Intelligent power grid optimal scheduling method based on distributed accelerated convergence algorithm

Abstract

The invention discloses an intelligent power grid optimal scheduling method based on a distributed accelerated convergence algorithm. The method comprises the following steps: carrying out system modeling and decomposition; carrying out agent design and initialization, local optimization and information exchange; convergence is accelerated in a distributed mode, meanwhile, an adaptive adjustment mechanism is introduced, and acceleration parameters are dynamically adjusted according to the convergence history of the algorithm; setting a conflict detection solution mechanism; globally coordinating and updating; and judging a termination condition. The method has good fault tolerance and adaptability, can be suitable for real-time optimization scheduling of a large-scale smart power grid, effectively improves the calculation speed and accuracy, reduces the operation and maintenance cost, and enhances the robustness of the system.

Description

Technical Field

[0001] The present invention belongs to the technical field of smart grids, and in particular relates to a smart grid optimization scheduling method based on a distributed accelerated convergence algorithm. Background Art

[0002] With the development of smart grid technology, the power grid system has become more complex and dynamic. In order to balance the supply and demand of electricity, reduce operating costs, improve energy efficiency and ensure the safety of the power grid, it is necessary to optimize the power grid. However, the traditional centralized optimization scheduling method can no longer meet the growing demand for smart grid scheduling due to its shortcomings such as large computational load, poor real-time performance and difficulty in expansion. Distributed optimization algorithms are gaining attention due to their advantages in computational efficiency, scalability and robustness. Such algorithms can achieve collaborative optimization of the entire network by performing local calculations in different areas of the power grid and exchanging information through communication networks. However, the existing distributed algorithms still have room for improvement in terms of convergence speed and real-time performance.

[0003] Therefore, a new smart grid optimization scheduling method based on distributed accelerated convergence algorithm is urgently needed. Summary of the invention

[0004] The purpose of this invention is to address the above problems and propose a smart grid optimization scheduling method based on a distributed accelerated convergence algorithm (D2-SAGA algorithm), aiming to improve the existing technology, accelerate the convergence speed of the optimization process, and ensure the accuracy and robustness of the optimization results.

[0005] In order to solve at least one of the above technical problems, according to one aspect of the present invention, a smart grid optimization scheduling method based on a distributed accelerated convergence algorithm is provided, comprising the following steps:

[0006] System modeling and decomposition;

[0007] Agent design and initialization, local optimization and information exchange;

[0008] Distributed acceleration convergence, while introducing an adaptive adjustment mechanism to dynamically adjust acceleration parameters based on the convergence history of the algorithm;

[0009] Set up conflict detection and resolution mechanisms;

[0010] Global coordination and update; coordinate the calculation results between agents through a central control center or a distributed consensus algorithm to ensure global optimization; update the local calculation results of the agent based on global information;

[0011] Termination condition determination: when the algorithm reaches the preset convergence criterion or number of iterations, the optimization process is terminated and the final power grid dispatching plan is output.

[0012] Furthermore, the specific steps of system modeling and decomposition are: modeling the smart grid optimization problem as a multi-objective optimization problem; using graph theory to decompose the grid topology into various sub-areas, each of which is responsible for local optimization calculations by an agent.

[0013] Furthermore, the specific steps of agent design and initialization, local optimization and information exchange are as follows:

[0014] Each agent is designed with its own objective function, constraints and state variables; Initialize the state variables of each agent;

[0015] Local optimization is performed independently by each grid node, and global coordination is gradually achieved by optimizing state variables and exchanging information between neighboring nodes.

[0016] Furthermore, the distributed acceleration convergence introduces an adaptive adjustment mechanism, and the specific steps of dynamically adjusting the acceleration parameters according to the convergence history of the algorithm are as follows:

[0017] Take each node i as a vector Start and update, for all t ≥ 0, first calculate the local gradient estimator:

[0018]

[0019] in, is a gradient list of batch samples. The gradient estimator only needs to calculate the gradient value once each time it calculates the gradient estimator. However, the gradient estimator is an offline gradient estimator and needs to store a gradient list. Each gradient in the list corresponds to each sample. If the sample is extracted in this iteration And calculated the latest gradient value of the sample Then the historical gradient of the sample corresponding to the gradient list needs to be Use New Gradient Replace;

[0020] Select the exact diffusion iteration method:

[0021]

[0022] Finally, information exchange between nodes is performed:

[0023]

[0024] where α t is the step length, is the sample randomly drawn from the local node i at the tth iteration, is the sample randomly drawn from node i at the tth iteration. The gradient value when , and satisfies w ij is the communication weight value from node i to node j in the communication network.

