A power distribution network networking type energy storage planning and deployment method and system

Abstract

The application discloses a power distribution network network-configuration type energy storage planning and deployment method and system, and the method comprises the following steps: obtaining historical meteorological disaster data, power grid data and energy storage parameter data of a target area; based on the historical meteorological disaster data, power grid data and energy storage parameter data, a three-layer coupled random optimization model for optimizing energy storage configuration, system operation under disasters and fault repair processes is constructed; a multi-agent deep reinforcement learning algorithm is used to solve the three-layer coupled random optimization model, and an optimal network-configuration type energy storage deployment scheme and a collaborative recovery strategy are output. The method realizes fine optimization of the whole-chain resilience of the power grid in extreme disasters, and significantly improves the economy and robustness of the planning scheme.

Description

Technical Field

[0001] This invention belongs to the field of power system planning and operation technology, and in particular relates to a method and system for planning and deploying grid-type energy storage in distribution networks. Background Technology

[0002] Traditional power distribution network planning methods primarily focus on economic efficiency and reliability under normal operating conditions, lacking quantitative consideration and targeted optimization of system resilience under extreme disasters. Meanwhile, existing energy storage system planning tends to emphasize its economic value in peak shaving and valley filling and renewable energy integration under normal conditions, failing to fully utilize its ability to "build a network" as an independent voltage and frequency source during disasters. This results in low energy storage utilization rates and insufficient support for power restoration during disasters.

[0003] Existing research on improving power grid resilience under disasters often employs deterministic fault scenarios or simplified recovery logic in its modeling, failing to accurately depict the spatiotemporal randomness of disaster intensity distribution, the failure probability of equipment based on vulnerability curves, and the dynamic scheduling process of repair resources. At the planning and solution level, traditional methods such as two-stage robust optimization or stochastic programming are frequently used. These methods are complex, involve many variables, and result in low computational efficiency, making it difficult to integrate with advanced intelligent algorithms for rapid and high-quality decision-making.

[0004] Therefore, there is an urgent need to propose a smart planning and deployment method and system for distribution network-based energy storage that can accurately model the entire disaster process, deeply integrate planning and operation recovery strategies, and efficiently solve the problem, so as to output an energy storage deployment scheme that is both highly resilient and economical. Summary of the Invention

[0005] This invention aims to overcome the problems faced by existing distribution network energy storage planning methods in extreme disaster scenarios, such as imprecise resilience modeling, disconnect between planning and operation recovery strategies, and low solution efficiency due to model complexity. It provides an intelligent planning method and system that can output a highly resilient grid-type energy storage deployment scheme and its collaborative recovery strategy.

[0006] In a first aspect, the present invention provides a method for planning and deploying grid-based energy storage in power distribution networks, comprising:

[0007] Acquire historical meteorological disaster data, power grid data, and energy storage parameter data for the target area;

[0008] Based on the historical meteorological disaster data, power grid data, and energy storage parameter data, a three-layer coupled stochastic optimization model is constructed to collaboratively optimize energy storage configuration, system operation under disasters, and fault repair processes. The three-layer coupled stochastic optimization model includes:

[0009] A high-level planning model that uses the location and capacity of energy storage systems as decision variables and aims to minimize the expected total life-cycle cost of the distribution network.

[0010] This is a mid-level operation model that takes the energy storage configuration of the upper-level planning model as input, aims to minimize the sum of system operating cost and load loss penalty cost under extreme disaster scenarios, and satisfies the power grid safe operation constraints and the operation constraints of grid-type energy storage during the post-disaster recovery period.

[0011] A lower-level recovery scheduling model that aims to minimize the resource scheduling cost for repairing faulty components and satisfies the logical constraints of repair team scheduling and fault repair progress.

[0012] The three-layer coupled stochastic optimization model is solved using a multi-agent deep reinforcement learning algorithm, and the optimal grid-type energy storage deployment scheme and collaborative recovery strategy are output.

