Multi-microgrid collaborative low-carbon scheduling method and device

Abstract

The invention is suitable for the technical field of power grid energy collaborative scheduling, and provides a multi-microgrid collaborative low-carbon scheduling method and device, and the method comprises the steps: obtaining the power related data of microgrids and a distribution network, and the operation state data of energy equipment in each microgrid, constructing a collaborative double-layer low-carbon scheduling model between a multi-energy coupling microgrid and the distribution network, the upper-layer model is a distribution network and multiple micro-grids cooperative scheduling model in a Stackelberg equilibrium interaction mode, and the lower-layer model is a scheduling model of each energy device in the micro-grids under a generative adversarial network considering source side uncertainty; and based on a multi-agent depth deterministic gradient algorithm and a distributed robust optimization method, solving the scheduling model to obtain a collaborative scheduling result by taking transaction cost minimization and carbon emission minimization as well as operation cost minimization and income maximization of each energy device in each microgrid as targets. According to the invention, cooperation with other micro-grids can be considered, and specific conditions in the micro-grids are considered in a refined manner, so that the energy utilization efficiency is improved.

Description

Technical Field

[0001] This application belongs to the technical field of grid energy collaborative scheduling, and particularly relates to a multi-microgrid collaborative low-carbon scheduling method and device. Background Art

[0002] In recent years, with the booming development of China's new energy industry, the utilization of new energy and the microgrid proposed to promote the consumption of new energy have received extensive attention. However, new energy is greatly affected by natural factors and has great uncertainty. Taking solar energy and wind energy as examples, the uncertainty of light intensity and wind speed changes will greatly increase the difficulty of equipment scheduling in the microgrid, increase the operating cost of the microgrid, and the disorder of energy utilization in the microgrid will also hinder the low-carbonization of the microgrid. Moreover, in the context of the increasingly widespread application of microgrids, the coordinated operation and scheduling among multiple microgrids have also become a major problem.

[0003] Therefore, coordinating the scheduling of multiple microgrids through the distribution network and making full use of load flexibility in the single microgrid scheduling process can effectively alleviate the above problems.

[0004] Existing research mainly focuses on the application of renewable resources in the microgrid and involves the modeling method of basic energy equipment. Regarding the scheduling problem of multiple microgrids, some research proposes to solve it through a multi-microgrid-distribution network hierarchical collaborative scheduling strategy based on a multi-agent algorithm, and some research proposes to refine the scheduling in the microgrid by using the renewable resources, load characteristics and other geographical environment characteristics in the system, but the energy utilization efficiency is not ideal. Summary of the Invention

[0005] The embodiments of this application provide a multi-microgrid collaborative low-carbon scheduling method and device, which can consider the cooperation with other microgrids and at the same time refine the specific situation inside the microgrid to improve the energy utilization efficiency.

[0006] This application is implemented through the following technical solutions:

[0007] In a first aspect, the embodiments of this application provide a multi-microgrid collaborative low-carbon scheduling method, including:

[0008] Obtain the power-related data of the microgrid and the distribution network and the operation status data of the energy equipment in each microgrid;

[0009] Based on the power-related data and the operation status data, construct a collaborative two-layer low-carbon scheduling model between the multi-energy coupling microgrid and the distribution network; the upper model of the collaborative two-layer low-carbon scheduling model is a collaborative scheduling model between the distribution network and multiple microgrids in the Stackelberg equilibrium interaction mode, and the lower model is a scheduling model of each energy equipment in the microgrid under the generative adversarial network considering the source-side uncertainty;

[0010] Based on the multi-agent deep deterministic gradient algorithm and the distributionally robust optimization method, with the minimization of the trading cost between the microgrid and the distribution network and the lowest carbon emissions as the objectives of the upper-layer model, and the minimization of the operating cost and the maximization of the revenue of each energy device in each microgrid as the objectives of the lower-layer model, the collaborative two-layer low-carbon scheduling model is solved to obtain the collaborative scheduling result.

[0011] In a second aspect, an embodiment of the present application provides a multi-microgrid collaborative low-carbon scheduling device, which applies the multi-microgrid collaborative low-carbon scheduling method as in the first aspect, and includes:

[0012] A data acquisition module, configured to acquire power-related data between the microgrid and the distribution network and the operating status data of energy devices in each microgrid;

[0013] A model construction module, configured to construct a collaborative two-layer low-carbon scheduling model between the multi-energy-coupled microgrid and the distribution network based on the power-related data and the operating status data; the upper-layer model of the collaborative two-layer low-carbon scheduling model is a collaborative scheduling model between the distribution network and multiple microgrids in the Stackelberg equilibrium interaction mode, and the lower-layer model is a scheduling model of each energy device in the microgrid under the generative adversarial network considering source-side uncertainty;

[0014] A model solving module, configured to solve the collaborative two-layer low-carbon scheduling model based on the multi-agent deep deterministic gradient algorithm and the distributionally robust optimization method, with the minimization of the trading cost between the microgrid and the distribution network and the lowest carbon emissions as the objectives of the upper-layer model, and the minimization of the operating cost and the maximization of the revenue of each energy device in each microgrid as the objectives of the lower-layer model, to obtain the collaborative scheduling result.

[0015] The beneficial effects of the embodiments of the present application are as follows:

[0016] The embodiments of the present application construct a collaborative two-layer low-carbon scheduling model between the multi-energy-coupled microgrid and the distribution network. With the aid of this model, the distribution network can perform collaborative scheduling on multiple microgrids according to the different operating states of each microgrid. This enables energy cooperation and mutual assistance between different microgrids, thereby reducing the system operating cost and improving the energy utilization efficiency. At the same time, the proposed multi-energy-coupled microgrid architecture model can comprehensively consider the characteristics of various energies and loads in the system, enhancing the low-carbon and economic nature of the scheduling.

[0017] The multi-microgrid-distribution network hierarchical collaborative scheduling strategy based on the multi-agent algorithm and the distributionally robust optimization method model based on the uncertainty generative adversarial network fully consider the uncertainty and time correlation of the entire system, enabling the corresponding strategy to be applied to the actual system. Finally, the proposed multi-agent deep deterministic gradient algorithm realizes the optimal scheduling of each microgrid, promotes the consumption of renewable energy and the optimal integration of dispatchable resources while reducing the operating cost, and maximally schedules the flexible resources of the system.

[0018] Compared with the case where uncertainty and time correlation are considered simultaneously, the total costs of both the entire distribution network-multi-microgrid system and the individual microgrid systems are significantly reduced, and the carbon emissions are also effectively controlled. Therefore, the collaborative two-layer low-carbon scheduling model between the multi-energy-coupled microgrid and the distribution network has better economic and environmental performance.

