Optimal distribution method and system for distribution network resources

Abstract

The invention relates to an optimal distribution method and system for distribution network resources, and the method comprises the following steps: S1, fusing user power failure information with ECS event information, positioning the incidence relation between a user side and a distribution network power failure event, fusing the synchronous phasor of a PMU and the operation state of an SCADA, and constructing a global data set; s2, based on the global data set, bringing recovery speed, electric energy loss, important user priority, distributed power utilization and environmental benefit targets into an optimization framework, and establishing a dynamic multi-target optimization model; s3, obtaining an optimal strategy based on a multilayer optimization algorithm according to the dynamic multi-objective optimization model; s4, dividing the distribution network into a plurality of functional areas according to the optimal strategy, and respectively making specific scheduling measures; and S5, based on a specific scheduling measure, issuing an optimization result to an equipment layer through the DMS, monitoring whether a scheduling action is completed according to a plan, and obtaining execution feedback. According to the invention, the intelligence, efficiency and reliability of distribution network operation can be obviously improved.

Description

Technical Field

[0001] The present invention relates to the field of distribution network resource allocation, and in particular to a distribution network resource optimization allocation method and system. Background Art

[0002] With the continuous development of power systems and the widespread access of distributed generation (DG), the operating environment of distribution networks has become increasingly complex. In this context, the challenges faced by power grids mainly include the increase in user power supply reliability requirements, the diversification of load demand characteristics, the increase in volatility of distributed generation and energy storage equipment, and the balance between economic efficiency and environmental benefits in power grid operation. In addition, the traditional distribution network resource scheduling method has the following shortcomings: First, it is difficult to integrate multi-source heterogeneous data (such as power outage event information, PMU synchronous phasors, SCADA operating status, etc.), resulting in obvious deficiencies in the precise positioning of power outage events and the comprehensive grasp of the entire network situation; second, the existing scheduling strategies are mostly based on static optimization, which makes it difficult to dynamically respond to real-time changes in the operating status of the power grid; third, there is a lack of a multi-objective optimization framework, which makes it difficult to simultaneously take into account multiple requirements such as power supply recovery speed, power loss, priority of important users, utilization of distributed generation and environmental benefits; fourth, the feedback mechanism between the execution layer and the optimization layer is weak, making it difficult to form a closed-loop control process. Summary of the invention

[0003] In order to solve the above problems, the purpose of the present invention is to provide a distribution network resource optimization allocation method and system, which integrates multi-source data, dynamic multi-objective optimization and intelligent control to optimize the distribution network resource allocation method, and significantly improve the intelligence, efficiency and reliability of distribution network operation.

[0004] To achieve the above object, the present invention adopts the following technical solutions:

[0005] A method for optimizing the allocation of distribution network resources comprises the following steps:

[0006] S1: Integrate the user power outage information with the ECS event information, locate the correlation between the user side and the distribution network power outage event, and integrate the PMU's synchronous phasor and the operating status of SCADA to build a global data set;

[0007] S2: Based on the global data set, the restoration speed, power loss, priority of important users, utilization of distributed power sources, and environmental benefit goals are incorporated into the optimization framework to establish a dynamic multi-objective optimization model;

[0008] S3: Obtain the optimal strategy based on the dynamic multi-objective optimization model and multi-layer optimization algorithm;

[0009] S4: According to the optimal strategy, the distribution network is divided into several functional areas, and specific scheduling measures are formulated for each area;

[0010] S5: Based on specific scheduling measures, the optimization results are sent to the device layer through DMS to monitor whether the scheduling actions are completed as planned and obtain execution feedback.