[0013] Secondly, the present invention provides a distribution network-based energy storage planning and deployment system, comprising:

[0014] The acquisition module is configured to acquire historical meteorological disaster data, power grid data, and energy storage parameter data for the target area.

[0015] The construction module is configured to build a three-layer coupled stochastic optimization model based on the historical meteorological disaster data, power grid data, and energy storage parameter data. This model collaboratively optimizes energy storage configuration, system operation under disasters, and fault repair processes. The three-layer coupled stochastic optimization model includes:

[0016] A high-level planning model that uses the location and capacity of energy storage systems as decision variables and aims to minimize the expected total life-cycle cost of the distribution network.

[0017] This is a mid-level operation model that takes the energy storage configuration of the upper-level planning model as input, aims to minimize the sum of system operating cost and load loss penalty cost under extreme disaster scenarios, and satisfies the power grid safe operation constraints and the operation constraints of grid-type energy storage during the post-disaster recovery period.

[0018] A lower-level recovery scheduling model that aims to minimize the resource scheduling cost for repairing faulty components and satisfies the logical constraints of repair team scheduling and fault repair progress.

[0019] The solution module is configured to use a multi-agent deep reinforcement learning algorithm to solve the three-layer coupled stochastic optimization model and output the optimal grid-type energy storage deployment scheme and collaborative recovery strategy.

[0020] Thirdly, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the distribution network-based energy storage planning and deployment method according to any embodiment of the present invention.

[0021] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the steps of the distribution network-type energy storage planning and deployment method according to any embodiment of the present invention.

[0022] The distribution network-based energy storage planning and deployment method and system of this application first collects disaster, power grid, and energy storage parameter data of the target area and generates a disaster scenario set including random faults and repair time. Then, a three-layer coupled stochastic optimization model of "planning-operation-recovery" is established. The upper layer of this three-layer coupled stochastic optimization model makes decisions on energy storage site selection and capacity setting to minimize the expected total life cycle cost, the middle layer simulates the system operation of grid-based energy storage under disasters, and the lower layer optimizes the dynamic scheduling of repair resources. Then, a multi-agent deep reinforcement learning algorithm is used to collaboratively solve the above complex model and output the optimal energy storage deployment scheme and collaborative recovery strategy. Finally, closed-loop optimization is formed through simulation verification and online updates, realizing fine optimization of the entire chain of power grid resilience of "prevention-resistance-recovery" under extreme disasters, and significantly improving the economy and robustness of the planning scheme. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 A flowchart illustrating a method for planning and deploying grid-type energy storage in a power distribution network, as provided in an embodiment of the present invention;

[0025] Figure 2 This is a structural block diagram of a distribution network-based energy storage planning and deployment system provided in an embodiment of the present invention;

[0026] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Please see Figure 1 The diagram shows a flowchart of a distribution network-based energy storage planning and deployment method according to this application.

[0029] like Figure 1 As shown, the specific steps for planning and deploying grid-based energy storage in power distribution networks include:

[0030] Step S101: Obtain historical meteorological disaster data, power grid data, and energy storage parameter data for the target area.

[0031] In this step,

[0032] Step S102: Based on the historical meteorological disaster data, power grid data, and energy storage parameter data, construct a three-layer coupled stochastic optimization model for collaboratively optimizing energy storage configuration, system operation under disasters, and fault repair process.

[0033] In this step, the three-layer coupled stochastic optimization model includes:

[0034] A high-level planning model that uses the location and capacity of the energy storage system as decision variables and aims to minimize the expected total cost of the distribution network throughout its entire life cycle.

[0035] Specifically, the objective function of the upper-level planning model is:

[0036] ,

[0037] ,

[0038] In the formula, This represents the expected total cost over the entire lifecycle of the distribution network. The annual value cost of energy storage system investment, The system operating cost in scenario k, The load loss penalty cost in scenario k. The cost of resource scheduling for fault repair in scenario k. For a collection of energy storage installation nodes, Cost per unit energy capacity The rated capacity for energy storage for node i. Cost per unit power capacity The rated power for energy storage at node i. For fixed installation costs per site, This is a binary decision variable, indicating whether energy storage should be installed at node i, with 1 for installation and 0 for no installation. The discount rate is... For the planning cycle.