[0025] Furthermore, the specific steps for setting the conflict detection and resolution mechanism are:

[0026] Real-time monitoring of power grid status: Use sensors and monitoring equipment to collect line power flow P in real time line , node voltage V i , power generation P gen and load demand P load ; Check if there is any conflict: Line overload: Power flow P line >P max ; Voltage out of limit: Node voltage V i Does not meet V min ≤V i ≤V max ; Power imbalance: The deviation between the generated power and the load demand exceeds the tolerance |P gen -P load |>∈;

[0027] Conflict classification and prioritization: Classify issues according to the type of conflict detected and assign priorities;

[0028] Set conflict resolution strategies: power allocation adjustment, for overloaded lines or over-limit nodes, redistribute edge rates through optimization algorithms to reduce overload or over-limit levels; load reduction, when power is unbalanced or severely overloaded, give priority to reducing non-critical loads to ensure normal power supply to critical loads; introduce slack variables, introduce slack variables in the optimization model, and relax constraints;

[0029] Verification after conflict resolution: After the conflict is resolved, recalculate the grid state to verify whether all constraints are met; if there is still a conflict, continue iterative adjustment until the conflict is resolved or the optimization termination condition is reached.

[0030] Furthermore, the specific steps for global coordination and updating are:

[0031] Global information collection: each node sends its local optimization results to the central control center through the communication network, or gradually aggregates them through neighboring nodes to achieve distributed information sharing;

[0032] Construct a global objective function. Construct a global objective function based on the state of the entire network.

[0033] Global optimization algorithm: The central control center constructs the global optimization problem and broadcasts the optimization results to each node after solving it. Each node gradually achieves the consistency of the global state by exchanging information with neighboring nodes.

[0034] Global coordination result distribution: distribute the global optimization results to each node, and each node adjusts its operation status accordingly;

[0035] Dynamic adjustment and feedback, real-time monitoring of the grid operation status, and detection of the actual effect of the coordination results; if new conflicts or imbalances are found, local or global optimization iterations are initiated.

[0036] Furthermore, the specific steps of determining the termination condition are:

[0037] Set the convergence criteria. When the change in the objective function meets the following conditions, the optimization is considered to have converged:

[0038] ΔJ global =J global,t+1 -J global,t |<∈ global ;

[0039] Verify whether all constraints are met; if not, continue optimization;

[0040] Iteration limit, if the maximum number of iterations T is reached max , forcefully terminate the optimization process;

[0041] Result output: output the final optimization result.

[0042] According to another aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, the steps of the smart grid optimization scheduling method based on a distributed accelerated convergence algorithm of the present invention are implemented.

[0043] According to another aspect of the present invention, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the smart grid optimization scheduling method based on a distributed accelerated convergence algorithm of the present invention when executing the program.

[0044] Compared with the prior art, the above method of the present invention has the following beneficial effects:

[0045] 1. The present invention reconstructs the real form of the coarse aggregate of the pavement material based on the discrete element theory, and distributes the dielectric properties of irregular shapes such as pores and mortar based on the random medium theory, so the model settlement results are more realistic;

[0046] 2. The present invention is a three-dimensional, multi-phase, inhomogeneous electromagnetic differential model that can obtain the full-section spectrum characteristics of highway structural defect echoes under inhomogeneous interference.

[0047] 1. Distributed computing: The use of distributed algorithms can disperse large-scale computing tasks to multiple computing nodes, which can significantly improve computing efficiency and reduce the computing pressure of a single node.

[0048] 2. Accelerated convergence: Through improved algorithm design, the convergence speed of the algorithm can be accelerated, and the solution to the optimization problem can be quickly found or the optimal solution can be approached, which is especially important for real-time or near real-time optimization scheduling.

[0049] 3. Scalability: Distributed algorithms usually have good scalability. When the scale of the power grid increases, the algorithm performance will not decrease significantly, which is crucial for the ever-expanding smart grid.

[0050] 4. Flexibility and robustness: When faced with changes in grid load, equipment failures, etc., distributed algorithms can adapt to these changes more flexibly and ensure the stable operation of the grid.

