supply and demand imbalance

[0247] and It was extracted from the tenders submitted by the followers. t This provides information on aggregate demand and aggregate supply in the real-time system. This is a classic market clearing signal: when demand exceeds supply, prices rise to suppress demand and incentivize supply; conversely, when demand exceeds supply, prices fall. This is the learning rate for that component.

[0248] Component Two: Profit Gradient Optimization

[0249] Is it the auctioneer's profit? tThe price gradient at any given moment. This is a profit-maximizing driver signal, guiding prices to adjust in the direction that maximizes the auctioneer's own profits the fastest. This is the learning rate for that component.

[0250] Component Three: Marginal Value Alignment

[0251] This is the dual variable (shadow price) of the energy balance constraint obtained from bidders. This term is a key innovation of this algorithm, representing feedback from market depth information. Its economic logic is: the auctioneer compares the prices they set with their own... and the bidder's internal true marginal value To adjust the price. If This indicates that the auctioneer's price was too low, meaning bidders would still have a profit margin even at that price, thus giving the auctioneer room to raise the price. Conversely, if... If the price is too high, it inhibits transactions, and the auctioneer should lower the price. This mechanism greatly accelerates the convergence process by driving the alignment of external prices with internal values ​​and avoids blindly searching for prices in non-economic intervals. This is the learning rate for that component.

[0252] 4.4 Convergence Check

[0253] The algorithm checks if the termination condition is met. The termination condition can be set as follows:

[0254] a) Continuous In this iteration, the auctioneer's profit increase Less than a preset minimum threshold ;

[0255] b) Price vector update magnitude Less than the threshold;

[0256] c) Number of iterations k Reaching the preset maximum number of iterations If the termination condition is not met, then let Return to step 4.2 to begin a new round of auction.

[0257] 4.5 Output Results

[0258] The algorithm converges and terminates when the termination condition is met. At this point, the price vector of the last round... and the corresponding scheduling scheme This is considered as the Stackelberg equilibrium solution of the entire two-level game problem and is output as the final optimal scheduling strategy.

[0259] To more intuitively illustrate the above solution process, Figure 3A flowchart of the solution based on the dynamic auctioneer algorithm is presented. The flowchart clearly illustrates the information interaction (price and tender documents) between the upper layer (auctioneer) and the lower layer (bidders), as well as the algorithm's iterative loop and convergence judgment logic.

[0260] and Figure 1 Corresponding to the method described above, this embodiment of the invention also provides a robust multi-micronet game optimization scheduling system based on a dynamic auctioneer algorithm, used for... Figure 1 The specific implementation of the method, a robust game optimization scheduling system for multi-micronets based on a dynamic auctioneer algorithm provided in this embodiment of the invention, can be applied to computer terminals or various mobile devices, specifically including:

[0261] The system modeling module is used to construct mathematical models of multi-microgrid integrated energy systems.

[0262] The two-layer model construction module is used to construct a two-layer robust optimization model consisting of an upper-layer leader model and a lower-layer follower alliance model. The upper-layer leader model aims to maximize the total profit of the system operator, and the decision variable is the price signal vector. The lower-layer follower alliance model is a robust optimization model that aims to maximize the alliance's profit under the worst-case uncertainty scenario, and the decision variable is the operating strategy of all controllable resources within the alliance.

[0263] The constraint configuration module is used to set constraints covering power balance, thermal power balance, cold power balance, power transmission between microgrids, power transmission between distribution network and microgrid, equipment output, SES operation, output of external wind farms, and reserve capacity.

[0264] The equilibrium solution module is used to iteratively solve the two-layer robust optimization model using the dynamic auctioneer algorithm. It outputs the optimization scheduling strategy corresponding to the Stackelberg equilibrium solution by initializing the price, generating and submitting the tender, driving price updates in three dimensions and performing convergence checks.

