A low-carbon scheduling method for multi-agent master-slave game of source, load and storage

Abstract

The present invention discloses a low-carbon scheduling method for the master-slave game of multiple subjects of source, load and storage. The method adopts the baseline method and stimulates the enthusiasm of source-load-storage subjects for energy conservation and emission reduction based on the reward-penalty ladder-type carbon transaction cost calculation model. The influence of dual incentives of cost and carbon compensation is considered to construct a demand response strategy of the source-load-storage system and adjust the subject's own strategy. The method describes the multi-subject low-carbon interaction mechanism based on the Stackelberg game, constructs the decision-making model of each source-load-storage subject, and solves the double-layer optimization by using the particle swarm algorithm and the Gurobi solver, so that each source-load-storage subject can reasonably adjust its own strategy, taking into account the environmental benefits while maximizing its own interests.

Description

Technical Field

[0001] The present invention relates to the field of flexible load interaction technology, and in particular to a low-carbon scheduling method for source-load-storage multi-agent master-slave game. Background Art

[0002] The country has been actively promoting the reform of energy structure to reduce energy loss, and has implemented a number of key projects, such as "electricity instead of coal, electricity instead of oil, electricity instead of gas", to improve the status of electricity in energy. With the access of a large number of new energy sources, the traditional power grid architecture has undergone tremendous changes, and the impact of random and intermittent distributed power sources, such as solar energy, on the power system has continued to increase. This has changed the direction of the flow, and the traditional unidirectional flow has become a bidirectional flow. The power grid has changed from "passive" to "active", showing the characteristics of multi-source.

[0004] Therefore, we urgently need to establish a system that highly integrates information, communication and control technologies to achieve coordinated control among "source-grid-load-storage-vehicle".

[0005] Through effective monitoring and optimal control of distributed power sources and diversified loads, the peak load and frequency regulation pressure of the power system can be reduced, thereby improving the operating efficiency of the distribution network. Distributed power generation has unique characteristics in terms of time. For example, photovoltaic power generation mainly occurs during the day. If these distributed photovoltaic resources are not reasonably and fully utilized, the peak-to-valley difference of the power system load may be enlarged, the difficulty of system peak load regulation may be increased, and even serious problems of abandoned light may occur.

[0006] In addition, there is a certain degree of randomness in the production capacity of new energy, which leads to large prediction errors and increases the risk of scheduling management.

[0007] How to effectively and reasonably manage the demand-side resources of the active distribution network has become an important focus of current research. Demand response refers to users responding to the electricity cost signals or other incentive mechanisms provided by the market to adjust their electricity consumption time and electricity consumption to achieve load peak shaving and valley filling, ensure the stable operation of the power system, and reduce operational risks. Under the current energy Internet structure, demand response is a core function in the smart power system, gradually developing in the direction of intelligence and automation, reducing the lag problems and additional costs caused by manual operations in the past.

[0008] Therefore, how to establish an accurate demand response model while encouraging users to actively respond to demand is an urgent problem to achieve low-carbon scheduling. Summary of the invention

[0009] Based on the problems raised by the above background technology, the purpose of the present invention is to provide a low-carbon scheduling method for the master-slave game of multiple subjects of source, load and storage, which solves the problem that establishing an accurate demand response model is an urgent problem to achieve low-carbon scheduling under the premise of encouraging users to actively respond to demand.

[0010] The present invention is achieved through the following technical solutions:

[0011] The first aspect of the present invention provides a low-carbon scheduling method for source-load-storage multi-agent master-slave game, comprising the following steps:

[0012] Establishing a reward and punishment ladder calculation model, and using the reward and punishment ladder calculation model to calculate carbon trading costs;

[0013] The demand response strategy considering electricity cost and carbon compensation constructs an interaction model between source-load-storage entities, and a decision model under the carbon trading cost is constructed based on the interaction model;

[0014] Based on Stackelberg game, an optimization model of source-load-storage subject under the decision model is established;

[0015] The optimization model is iteratively solved using an improved particle swarm algorithm and a solver to obtain the optimal low-carbon scheduling result.