[0019] For the beneficial effects of the second aspect of the embodiments, please refer to the beneficial effects of the first aspect of the embodiments, which will not be elaborated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0021] Figure 1 It is a hierarchical collaborative scheduling structure diagram of a distribution network-multi-microgrid multi-agent power grid provided by an embodiment of the present application;

[0022] Figure 2 It is a schematic flowchart of a multi-microgrid collaborative low-carbon scheduling method provided by an embodiment of the present application;

[0023] Figure 3 It is a schematic flowchart of constructing a collaborative two-layer low-carbon scheduling model between a multi-energy-coupled microgrid and a distribution network provided by an embodiment of the present application;

[0024] Figure 4 It is an internal scheduling structure diagram of a microgrid after the collaborative scheduling of a distribution network and a microgrid is completed provided by an embodiment of the present application;

[0025] Figure 5 It is a schematic structural diagram of a multi-microgrid collaborative low-carbon scheduling device provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] In the following description, specific details such as specific system structures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0027] It should be understood that when used in the specification of this application and the appended claims, the term "comprising" indicates the presence of the described features, wholes, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0028] It should also be understood that the term "and / or" used in the specification of this application and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0029] In addition, in the description of the specification of this application and the appended claims, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0030] Reference to "one embodiment", "an embodiment", or "some embodiments" etc. described in the specification of this application means that a specific feature, structure, or characteristic described in connection with the embodiment is included in one or more embodiments of this application. Thus, statements such as "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments" etc. that appear in different places in this specification do not necessarily all refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "comprising", "including", "having", and their variants all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0031] To enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the accompanying drawings and specific implementation manners. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0032] Refer to Figure 1As shown in the figure, it is the hierarchical collaborative scheduling structure diagram of the distribution network - multi - micro - grid multi - agent power grid for the multi - micro - grid collaborative low - carbon scheduling method and device of the present invention. The main grid includes power plants and various power transmission lines. The distribution network 101 is the main network for power distribution, which is abbreviated as the distribution network in the present invention and is responsible for delivering electric energy from the transmission network to end - users. The multi - energy micro - grid 102 is a local power grid composed of multiple types of energy, abbreviated as the micro - grid in the present invention, such as solar energy, wind energy, energy storage devices, etc. Energy exchange occurs between the distribution network and these scattered micro - grids, including the input and output of electric energy, as well as the coordination mechanism between the two sides. In a network containing multiple micro - energy systems such as photovoltaic, wind power, batteries, electrical appliances, power - to - gas devices, and gas turbines, it is necessary to reasonably allocate various energy resources.

[0033] Figure 2 It is a schematic flowchart of the multi - micro - grid collaborative low - carbon scheduling method based on the overall architecture diagram of the above - mentioned distributed optical - storage - heat coupling system provided by an embodiment of the present application. Referring to Figure 2 This multi - micro - grid collaborative low - carbon scheduling method includes:

[0034] Step 201, obtain the power - related data of the micro - grid and the distribution network and the operation status data of the energy devices in each micro - grid.

[0035] Exemplarily, the power - related data of the micro - grid and the distribution network may include real - time electricity load, electricity price information, environmental cost indicators, and the operation status data of energy devices such as photovoltaic, wind power, batteries, electrical appliances, power - to - gas devices, and gas turbines in each micro - grid.

[0036] Step 202, based on the power - related data and the operation status data, construct a collaborative two - layer low - carbon scheduling model between the multi - energy - coupled micro - grid and the distribution network.

[0037] The collaborative two - layer low - carbon scheduling model between the multi - energy - coupled micro - grid and the distribution network is a two - layer model. The upper - layer model of the collaborative two - layer low - carbon scheduling model is a collaborative scheduling model of the distribution network and multiple micro - grids in the Stackelberg equilibrium interaction mode, and the lower - layer model is a scheduling model of each energy device in the micro - grid under the generative adversarial network considering source - side uncertainty.

[0038] Exemplarily, the goal of the upper - layer model is to minimize the transaction cost and carbon emissions between the micro - grid and the distribution network, and the goal of the lower - layer model is to minimize the operation cost and maximize the revenue of each energy device in each micro - grid. By constructing such a collaborative two - layer low - carbon scheduling model, the requirements of the distribution network and the micro - grid can be comprehensively considered to achieve the collaborative low - carbon scheduling of multi - micro - grids.

[0039] Step 203: Based on the multi-agent deep deterministic gradient algorithm and the distributionally robust optimization method, with the minimization of the trading cost between the microgrid and the distribution network and the lowest carbon emissions as the objectives of the upper-layer model, and the minimization of the operating cost and maximization of the revenue of each energy device within each microgrid as the objectives of the lower-layer model, solve the collaborative two-layer low-carbon scheduling model to obtain the collaborative scheduling result.

[0040] Exemplarily, according to the collaborative scheduling result, schedule the energy devices within each microgrid. Specifically, based on the power interaction result between the distribution network and the microgrid obtained from the upper-layer model and the scheduling strategy of each energy device within the microgrid obtained from the lower-layer model, perform real-time scheduling on energy devices such as photovoltaic power, wind power, storage batteries, electrical appliances, power-to-gas devices, and gas turbines.

[0041] Exemplarily, this embodiment can monitor the operating status of the energy devices within each microgrid in real time and feedback the monitoring result to Step 201 to update the power-related data and the operating status data. Through real-time monitoring and data feedback, the multi-microgrid collaborative low-carbon scheduling method can adapt to the changes in the operating status of the energy devices and the external environment, thereby continuously optimizing the scheduling strategy to ensure that the multi-microgrid system always maintains an efficient and low-carbon operating state.

[0042] Through the constructed collaborative two-layer low-carbon scheduling model between the multi-energy-coupled microgrid and the distribution network, the distribution network can perform collaborative scheduling on multiple microgrids according to the different operating statuses of each microgrid. This enables regional energy collaboration and mutual assistance between different microgrids, thereby reducing the system operating cost and improving the energy utilization efficiency. At the same time, the proposed multi-energy-coupled microgrid architecture model can comprehensively consider the characteristics of various energy sources and loads within the system, enhancing the low-carbon and economic performance of the scheduling.

[0043] The multi-microgrid - distribution network hierarchical collaborative scheduling strategy based on the multi-agent algorithm and the distributionally robust optimization method model based on the uncertainty generative adversarial network fully consider the uncertainty and time correlation of the entire system, enabling the corresponding strategy to be applied to the actual system. Finally, the proposed multi-agent deep deterministic gradient algorithm realizes the optimal scheduling of each microgrid, promoting the consumption of renewable energy and the optimal integration of dispatchable resources while reducing the operating cost, and maximizing the scheduling of the flexible resources of the system.

[0044] Compared with the situation of simultaneously considering uncertainty and time correlation, the total cost of the entire distribution network - multi-microgrid system and individual microgrid systems has been significantly reduced, and the carbon emissions have also been effectively controlled. Therefore, the collaborative two-layer low-carbon scheduling model between the multi-energy-coupled microgrid and the distribution network has better economic and environmental performance.

[0045] In one embodiment, based on power-related data and operating status data, a collaborative two-layer low-carbon scheduling model between a multi-energy coupling microgrid and a distribution network is constructed. See Figure 3 , step 202 includes:

[0046] Step 2021, based on power-related data and operating status data, construct a collaborative scheduling model between the distribution network and multiple microgrids in the Stackelberg equilibrium interaction mode.

[0047] Exemplarily, step 2021 includes:

[0048] Step 20211, based on power-related data and operating status data, construct a collaborative scheduling model between the distribution network and multiple microgrids.

[0049] The objective function of the collaborative scheduling model between the distribution network and multiple microgrids is expressed as:

[0050]

[0051] where, C DSO is the operating cost of the distribution network; is the electricity selling price of the distribution network to the transmission network at time t; is the electricity purchasing price of the distribution network from the transmission network at time t; is the distribution of the grid-connected electricity volume of the distribution network selling electricity to the transmission network at time t; is the distribution of the grid-connected electricity volume of the distribution network purchasing electricity from the transmission network at time t; is the electricity selling price of the distribution network to microgrid i at time t; is the electricity purchasing price of the distribution network from microgrid i at time t; is the electricity selling volume of the distribution network to microgrid i at time t; is the electricity purchasing volume of the distribution network from microgrid i at time t; K is the number of distribution networks of the subordinate microgrid i; is the power consumption at node j at time t; n is the number of branches in the power grid; C total is the carbon emission cost during the operation of the distribution network. The trading prices between the distribution network and each microgrid are and satisfy is the minimum trading price between the distribution network and each microgrid, is the maximum trading price between the distribution network and each microgrid.