[0039] The constraint expressions for the upper-level planning model are:

[0040] ,

[0041] ,

[0042] ,

[0043] ,

[0044] ,

[0045] In the formula, This represents the lower limit of the energy storage capacity of node i. This is a binary decision variable, indicating whether energy storage is installed at node i (1 for installation, 0 for no installation). The rated energy stored for node i This represents the upper limit of the energy storage capacity of node i. For the set of energy storage installation nodes, This represents the lower limit of the energy storage power of node i. The rated power for energy storage at node i. This represents the upper limit of energy storage power for node i. The maximum permissible power-energy ratio for energy storage. To allow the maximum amount of energy storage to be deployed, This represents the upper limit of the total energy storage capacity allowed by the system.

[0046] This is a mid-level operation model that takes the energy storage configuration of the upper-level planning model as input, aims to minimize the sum of system operating costs and load loss penalty costs under extreme disaster scenarios, and satisfies the power grid safety operation constraints and the operation constraints of grid-based energy storage during the post-disaster recovery period.

[0047] Specifically, the objective function of the middle-level operating model is:

[0048] ,

[0049] In the formula, The system operating cost in scenario k, The load loss penalty cost under scenario k, For time step, The price at which electricity is purchased from the upper-level power grid or emergency power source at time t. Let be the active power purchased from an external power source at time t. This is the network loss cost coefficient. branch road The resistance, For scenario k, the branch through which the flow occurs at time t The current amplitude, This is a set of time steps for the post-disaster recovery period. For the set of all nodes in the distribution network, For branch road collection, The penalty cost for reducing unit load. Let be the load reduction amount at node i at time t in scenario k.

[0050] The constraints for safe operation of the power grid include node power balance constraints and branch power flow and voltage constraints. The expression for the node power balance constraint is as follows:

[0051] ,

[0052] ,

[0053] In the formula, Let be the active power of branch m at node i at time t. The active power output of the distributed power source at node i. Let be the active load of node i. Net active power is injected into the energy storage of node i. Let be the reactive power of branch m at node i at time t. For node i, the reactive power output of the distributed power source. Let i be the reactive load. Net reactive power is injected into the energy storage of node i. For the set of all nodes in the distribution network, For branch road collection, This is a set of time steps for the post-disaster recovery period;

[0054] The expressions for the branch power flow and voltage constraints are as follows:

[0055] ,

[0056] ,

[0057] ,

[0058] In the formula, Let i be the voltage amplitude at node i. Let j be the voltage amplitude at node j. Let (i,j) be the resistance of branch (i,j). Let be the active power of branch (i,j). Let (i,j) be the reactance of branch (i,j). Let (i,j) be the reactive power of the branch. Let (i,j) be the current in branch (i,j). This is the lower limit of the node voltage amplitude. This represents the upper limit of the node voltage amplitude.

[0059] The expression for the operational constraints of the grid-type energy storage is:

[0060] ,

[0061] ,

[0062] ,

[0063] ,

[0064] ,

[0065] ,

[0066] ,

[0067] In the formula, Net active power is injected into the energy storage of node i. For node i, the energy storage and discharge power. The energy storage charging power for node i, This is a charging status indicator (1 for charging). The rated power for energy storage at node i. Let i be the state of charge of the energy storage at time t. Let i be the state of charge of the energy storage at time t-1. For energy storage charging efficiency, For energy storage and discharge efficiency, The rated capacity for energy storage for node i. For time step, This is the lower limit of SOC. This is the upper limit of SOC. for The state of charge of energy storage i at any given time. This represents the initial state of charge of the energy storage. The limit for SOC variation at the beginning and end of the cycle. To enable network 0 / 1 variables, This is the lower limit for reactive power adjustment under grid-connected mode. Net reactive power is injected into the energy storage of node i. This is the upper limit for reactive power adjustment under the grid-connected mode.