[0011] Further, S1 is specifically:

[0012] Obtain user power outage information and ECS event information, the power outage information includes user-level smart meter data, the user location of the power outage (x u ,y u ), user category, fault time T reported by the user ureport ;

[0013] The ECS event information includes distribution network fault event data, including the fault occurrence device, fault type and its impact range, time information, fault start time T fstart , Estimated repair time T frestore ;

[0014] Obtain voltage phasors, currents, and node voltage fluctuations through PMU; obtain the operating status of switches and circuit breakers, and real-time operating data of power equipment through SCADA;

[0015] The physical topological relationship between the user and the power-off device is determined through GIS information, and the associated path from user u to device e is calculated:

[0016]

[0017] in, are all paths from user to device, w(p) is the path weight;

[0018] Determine the cause and effect relationship of the fault based on the ECS fault time Tfstart and the user reporting time Tureport;

[0019] Use PMU synchronization data to dynamically correct the electrical quantity of the node, and update the operating status of the faulty equipment in combination with the SCADA equipment status;

[0020] Eliminate redundant data and generate a unified global data set Dg lobal .

[0021] Furthermore, the dynamic multi-objective optimization model is specifically:

[0022] Min Z=w1F1+w2F2-w3F3-w4F4-w5F5;

[0023] Among them, w1, w2, w3, w4, and w5 are weight coefficients; F1 is to maximize the recovery speed; F2 is to minimize the power loss; F3 is to give priority to the restoration of important users; F4 is to optimize the utilization of distributed power sources; F5 is to maximize the environmental benefits;

[0024]

[0025] F4=Ση DG ·P DG ;

[0026] F5=α·(P DG / P total );

[0027] in, is the planned recovery time for user u; is the power outage start time for user u, U is the set of all users, and U im is an important user set; I ij is the line current between nodes i and j; R ij is the line resistance between nodes i and j; W u is the weight of important users; T max is the maximum recovery time; η DG is the utilization efficiency coefficient of distributed power source; P DG is the output power of distributed power generation; α is the carbon emission reduction benefit of unit load reduction; P total Restore total power to the system;

[0028] The constraints include:

[0029] Power flow balance constraint, node power must satisfy the balance equation:

[0030]

[0031] in, is the input active power of node i; Active power demand of the load at node i; are the input reactive power and load active power requirements of node i respectively; θ i is the voltage phase angle of node i;

[0032] Voltage stability constraints, the voltage values ​​of all nodes should meet the allowable range:

[0033]

[0034] Line capacity constraints, line power cannot exceed its rated capacity:

[0035]

[0036] Distributed power generation operation constraints: the power output of distributed power generation cannot exceed its maximum capacity:

[0037]

[0038] Scheduling time constraints, the recovery task must be completed within the specified time T max Completed within:

[0039]

[0040] Further, the multi-layer optimization algorithm is as follows:

[0041] Use genetic algorithm to search the solution space globally and obtain the optimal solution candidate area;

[0042] In the optimal solution candidate area obtained by genetic algorithm search, particle swarm optimization algorithm is used for local search to reduce local oscillation and improve solution accuracy;

[0043] By introducing a deep learning model to dynamically evaluate the indicators of the current optimal solution, the weights of the multi-objective optimization model are adjusted to adapt to system changes;

[0044] When the fitness convergence condition is met or the maximum number of iterations is reached, the optimal scheduling strategy is output.

[0045] Furthermore, a genetic algorithm is used to perform a global search of the solution space and preliminarily explore the optimal solution candidate area, as follows:

[0046] Use random initialization to generate the initial population, and set the initial population size to N pop :

[0047] Each individual is a solution vector X, containing decision variables:

[0048] X = [X1, X2, X3];

[0049] Among them, X1 is the switch operation plan; X2 is the user recovery order priority; X3 is the distributed power output distribution;

[0050] The objective function is used as the fitness value evaluation indicator;

[0051] Fitness=Min Z=w1F1+w2F2-w3F3-w4F4-w5F5

[0052] According to the fitness value Fitness, use probability P i′ Select individual:

[0053]

[0054] Among them, Fitness i′ is the fitness value of individual i′, which represents the evaluation result of the optimization function;

[0055] Gene exchange is performed on selected individuals to generate the next generation;

[0056] Randomly change some gene values ​​to enhance population diversity;