[0052]

[0053]

[0054] where, is the cost generated by unit carbon emission; E con 、E line 、Eope , E dec respectively represent the carbon emissions of construction, line loss, operation and maintenance, and decommissioning disposal; M a , P loss , P j , V k , M l respectively represent the building material weight, line loss electricity, operation and maintenance power consumption, gas volume, and number of decommissioned devices; F a , F ele , F j , F k , F l are the corresponding carbon emission factors of the building material weight, line loss electricity, operation and maintenance power consumption, gas volume, and number of decommissioned devices respectively.

[0055] The relationship constraint between the distribution network and the microgrid is expressed as:

[0056]

[0057] Among them, S j,t represents the complex power injection of node j at time period t; P j,t is the active power of node j at time period t; Q j,t is the reactive power of node j at time period t; is the power loss of node j at time node t; V j,t is the head voltage of the branch from node j to node l at time period t; V l,t is the tail voltage of the branch from node j to node l at time period t; G jl is the conductance of the branch from node j to node l; B jl is the susceptance of the branch from node j to node l; is the complex representation of the phase difference of the head and tail voltages of the branch from node j to node l, ω jl is the angular frequency from node j to node l; among them is the minimum active power of node j, is the maximum active power of node j. The system tolerable voltage satisfies V j,t ∈{V min , V max}, V l,t ∈{V min , V max}.

[0058] Step 20212, based on the power-related data and operation status data, construct a Stackelberg equilibrium interaction mode.

[0059] Exemplarily, based on the power-related data and operation status data, constructing a Stackelberg equilibrium interaction mode includes:

[0060] First, based on power-related data and operating status data, construct an environment set between the distribution network and multiple microgrids.

[0061] The environment set includes the electric energy sold by the distribution network to the microgrid and the power purchase price of the distribution network. For the distribution network, the microgrid is the environment, which includes the amount of electric energy sold to the microgrid and the power purchase price of the distribution network

[0062]

[0063] Among them, if then At this time, the distribution network sells electricity to the microgrid i at time t; when At this time, the distribution network purchases electricity from the microgrid i at time t; for the compliant microgrid, the distribution network is the environment, and the environment set includes the transaction electricity price of the distribution network selling electricity to and purchasing electricity from the microgrid i at time t and the estimated renewable energy output of the microgrid i at time t and the estimated user-side load The state of the two-way power flow device can be expressed as follows:

[0064]

[0065] Secondly, based on the environment set, construct an action set between the distribution network and the multi-energy microgrid.

[0066] From the agent environment set, the action sets of the dominant distribution network and the compliant microgrid are as follows:

[0067]

[0068] Among them, is the total power of renewable energy; is the charge and discharge power of the energy storage device.

[0069] Then, construct the reward functions of the distribution network and the microgrid in each time period.

[0070] For the actual operation of the system, it is necessary to consider the operating constraints. The reward function should reflect the benefits of each agent and guide its learning in the allowed area. Incorporate the constraints of the system (system tolerable voltage, system power limit, two-way power flow device state constraint, system power exchange constraint) as penalty functions into the reward function, which are respectively represented as V P 、 S P 、 The reward functions of the distribution network and the microgrid in each time period are as follows:

[0071]

[0072] Among them, is the reward function of the distribution network at time t; is the reward function of microgrid i at time t; δ S and δ ex all represent penalty coefficients. If the actions shown by the agent satisfy all the constraints, then the distribution network and microgrid i can obtain their respective additional rewards at time t and

[0073] Finally, based on the environment set, action set, and reward function, a Stackelberg equilibrium interaction mode between the distribution network and multiple microgrids is constructed using Markov decision-making.

[0074] Under the conditions of the above environment set and action set, the reward functions of the distribution network and microgrids at each time period are formed, and a Stackelberg equilibrium interaction mode between the distribution network and multiple microgrids is formed.

[0075] The Markov decision process (MDP) is a mathematical framework for modeling decision-making. When constructing the Stackelberg equilibrium interaction mode between the distribution network and multiple microgrids, the environment set defines the state space of the system, including information such as the electric energy sold by the distribution network to the microgrids, the power purchase price, the estimated renewable energy output of the microgrids, the estimated user-side load, and the status of the bidirectional energy flow equipment. The action set determines the actions that the distribution network (the leader) and microgrids (the followers) can take, such as the power selling and power purchase decisions of the distribution network, and the renewable energy utilization of the microgrids and the charging and discharging power of the energy storage devices. The reward function reflects the benefits of each agent (the distribution network and microgrids) under specific actions, and considers the constraints of system operation as a penalty function.

[0076] Step 20213, using the Stackelberg equilibrium interaction mode, optimize the coordinated scheduling model of the distribution network and multiple microgrids to obtain the coordinated scheduling model of the distribution network and multiple microgrids under the Stackelberg equilibrium interaction mode.

[0077] Exemplarily, at each time step, the distribution network (the leader) selects an action (from the action set) according to the current environmental state (from the environmental set). After observing the action of the distribution network, the microgrid (the follower) then selects its own action according to the current environmental state. Then, the system calculates the reward (based on the reward function) according to the actions of the distribution network and the microgrid and the environmental state. This process is repeated continuously. At each time step, the agent selects an action according to the current state to maximize its long-term cumulative reward. As time goes by, the system gradually converges to an equilibrium state, i.e., the Stackelberg equilibrium. In this equilibrium state, the distribution network, as the leader, has selected the optimal strategy (considering the reaction of the microgrid), and the microgrid, as the follower, has also selected its own optimal strategy according to the strategy of the distribution network. Moreover, neither party has the motivation to unilaterally change its strategy. In this way, the collaborative scheduling model of the distribution network and multiple microgrids is optimized by using the Stackelberg equilibrium interaction mode between the distribution network and multiple microgrids.

[0078] Step 2022: Based on the transaction electricity quantity and electricity price between multiple microgrids and the distribution network output by the collaborative scheduling model of the distribution network and multiple microgrids, construct a scheduling model for each energy device in the microgrid under a generative adversarial network considering source-side uncertainty.

[0079] Exemplarily, step 2022 includes:

[0080] Step 20221: Based on the transaction electricity quantity and electricity price between multiple microgrids and the distribution network output by the collaborative scheduling model of the distribution network and multiple microgrids, construct a scheduling model for each energy device in the microgrid.

[0081] Exemplarily, there is an independent line between each microgrid and the distribution network. After the distribution network stipulates the relevant fees according to the estimated operation conditions of each microgrid, each microgrid can refer to the distribution network pricing for more refined scheduling of the internal energy facilities. See Figure 4 , Figure 4 For the internal scheduling structure diagram of the microgrid after the collaborative scheduling of the distribution network and the microgrid, an example of the internal gas transmission and energy transmission of the microgrid is given.

[0082] The scheduling model of each energy device in the microgrid includes the internal refined scheduling model of multiple microgrids after the collaboration of the distribution network - multiple microgrids, the power supply end model, the load model, and the model of the bidirectional energy flow device.

[0083] Based on the optimal scheduling model of the multi-energy coupling microgrid oriented to low-carbon economy, the objective function of the internal refined scheduling model of multiple microgrids after the collaboration of the distribution network - multiple microgrids can be expressed as:

[0084]

[0085] where, F ope is the operating cost of a single microgrid; It is the carbon reward and punishment in the microelectronics system.