[0068] A lower-level recovery scheduling model aims to minimize the resource scheduling cost for repairing faulty components and satisfies the logical constraints of repair team scheduling and fault repair progress.

[0069] Specifically, the objective function of the lower-level recovery scheduling model is:

[0070] ,

[0071] In the formula, Let k be the resource scheduling cost for fault repair. To allow the repair team to assemble, This is a set of time steps for the post-disaster recovery period. Let r be the unit movement cost of the team. Let r be the distance traveled by team r at time t. Let r be the unit work cost of team r. This represents the average working hours of the team.

[0072] The expression for the logical constraints on the dispatching of emergency repair teams and the progress of fault repair is as follows:

[0073] ,

[0074] ,

[0075] In the formula, To determine if the variable was successfully repaired, check whether component e was repaired at time t (1 for yes). To plan a set of time periods, To extend the repair time, For components The moment the emergency repairs began Whether team r is assigned a repair component at time t. , To allow the repair team to assemble, For team r, there is a set of repairable components.

[0076] Step S103: The three-layer coupled stochastic optimization model is solved using a multi-agent deep reinforcement learning algorithm to output the optimal grid-type energy storage deployment scheme and collaborative recovery strategy.

[0077] In this step, a planning agent is constructed, wherein the observation space of the planning agent is the static characteristics of the power grid, which include topological connectivity, historical average load, installed capacity of distributed power sources, and spatial distribution of disaster risks.

[0078] The action space of the planning agent is the energy storage deployment scheme, which outputs the installation status, rated energy capacity and rated power capacity of each candidate node.

[0079] Construct an operation-recovery collaborative intelligent agent, wherein the observation space of the operation-recovery collaborative intelligent agent is the post-disaster dynamic state, which includes real-time topological connectivity, load and power output, energy storage charging status, and the location and working status of the emergency repair team.

[0080] The action space of the operation-recovery collaborative intelligent agent is the collaborative control command, which jointly outputs commands for energy storage charging and discharging power, network switching operation, load reduction, and emergency repair task allocation.

[0081] Initialize the policy network parameters of the planning agent and the policy network parameters of the run-recovery cooperative agent;

[0082] In each training round, a disaster scenario is randomly selected, and the planning agent generates an energy storage configuration scheme.

[0083] Based on the energy storage configuration scheme and the selected scenario, the operation-recovery collaborative intelligent agent simulates the post-disaster operation and recovery process and generates a sequence of control and scheduling actions.

[0084] Based on the system state trajectory generated by the agent's interaction, the joint reward is calculated, and the expression for the joint reward is:

[0085] ,

[0086] ,

[0087] In the formula, For joint awards, The annual value cost weighting coefficient is used. The annual value cost of energy storage system investment, For system operation and compliance loss cost weighting coefficients, The system operating cost in scenario k, The load loss penalty cost in scenario k. These are the weighting coefficients. Let k be the resource scheduling cost for fault repair. For resilience reward weighting coefficient, As a resilience bonus value, It is a collection of typical extreme disaster scenarios. The weight of the resilience index is the power loss value. This represents the power loss value in scenarios without energy storage solutions. As a weight for the resilience index of average power restoration time, For scenario k without energy storage, the average system power recovery time is... For the scene The power loss value under the following conditions For the scene Average power restoration time of the system;

[0088] By using reinforcement learning algorithms, the policy network parameters of the planning agent and the policy network parameters of the run-recovery cooperative agent are updated simultaneously based on the joint reward to maximize the long-term cumulative reward.

[0089] The system is repeatedly updated. After the policy network parameters converge, the optimal grid-type energy storage deployment scheme is decoded from the planning agent, and the collaborative recovery strategy for various disaster scenarios is decoded from the operation-recovery collaborative agent strategy.