[0051] 5. Support multi-objective optimization: Distributed algorithms can set multiple optimization goals, such as cost minimization, emission reduction, service quality improvement, etc., and achieve comprehensive optimization by balancing these goals.

[0052] 6. Improve system safety: Through effective optimization and scheduling, the risk of system operation can be reduced, the safety of the entire power grid can be improved, and large-scale power outages can be avoided. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0054] Figure 1 It is a flow chart of a method in a preferred embodiment provided by the present invention;

[0055] Figure 2 In the preferred embodiment provided by the present invention;

[0056] Figure 3 It is a preferred embodiment provided by the present invention. DETAILED DESCRIPTION

[0057] To make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the embodiments of the present invention will be described clearly and completely in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them.

[0058] Unless otherwise defined, technical or scientific terms used herein shall have the same meaning as commonly understood by one of ordinary skill in the art to which the invention belongs.

[0059] Example 1: Figure 1-3 As shown, a smart grid optimization scheduling method based on a distributed accelerated convergence algorithm includes the following steps:

[0060] Step 1: System modeling and decomposition;

[0061] First, the smart grid optimization problem is modeled as a multi-objective optimization problem. Then, the grid topology is decomposed into sub-areas using graph theory, and each sub-area is responsible for local optimization calculations by an agent.

[0062] Step 2: Agent design and initialization, local optimization and information exchange;

[0063] Each agent is designed with its own objective function, constraints and state variables. Initialize the state variables of each agent, including power generation, consumption, storage and transmission.

[0064] Local optimization and information exchange:

[0065] Local optimization is performed independently by each grid node, which gradually achieves global coordination by optimizing its state variables such as power output and exchanging information between neighboring nodes.

[0066] The specific process is as follows:

[0067] 1. Node initialization

[0068] Each node (such as a power station, substation or distributed energy) initializes the state variable x0 (such as power output, storage status, etc.) and reads the initial parameters of the node, including: power demand forecast D i , power generation cost coefficient a i ,b i ,c i 、Power upper and lower limits P min ,P max , boundary power flow P line Etc. Set the learning rate α, convergence threshold ∈, maximum number of iterations T max .

[0069] 2. Construct optimization objective function

[0070] Each node constructs an objective function based on its function and constraints. For example, the goal of a power station is to minimize the cost of generating electricity and

[0071] Meeting power output limits:

[0072] J i (x) = a i x 2 +b i x+c i +λ[max(0,xPmax )2+max(0,P min -x) 2 ]

[0073] Among them, a i ,b i ,c i is the power generation cost coefficient, and λ is the penalty coefficient.

[0074] 3. Gradient descent for optimization

[0075]

[0076] Use the gradient descent method to gradually optimize the power output, and the update formula is:

[0077] The gradient formula is:

[0078]

[0079] Wherein, H(·) is a step function, which is used to detect whether the power exceeds the upper and lower limits.

[0080] 4. Information exchange between nodes

[0081] Each node sends its own boundary power flow P to neighboring nodes through the communication network line and the current electricity price λ, and receives relevant information of neighbor nodes at the same time. Based on the received neighbor information, the objective function and boundary constraints are dynamically adjusted to ensure that local optimization is consistent with global coordination.

[0082] 5. Termination Conditions

[0083] The optimization stops when any of the following conditions is met:

[0084] Gradient norm

[0085] The change of objective function |J i (x t+1 )-J i (x t )|<δ;

[0086] Reach the maximum number of iterations T max .

[0087] Step 3: Distributed accelerated convergence;

[0088] A distributed accelerated convergence algorithm is used to improve the convergence speed of the algorithm. On this basis, an adaptive adjustment mechanism is introduced to dynamically adjust the acceleration parameters according to the convergence history of the algorithm. The distributed accelerated convergence algorithm proposes a distributed random algorithm that can eliminate random variance and overcome data heterogeneity. The algorithm combines the advantages of the exact diffusion algorithm and the variance reduction algorithm. The calculation process is as follows:

[0089] Take each node i as a vector Start and update, for all t ≥ 0, first calculate the local gradient estimator:

[0090]

[0091] The function of this gradient estimator is to accelerate the convergence speed of the algorithm while eliminating random variance and improving the convergence stability. is a gradient list of batch samples. The gradient estimator only needs to calculate the gradient value once each time it calculates the gradient estimator. However, the gradient estimator is an offline gradient estimator and needs to store a gradient list. Each gradient in the list corresponds to each sample. If the sample is extracted in this round of iteration And calculated the latest gradient value of the sample Then the historical gradient of the sample corresponding to the gradient list needs to be Using the new gradient, Replace it, the replacement diagram is as follows Figure 2 shown.