[0086] The operating cost of a single microgrid in a multi - energy - coupled microgrid system can be quantified as the following objective function:

[0087]

[0088] Among them, F om and F buy are the equipment operation and maintenance costs of the multi - energy - coupled microgrid system and the costs of purchasing energy from other microgrids; f om,pv 、f om,pw 、f om,ccs 、f om,p2hg 、f om,gt 、f om,b are the operation and maintenance costs per unit of energy generated or consumed by photovoltaic, wind power, carbon capture devices, power - to - gas devices, combined heat and power, and boilers respectively; P t pv 、P t pw 、P t CCS 、P t P2H 、P t P2G 、 respectively represent the amounts of energy generated or consumed by the corresponding energy - generating or - consuming devices of photovoltaic, wind power, carbon capture devices, power - to - gas devices, combined heat and power, and boilers in the t - th period; P t e 、 are the costs per unit of electricity and gas in the t - th period; f t e 、f t g are the total electricity and gas purchase quantities in the t - th period; represents the unit gas cost that can be saved by the power - to - gas device in the t - th period; f t p2hg represents the gas output of the power - to - gas device in the t - th period.

[0089] Regarding the carbon reward and punishment costs, in the context of a multi - energy - coupled microgrid, dominated by the carbon emissions of gas turbines and boilers, the indirect carbon emissions are related to the distribution network. The hierarchical incentive and constraint mechanism based on carbon emission quotas can be expressed as:

[0090]

[0091] Among them, is the carbon emission of the microgrid in the t - th period; ε e 、ε b 、ε gtis the measurement coefficient of unit equivalent carbon emissions, applicable to purchased electricity, boiler gas consumption, and gas turbine gas consumption; is the price of unit carbon emissions; is the carbon emission quota of a single microgrid; θ is the reward and punishment base of unit carbon emission rights; τ is the reward and punishment power factor of unit carbon emissions; γ is the size of the carbon emission reward and punishment increase coefficient; l is the interval of carbon emission steps.

[0092] Exemplarily, the power supply side model is introduced next. The power supply side model includes the average power of a wind turbine in a day, the average photovoltaic power of photovoltaic power generation, the model of the hydrogen and carbon dioxide reforming to methane system, and the models of gas turbines and waste heat recovery devices.

[0093] Taking wind power generation as an example, the most important influencing factor in the wind power generation process is that the wind speed conforms to the Weibull distribution. From this distribution, the average power of the wind turbine in a day can be obtained as follows:

[0094]

[0095] Among them, represents the average power of the wind turbine; P rated represents the rated power of the wind turbine; k represents the shape parameter of the Weibull distribution; λ represents the scale parameter; Γ is the gamma function; reflects the relationship between the wind speed distribution and the rated wind speed, v rated is the rated wind speed.

[0096] Taking photovoltaic power generation as an example, the operation of the photovoltaic unit is mainly related to solar radiation and temperature. The average photovoltaic power of photovoltaic power generation is as follows:

[0097]

[0098] Among them, is the average photovoltaic power; S is the total area of the photovoltaic panels; η is the photoelectric conversion efficiency; I max is the maximum light intensity; B(α,β) is the Beta function, and α and β are the parameters related to the Beta distribution. is the temperature coefficient; T W represents the actual working temperature of the photovoltaic module; T rate represents the rated working temperature of the photovoltaic module.

[0099] Taking the hydrogen and carbon dioxide reforming to methane system as an example, the main body of the gas reforming to methane system in each multi-energy coupling microgrid consists of renewable energy power generation devices (i.e., photovoltaic and wind turbines), a carbon capture device, and an electrolysis to gas equipment. As for the electrolysis to gas equipment, it is only put into use when the renewable energy power generation exceeds the load. The model of the entire hydrogen and carbon dioxide reforming to methane system is expressed as:

[0100]

[0101] Among them, is the power of the carbon capture device at time t; P CCS,F and are the fixed power and operating power of the carbon capture device at time t; is the volume of CO2 sequestered at time t; μ CCS is the energy consumption coefficient of the carbon collection device. is the amount of hydrogen produced by electrolyzing water at time t; is the power of hydrogen production by electrolyzing water at time t; η P2H and are the hydrogen production efficiency and the calorific value of hydrogen. P ι P2G is the power of the methane production device at time t; is the volume of CO2 used by the methane production device at time t; is the volume of CO2 required to produce a unit volume of methane; η P2G is the efficiency of the methane production device; P ι PW and P ι PV are the wind power and photovoltaic power for supplying energy to the power-to-gas device at time t; is the volume of methane generated at time t; h g is the calorific value of methane.

[0102] When each device in the hydrogen and carbon dioxide reforming to methane system operates in coordination, part of the CO2 required by the methane production device is supplied by the carbon capture device, and the balance is input from the outside. Specifically as follows:

[0103]

[0104] Among them, is the amount of CO2 input from the outside at time t.

[0105] Exemplarily, taking a gas turbine and a waste heat recovery device as an example, the cogeneration part consists of a gas turbine and a waste heat recovery unit. The gas turbine generates electricity and discharges warm gas, and the waste heat recovery device supplies heat through the warm gas. The models of the gas turbine and the waste heat recovery device are as follows:

[0106]

[0107] Among them, P t gt,e is the output electric power of the gas turbine at time t; V t gt,g is the gas consumption per unit time of the gas turbine at time t; is the output thermal power of the gas turbine at time t; τhb is the waste heat recovery transfer coefficient; τ gt,e and τ gt,h are the electrical and heat transfer coefficients of the gas turbine; V t gt,g is the gas consumption of the gas turbine in the t period; the electrical power and heat power of the gas turbine in this device shall not be higher than the bearing capacity of the device, that is, 0 ≤ P t gt,e ≤ P gt,e,max and P gt,e,max is the maximum electrical power output by the gas turbine, is the maximum heat power output by the gas turbine.

[0108] Exemplarily, the load model will be introduced next.

[0109] Taking the boiler as an example, the boiler uses gas or electricity for heating. The output model of the boiler can be expressed as follows:

[0110]

[0111] Among them, is the thermal energy output by the boiler in the t period; τ b represents the boiler transfer coefficient; W t b,g is the energy consumption of the boiler in the t period; the thermal energy output by the boiler in the t period must be within the tolerable range of the device, that is H b,h,max is the maximum thermal energy output by the boiler.

[0112] Exemplarily, the model of the bi-directional energy flow device will be introduced next.

[0113] The relevant formulas of bi-directional energy flow devices represented by energy storage batteries, gas storage tanks, heat storage tanks, etc. in multi-energy flow devices are expressed as follows:

[0114]

[0115] Among them, is the energy storage status of the energy storage unit in the t period; ξ bat,char and ξ bat,dis are the charge and discharge rates of the bi-directional energy flow device; D bat,min and D bat,max are the lower and upper limits of the capacity of the energy storage unit; W t char and W t dis are the energy charged and discharged by the energy storage unit in the time period t; W char,max and W dis,max are the upper limits of the charging / discharging energy of the bi-directional energy flow device; indicates that the operating state of the dual-energy flow device in the t period is charging; Indicates that the operating state of the dual-energy flow device in the t period is energy release; λ bat,loss is the self-energy release rate of the dual-energy flow device.

[0116] For the multi-energy-coupled microgrid, the constraint conditions of the scheduling models of various energy devices in the microgrid are set, including the electrical load balance constraint and the gas load balance constraint.

[0117] Exemplarily, the electrical load balance constraint is introduced. The electrical load balance limit of the multi-energy-coupled microgrid is as follows:

[0118] 0 ≤ P t e ≤ P e,max (30)

[0119]

[0120] Among them, P t eb,e is the electric boiler power in the t period; P e,max is the maximum electric load power that the system can bear; P t CCS is the power of the carbon capture device; P t dis and P t char are the discharge and charge powers of the dual-energy flow device.