[0016] In the above technical solution, the carbon trading cost is first calculated through the reward and punishment ladder calculation model, and the carbon trading cost is used in the construction of the subsequent decision-making model. Its purpose is to use the baseline method to determine the free carbon emission quota of each source-load-storage entity, and use the reward and punishment ladder carbon trading model based on the free carbon emission quota to stimulate the enthusiasm of the source-load-storage entity to save energy and reduce emissions.

[0017] The purpose of the demand response strategy considering electricity costs and carbon compensation is to autonomously adjust the energy use strategy based on the energy supply cost and maximize the benefits, promote the consumption of new energy, and relieve equipment pressure. On this basis, an interaction model between source-load-storage entities is constructed to determine the interaction form and content between source-load-storage entities. Based on the interaction model, a decision-making model between various entities under carbon trading costs and their response constraints are constructed.

[0018] Based on the Stackelberg game, an optimization model of the source-load-storage subject under the decision-making model is established, and then the process of the source-load-storage subject pursuing the optimal goal is described. In this optimization model, the master-slave relationship of the source-load-storage subject is established, and the goal and participation data of the Stackelberg game are established based on this master-slave relationship.

[0019] Then, the improved particle swarm algorithm and solver are used to iteratively solve the optimization model, and then a low-carbon scheduling method is obtained, so that each subject can reasonably adjust its own strategy and take into account the cost and environmental benefits of the source-load-storage system.

[0020] In an optional embodiment, the carbon trading cost is calculated using the reward and punishment ladder calculation model, including:

[0021] The baseline method is used to determine the free carbon emission quota of each source-load-storage entity;

[0022] Acquire the actual carbon emissions of the energy management subject and the length of the carbon emissions interval, and compare the actual carbon emissions of the energy management subject with the free carbon emissions quota and the length of the carbon emissions interval;

[0023] The reward and punishment ladder calculation model is used to calculate the carbon trading cost of the ladder corresponding to the comparison result.

[0024] In an optional embodiment, the source-load-storage subject includes: an energy management subject, an energy supply subject, an energy storage subject and a user;

[0025] The demand response strategy considering electricity cost and carbon compensation builds an interaction model between source-load-storage entities, including:

[0026] Establish an interactive relationship between the energy management subject and the energy supply subject; establish an interactive relationship between the energy management subject and the energy storage subject; establish an interactive relationship between the energy management subject and the user.

[0027] In an optional embodiment, the decision model of the energy management subject is constructed as follows:

[0028] Determine the energy supply cost of the energy management subject to provide energy to the user and the energy storage subject, and the interaction cost between the energy management subject and the energy storage subject and the power grid;

[0029] Obtain penalty costs and carbon compensation functions for cooling interruptions;

[0030] The decision objective function of the energy management subject is constructed by comprehensively considering the energy supply cost, the interaction cost, the penalty cost and the carbon compensation function.

[0031] In an optional embodiment, the decision model of the energy supply entity is constructed as follows:

[0032] Determine the energy supply income of the energy supply entity and the carbon trading cost of the energy supply entity;

[0033] The decision objective function of the energy supply entity is constructed by comprehensively considering the energy supply income, the carbon transaction cost of the energy supply entity and the carbon compensation function.

[0034] In an optional embodiment, the decision model of the energy storage entity is constructed as follows:

[0035] Obtaining the energy release income and charging cost of the energy storage entity, as well as the operation and maintenance cost of the energy storage entity;

[0036] The decision objective function of the energy storage entity is constructed by comprehensively considering the energy release benefit, the energy charging cost and the operation and maintenance cost of the energy storage entity.

[0037] In an optional embodiment, the user's decision model is constructed as follows:

[0038] Determine the user satisfaction index and energy acquisition function based on the electricity and cooling energy consumed by the user;

[0039] The user's decision objective function is constructed by integrating the user satisfaction index, the energy acquisition function and the carbon compensation function.