[0265] In summary, the core of this invention is a two-tiered optimization framework based on Stackelberg master-slave game theory. In this framework, the System Operator (SO) acts as the Leader, aiming to maximize the total profit of the entire system by formulating time-of-use energy prices and demand response incentive strategies. A cooperative alliance composed of multiple heterogeneous microgrids acts as Followers, responding to price signals issued by the Leader and collaboratively optimizing their collective operating strategies to minimize their own operating costs under the worst-case uncertainty scenarios.

[0266] The key innovation of this invention lies in proposing a novel Dynamic Auctioneer Algorithm (DAA) for solving the aforementioned two-level game model. This algorithm replaces the traditional heuristic solver, cleverly reconstructing the complex optimization process into a dynamic auction process. This reconstruction is not merely a computational substitution; it imbues the process of solving the master-slave game equilibrium with economic implications. Within this framework, the price signal issued by the leader is considered the "auction price" set by the auctioneer, while the optimal scheduling scheme calculated by the follower alliance is equivalent to its "bid," reflecting its true valuation of energy services under the current price system. This analogy provides a solid theoretical and practical foundation for designing efficient price update mechanisms and ensuring the algorithm converges to the Stackelberg equilibrium, enabling the entire iterative optimization process to simulate a realistic market clearing mechanism.

[0267] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0268] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A robust game optimization scheduling method for multi-micronets based on a dynamic auctioneer algorithm, characterized in that, Includes the following steps: S1. Construct a mathematical model for a multi-microgrid integrated energy system; S2. Construct a two-layer robust optimization model consisting of an upper-level leader model and a lower-level follower alliance model. The upper-level leader model aims to maximize the total profit of the system operator, and the decision variable is the price signal vector. The lower-level follower alliance model is a robust optimization model that aims to maximize the alliance's profit under the worst-case uncertainty scenario, and the decision variable is the operating strategy of all controllable resources within the alliance. S3. Establish constraints covering power balance, thermal power balance, cold power balance, power transmission between microgrids, power transmission between distribution network and microgrid, equipment output, SES operation, output of external wind farms, and reserve capacity. S4. The dynamic auctioneer algorithm is used to iteratively solve the two-layer robust optimization model. By initializing the price, generating and submitting the tender, driving the price update in three dimensions and checking the convergence, the optimization scheduling strategy corresponding to the Stackelberg equilibrium solution is output. S1 specifically includes the following steps: S11. Based on the system operator as the leader and the cooperative alliance of multiple heterogeneous microgrids, a shared energy storage power station and a centralized renewable energy power station as followers, establish an architecture model for a multi-microgrid integrated energy system. S12. Take into account the user's IDR behavior and conduct modeling for demand response according to the three different load types: electrical load, thermal load, and cooling load. S13. Considering the uncertainties of wind, solar and load, a model based on probability distribution and scenario generation is adopted for modeling, and finally typical scenarios of wind power, photovoltaic and electric heating cooling load are obtained. The specific steps of S12 are as follows: S121. Incorporate price-based electricity demand response into microgrids 1 and 3, and adjust the electricity load elasticity coefficient. The expression is: Electricity price elasticity matrix The expression is: Electricity load after demand response Represented as: In the formula: , These represent electricity consumption. and electricity price The relative increment; This indicates the electricity consumption before optimization. This indicates the change in electricity consumption after the controllable load power consumption status has been adjusted. Represents the elastic coefficient. and These represent the electricity price and its change, respectively. S122. In microgrid 2, the incentive-based load demand response is adopted, and the calculation method is as follows: In the formula: , , These represent the base load, the load that can be reduced, and the load that can be transferred, respectively. This indicates the total amount of load that can be transferred within the scheduling period; and These represent the maximum power values ​​for transferable and load-reducible loads, respectively. and Corresponding transfer compensation unit price and reduction compensation unit price; This indicates the compensation generated by an incentive-based demand response; This represents the load power under incentive-driven demand response. T Indicates the total number of scheduling periods; The heat sources for S123, heat load 2, and heat load 3 are supplied by gas turbines and gas boilers, and an excitation-based heat load demand response mechanism is adopted. The calculation method for transferable heat load is as follows: In the formula: Indicates the heat load that can be transferred; and These represent the start and end times of the heat load transfer, respectively. Indicates the duration of continuous operation of the load; and These respectively represent the operating status and start / stop status of the heat load; express t Transferable heat load at any given time; In the formula: and These represent the base heat load and the heat load after demand response adjustment, respectively. and These represent the prices for heat load transfer and reduction, respectively. This indicates the benefits derived from heat load demand response; This indicates the starting point at which the heat load can be reduced. This indicates the end point at which the heat load can be reduced. express t The heat load can be reduced at any time. This indicates that the total heat load can be reduced within the scheduling cycle. This represents the minimum power required to reduce the heat load. This indicates the maximum power that can reduce the heat load; S124. The cooling load is equivalent to a cooling demand response with adjustment capabilities, expressed as follows: In the formula: The voting value represents the feeling of hot or cold. This indicates the cooling load power after demand response. C represents the heat loss rate, and C represents the heat capacity. , They represent t Time period t Indoor temperature during the -1 hour period express t Outdoor temperature during the time period; S Indicates the cooling area. Indicates the scheduling time interval.