[0121] Exemplarily, the gas load balance constraint is introduced. The gas load balance limit of the multi-energy-coupled microgrid is as follows:

[0122]

[0123] Among them, C g,max is the limit value of the gas load that the system can bear; and represent the gas discharge volume and gas storage volume of the gas storage tank in the two-way energy flow device in the system; and are the gas consumption of the gas boiler and gas turbine per unit time in the system.

[0124] Step 20222, based on the transaction electricity quantity and electricity price between multiple microgrids and the distribution network output by the coordinated scheduling model of the distribution network and multiple microgrids, construct a generative adversarial network considering the uncertainty on the source side.

[0125] Exemplarily, step 20222 includes:

[0126] First, based on the transaction electricity quantity and electricity price between multiple microgrids and the distribution network output by the coordinated scheduling model of the distribution network and multiple microgrids, use temporal logic to reduce the dimension of the uncertainty data on the source side and obtain the low-dimensional uncertainty data on the source side.

[0127] Exemplarily, after the coordinated dispatching of the distribution network and the microgrid in sub-step 2021, the amount of electric energy traded between each micro energy network and other micro energy networks and the electricity price have been determined. Each micro energy network will schedule the internal energy facilities through an adversarial network for uncertainty based on the dispatching result. In the autoencoder network of the adversarial network for uncertainty, let the temporal vector I = {i1, …, iε} reduce the dimension in the embedding network of the autoencoder network in time series, and then convert the temporal vector into a high-dimensional vector through the recovery network of the autoencoder network, so as to establish a latent space for dimension reduction to learn the temporal characteristics of the data.

[0128] Using the reconstruction loss function X R Precisely reduce the dimension of the original data. The dimension-reduced source-side uncertainty data can be expressed as:

[0129]

[0130] where m(i) represents the original high-dimensional static vector; represents the high-dimensional static vector after dimension reduction and then recovery; y t (i) is the high-dimensional dynamic vector; represents the high-dimensional dynamic vector after dimension reduction and then recovery.

[0131] Secondly, based on the dimension-reduced source-side uncertainty data, construct the generator of the adversarial network within the microgrid.

[0132] Exemplarily, in the adversarial network, the generator processes the temporal vector space Y T with the generation function, and then the dynamic generation vector Y i:N can be obtained. Jointly encode all the data output by the generation function and the input function in the latent space, and finally generate the temporal feature generation network according to the time-series data The task of judging whether the input data is accurate is performed by the discriminator in the adversarial network through the adversarial loss X S and then judge the difference between the input data and the output data through the supervision loss X U where:

[0133]

[0134] where v n represents the classification of the uncertainty data; w n represents the supervision item of the generator function.

[0135] Then, use the Adam optimizer to update the network parameters of the adversarial network, and use the updated network parameters to train the generator to obtain the trained network parameters.

[0136] Exemplarily, the network parameters are updated using the Adam optimizer to train the model, with a learning rate of 0.001, and the peak number of training times is denoted as S. The parameters of the input, recovery, generator, and discriminator are χ e , χ r , χ g and χ d ; then calculate the gradients of the parameters in each network and Finally, update the network parameters using the optimizer.

[0137] Based on the trained network parameters, calculate the estimated error vector of the source-side uncertainty data, and use the estimated error vector to correct the energy scheduling modes in multiple microgrids to form a generative adversarial network considering source-side uncertainty.

[0138] Exemplarily, the predicted data set is obtained through the above steps containing the output of photovoltaic and wind power the power of the carbon capture device the output of the gas reforming to methane equipment the electrical load the heat load the charging and discharging of the bi-directional energy flow device the electricity price and the gas price as well as the carbon price Define that the estimated error of all uncertainty data within a day is represented by the vector , and represent the true value and the estimated value of various data in the data set , and express all possible results included in as the support set and represent respectively 's minimum and maximum values. The uncertainty set describes 's range of the joint probability distribution r. Due to the influence of natural environmental factors, the data collected in the multi-energy coupling microgrid is time-related. Therefore, the time covariance condition is introduced into the uncertainty set to characterize the time-related property of the data:

[0139]

[0140] where, t1 and t2 are two time periods within the same day; x z is 's mean vector, is the set of vectors ; and is x z for the t1 and t2 terms of is the time covariance.

[0141] Step 20223: Use a generative adversarial network considering the uncertainty on the source side to optimize the scheduling model of each energy device in the microgrid, and obtain the scheduling model of each energy device in the microgrid under the generative adversarial network considering the uncertainty on the source side.

[0142] Based on the transaction electricity quantity and electricity price output by the collaborative scheduling model of the distribution network and multiple microgrids, the above content uses temporal logic to reduce the dimensionality of the uncertainty data on the source side to obtain low-dimensional data. Then, a generator of the generative adversarial network in the microgrid is constructed based on the low-dimensional data. Next, the Adam optimizer is used to update the network parameters and train the generator to obtain the trained network parameters. After that, by calculating the prediction error vector of the uncertainty data on the source side, the energy scheduling modes in multiple microgrids are corrected to form a generative adversarial network considering the uncertainty on the source side. Finally, this generative adversarial network is used to optimize the scheduling model of each energy device in the microgrid, so as to obtain the scheduling model of each energy device in the microgrid under the generative adversarial network considering the uncertainty on the source side.

[0143] In this embodiment, in the Stackelberg equilibrium interaction mode, the distribution network can play a leading role in guiding multiple microgrids to perform reasonable resource allocation and scheduling. The distribution network can make decisions first according to factors such as the overall power demand and network conditions, and the microgrids then make responsive adjustments according to the decisions of the distribution network, which helps to improve the stability of the entire power system and reduce the power fluctuations that may occur due to uncoordinated scheduling. At the same time, through this collaborative scheduling model, the resources between the distribution network and the microgrids can be better integrated, the energy utilization efficiency can be improved, and the energy loss can be reduced. For example, the distribution network can reasonably allocate the load demand to each microgrid, avoiding the situation where some microgrids are overloaded while other microgrids have idle resources, and realizing the optimal allocation of power resources within the entire system.

[0144] Optimization of the scheduling model of each energy device in a microgrid using a generative adversarial network considering source-side uncertainties. In view of uncertainties such as the intermittency of renewable energy generation on the source side, the generative adversarial network can effectively model and handle such uncertainties. On the one hand, it can improve the flexibility of energy device scheduling in the microgrid. When facing fluctuations in source-side energy supply, energy devices can adjust their operating states in a timely manner according to the optimization strategy provided by the generative adversarial network. For example, energy storage devices can perform charge and discharge operations more intelligently to smooth out the fluctuations in renewable energy generation. On the other hand, it helps to improve the reliability of the microgrid. Through optimized scheduling, even in the case of large uncertainties on the source side, the microgrid can still ensure stable internal power supply, reduce the risk of power outages, and while meeting the internal load demand, maximize the utilization of local renewable energy, reduce dependence on the external power grid, thereby achieving a double improvement in economic and environmental benefits.

[0145] In one embodiment, a process of solving a collaborative two-layer low-carbon scheduling model is introduced based on a multi-agent deep deterministic gradient algorithm and a distributionally robust optimization method, with the goal of minimizing the transaction cost between the microgrid and the distribution network and minimizing carbon emissions as the upper-layer model, and the goal of minimizing the operating cost and maximizing the revenue of each energy device in each microgrid as the lower-layer model. Step 203 includes:

[0146] Step 2031, based on the multi-agent deep deterministic gradient algorithm, with the goal of minimizing the transaction cost between the microgrid and the distribution network and minimizing carbon emissions as the upper-layer model, solve the collaborative scheduling model of the distribution network and multiple microgrids in the Stackelberg equilibrium interaction mode to obtain the optimal state variable parameters of the collaborative scheduling model of the distribution network and multiple microgrids.