[0040] In an optional embodiment, establishing the source-load-storage subject under the decision model based on the Stackelberg game includes:

[0041] The energy management subject is used as the leader of the Stackelberg game, and the energy supply subject, the energy storage subject and the user are used as followers of the Stackelberg game to establish a master-slave relationship among the participants;

[0042] The energy cost of the energy management subject, the output power of the functional equipment of the energy supply subject, the charging and discharging power of the energy storage subject and the adjustable load of the user are used as the strategy of the Stackelberg game;

[0043] The decision-making model of each source-load-storage entity is used as the effectiveness of the Stackelberg game;

[0044] Establish an optimization model based on the master-slave relationship, strategy and performance of the participants.

[0045] In an optional embodiment, the optimization model is iteratively solved using an improved particle swarm algorithm and a solver, including:

[0046] Dividing the optimization model into an upper optimization model and a lower optimization model;

[0047] The upper layer optimization model is solved by using an improved particle swarm algorithm, and the lower layer optimization model is solved by using a solver;

[0048] The upper layer optimization model and the lower layer optimization model are interactively iterated to obtain an optimization result.

[0049] In an optional embodiment, solving the upper optimization model using an improved particle swarm algorithm includes:

[0050] A. Initialize particle position, velocity and population size, and create an initial particle swarm;

[0051] B. Calculate the fitness function value of each particle in the initial particle swarm;

[0052] C. Determine the individual optimal position and the global optimal position according to the fitness function value;

[0053] D. Determine whether the end condition is met. If not, update the speed and position of the particles in the initial particle swarm according to the individual optimal position and the global optimal position, and repeat steps B to D.

[0054] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0055] Fully considering the source-load-storage system, limiting carbon emissions based on the reward-penalty ladder carbon trading mechanism, proposing a demand response strategy that considers the dual incentives of cost-load-carbon compensation, with the energy management subject as the leader, the energy supply subject, the energy storage subject and the user as followers, establishing a multi-subject low-carbon interaction mechanism based on the Stackelberg game, and constructing a decision-making model for each subject. This allows each subject to reasonably adjust its own strategy, taking into account the cost and environmental benefits of the source-load-storage system. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other relevant drawings can be obtained based on these drawings without creative work. In the drawings:

[0057] Figure 1 A schematic flow chart of a low-carbon scheduling method for source-load-storage multi-agent master-slave game provided in Example 1 of the present invention;

[0058] Figure 2 A schematic diagram of the structure of a multi-source flexible load group multi-agent game interaction framework provided in Example 1 of the present invention;

[0059] Figure 3 A schematic diagram of the flow of the interactive control strategy for the distribution network provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0060] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0061] Example 1

[0062] like Figure 1 As shown, Embodiment 1 of the present invention provides a low-carbon scheduling method for source-load-storage multi-agent master-slave game, which includes the following steps:

[0063] Step S1: Establish a reward and punishment ladder calculation model, and use the reward and punishment ladder calculation model to calculate the carbon trading cost.

[0064] The baseline method is used to determine the free carbon emission quota of each entity. It is assumed that the electricity obtained by the source-load-storage system (SSLS) from the external power grid is thermal power generation, and the initial allocation of carbon emissions is the electricity obtained from the outside by the energy management entity.

[0065] Among them, the electricity obtained from the outside by the energy management subject is:

[0066]

[0067] Where: D GPMO is the electricity obtained from the outside, δ p is the carbon emission allocation per unit of electricity, It is the amount of electricity obtained by the energy management entity from the external power grid.

[0068] In order to stimulate the enthusiasm of various entities to save energy and reduce emissions, a reward and punishment ladder is constructed as shown in the following formula:

[0069]

[0070] Where: F c is the carbon trading cost; D r The actual total carbon emissions of the energy management entity; D f is the free carbon emission quota; μ and λ are the reward coefficient and penalty coefficient respectively; c is the carbon transaction cost; h is the length of the carbon emission interval.