[0092] In order to overcome the influence of data heterogeneity, the gradient descent iteration of the traditional distributed stochastic gradient descent (DSGD) Change to an accurate diffusion iteration method with correction function:

[0093]

[0094] This iterative method can eliminate the influence of data heterogeneity and improve convergence accuracy. Finally, information exchange between nodes is performed:

[0095]

[0096] where α t is the step length, is the sample randomly drawn from the local node i at the tth iteration, is the sample randomly drawn from node i at the tth iteration. The gradient value when , and satisfies w ij is the communication weight value from node i to node j in the communication network. For a random communication network, Figure 3 shown. Figure 3 Each node (represented by a circle) represents a grid unit, such as a power station, substation, or load node. The number inside the node represents the node number in the star topology, which is used to uniquely identify the grid unit and may be related to its physical location or logical order. The connection between nodes represents the information exchange path, and the weight w on the connection line ij , w ij Reflects the information exchange intensity or dependency between node i and node j.

[0097] Step 4: Conflict detection and resolution;

[0098] The goal of conflict detection and resolution is to ensure that the power grid operates safely and meets all constraints (such as power flow limit, voltage range, power balance). The specific process is as follows:

[0099] 1. Real-time monitoring of power grid status

[0100] Use sensors and monitoring equipment (such as PMU, wide area measurement unit) to collect line power flow P in real time line , node voltage V i , power generation P gen and load demand P load .

[0101] Check for conflicts with:

[0102] Line overload: Power flow P line >P max .

[0103] Voltage out of limit: Node voltage V i Does not meet V min ≤V i ≤V max .

[0104] Power imbalance: The deviation between the generated power and the load demand exceeds the tolerance |P gen -P load |>∈.

[0105] 2. Conflict classification and prioritization

[0106] Issues are categorized and prioritized based on the type of conflict detected:

[0107] Level 1 conflict (high priority): The line is severely overloaded, the voltage exceeds the limit (such as V i >1.1V nominal ), serious power imbalance and failure of key equipment.

[0108] Second level conflict (medium priority): Slight line overload, voltage deviation from normal range (such as 0.95V nominal ≤Vi <V min ).

[0109] Level 3 conflict (low priority): Minor power imbalance or fluctuation of non-critical load.

[0110] 3. Conflict Resolution Strategies

[0111] Power allocation adjustment: For overloaded lines or out-of-limit nodes, the edge rate is redistributed through optimization algorithms to reduce the overload or out-of-limit degree.

[0112] Load shedding: In the event of power imbalance or severe overload, non-critical loads (such as commercial lighting) are prioritized for shedding to ensure normal power supply to critical loads (such as medical facilities).

[0113] Introducing slack variables: Introducing slack variables in the optimization model to relax the constraints, for example:

[0114] P line ≤P max +∈where ∈≥0 is a slack variable whose value is dynamically adjusted according to the conflict situation.

[0115] 4. Verify after conflict resolution

[0116] After the conflict is resolved, recalculate the grid state to verify that all constraints are met

[0117] If conflicts still exist, continue iterating until the conflicts are resolved or the optimization termination condition is reached.

[0118] Step 5: Global coordination and update;

[0119] The calculation results of each agent are coordinated through a central control center or a distributed consensus algorithm to ensure global optimization. The local calculation results of the agent are updated according to the global information.

[0120] 1. Global information collection

[0121] Each node sends its local optimization results (such as power output, electricity price, etc.) to the central control center through the communication network, or gradually aggregates them through neighboring nodes to achieve distributed information sharing.

[0122] The information includes: local objective function value, constraint status (such as power flow, voltage), etc.

[0123] 2. Constructing the global objective function

[0124] According to the state of the entire network, a global objective function is constructed, for example:

[0125]

[0126] Among them, J i(x) is the local objective function of node i, and β is the penalty coefficient.