[0147] Exemplarily, the algorithm adopts a policy-value framework with two sets of neural networks: a policy network (π network) and a value network (V network). A stochastic policy is used to implement the agent's behavior, and a deterministic gradient policy is used to update the parameters. The value network obtains the global state of the system and evaluates the system behavior, while the policy network operates in the local environment. When the algorithm executes, the policy network generates a policy from the initial random policy, and then the value network evaluates this policy and calculates the V-function value, directing the action network to change the network parameters according to the evaluation value, and continuously performing the above actions to gradually increase the cumulative reward obtained by the policy.

[0148]

[0149]

[0150] where j i ′ ,t is the target reward; is the target evaluation network V-function value; is the strategy of the target action network; w is the error generated by the evaluation network; is the V-function value of the evaluation network in reality; f represents the experience of the sample; and represent the gradients of the value network and the policy network; is the gradient of the network policy.

[0151] Input the state variable parameters of the distribution network agent and each microgrid agent and into the neural network. After continuously updating the two neural networks through the above training steps, the neural network will output the optimal solutions of the corresponding state variable parameters.

[0152] Step 2032: Based on the distributionally robust optimization method, with the goal of minimizing the operating cost and maximizing the revenue of each energy device in each microgrid as the lower-level model, solve the scheduling model of each energy device in the microgrid under the generative adversarial network considering source-side uncertainty to obtain the scheduling results of each energy device in the microgrid.

[0153] Exemplarily, unify each constraint condition of the scheduling model of each energy device in the microgrid in Step 20221 as:

[0154]

[0155] Among them, and are coefficient matrices; the vector y is all the decision variables of the first stage; P0 is the output power of each energy device, and q represents the parameter vector.

[0156] Based on the energy scheduling results in the multi-energy network of the generative adversarial network output in Step 20222, starting from the source-side uncertainty, schedule the flexible load in the case of prediction deviation to ensure the stable operation of the system. Let g0 be the result of the last scheduling of each energy device in the multi-energy-coupled microgrid, is the result of the next scheduling of each energy device in the multi-energy-coupled microgrid; is the adjustment rate matrix, is 's adjustment rate vector, then:

[0157]

[0158] After the above steps, the total operating cost of the multi-energy-coupled microgrid is transformed into:

[0159]

[0160] Among them, h1 represents the cost of the first-stage scheduling, and They are the adjustment amounts of the combined trading cost, operation and maintenance cost, gas cost, curtailment penalty cost of wind and light, power purchase cost, and carbon reward and penalty cost of the microgrid, respectively. represents the uncertainty set The minimum expected value of the adjustment cost under the worst-case distribution, where r represents the joint probability distribution of.

[0161] After uncertainty correction, the constraint conditions of the multi-energy coupled microgrid are transformed into:

[0162]

[0163] Among them, and are coefficient matrices; is related to the coefficient vector related to.

[0164] Under the above constraint conditions, the output result of the objective function of the multi-energy coupled microgrid is:

[0165]

[0166] Among them, χ T is the coefficient vector related to y; and are related to the coefficient vectors related to, then Formulas (41) and (45) can be regarded as the abbreviated forms of the entire multi-energy coupled microgrid model.

[0167] Step 2033, based on the optimal state variable parameters of the coordinated scheduling model of the distribution network and multiple microgrids and the scheduling results of each energy device in the microgrid, obtain the coordinated scheduling result.

[0168] In summary, the multi-microgrid collaborative low-carbon scheduling method provided by the embodiments of the present application, with the help of the collaborative two-layer low-carbon scheduling model between the multi-energy coupled microgrid and the distribution network, the distribution network can carry out collaborative scheduling of multiple microgrids according to the different operating states of each microgrid. This enables energy collaborative mutual assistance between different microgrids, thereby reducing the system operation cost and improving the energy utilization efficiency. At the same time, the proposed multi-energy coupled microgrid architecture model can comprehensively consider the characteristics of various energies and loads in the system, improving the low-carbon and economic nature of the scheduling.

[0169] The hierarchical collaborative scheduling strategy of multi-microgrid and distribution network based on multi-agent algorithm and the distributionally robust optimization method model based on the uncertainty generative adversarial network fully consider the uncertainty and time correlation of the entire system, enabling the corresponding strategy to be applied to the actual system. Finally, the proposed multi-agent deep deterministic gradient algorithm realizes the optimal scheduling of each microgrid, reduces the operating cost, promotes the consumption of renewable energy and the optimal integration of dispatchable resources, and maximally schedules the flexible resources of the system.

[0170] Compared with the situation considering both uncertainty and time correlation, the total costs of the entire distribution network-multi-microgrid system and individual microgrid systems are significantly reduced, and the carbon emissions are also effectively controlled. Therefore, the collaborative two-layer low-carbon scheduling model between multi-energy coupled microgrids and distribution networks has better economic and environmental performance.

[0171] See Figure 5 , the embodiment of the present application provides a multi-microgrid collaborative low-carbon scheduling device, which executes the multi-microgrid collaborative low-carbon scheduling method as described in the above embodiment, including a data acquisition module 301, a model construction module 302, and a model solving module 303.

[0172] The data acquisition module 301 is used to acquire the power-related data of the microgrid and the distribution network and the operating status data of the energy devices in each microgrid;

[0173] The model construction module 302 is used to construct a collaborative two-layer low-carbon scheduling model between multi-energy coupled microgrids and distribution networks based on the power-related data and the operating status data; the upper layer model of the collaborative two-layer low-carbon scheduling model is a collaborative scheduling model of the distribution network and multiple microgrids in the Stackelberg equilibrium interaction mode, and the lower layer model is a scheduling model of each energy device in the microgrid under the generative adversarial network considering source-side uncertainty;

[0174] The model solving module 303 is used to solve the collaborative two-layer low-carbon scheduling model based on the multi-agent deep deterministic gradient algorithm and the distributionally robust optimization method, with the goal of minimizing the transaction cost between the microgrid and the distribution network and the lowest carbon emissions for the upper layer model, and the goal of minimizing the operating cost and maximizing the revenue of each energy device in each microgrid for the lower layer model, so as to obtain the collaborative scheduling result.

[0175] Exemplarily, the model construction module 302 is specifically used for:

[0176] Construct a collaborative scheduling model of the distribution network and multiple microgrids in the Stackelberg equilibrium interaction mode based on the power-related data and the operating status data;

[0177] Based on the transaction electricity quantity and electricity price between multiple microgrids and the distribution network output by the collaborative scheduling model of the distribution network and multiple microgrids, a scheduling model for each energy device in the microgrid is constructed under the generative adversarial network considering source-side uncertainty.

[0178] Exemplarily, the objective function of the collaborative scheduling model of the distribution network and multiple microgrids is expressed as:

[0179]

[0180] Among them, C DSO is the operating cost of the distribution network; is the electricity selling price of the distribution network to the transmission network at time t; is the electricity purchasing price of the distribution network from the transmission network at time t; is the allocation of the grid-connected electricity quantity of the distribution network selling electricity to the transmission network at time t; is the allocation of the grid-connected electricity quantity of the distribution network purchasing electricity from the transmission network at time t; is the electricity selling price of the distribution network to microgrid i at time t; is the electricity purchasing price of the distribution network from microgrid i at time t; is the electricity quantity sold by the distribution network to microgrid i at time t; is the electricity quantity purchased by the distribution network from microgrid i at time t; K is the number of distribution networks of the affiliated microgrid i; is the power consumption at node j at time t; n is the number of branches in the power grid; C total is the carbon emission cost during the operation of the distribution network;

[0181] The relationship constraint between the distribution network and the microgrid is expressed as:

[0182]

[0183] Among them, S j,t represents the complex power injection at node j at time t; P j,t is the active power at node j at time t; Q j,t is the reactive power at node j at time t; is the power loss at node j at time t; V j,t is the head voltage of the branch from node j to node l at time t; V l,t is the tail voltage of the branch from node j to node l at time t; G jl is the conductance of the branch from node j to node l; B jl is the susceptance of the branch from node j to node l; is the complex representation of the phase difference of the head and tail voltages of the branch from node j to node l, ω jl is the angular frequency from node j to node l; among them is the minimum active power at node j, is the maximum active power on node j.