[0071] Among them, when the actual total carbon emissions of the energy management subject are D r Lower than free carbon emission quota D f If the carbon emission right is insufficient, a certain reward will be given; otherwise, insufficient carbon emission rights will be obtained.

[0072] Step S2: Considering the demand response strategy of electricity cost and carbon compensation, an interaction model between source-load-storage entities is constructed, and a decision model under the carbon trading cost is constructed based on the interaction model.

[0073] Among them, demand response strategies that consider electricity costs and carbon compensation include:

[0074] 1. When the load is in the valley period, the energy management subject guides users to increase energy consumption. At this time, the functional cost is in the valley period, and users adjust their energy consumption strategies according to their own satisfaction indicators. On the basis of cost incentives, the increase in electric load and cooling load in response to cost incentives is converted into carbon compensation rewards to users, stimulating users to adjust the optimal adjustable load range.

[0075] 2. When the load is in normal period, the user's load adjustment is small, there is only less carbon compensation or almost no carbon compensation reward, and the user's actual energy load is close to the demand load.

[0076] 3. When the load is at the peak stage, the energy management entity sets a higher functional cost in order to guide users to actively reduce energy consumption. The energy consumption strategy is adjusted according to the user satisfaction index, and the changes in electric load and cooling load reduced by the response cost incentive are converted into carbon compensation rewards to users, stimulating users to adjust the optimal adjustable load range.

[0077] Based on the above demand response strategy, an interaction model between source, load and storage is constructed, such as Figure 2 As shown in the figure, the energy management subject can be regarded as an agent with two-way energy flow, which is responsible for the source-load-storage power balance, while considering the initiative and decision-making ability of the three parties of source-load-storage, and formulating the energy provision cost and energy acquisition cost with the goal of maximizing net profit.

[0078] When the amount of electricity obtained by the energy management subject cannot meet the needs of users, it is necessary to obtain energy from the external power grid and bear the corresponding carbon emission costs. The energy supply subject optimizes the output of each device with the maximum energy supply benefit as the objective function. The energy storage subject optimizes its own charging and discharging power through low charging and high discharging between the energy management subject and the user based on the cost information. Users comprehensively consider the cost of purchasing energy, energy comfort and carbon compensation to adjust their energy demand to maximize the comprehensive benefits.

[0079] In this embodiment, the energy management subject formulates the cost of obtaining and providing energy based on the time-sharing electricity cost, considers the output characteristics of each power device and the energy consumption characteristics of the user, guides the output of each device and adjusts the energy consumption demand of the user, and its benefits come from the energy mutual assistance between other subjects. Specifically, its decision objective function is constructed as follows:

[0080]

[0081] Where: superscript t represents time period t, and They are the energy supply benefits provided to users and energy storage entities respectively; and are the interaction costs between the energy management entity, the energy storage entity, and the power grid; is the penalty cost of cooling interruption; β1 and They are the weight coefficient of the carbon compensation for users undertaken by the energy management entity and the carbon compensation function for users; is the carbon trading cost of electricity obtained from the outside; T is the total time period, which is 24 hours in this embodiment.

[0082] In an optional embodiment, the above costs include:

[0083]

[0084] Where: and They are the actual electric and cooling power on the energy consumption side, respectively; and They are respectively the electricity and cooling power obtained by the energy management subject from the energy supply subject; The power supplied by the energy management entity to the energy storage manager; and are the costs of electricity and cooling energy provided to the energy management subject, respectively; and They are respectively the cost of obtaining electric energy and cooling energy from the energy supply entity by the energy management entity; and are the costs of the energy management entity providing power to the grid and obtaining power; c is the cooling interruption penalty coefficient.

[0085] Furthermore, in order to ensure the cooling interruption penalty coefficient, the energy acquisition and energy supply costs meet the following constraints:

[0086]

[0087] Where: and are the minimum and maximum costs of cooling energy, respectively; and To provide average electricity and cooling costs.

[0088] Furthermore, the energy management entity must also meet the interactive power constraints with the external power grid connection line.