[0057] Select a new generation of population according to fitness and obtain the optimal solution candidate area

[0058] Furthermore, in the optimal solution candidate area obtained by the genetic algorithm search, the particle swarm optimization algorithm is used for local search to reduce local oscillation and improve the solution accuracy, as follows:

[0059] Generate the initial population of PSO with the optimal solution of genetic algorithm as the center

[0060]

[0061] Among them, δ is a small-scale random disturbance;

[0062] Particle updates include velocity updates and position updates:

[0063] Speed ​​Update:

[0064]

[0065] Location Updates:

[0066] x ij (t+1)=x ij (t)+υ ij (t+1);

[0067] At each iteration, the position fitness of each particle is calculated:

[0068]

[0069] Update the particle's historical optimal position:

[0070]

[0071] Update the global optimal solution:

[0072]

[0073] Output the optimal solution searched by the particle swarm optimization algorithm.

[0074] Furthermore, by introducing a deep learning model to dynamically evaluate the indicators of the current optimal solution, the weights of the multi-objective optimization model are adjusted to adapt to system changes, as follows

[0075] The deep learning model LSTM is used to predict load demand and recovery time. The output prediction indicators will be used for the real-time state factor ψ i (t) calculation;

[0076] The evaluation results of deep learning prediction are put into the weight adjustment formula to dynamically adjust the weight w according to the real-time system status factor. i (t):

[0077]

[0078] The genetic algorithm and particle swarm optimization algorithm recalculate the fitness value based on the updated weights; and output the currently optimized scheduling plan.

[0079] Furthermore, S4 is specifically as follows: according to the geographical location, load distribution, equipment type, user demand priority and distributed power access of the distribution network, the distribution network is divided into several independent or semi-independent functional areas; each area is classified according to its control objectives and load characteristics;

[0080] Formulate dispatching measures based on functional classification within the region;

[0081] Comprehensively coordinate the allocation of power resources among regions based on the optimal strategy to ensure overall power restoration efficiency and stability;

[0082] Local control and independent scheduling are adopted within each region, and overall optimization is achieved through global coordination between regions to ensure efficient energy scheduling and target coordination.

[0083] Furthermore, S5 is specifically:

[0084] Based on specific scheduling measures, the DMS system is used to send instructions to the device layer;

[0085] Monitor the actual execution process of the device in real time and obtain the execution status;

[0086] Check whether the instructions are executed as planned and determine whether the plan needs to be adjusted;

[0087] The collection device uploads the execution feedback data and compares it with the planned results. If there is any deviation, the rescheduling logic is triggered in real time, and the adjusted scheduling results are reissued to form a closed-loop control.

[0088] A distribution network resource optimization allocation system, characterized in that it includes a processor, a memory and a computer program stored in the memory, and when the processor executes the computer program, it specifically executes the steps in the distribution network resource optimization allocation method as described above.

[0089] The present invention has the following beneficial effects:

[0090] 1. The present invention integrates multi-source data, dynamic multi-objective optimization and intelligent control to optimize the distribution network resource allocation method, which significantly improves the intelligence, efficiency and reliability of distribution network operation;

[0091] 2. The present invention integrates user power outage information and ECS event information, not only can the power outage user be located quickly and accurately, but also can associate the causal relationship between the power outage event and user demand, making the problem location more accurate and shortening the fault location and analysis time;

[0092] 3. The present invention uses a multi-layer optimization algorithm, which can not only ensure the ability to explore the global solution space in complex multi-objective optimization problems, but also quickly converge in the decision-making area. By introducing deep learning to dynamically evaluate the current optimal solution, the efficiency and decision-making quality of multi-objective optimization can be significantly improved, and the intelligent scheduling of dynamic power systems can be realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0093] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0094] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0095] refer to Figure 1 In this embodiment, a method for optimizing the allocation of distribution network resources is provided, comprising the following steps:

[0096] S1: Integrate the user power outage information with the ECS event information, locate the correlation between the user side and the distribution network power outage event, and integrate the PMU's synchronous phasor and the operating status of SCADA to build a global data set;

[0097] S2: Based on the global data set, the recovery speed, power loss, important user priority, distributed power utilization, and environmental benefit goals are incorporated into the optimization framework to establish a dynamic multi-objective optimization model to guide the optimization scheduling operation and resource allocation decision;

[0098] S3: Obtain the optimal strategy based on the dynamic multi-objective optimization model and multi-layer optimization algorithm;

[0099] S4: According to the optimal strategy, the distribution network is divided into several functional areas, and specific scheduling measures are formulated for each area (such as enabling distributed power sources, switching, load reduction, etc.);

[0100] S5: Based on specific scheduling measures, the optimization results are sent to the device layer through DMS to monitor whether the scheduling actions are completed as planned and obtain execution feedback.

[0101] In this embodiment, S1 is specifically:

[0102] Obtain user power outage information and ECS event information, the power outage information includes user-level smart meter data (power outage status, voltage measurement, current feedback, etc.), power outage user location (x u,y u ), user category (residential, commercial, important load), fault time T reported by the user ureport ;

[0103] The ECS event information includes distribution network fault event data, including the fault-occurring device (switch, transformer, line), fault type and its impact range, time information, fault start time T fstart , Estimated repair time T frestore ;

[0104] The voltage phasor, current and node voltage fluctuations are obtained through PMU; the operating status (closing / opening) of switches and circuit breakers and real-time operating data of power equipment (such as load power P, QP, Q, voltage VV, current II, etc.) are obtained through SCADA;

[0105] The physical topological relationship between users and power outage equipment (transformers, medium-voltage lines, switches) is determined through GIS information, and the associated path from user u to equipment e is calculated:

[0106]

[0107] in, are all paths from user to device, w(p) is the path weight;

[0108] Determine the cause and effect relationship of the fault (such as user reporting lag calculation) based on the ECS fault time Tfstart and the user reporting time Tureport;

[0109] Use PMU synchronization data to dynamically correct the electrical quantity of the node, and update the operating status of the faulty equipment in combination with the SCADA equipment status;

[0110] Eliminate redundant data and generate a unified global data set Dg lobal .

[0111] In this embodiment, the dynamic multi-objective optimization model is specifically:

[0112] Min Z=w1F1+w2F2-w3F3-w4F4-w5F5;

[0113] Among them, w1, w2, w3, w4, and w5 are weight coefficients; F1 is to maximize the recovery speed; F2 is to minimize the power loss; F3 is to give priority to the restoration of important users; F4 is to optimize the utilization of distributed power sources; F5 is to maximize the environmental benefits;

[0114]

[0115] F4=Ση DG ·P DG ;

[0116] F5=α·(P DG / P total );

[0117] in, is the planned recovery time for user u; is the power outage start time for user u, U is the set of all users, and U im is an important user set; I ij is the line current between nodes i and j; R ij is the line resistance between nodes i and j; W u is the weight of important users; T max is the maximum recovery time; η DG is the utilization efficiency coefficient of distributed power source; P DG is the output power of distributed power generation; α is the carbon emission reduction benefit of unit load reduction; P total Restore total power to the system;

[0118] The constraints include:

[0119] Power flow balance constraint, node power must satisfy the balance equation:

[0120]

[0121] in, is the input active power of node i; P i f Active power demand of the load at node i; are the input reactive power and load active power requirements of node i respectively; θ i is the voltage phase angle of node i;

[0122] Voltage stability constraint: the voltage values ​​of all nodes should meet the allowable range:

[0123]

[0124] Line capacity constraints, line power cannot exceed its rated capacity:

[0125]

[0126] Distributed power generation operation constraints: the power output of distributed power generation cannot exceed its maximum capacity:

[0127]

[0128] Scheduling time constraints, the recovery task must be completed within the specified time T max Completed within:

[0129]

[0130] In this embodiment, the multi-layer optimization algorithm is specifically:

[0131] Use genetic algorithm to search the solution space globally and obtain the optimal solution candidate area;

[0132] In the optimal solution candidate area obtained by genetic algorithm search, particle swarm optimization algorithm is used for local search to reduce local oscillation and improve solution accuracy;

[0133] By introducing a deep learning model to dynamically evaluate the indicators of the current optimal solution, the weights of the multi-objective optimization model are adjusted to adapt to system changes;

[0134] When the fitness convergence condition is met or the maximum number of iterations is reached, the optimal scheduling strategy is output.