[0184] Exemplarily, the scheduling model of each energy device in the microgrid includes an internal fine scheduling model of multiple microgrids after the cooperation of the distribution network and multiple microgrids, a power supply end model, a load model, and a model of bidirectional energy flow devices.

[0185] Exemplarily, in the model construction module 302, based on power-related data and operation status data, a cooperative scheduling model of the distribution network and multiple microgrids in the Stackelberg equilibrium interaction mode is constructed, including:

[0186] Based on power-related data and operation status data, construct a cooperative scheduling model of the distribution network and multiple microgrids;

[0187] Based on power-related data and operation status data, construct a Stackelberg equilibrium interaction mode;

[0188] Using the Stackelberg equilibrium interaction mode, optimize the cooperative scheduling model of the distribution network and multiple microgrids to obtain a cooperative scheduling model of the distribution network and multiple microgrids in the Stackelberg equilibrium interaction mode.

[0189] Exemplarily, in the model construction module 302, based on power-related data and operation status data, a Stackelberg equilibrium interaction mode is constructed, including:

[0190] Based on power-related data and operation status data, construct an environment set between the distribution network and multiple microgrids; the environment set includes the electric energy sold by the distribution network to the microgrids and the power purchase price of the distribution network;

[0191] Based on the environment set, construct an action set between the distribution network and multi-energy microgrids;

[0192] Construct a reward function for the distribution network and microgrids in each time period;

[0193] Based on the environment set, action set, and reward function, use Markov decision to construct a Stackelberg equilibrium interaction mode between the distribution network and multiple microgrids.

[0194] Exemplarily, in the model construction module 302, based on the transaction power and electricity price between multiple microgrids and the distribution network output by the cooperative scheduling model of the distribution network and multiple microgrids, a scheduling model of each energy device in the microgrid under the generative adversarial network considering source-side uncertainty is constructed, including:

[0195] Based on the transaction power and electricity price between multiple microgrids and the distribution network output by the cooperative scheduling model of the distribution network and multiple microgrids, construct a scheduling model of each energy device in the microgrid;

[0196] Based on the transaction power and electricity price between multiple microgrids and the distribution network output by the cooperative scheduling model of the distribution network and multiple microgrids, construct a generative adversarial network considering source-side uncertainty;

[0197] Using a generative adversarial network that takes into account the uncertainty on the source side, optimize the scheduling models of various energy devices in the microgrid to obtain the scheduling models of various energy devices in the microgrid under the generative adversarial network that considers the uncertainty on the source side.

[0198] Exemplarily, in the model construction module 302, based on the transaction electricity quantities and electricity prices between multiple microgrids and the distribution network output by the collaborative scheduling model of the distribution network and multiple microgrids, construct a generative adversarial network that takes into account the uncertainty on the source side, including:

[0199] Based on the transaction electricity quantities and electricity prices between multiple microgrids and the distribution network output by the collaborative scheduling model of the distribution network and multiple microgrids, use temporal logic to reduce the dimension of the uncertainty data on the source side to obtain low-dimensional uncertainty data on the source side;

[0200] Based on the low-dimensional uncertainty data on the source side, construct a generator of the generative adversarial network in the microgrid;

[0201] Use the Adam optimizer to update the network parameters of the generative adversarial network, and use the updated network parameters to train the generator to obtain the trained network parameters;

[0202] Based on the trained network parameters, calculate the estimated error vector of the uncertainty data on the source side, and use the estimated error vector to correct the energy scheduling modes in multiple microgrids to form a generative adversarial network that takes into account the uncertainty on the source side.

[0203] Exemplarily, the model solving module 303 is specifically used for:

[0204] Based on the multi-agent deep deterministic gradient algorithm, with the minimization of the transaction cost between the microgrid and the distribution network and the lowest carbon emissions as the objectives of the upper-layer model, solve the collaborative scheduling model of the distribution network and multiple microgrids in the Stackelberg equilibrium interaction mode to obtain the optimal state variable parameters of the collaborative scheduling model of the distribution network and multiple microgrids;

[0205] Based on the distributionally robust optimization method, with the minimization of the operating costs and the maximization of the benefits of each energy device in each microgrid as the objectives of the lower-layer model, solve the scheduling models of various energy devices in the microgrid under the generative adversarial network that takes into account the uncertainty on the source side to obtain the scheduling results of various energy devices in the microgrid;

[0206] Based on the optimal state variable parameters of the collaborative scheduling model of the distribution network and multiple microgrids and the scheduling results of various energy devices in the microgrid, obtain the collaborative scheduling results.

[0207] It should be noted that although several units / modules or sub-units / modules of the multi-microgrid collaborative low-carbon scheduling device are mentioned in the above detailed description, this division is merely exemplary and not mandatory. In fact, according to the embodiments of the present application, the features and functions of two or more of the above-described units / modules can be embodied in one unit / modules. Conversely, the features and functions of one unit / modules described above can be further divided and embodied by multiple unit / modules.

[0208] In addition, although the operations of the method of the present application are described in a specific order in the drawings, this does not require or imply that these operations must be performed in that specific order, or that all of the shown operations must be performed to achieve the desired result. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step for execution, and / or one step may be decomposed into multiple steps for execution.

[0209] The present application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the multi-microgrid collaborative low-carbon scheduling method provided in the above embodiments of the present application.

[0210] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For parts not detailed or recorded in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0211] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A multi-microgrid collaborative low-carbon scheduling method, characterized in that, Including: Obtaining power-related data of the microgrid and the distribution network, as well as operation status data of energy devices within each microgrid; Based on the power-related data and the operation status data, constructing a collaborative two-layer low-carbon scheduling model between the multi-energy-coupled microgrid and the distribution network; the upper-layer model of the collaborative two-layer low-carbon scheduling model is a collaborative scheduling model between the distribution network and multiple microgrids in the Stackelberg equilibrium interaction mode, and the lower-layer model is a scheduling model of each energy device within the microgrid under a generative adversarial network considering source-side uncertainty; Based on the multi-agent deep deterministic gradient algorithm and the distributionally robust optimization method, with the minimization of the transaction cost between the microgrid and the distribution network and the lowest carbon emissions as the objectives of the upper-layer model, and the minimization of the operation cost and the maximization of the revenue of each energy device within each microgrid as the objectives of the lower-layer model, solving the collaborative two-layer low-carbon scheduling model to obtain a collaborative scheduling result.

2. The multi-microgrid collaborative low-carbon scheduling method according to claim 1, wherein, The constructing of the collaborative two-layer low-carbon scheduling model between the multi-energy-coupled microgrid and the distribution network based on the power-related data and the operation status data includes: Based on the power-related data and the operation status data, constructing the collaborative scheduling model between the distribution network and multiple microgrids in the Stackelberg equilibrium interaction mode; Based on the transaction power and electricity price between multiple microgrids and the distribution network output by the collaborative scheduling model between the distribution network and multiple microgrids, constructing the scheduling model of each energy device within the microgrid under a generative adversarial network considering source-side uncertainty.