[0090] In summary, the method of this application constructs a three-layer coupled stochastic optimization model of "planning-operation-recovery", which integrates the "grid construction" capability of energy storage, the spatiotemporal randomness of disasters, and the dynamic scheduling of emergency repair resources into the optimization framework. This three-layer coupled stochastic optimization model ensures that the planning scheme can not only optimally configure energy storage to prevent and resist disaster impacts, but also seamlessly connect with the optimal post-disaster operation and recovery strategy, fundamentally improving the overall resilience and investment efficiency of the power grid.

[0091] For three-layer coupled stochastic optimization models that are difficult to solve directly using traditional methods, a planning and operation-recovery collaborative agent was designed. Guided by a joint reward function, the agent learns collaboratively and can efficiently explore the solution space and learn the optimal strategy directly from its interaction with the environment. This greatly improves the solution efficiency and decision quality, making it possible to move from "offline simulation optimization" to "online intelligent decision-making" for resilient planning of complex systems.

[0092] Please see Figure 2 The diagram shows a structural block diagram of a distribution network-type energy storage planning and deployment system according to this application.

[0093] like Figure 2 As shown, the distribution network-type energy storage planning and deployment system 200 includes an acquisition module 210, a construction module 220, and a solution module 230.

[0094] The acquisition module 210 is configured to acquire historical meteorological disaster data, power grid data, and energy storage parameter data of the target area; the construction module 220 is configured to construct a three-layer coupled stochastic optimization model based on the historical meteorological disaster data, power grid data, and energy storage parameter data, which coordinates and optimizes energy storage configuration, system operation under disaster, and fault repair process. The three-layer coupled stochastic optimization model includes: an upper-layer planning model with the location and capacity of the energy storage system as decision variables and the goal of minimizing the expected value of the total life cycle cost of the distribution network; a middle-layer operation model with the energy storage configuration of the upper-layer planning model as input, aiming to minimize the sum of system operation cost and load loss penalty cost under extreme disaster scenarios, and satisfying the power grid safety operation constraints and grid-type energy storage operation constraints during the post-disaster recovery period; and a lower-layer recovery scheduling model with the goal of minimizing the emergency repair resource scheduling cost of faulty components and satisfying the logical constraints of emergency repair team scheduling and fault repair progress; and a solution module 230 is configured to use a multi-agent deep reinforcement learning algorithm to solve the three-layer coupled stochastic optimization model and output the optimal grid-type energy storage deployment scheme and collaborative recovery strategy.

[0095] It should be understood that Figure 2 The modules and references described in the document Figure 1 The steps described in the text correspond to those in the method described above. Therefore, the operations, features, and corresponding technical effects described above also apply to the method described in the text. Figure 2 The various modules in the document will not be described in detail here.

[0096] In other embodiments, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the distribution network-type energy storage planning and deployment method in any of the above method embodiments.

[0097] In one embodiment, the computer-readable storage medium of the present invention stores computer-executable instructions, which are configured as follows:

[0098] Acquire historical meteorological disaster data, power grid data, and energy storage parameter data for the target area;

[0099] Based on the historical meteorological disaster data, power grid data, and energy storage parameter data, a three-layer coupled stochastic optimization model is constructed to collaboratively optimize energy storage configuration, system operation under disasters, and fault repair processes. The three-layer coupled stochastic optimization model includes:

[0100] A high-level planning model that uses the location and capacity of energy storage systems as decision variables and aims to minimize the expected total life-cycle cost of the distribution network.

[0101] This is a mid-level operation model that takes the energy storage configuration of the upper-level planning model as input, aims to minimize the sum of system operating cost and load loss penalty cost under extreme disaster scenarios, and satisfies the power grid safe operation constraints and the operation constraints of grid-type energy storage during the post-disaster recovery period.

[0102] A lower-level recovery scheduling model that aims to minimize the resource scheduling cost for repairing faulty components and satisfies the logical constraints of repair team scheduling and fault repair progress.

[0103] The three-layer coupled stochastic optimization model is solved using a multi-agent deep reinforcement learning algorithm, and the optimal grid-type energy storage deployment scheme and collaborative recovery strategy are output.