2. The robust game optimization scheduling method for multi-micronets based on the dynamic auctioneer algorithm according to claim 1, characterized in that, The specific steps of S13 are as follows: Wind speed follows a Weibull distribution, and its probability density function is: In the formula: v For wind speed, , These are the parameters of the Weibull distribution; In the formula: It signifies that one has contributed merit and effort. Indicates the cut-in wind speed. Indicates the cut-out wind speed. Indicates the rated wind speed of the unit. Indicates the rated output power; The output of photovoltaic power generation is related to the intensity of solar radiation, which follows a Beta distribution with the following probability density function: In the formula: r Indicates light intensity. This indicates the maximum light intensity. , The coefficient of the Beta distribution; The prediction errors for electricity, heat, and cooling loads follow a normal distribution: In the formula: L This represents the actual value of the load. This represents the predicted load value. The standard deviation of the prediction error; Taking into account the uncertainties of wind, solar, and load, Latin hypercube sampling technology is used to generate multiple scenarios, and clustering is performed through clustering algorithms to finally obtain typical scenarios of wind power, photovoltaic, and electric heating and cooling loads.

3. The robust game optimization scheduling method for multi-micronets based on the dynamic auctioneer algorithm according to claim 1, characterized in that, S2 specifically includes the following steps: S21. The objective function of the upper-level leader model is to maximize the interactive benefits of the system operators. : In the formula: This indicates that the leaders have formulated a micro-network. i Internal transaction price for electricity sales, microgrid i Electricity sold to the distribution network This indicates the cost price of electricity purchased from an external power grid. microgrid i Purchased power from the distribution network; and These represent the total compensation expenditure paid by the leader to incentivize followers to participate in electricity and heat demand response, respectively, with the unit price being a decision variable for the leader. N Indicates the number of microgrids; The decision variables in the upper-level leader model are a set of price signal vectors. By issuing price signal vectors, the upper-level leader guides the behavior of the lower-level alliance in order to maximize its own profits. S22. The objective function of the lower-level follower alliance model is to maximize the total revenue of the lower-level alliance. : In the formula: z Represents a vector of uncertain scenarios. Z Represents a set of uncertainties. This represents the total revenue of the alliance. Indicates the total cost of the alliance; Specifically, it can be elaborated as follows: In the formula: , They represent microgrids respectively. i exist t The compensation benefits obtained from electricity and heat demand response at all times; Indicates power generation equipment g Unit power generation subsidy, Indicates power generation equipment g exist t Constant effort This indicates that the leaders have formulated a micro-network. i Internal transaction price for electricity sales, microgrid i From the power purchase capacity of the distribution network, Indicates the price of natural gas. microgrid i Gas purchase volume, Indicates power generation equipment g The unit operation and maintenance cost This represents the cost per unit of standby capacity. This indicates that the system is providing an increased reserve capacity; PV Indicates photovoltaic power generation, CHP Indicates combined heat and power (CHP). WT Indicates wind power generation; The decision variables in the lower-level follower alliance model are the operating strategies of all controllable resources within the alliance, which are specific responses to the leader's price signals.