3. The multi-microgrid collaborative low-carbon scheduling method according to claim 1, characterized in that The objective function of the collaborative scheduling model between the distribution network and multiple microgrids is expressed as: Among them, C DSO The operating cost of the distribution network; The electricity price sold by the distribution network to the transmission network during period t; The electricity price that the distribution network purchases from the transmission network during period t; It is the distribution of the amount of electricity sold by the distribution network to the transmission network during period t; It is the distribution of the amount of electricity that is online during the t period when the distribution network purchases electricity from the transmission network; is the electricity price sold by the distribution network to microgrid i during period t; is the price of electricity purchased by the distribution network from microgrid i during period t; is the amount of electricity sold by the distribution network to microgrid i during period t; is the amount of electricity purchased by the distribution network from microgrid i during period t; K is the number of distribution networks under the jurisdiction of microgrid i; is the power consumption at node j during period t; n is the number of branches in the power grid; C total The carbon emission cost during the operation of the distribution network; The relationship constraints between the distribution network and the microgrid are expressed as: Among them, S j,t represents the complex power injection of node j at time period t; P j,t is the active power of node j at time period t; Q j,t is the reactive power of node j at time period t; is the power loss of node j at time t; V j,t is the head voltage of the branch from node j to node l at time period t; V l,t is the tail voltage of the branch from node j to node l at time period t; G jl is the conductance of the branch from node j to node l; B jl is the susceptance of the branch from node j to node l; is the complex representation of the phase difference of the head and tail voltages of the branch from node j to node l, ω jl is the angular frequency from node j to node l; among them is the minimum active power of node j, is the maximum active power of node j.

4. The multi-microgrid collaborative low-carbon scheduling method according to claim 1, wherein The scheduling model of each energy device within the microgrid includes an internal fine scheduling model, a power supply end model, a load model, and a model of bidirectional energy flow devices of multiple microgrids after collaboration between the distribution network and multiple microgrids.

5. The multi-microgrid collaborative low-carbon scheduling method according to claim 2, wherein The constructing of the collaborative scheduling model between the distribution network and multiple microgrids in the Stackelberg equilibrium interaction mode based on the power-related data and the operation status data includes: Based on the power-related data and the operation status data, constructing a collaborative scheduling model between the distribution network and multiple microgrids; Based on the power-related data and the operation status data, constructing a Stackelberg equilibrium interaction mode; Using the Stackelberg equilibrium interaction mode to optimize the collaborative scheduling model between the distribution network and multiple microgrids to obtain the collaborative scheduling model between the distribution network and multiple microgrids in the Stackelberg equilibrium interaction mode.

6. The multi-microgrid collaborative low-carbon scheduling method according to claim 5, wherein The constructing of the Stackelberg equilibrium interaction mode based on the power-related data and the operation status data includes: Based on the power-related data and the operation status data, constructing an environment set between the distribution network and multiple microgrids; the environment set includes the electric energy sold by the distribution network to the microgrid and the power purchase price of the distribution network; Based on the environment set, constructing an action set between the distribution network and the multi-energy microgrid; Constructing a reward function for the distribution network and the microgrid in each time period; Based on the environment set, the action set, and the reward function, using Markov decision to construct a Stackelberg equilibrium interaction mode between the distribution network and multiple microgrids.

7. The multi-microgrid collaborative low-carbon scheduling method according to claim 2, wherein Construct the scheduling model of each energy device in the microgrid under the generative adversarial network considering the uncertainty of the source side based on the transaction electricity quantity and electricity price between the multi - microgrid and the distribution network output by the collaborative scheduling model of the distribution network and multiple microgrids, including: Construct the scheduling model of each energy device in the microgrid based on the transaction electricity quantity and electricity price between the multiple microgrids and the distribution network output by the collaborative scheduling model of the distribution network and multiple microgrids; Construct a generative adversarial network considering the uncertainty of the source side based on the transaction electricity quantity and electricity price between the multiple microgrids and the distribution network output by the collaborative scheduling model of the distribution network and multiple microgrids; Use the generative adversarial network considering the uncertainty of the source side to optimize the scheduling model of each energy device in the microgrid, and obtain the scheduling model of each energy device in the microgrid under the generative adversarial network considering the uncertainty of the source side.

8. The multi-microgrid collaborative low-carbon scheduling method according to claim 7, wherein, The constructing of a generative adversarial network considering the uncertainty of the source side based on the transaction electricity quantity and electricity price between the multiple microgrids and the distribution network output by the collaborative scheduling model of the distribution network and multiple microgrids includes: Based on the transaction electricity quantity and electricity price between the multiple microgrids and the distribution network output by the collaborative scheduling model of the distribution network and multiple microgrids, use temporal logic to reduce the dimension of the uncertainty data on the source side, and obtain the low - dimensional uncertainty data on the source side; Based on the low - dimensional uncertainty data on the source side, construct the generator of the generative adversarial network in the microgrid; Use the Adam optimizer to update the network parameters of the generative adversarial network, and use the updated network parameters to train the generator to obtain the trained network parameters; Based on the trained network parameters, calculate the estimated error vector of the uncertainty data on the source side, and use the estimated error vector to correct the energy scheduling modes in multiple microgrids to form the generative adversarial network considering the uncertainty of the source side.

9. The multi-microgrid collaborative low-carbon scheduling method according to claim 1, wherein The solving of the collaborative two - layer low - carbon scheduling model with the goal of minimizing the transaction cost between the microgrid and the distribution network and minimizing carbon emissions as the upper - layer model and the goal of minimizing the operating cost and maximizing the revenue of each energy device in each microgrid as the lower - layer model based on the multi - agent deep deterministic gradient algorithm and the distributionally robust optimization method to obtain the collaborative scheduling result includes: Based on the multi - agent deep deterministic gradient algorithm, with the goal of minimizing the transaction cost between the microgrid and the distribution network and minimizing carbon emissions as the upper - layer model, solve the collaborative scheduling model of the distribution network and multiple microgrids in the Stackelberg equilibrium interaction mode to obtain the optimal state variable parameters of the collaborative scheduling model of the distribution network and multiple microgrids; Based on the distributionally robust optimization method, with the goal of minimizing the operating cost and maximizing the revenue of each energy device in each microgrid as the lower - layer model, solve the scheduling model of each energy device in the microgrid under the generative adversarial network considering the uncertainty of the source side to obtain the scheduling results of each energy device in the microgrid; Based on the optimal state variable parameters of the collaborative scheduling model of the distribution network and multiple microgrids and the scheduling results of each energy device in the microgrid, obtain the collaborative scheduling result.

10. A multi - microgrid collaborative low - carbon scheduling device, which applies the multi - microgrid collaborative low - carbon scheduling method according to any one of claims 1 to 9, includes: A data acquisition module, configured to acquire power-related data of the microgrid and the distribution network, as well as the operating status data of the energy devices within each microgrid; A model construction module, configured to construct a collaborative two-layer low-carbon scheduling model between the multi-energy-coupled microgrid and the distribution network based on the power-related data and the operating status data; the upper-layer model of the collaborative two-layer low-carbon scheduling model is a collaborative scheduling model between the distribution network and multiple microgrids in the Stackelberg equilibrium interaction mode, and the lower-layer model is a scheduling model of each energy device within the microgrid under a generative adversarial network considering source-side uncertainty; A model solving module, configured to solve the collaborative two-layer low-carbon scheduling model based on the multi-agent deep deterministic gradient algorithm and the distributionally robust optimization method, with the minimization of the transaction cost between the microgrid and the distribution network and the lowest carbon emission as the objectives of the upper-layer model, and the minimization of the operating cost and the maximization of the revenue of each energy device within each microgrid as the objectives of the lower-layer model, to obtain a collaborative scheduling result.

Images