[0127] 3. Global Optimization Algorithm

[0128] Central optimization algorithm: The central control center constructs the global optimization problem and broadcasts the optimization results to each node after solving it.

[0129] Distributed consensus algorithm: Each node gradually achieves consistency in the global state by exchanging information with neighboring nodes.

[0130] 4. Global coordination results distribution

[0131] The global optimization results (such as power allocation, voltage adjustment, etc.) are distributed to each node, and each node adjusts its operating status accordingly.

[0132] 5. Dynamic adjustment and feedback

[0133] Monitor the grid operation status in real time and detect the actual effect of the coordination results. If new conflicts or imbalances are found, start local or global optimization iterations.

[0134] Step 6: Determine the termination condition;

[0135] When the algorithm reaches the preset convergence criterion or number of iterations, the optimization process is terminated and the final power grid dispatching plan is output.

[0136] The termination conditions of the optimization process are as follows:

[0137] 1. Convergence Criteria

[0138] The optimization is considered to have converged when the objective function changes to the following conditions:

[0139] ΔJ global =J global,t+1 -J global,t |<∈ global

[0140] 2. Constraint Satisfaction

[0141] Verify that all constraints are met (such as power flow, voltage range, etc.). If not, continue optimization.

[0142] 3. Iteration limit

[0143] If the maximum number of iterations T is reached max , forcibly terminate the optimization process.

[0144] 4. Result Output

[0145] Output the final optimization results, including power distribution, electricity price, voltage status, power generation cost, line loss, etc.

[0146] In summary, the distributed accelerated convergence algorithm of the present invention adopts advanced optimization theory and control strategy to ensure the rapid convergence and high stability of the algorithm. At the same time, the present invention has good fault tolerance and adaptability, can be applied to the real-time optimization scheduling of large-scale smart grids, effectively improves the calculation speed and accuracy, reduces the operation and maintenance costs, and enhances the robustness of the system.

[0147] Embodiment 2:

[0148] The computer-readable storage medium of this embodiment stores a computer program thereon, and when the program is executed by a processor, the steps in the smart grid optimization scheduling method based on a distributed accelerated convergence algorithm of Example 1 are implemented.

[0149] The computer-readable storage medium of this embodiment may be an internal storage unit of the terminal, such as a hard disk or memory of the terminal; the computer-readable storage medium of this embodiment may also be an external storage device of the terminal, such as a plug-in hard disk, a smart memory card, a secure digital card, a flash memory card, etc. equipped on the terminal; further, the computer-readable storage medium may also include both an internal storage unit of the terminal and an external storage device.

[0150] The computer-readable storage medium of this embodiment is used to store computer programs and other programs and data required by the terminal. The computer-readable storage medium can also be used to temporarily store data that has been output or is to be output.

[0151] Embodiment 3:

[0152] The computer device of this embodiment includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of the smart grid optimization scheduling method based on a distributed accelerated convergence algorithm of Embodiment 1 are implemented.

[0153] In this embodiment, the processor may be a central processing unit, or other general-purpose processors, digital signal processors, application-specific integrated circuits, readily available programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A portion of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0154] Those skilled in the art will appreciate that the disclosed content of the embodiments may be provided as methods, systems, or computer program products. Therefore, the present solution may adopt the form of hardware embodiments, software embodiments, or embodiments combining software and hardware. Moreover, the present solution may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) containing computer-usable program codes.

[0155] The present solution is described with reference to the flowchart and / or block diagram of the method and computer program product according to the embodiment of the present solution. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions; these computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0156] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0157] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0158] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0159] The examples described in the present invention are merely descriptions of the preferred implementation modes of the present invention, and are not intended to limit the concept and scope of the present invention. Without departing from the design concept of the present invention, various modifications and improvements made to the technical solutions of the present invention by engineers and technicians in this field should all fall within the protection scope of the present invention.

Claims

1. A smart grid optimization dispatching method based on a distributed accelerated convergence algorithm, characterized in that: The steps include: System modeling and decomposition; Agent design and initialization, local optimization and information exchange; Distributed acceleration convergence, while introducing an adaptive adjustment mechanism to dynamically adjust acceleration parameters based on the convergence history of the algorithm; Set up conflict detection and resolution mechanisms; Global coordination and update; coordinate the calculation results between agents through a central control center or a distributed consensus algorithm to ensure global optimization; Update the agent's local computation results based on global information; Termination condition determination: when the algorithm reaches the preset convergence criterion or number of iterations, the optimization process is terminated and the final power grid dispatching plan is output.