[0089]

[0090] Where: and Provide and obtain electric power to the external power grid for energy management entities; and It is a 0-1 variable, indicating the state of the energy management subject providing or obtaining electric energy from the external power grid; and To provide and obtain upper limit of electrical power.

[0091] In an optional embodiment, the energy supply entity optimizes the output of each controllable device by maximizing its own profit according to the energy cost set by the energy management entity, and its objective function is:

[0092]

[0093] Where: Energy supply income for energy suppliers; is the carbon trading cost of the energy supplier; β2 is the weight coefficient of the energy supplier's carbon compensation for users.

[0094] Furthermore, the power and cooling output of the energy supplier in period t is subject to the following opportunity constraints:

[0095]

[0096] Where: p r {} is the probability of the event happening, γ is a random variable that follows a normal distribution, and ξ is the confidence level.

[0097] Furthermore, the energy constraint is described by chance constraint, and the deterministic equivalence class method is used to process the chance constraint and convert it into a deterministic constraint:

[0098]

[0099] K ξ =φ -1 (1-ξ)

[0100] Where: π i,t is the Lagrange multiplier corresponding to the inequality constraint; φ -1 is the inverse function of the random variable distribution function φ. -1 In the case of multiple values, select the maximum value, that is:

[0101] K ξ = sup{K|K=φ -1 (1-ξ)}

[0102] In the formula, sup{} represents the minimum upper bound of the set.

[0103] In a possible embodiment, the energy storage entity realizes arbitrage between the energy management entity and the user through "low charging and high discharging", and its objective function is:

[0104]

[0105] Where: and They are the energy release benefit and charging cost of energy storage respectively; It is the operation and maintenance cost of the energy storage entity.

[0106] Furthermore, considering the energy loss and efficiency of charging and discharging, the energy storage state constraints of the energy storage device satisfy:

[0107]

[0108] Where: η bes and η ces are the charging and discharging efficiencies of the energy storage device, respectively; and are the minimum and maximum values ​​respectively; μ es is the energy self-loss rate of the energy storage device.

[0109] In a possible embodiment, user satisfaction indicators are considered Energy acquisition function and carbon compensation function The user load adjustment optimization objective function is:

[0110]

[0111] Furthermore, A commonly used quadratic function is used to represent it. and The expression is:

[0112]

[0113] Where: i∈{e,c}; v i and u i are the preference coefficients of users for consuming electricity and cooling energy, respectively.

[0114] Furthermore, the user's energy load is composed of fixed load and adjustable load The former has higher requirements on reliability, while the latter can be freely adjusted by users according to the energy supply cost. The degree of change is related to the user's energy satisfaction and meets the following constraints.

[0115]

[0116] Where: Indicates the period t The maximum value of It represents the total transferable load of electricity and cooling load in period t.

[0117] Step S3: establishing an optimization model of the source-load-storage subject under the decision model based on Stackelberg game.

[0118] Stackelberg game describes the decision-making process of energy management entities, energy supply entities, energy storage entities and users in pursuit of their respective optimal goals.

[0119] Energy supply entities, energy storage entities and users act as followers and optimize their own strategies according to the cost strategy of the leader energy management entity, and finally obtain the optimal solution of Stackelberg equilibrium. The Stackelberg game model can be expressed as:

[0120] Φ={PMO;{PGO,ESO,USER};p PMO ; {W PGO ,

[0121] D ESO ,L user}; E PMO ; {E PGO ,E ESO ,E user}}

[0122] The model includes actors, strategies, and utilities.

[0123] The participants include the leader energy management entity and the follower energy supply entity, energy storage entity, and users;

[0124] The strategy includes the energy management subject's acquisition of energy, provision of energy cost p PMO , the output power of the energy supply equipment in the energy supply subject is W PGO , the charging and discharging power D in the energy storage body ESO and the user's adjustable electricity and cooling load L user ;

[0125] The utility of each participant is the objective function of the aforementioned entities.