[0104] Computer-readable storage media may include a stored program area and a stored data area, wherein the stored program area may store an operating system and an application program required for at least one function; the stored data area may store data created based on the use of the distribution network-based energy storage planning and deployment system. Furthermore, the computer-readable storage medium may include high-speed random access memory, and may also include memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the computer-readable storage medium may optionally include memory remotely located relative to a processor, and these remote memories can be connected to the distribution network-based energy storage planning and deployment system via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0105] Figure 3 This is a schematic diagram of the structure of the electronic device provided in the embodiment of the present invention, such as... Figure 3 As shown, the device includes a processor 310 and a memory 320. The electronic device may also include an input device 330 and an output device 340. The processor 310, memory 320, input device 330, and output device 340 can be connected via a bus or other means. Figure 3 Taking a bus connection as an example, the memory 320 is the computer-readable storage medium described above. The processor 310 executes various server functions and data processing by running non-volatile software programs, instructions, and modules stored in the memory 320, thereby realizing the distribution network-based energy storage planning and deployment method described in the above embodiment. The input device 330 can receive input digital or character information and generate key signal inputs related to user settings and function control of the distribution network-based energy storage planning and deployment system. The output device 340 may include a display screen or other display device.

[0106] The aforementioned electronic device can execute the method provided in the embodiments of the present invention, and has the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the method provided in the embodiments of the present invention.

[0107] In one implementation, the above-described electronic device is applied in a distribution network-based energy storage planning and deployment system for a client, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:

[0108] Acquire historical meteorological disaster data, power grid data, and energy storage parameter data for the target area;

[0109] Based on the historical meteorological disaster data, power grid data, and energy storage parameter data, a three-layer coupled stochastic optimization model is constructed to collaboratively optimize energy storage configuration, system operation under disasters, and fault repair processes. The three-layer coupled stochastic optimization model includes:

[0110] A high-level planning model that uses the location and capacity of energy storage systems as decision variables and aims to minimize the expected total life-cycle cost of the distribution network.

[0111] This is a mid-level operation model that takes the energy storage configuration of the upper-level planning model as input, aims to minimize the sum of system operating cost and load loss penalty cost under extreme disaster scenarios, and satisfies the power grid safe operation constraints and the operation constraints of grid-type energy storage during the post-disaster recovery period.

[0112] A lower-level recovery scheduling model that aims to minimize the resource scheduling cost for repairing faulty components and satisfies the logical constraints of repair team scheduling and fault repair progress.

[0113] The three-layer coupled stochastic optimization model is solved using a multi-agent deep reinforcement learning algorithm, and the optimal grid-type energy storage deployment scheme and collaborative recovery strategy are output.