[0135] In this embodiment, a global search of the solution space is performed using a genetic algorithm to preliminarily explore the optimal solution candidate region, as follows:

[0136] Use random initialization to generate the initial population, and set the initial population size to N pop :

[0137] Each individual is a solution vector X, containing decision variables:

[0138] X = [X1, X2, X3];

[0139] Among them, X1 is the switch operation scheme (on / off state); X2 is the user recovery order priority; X3 is the distributed power output distribution;

[0140] The objective function is used as the fitness value evaluation indicator;

[0141] Fitness=Min Z=w1F1+w2F2-w3F3-w4F4-w5F5

[0142] According to the fitness value Fitness, use probability P i′ Select individual:

[0143]

[0144] Among them, Fitness i′ is the fitness value of individual i′, which represents the evaluation result of the optimization function;

[0145] Gene exchange is performed on selected individuals to generate the next generation;

[0146] Randomly change some gene values ​​to enhance population diversity;

[0147] Select a new generation of population according to fitness and obtain the optimal solution candidate area

[0148] In this embodiment, in the optimal solution candidate area obtained by the genetic algorithm search, the particle swarm optimization algorithm is used to perform local search to reduce local oscillation and improve the solution accuracy, as follows:

[0149] Generate the initial population of PSO with the optimal solution of genetic algorithm as the center

[0150]

[0151] Among them, δ is a small-scale random disturbance;

[0152] Particle updates include velocity updates and position updates:

[0153] Speed ​​Update:

[0154]

[0155] Location Updates:

[0156] x ij (t+1)=x ij (t)+υ ij (t+1);

[0157] At each iteration, the position fitness of each particle is calculated:

[0158]

[0159] Update the particle's historical optimal position:

[0160]

[0161] Update the global optimal solution:

[0162]

[0163] Output the optimal solution searched by the particle swarm optimization algorithm.

[0164] In this embodiment, by introducing a deep learning model to dynamically evaluate the indicators of the current optimal solution, the weights of the multi-objective optimization model are adjusted to adapt to system changes, as follows

[0165] The deep learning model LSTM is used to predict load demand and recovery time. The output prediction indicators will be used for the real-time state factor ψ i (t) calculation;

[0166] The evaluation results of deep learning prediction are put into the weight adjustment formula to dynamically adjust the weight w according to the real-time system status factor.i (t):

[0167]

[0168] The genetic algorithm and particle swarm optimization algorithm recalculate the fitness value based on the updated weights; and output the currently optimized scheduling plan.

[0169] In this embodiment, the distribution network is divided into several independent or semi-independent functional areas according to the geographical location, load distribution, equipment type, user demand priority and distributed power access of the distribution network; each area is classified according to its control target and load characteristics;

[0170] In this embodiment, the functional areas are classified as follows:

[0171]

[0172]

[0173] Formulate dispatching measures based on the functional classification within the area (such as priority recovery area, important user area, DG master control area, etc.);

[0174] Comprehensively coordinate the allocation of power resources among regions based on the optimal strategy to ensure overall power restoration efficiency and stability;

[0175] Local control and independent scheduling are adopted within each region, and overall optimization is achieved through global coordination between regions to ensure efficient energy scheduling and target coordination.