2. The method according to claim 1, characterized in that The specific steps of system modeling and decomposition are: modeling the smart grid optimization problem as a multi-objective optimization problem; using graph theory to decompose the grid topology into various sub-areas, each of which is responsible for local optimization calculations by an agent.

3. The method according to claim 2, characterized in that The specific steps of agent design and initialization, local optimization and information exchange are: Each agent is designed with its own objective function, constraints and state variables; Initialize the state variables of each agent; Local optimization is performed independently by each grid node, and global coordination is gradually achieved by optimizing state variables and exchanging information between neighboring nodes.

4. The method according to claim 3, characterized in that Distributed acceleration convergence, while introducing an adaptive adjustment mechanism, dynamically adjusts the acceleration parameters according to the convergence history of the algorithm. The specific steps are: Take each node i as a vector Start and update, for all t ≥ 0, first calculate the local gradient estimator: in, is a gradient list of batch samples. The gradient estimator only needs to calculate the gradient value once each time it calculates the gradient estimator. However, the gradient estimator is an offline gradient estimator and needs to store a gradient list. Each gradient in the list corresponds to each sample. If the sample is extracted in this iteration And calculated the latest gradient value of the sample Then the historical gradient of the sample corresponding to the gradient list needs to be Use New Gradient Replace; Select the exact diffusion iteration method: Finally, information exchange between nodes is performed: where α t is the step length, is the sample randomly drawn from the local node i at the tth iteration, is the sample randomly drawn from node i at the tth iteration. The gradient value when , and satisfies w ij is the communication weight value from node i to node j in the communication network.

5. The method according to claim 4, characterized in that The specific steps to set up the conflict detection and resolution mechanism are: Real-time monitoring of power grid status: Use sensors and monitoring equipment to collect line power flow P in real time line , node voltage V i , power generation P gen and load demand P load ; Check if there is any conflict: Line overload: Power flow P line >P max ; Voltage out of limit: Node voltage V i Does not meet V min ≤V i ≤V max ; Power imbalance: The deviation between the generated power and the load demand exceeds the tolerance |P gen -P load |>∈; Conflict classification and prioritization: Classify issues according to the type of conflict detected and assign priorities; Set conflict resolution strategy: adjust power distribution, and redistribute edge rates for overloaded lines or over-limited nodes through optimization algorithms to reduce the overload or over-limit degree; Load reduction: When power is unbalanced or seriously overloaded, non-critical loads are reduced first to ensure normal power supply to critical loads. Slack variables are introduced into the optimization model to relax constraints. Verification after conflict resolution: After the conflict is resolved, recalculate the grid state to verify whether all constraints are met; if there is still a conflict, continue iterative adjustment until the conflict is resolved or the optimization termination condition is reached.

6. The method according to claim 5, characterized in that The specific steps for global coordination and updating are: Global information collection: each node sends its local optimization results to the central control center through the communication network, or gradually aggregates them through neighboring nodes to achieve distributed information sharing; Construct a global objective function. Construct a global objective function based on the state of the entire network. Global optimization algorithm: The central control center constructs the global optimization problem and broadcasts the optimization results to each node after solving it. Each node gradually achieves the consistency of the global state by exchanging information with neighboring nodes. Global coordination result distribution: distribute the global optimization results to each node, and each node adjusts its operation status accordingly; Dynamic adjustment and feedback, real-time monitoring of the grid operation status, and detection of the actual effect of the coordination results; if new conflicts or imbalances are found, local or global optimization iterations are initiated.

7. The method according to claim 6, characterized in that The specific steps for determining the termination condition are: Set the convergence criteria. When the change in the objective function meets the following conditions, the optimization is considered to have converged: ΔJ global =J global,t+1 -J global,t |<∈ global ; Verify whether all constraints are met; if not, continue optimization; Iteration limit, if the maximum number of iterations T is reached max , forcefully terminate the optimization process; Result output: output the final optimization result.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the smart grid optimization scheduling method based on a distributed accelerated convergence algorithm as described in any one of claims 1 to 7 are implemented.

9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the smart grid optimization scheduling method based on a distributed accelerated convergence algorithm as described in any one of claims 1 to 7 are implemented.

Images