[0126] In the game process, when any interested party cannot unilaterally change the strategy of the equilibrium solution to obtain benefits, it means that the game reaches the Stackelberg equilibrium, that is, it satisfies the following conditions, then is the equilibrium of the Stackelberg game.

[0127]

[0128] Step S4: using an improved particle swarm algorithm and a solver to iteratively solve the optimization model to obtain an optimal low-carbon scheduling result.

[0129] The optimization model is divided into two layers, the upper layer optimization model is solved by using the improved particle swarm algorithm, and the lower layer optimization model is solved by using the solver. The two solving processes are then iterated to obtain the optimization result.

[0130] like Figure 3 As shown in the figure, the improved particle swarm algorithm is used to solve the upper optimization model, including:

[0131] (1) Initialize the particle position, velocity, population size and other parameters in the algorithm;

[0132] (2) Determine the individual optimal position and the global optimal particle, and update the excellent particles during the optimization process.

[0133] First, the parameters are initialized, including the setting of the maximum number of iterations and the population size, and an initial particle swarm is created, each particle having a random initial position and velocity.

[0134] Then, the fitness of each particle is calculated, and the individual optimal position and the global optimal particle are determined based on the fitness.

[0135] Among them, when determining the individual optimal position and the global optimal particle, it includes: eliminating the extreme particles existing in the excellent particles in the optimization process, calculating the crowding distance of the particles, and taking the particle with the largest crowding distance as the global leader. When selecting the individual optimal position, the excellent particles in the optimization process are also needed, and the individual particle reinforcement learning factor is introduced to speed up the speed of the particles flying towards the frontier solution. When selecting, the solution of the dominant area is taken as the individual leader, and its flight mode is constructed to improve the convergence of the solution set.

[0136] In this embodiment, the specific selection steps are as follows:

[0137] 1. Give priority to selecting external archive particles in the dominance area and use them as corresponding individual leaders;

[0138] 2. When there is no non-inferior solution in the dominance area, determine whether there is a dominance relationship between the leader and the current individual;

[0139] 3. When there is a dominance relationship and the current individual dominates the individual guide, the individual guide is directly updated to the current particle; when there is a dominance relationship and the individual guide dominates the current individual, the individual guide particle is kept unchanged; when there is no dominance relationship between the two, a random selection is made between the guide particle and the current individual, and the selection result is the new individual guide particle.

[0140] Finally, check whether the maximum number of iterations or other end conditions are reached. If they are met, the algorithm ends; if not, the speed and position of the particle are updated according to the individual and global optimal positions, the fitness of the updated particle is calculated again, and the local optimal position and global optimal position of the particle are updated.

[0141] In this embodiment, the lower layer optimization model is solved by using a solver, including:

[0142] The energy supply entity, energy storage entity and user call the Gurobi solving tool to calculate the optimal output of the energy supply entity's energy supply equipment, the optimal charging and discharging power of the energy storage entity's energy storage device and the user's adjustable electric and cold load, and send the optimization results to the upper-level leaders.

[0143] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A low-carbon scheduling method for multi-agent master-slave game of source, load and storage, characterized in that: The steps include: Establishing a reward and punishment ladder calculation model, and using the reward and punishment ladder calculation model to calculate carbon trading costs; The demand response strategy considering electricity cost and carbon compensation constructs an interaction model between source-load-storage entities, and a decision model under the carbon trading cost is constructed based on the interaction model; Based on Stackelberg game, an optimization model of source-load-storage subject under the decision model is established; The optimization model is iteratively solved using an improved particle swarm algorithm and a solver to obtain the optimal low-carbon scheduling result.

2. A low-carbon scheduling method for source-load-storage multi-agent master-slave game according to claim 1, characterized in that: The carbon trading costs are calculated using the reward and punishment ladder calculation model, including: The baseline method is used to determine the free carbon emission quota of each source-load-storage entity; Acquire the actual carbon emissions of the energy management subject and the length of the carbon emissions interval, and compare the actual carbon emissions of the energy management subject with the free carbon emissions quota and the length of the carbon emissions interval; The reward and punishment ladder calculation model is used to calculate the carbon trading cost of the ladder corresponding to the comparison result.