[0114] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0115] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for planning and deploying grid-type energy storage in power distribution networks, characterized in that, include: Acquire historical meteorological disaster data, power grid data, and energy storage parameter data for the target area; Based on the historical meteorological disaster data, power grid data, and energy storage parameter data, a three-layer coupled stochastic optimization model is constructed to collaboratively optimize energy storage configuration, system operation under disasters, and fault repair processes. The three-layer coupled stochastic optimization model includes: A high-level planning model that uses the location and capacity of energy storage systems as decision variables and aims to minimize the expected total life-cycle cost of the distribution network. This is a mid-level operation model that takes the energy storage configuration of the upper-level planning model as input, aims to minimize the sum of system operating cost and load loss penalty cost under extreme disaster scenarios, and satisfies the power grid safe operation constraints and the operation constraints of grid-type energy storage during the post-disaster recovery period. A lower-level recovery scheduling model that aims to minimize the resource scheduling cost for repairing faulty components and satisfies the logical constraints of repair team scheduling and fault repair progress. A multi-agent deep reinforcement learning algorithm is used to solve the three-layer coupled stochastic optimization model, outputting the optimal grid-type energy storage deployment scheme and collaborative recovery strategy, specifically including: Construct a planning intelligent agent, wherein the observation space of the planning intelligent agent is the static characteristics of the power grid, which include topological connection relationships, historical average load, installed capacity of distributed power sources, and spatial distribution of disaster risks; The action space of the planning agent is the energy storage deployment scheme, which outputs the installation status, rated energy capacity and rated power capacity of each candidate node. Construct an operation-recovery collaborative intelligent agent, wherein the observation space of the operation-recovery collaborative intelligent agent is the post-disaster dynamic state, which includes real-time topological connectivity, load and power output, energy storage charging status, and the location and working status of the emergency repair team. The action space of the operation-recovery collaborative intelligent agent is the collaborative control command, which jointly outputs commands for energy storage charging and discharging power, network switching operation, load reduction, and emergency repair task allocation. Initialize the policy network parameters of the planning agent and the policy network parameters of the run-recovery cooperative agent; In each training round, a disaster scenario is randomly selected, and the planning agent generates an energy storage configuration scheme. Based on the energy storage configuration scheme and the selected scenario, the operation-recovery collaborative intelligent agent simulates the post-disaster operation and recovery process and generates a sequence of control and scheduling actions. Calculate the joint reward based on the system state trajectory generated by the agent's interaction; By using reinforcement learning algorithms, the policy network parameters of the planning agent and the policy network parameters of the run-recovery cooperative agent are updated simultaneously based on the joint reward to maximize the long-term cumulative reward. The system is repeatedly updated. After the policy network parameters converge, the optimal grid-type energy storage deployment scheme is decoded from the planning agent, and the collaborative recovery strategy for various disaster scenarios is decoded from the operation-recovery collaborative agent strategy.

2. The method for planning and deploying grid-type energy storage in a distribution network according to claim 1, characterized in that, The objective function of the upper-level planning model is: ， ， In the formula, This represents the expected total cost over the entire lifecycle of the distribution network. The annual value cost of energy storage system investment, The system operating cost in scenario k, The load loss penalty cost in scenario k. The cost of resource scheduling for fault repair in scenario k. For a collection of energy storage installation nodes, Cost per unit energy capacity The rated capacity for energy storage for node i. Cost per unit power capacity The rated power for energy storage at node i. For fixed installation costs per site, This is a binary decision variable, indicating whether energy storage should be installed at node i, with 1 for installation and 0 for no installation. The discount rate is... This is the discount rate.

3. The method for planning and deploying grid-type energy storage in a distribution network according to claim 1, characterized in that, The objective function of the intermediate-level operating model is: ， In the formula, The system operating cost in scenario k, The load loss penalty cost in scenario k. For time step, The price at which electricity is purchased from the upper-level power grid or emergency power source at time t. Let be the active power purchased from an external power source at time t. This is the network loss cost coefficient. branch road The resistance, For scenario k, the branch through which the flow occurs at time t The current amplitude, This is a set of time steps for the post-disaster recovery period. For the set of all nodes in the distribution network, For branch road collection, The penalty cost for reducing unit load. Let be the load reduction amount at node i at time t in scenario k.

4. The method for planning and deploying grid-type energy storage in a distribution network according to claim 1, characterized in that, The power grid safety operation constraints include node power balance constraints and branch power flow and voltage constraints. The expression for the node power balance constraint is as follows: ， ， In the formula, Let be the active power of branch m at node i at time t. The active power output of the distributed power source at node i. Let be the active load of node i. Net active power is injected into the energy storage of node i. Let be the reactive power of branch m at node i at time t. For node i, the reactive power output of the distributed power source. Let i be the reactive load. Net reactive power is injected into the energy storage of node i. For the set of all nodes in the distribution network, For branch road collection, This is a set of time steps for the post-disaster recovery period; The expressions for the branch power flow and voltage constraints are as follows: ， ， ， In the formula, Let i be the voltage amplitude at node i. Let j be the voltage amplitude at node j. Let (i,j) be the resistance of branch (i,j). Let be the active power of branch (i,j). Let (i,j) be the reactance of branch (i,j). Let (i,j) be the reactive power of the branch. Let (i,j) be the current in branch (i,j). This is the lower limit of the node voltage amplitude. This represents the upper limit of the node voltage amplitude. The expression for the operational constraints of the grid-type energy storage is: ， ， ， ， ， ， ， In the formula, Net active power is injected into the energy storage of node i. For node i, the energy storage and discharge power. The energy storage charging power for node i, This is a charging status indicator. The rated power for energy storage at node i Let i be the state of charge of the energy storage at time t. Let i be the state of charge of the energy storage at time t-1. For energy storage charging efficiency, For energy storage and discharge efficiency, The rated capacity for energy storage for node i. For time step, This is the lower limit of SOC. This is the upper limit of SOC. for The state of charge of energy storage i at any given time. This represents the initial state of charge of the energy storage. The limit for SOC variation at the beginning and end of the cycle. To enable network 0 / 1 variables, This is the lower limit for reactive power adjustment under grid-connected mode. Net reactive power is injected into the energy storage of node i. This is the upper limit for reactive power adjustment under the grid-connected mode.