4. The robust game optimization scheduling method for multi-micronets based on the dynamic auctioneer algorithm according to claim 1, characterized in that, S4 specifically includes the following steps: S41, System operator sets initial price vector and iteration parameters; S42. The system operator broadcasts the initial price vector, solves for the optimal scheduling scheme and the dual variables corresponding to the power balance constraints, and submits the solution. S43. The system operator uses a three-dimensional hybrid driving strategy to update prices; S44. If the preset convergence condition is met, output the current price vector and the corresponding scheduling scheme as the Stackelberg equilibrium solution; otherwise, return to S42 and repeat the iteration.

5. The robust game optimization scheduling method for multi-micronets based on the dynamic auctioneer algorithm according to claim 4, characterized in that, The specific steps of S42 are as follows: S421, The system operator broadcasts the current number to followers. k Price vector of the wheel ; S422. The follower uses the price vector as the known input parameter of the lower-level robust optimization model and transforms it into a single-level, equivalent mixed-integer linear programming problem using KKT conditions and strong duality theory. S423. By solving the MILP problem, the follower obtains the optimal 24-hour cooperative scheduling scheme. and dual variables; S424, Followers will contain the optimal scheduling scheme. and key dual variables The complete information package is submitted to the system operator.

6. The robust game optimization scheduling method for multi-micronets based on the dynamic auctioneer algorithm according to claim 4, characterized in that, In S43, the three-dimensional hybrid driving strategy integrates three economic mechanisms: market supply and demand, profit orientation, and marginal cost alignment, expressed as: In the formula: and It was extracted from the tenders submitted by the followers. t Total demand and total supply information for the real-time system. for The learning rate; Is it the auctioneer's profit? t Price gradient at any time, for The learning rate; These are the dual variables of the power balance constraint obtained from the bidders. This indicates the price set by the system operator. for The learning rate.

7. The robust game optimization scheduling method for multi-micronets based on the dynamic auctioneer algorithm according to claim 4, characterized in that, The preset convergence conditions in S44 include: a) Continuous N In this iteration, the system operator's profit increase Less than a preset minimum threshold ; b) Price vector update magnitude Less than the threshold; c) Number of iterations k Reaching the preset maximum number of iterations If the termination condition is not met, then let k = k +1, return to S42, and begin a new round of auction.

8. A robust game optimization scheduling system for multi-micronets based on a dynamic auctioneer algorithm, performing the method as described in any one of claims 1-7, characterized in that, include: The system modeling module is used to construct mathematical models of multi-microgrid integrated energy systems. The two-layer model construction module is used to construct a two-layer robust optimization model consisting of an upper-layer leader model and a lower-layer follower alliance model. The upper-layer leader model aims to maximize the total profit of the system operator, and the decision variable is the price signal vector. The lower-layer follower alliance model is a robust optimization model that aims to maximize the alliance's profit under the worst-case uncertainty scenario, and the decision variable is the operating strategy of all controllable resources within the alliance. The constraint configuration module is used to set constraints covering power balance, thermal power balance, cooling power balance, power transmission between microgrids, power transmission between distribution network and microgrid, equipment output, SES operation, output of external wind farms, and reserve capacity. The equilibrium solution module is used to iteratively solve the two-layer robust optimization model using the dynamic auctioneer algorithm. It outputs the optimization scheduling strategy corresponding to the Stackelberg equilibrium solution by initializing the price, generating and submitting the tender, driving price updates in three dimensions and performing convergence checks.

Images