3. According to claim 1, a low-carbon scheduling method for source-load-storage multi-agent master-slave game is characterized in that: The source-load-storage subject includes: an energy management subject, an energy supply subject, an energy storage subject and a user; The demand response strategy considering electricity cost and carbon compensation builds an interaction model between source-load-storage entities, including: Establish an interactive relationship between the energy management subject and the energy supply subject; establish an interactive relationship between the energy management subject and the energy storage subject; establish an interactive relationship between the energy management subject and the user.

4. A low-carbon scheduling method for source-load-storage multi-agent master-slave game according to claim 3, characterized in that: The decision model of the energy management subject is constructed as follows: Determine the energy supply cost of the energy management subject to provide energy to the user and the energy storage subject, and the interaction cost between the energy management subject and the energy storage subject and the power grid; Obtain penalty costs and carbon compensation functions for cooling interruptions; The decision objective function of the energy management subject is constructed by comprehensively considering the energy supply cost, the interaction cost, the penalty cost and the carbon compensation function.

5. A low-carbon scheduling method for source-load-storage multi-agent master-slave game according to claim 4, characterized in that: The decision model of the energy supply entity is constructed as follows: Determine the energy supply income of the energy supply entity and the carbon trading cost of the energy supply entity; The decision objective function of the energy supply entity is constructed by comprehensively considering the energy supply income, the carbon transaction cost of the energy supply entity and the carbon compensation function.

6. A low-carbon scheduling method for source-load-storage multi-agent master-slave game according to claim 3, characterized in that: The decision model of the energy storage entity is constructed as follows: Obtaining the energy release income and charging cost of the energy storage entity, as well as the operation and maintenance cost of the energy storage entity; The decision objective function of the energy storage entity is constructed by comprehensively considering the energy release benefit, the energy charging cost and the operation and maintenance cost of the energy storage entity.

7. A low-carbon scheduling method for source-load-storage multi-agent master-slave game according to claim 4, characterized in that: The user's decision model is constructed as follows: Determine the user satisfaction index and energy acquisition function based on the electricity and cooling energy consumed by the user; The user's decision objective function is constructed by integrating the user satisfaction index, the energy acquisition function and the carbon compensation function.

8. The low-carbon scheduling method for source-load-storage multi-agent master-slave game according to claim 3 is characterized in that: The source-load-storage subject under the decision model is established based on Stackelberg game, including: The energy management subject is used as the leader of the Stackelberg game, and the energy supply subject, the energy storage subject and the user are used as followers of the Stackelberg game to establish a master-slave relationship among the participants; The energy cost of the energy management subject, the output power of the functional equipment of the energy supply subject, the charging and discharging power of the energy storage subject and the adjustable load of the user are used as the strategy of the Stackelberg game; The decision-making model of each source-load-storage entity is used as the effectiveness of the Stackelberg game; Establish an optimization model based on the master-slave relationship, strategy and performance of the participants.

9. The low-carbon scheduling method for source-load-storage multi-agent master-slave game according to claim 1 is characterized in that: The optimization model is iteratively solved using an improved particle swarm algorithm and a solver, including: Dividing the optimization model into an upper optimization model and a lower optimization model; The upper layer optimization model is solved by using an improved particle swarm algorithm, and the lower layer optimization model is solved by using a solver; The upper layer optimization model and the lower layer optimization model are interactively iterated to obtain an optimization result.

10. A low-carbon scheduling method for source-load-storage multi-agent master-slave game according to claim 9, characterized in that: The upper optimization model is solved by using an improved particle swarm algorithm, including: A. Initialize particle position, velocity and population size, and create an initial particle swarm; B. Calculate the fitness function value of each particle in the initial particle swarm; C. Determine the individual optimal position and the global optimal position according to the fitness function value; D. Determine whether the end condition is met. If not, update the speed and position of the particles in the initial particle swarm according to the individual optimal position and the global optimal position, and repeat steps B to D.

Images