5. The method for planning and deploying grid-type energy storage in a distribution network according to claim 1, characterized in that, The objective function of the lower-level recovery scheduling model is: ， In the formula, Let k be the resource scheduling cost for fault repair. To allow the repair team to assemble, This is a set of time steps for the post-disaster recovery period. Let r be the unit movement cost of the team. Let r be the distance traveled by team r at time t. Let r be the unit work cost of team r. This represents the average working hours of the team.

6. The method for planning and deploying grid-type energy storage in a distribution network according to claim 1, characterized in that, The expression for the logical constraints on the dispatching of the emergency repair team and the progress of fault repair is as follows: ， ， In the formula, To determine if the variable was successfully repaired, check whether component e has been repaired at time t. To plan a set of time periods, To extend the repair time, For components The moment the emergency repairs began Whether team r is assigned a repair component at time t. , To allow the repair team to assemble, For team r, there is a set of repairable components.

7. The method for planning and deploying grid-type energy storage in a distribution network according to claim 1, characterized in that, The expression for the joint reward is: ， ， In the formula, For joint awards, The weighting coefficient is the annual value cost. The annual value cost of energy storage system investment, For system operation and compliance loss cost weighting coefficients, The system operating cost in scenario k, The load loss penalty cost in scenario k. These are the weighting coefficients. Let k be the resource scheduling cost for fault repair. For resilience reward weighting coefficient, As a resilience bonus value, It is a collection of typical extreme disaster scenarios. The weight of the resilience index is the value of the power loss. This represents the power loss value in scenarios without energy storage solutions. As a weight for the resilience index of average power restoration time, For scenario k without energy storage, the average system power recovery time is... For the scene The power loss value under the following conditions For the scene Average power restoration time of the system.

8. A distribution network-based energy storage planning and deployment system, used to implement the method described in any one of claims 1-7, characterized in that, The system includes: The acquisition module is configured to acquire historical meteorological disaster data, power grid data, and energy storage parameter data for the target area. The construction module is configured to build a three-layer coupled stochastic optimization model based on the historical meteorological disaster data, power grid data, and energy storage parameter data. This model collaboratively optimizes energy storage configuration, system operation under disasters, and fault repair processes. The three-layer coupled stochastic optimization model includes: A high-level planning model that uses the location and capacity of energy storage systems as decision variables and aims to minimize the expected total life-cycle cost of the distribution network. This is a mid-level operation model that takes the energy storage configuration of the upper-level planning model as input, aims to minimize the sum of system operating cost and load loss penalty cost under extreme disaster scenarios, and satisfies the power grid safe operation constraints and the operation constraints of grid-type energy storage during the post-disaster recovery period. A lower-level recovery scheduling model that aims to minimize the resource scheduling cost for repairing faulty components and satisfies the logical constraints of repair team scheduling and fault repair progress. The solution module is configured to use a multi-agent deep reinforcement learning algorithm to solve the three-layer coupled stochastic optimization model and output the optimal grid-type energy storage deployment scheme and collaborative recovery strategy.

9. An electronic device, characterized in that, include: At least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method according to any one of claims 1 to 7.

Images