[0176] In this embodiment, S5 is specifically:

[0177] Based on specific dispatching measures, the DMS system is used to send instructions to the equipment layer (such as automatic switches, distributed power controllers, load controllers, etc.);

[0178] Monitor the actual execution process of the equipment in real time and obtain the execution status (such as switch status, load data, DG output power, etc.);

[0179] Check whether the instructions are executed as planned and determine whether the plan needs to be adjusted;

[0180] The collection device uploads execution feedback data (equipment status, real-time flow data, etc.) and compares it with the planned results. If there is a deviation, the re-scheduling logic is triggered in real time, and the adjusted scheduling results are re-issued to form a closed-loop control.

[0181] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0182] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of 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 produce 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.

[0183] 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 including 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.

[0184] 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.

[0185] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any technician familiar with the profession may use the above disclosed technical content to change or modify it into an equivalent embodiment with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention still belongs to the protection scope of the technical solution of the present invention.

Claims

1. A method for optimizing the allocation of distribution network resources, characterized in that: The following steps are involved: S1: Integrate the user power outage information with the ECS event information, locate the correlation between the user side and the distribution network power outage event, and integrate the PMU's synchronous phasor and the operating status of SCADA to build a global data set; S2: Based on the global data set, the restoration speed, power loss, priority of important users, utilization of distributed power sources, and environmental benefit goals are incorporated into the optimization framework to establish a dynamic multi-objective optimization model; S3: Obtain the optimal strategy based on the dynamic multi-objective optimization model and multi-layer optimization algorithm; S4: According to the optimal strategy, the distribution network is divided into several functional areas, and specific scheduling measures are formulated for each area; S5: Based on specific scheduling measures, the optimization results are sent to the device layer through DMS to monitor whether the scheduling actions are completed as planned and obtain execution feedback.

2. A method for optimizing the allocation of distribution network resources according to claim 1, characterized in that: The S1 is specifically: Obtain user power outage information and ECS event information, the power outage information includes user-level smart meter data, the user location of the power outage (x u ,y u ), user category, fault time T reported by the user ureport ; The ECS event information includes distribution network fault event data, including the fault occurrence device, fault type and its impact range, time information, fault start time T fstart , Estimated repair time T frestore ; Obtain voltage phasors, currents, and node voltage fluctuations through PMU; obtain the operating status of switches and circuit breakers, and real-time operating data of power equipment through SCADA; The physical topological relationship between the user and the power-off device is determined through GIS information, and the associated path from user u to device e is calculated: in, are all paths from user to device, w(p) is the path weight; Determine the cause and effect relationship of the fault based on the ECS fault time Tfstart and the user reporting time Tureport; Use PMU synchronization data to dynamically correct the electrical quantity of the node, and update the operating status of the faulty equipment in combination with the SCADA equipment status; Eliminate redundant data and generate a unified global data set Dg lobal .

3. A method for optimizing the allocation of distribution network resources according to claim 1, characterized in that: The dynamic multi-objective optimization model is specifically: Min Z=w1F1+w2F2-w3F3-w4F4-w5F5; Among them, w1, w2, w3, w4, and w5 are weight coefficients; F1 is to maximize the recovery speed; F2 is to minimize the power loss; F3 is to give priority to the restoration of important users; F4 is to optimize the utilization of distributed power sources; F5 is to maximize the environmental benefits; F4=S DG ·P DG ; F5=α·(P DG / P total ); in, is the planned recovery time for user u; is the power outage start time for user u, U is the set of all users, and U im is an important user set; I ij is the line current between nodes i and j; R ij is the line resistance between nodes i and j; W u is the weight of important users; T max is the maximum recovery time; η DG is the utilization efficiency coefficient of distributed power source; P DG is the output power of distributed power generation; α is the carbon emission reduction benefit of unit load reduction; P total Restore total power to the system; The constraints include: Power flow balance constraint, node power must satisfy the balance equation: in, is the input active power of node i; Active power demand of the load at node i; are the input reactive power and load active power requirements of node i respectively; θ i is the voltage phase angle of node i; Voltage stability constraint: the voltage values ​​of all nodes should meet the allowable range: Line capacity constraints, line power cannot exceed its rated capacity: Distributed power generation operation constraints: the power output of distributed power generation cannot exceed its maximum capacity: Scheduling time constraints, the recovery task must be completed within the specified time T max Completed within:

4. A method for optimizing the allocation of distribution network resources according to claim 1, characterized in that: The multi-layer optimization algorithm is specifically: Use genetic algorithm to search the solution space globally and obtain the optimal solution candidate area; In the optimal solution candidate area obtained by genetic algorithm search, particle swarm optimization algorithm is used for local search to reduce local oscillation and improve solution accuracy; By introducing a deep learning model to dynamically evaluate the indicators of the current optimal solution, the weights of the multi-objective optimization model are adjusted to adapt to system changes; When the fitness convergence condition is met or the maximum number of iterations is reached, the optimal scheduling strategy is output.

5. A method for optimizing the allocation of distribution network resources according to claim 4, characterized in that: The genetic algorithm is used to perform a global search of the solution space and preliminarily explore the optimal solution candidate area, as follows: Use random initialization to generate the initial population, and set the initial population size to N pop : Each individual is a solution vector X, containing decision variables: X = [X1, X2, X3]; Among them, X1 is the switch operation plan; X2 is the user recovery order priority; X3 is the distributed power output distribution; The objective function is used as the fitness value evaluation indicator; Fitness=Min Z=w1F1+w2F2-w3F3-w4F4-w5F5 According to the fitness value Fitness, the probability P is used i′ Select individual: Among them, Fitness i′ is the fitness value of individual i′, which represents the evaluation result of the optimization function; Gene exchange is performed on selected individuals to generate the next generation; Randomly change some gene values ​​to enhance population diversity; Select a new generation of population according to fitness and obtain the optimal solution candidate area 6. A method for optimizing the allocation of distribution network resources according to claim 5, characterized in that: In the optimal solution candidate area obtained by the genetic algorithm search, the particle swarm optimization algorithm is used for local search to reduce local oscillation and improve the solution accuracy, as follows: Generate the initial population of PSO with the optimal solution of genetic algorithm as the center Among them, δ is a small-scale random disturbance; Particle updates include velocity updates and position updates: Speed ​​Update: Location Updates: x ij (t+1)=x ij (t)+v ij (t+1); At each iteration, the position fitness of each particle is calculated: Update the particle's historical optimal position: Update the global optimal solution: Output the optimal solution searched by the particle swarm optimization algorithm.

7. A method for optimizing the allocation of distribution network resources according to claim 6, characterized in that: The deep learning model is introduced to dynamically evaluate the indicators of the current optimal solution, adjust the weights of the multi-objective optimization model, and adapt to system changes. The details are as follows The deep learning model LSTM is used to predict load demand and recovery time. The output prediction indicators will be used for the real-time state factor ψ i (t) calculation; The evaluation results of deep learning prediction are put into the weight adjustment formula to dynamically adjust the weight w according to the real-time system status factor. i (t): The genetic algorithm and particle swarm optimization algorithm recalculate the fitness value based on the updated weights; and output the currently optimized scheduling plan.

8. A method for optimizing the allocation of distribution network resources according to claim 1, characterized in that: S4 specifically includes: dividing the distribution network into a number of independent or semi-independent functional areas according to the geographical location, load distribution, equipment type, user demand priority and distributed power access of the distribution network; each area is classified according to its control target and load characteristics; Formulate dispatching measures based on functional classification within the region; Comprehensively coordinate the allocation of power resources among regions based on the optimal strategy to ensure overall power restoration efficiency and stability; Local control and independent scheduling are adopted within each region, and overall optimization is achieved through global coordination between regions to ensure efficient energy scheduling and target coordination.

9. A method for optimizing the allocation of distribution network resources according to claim 1, characterized in that: The S5 is specifically: Based on specific scheduling measures, the DMS system is used to send instructions to the device layer; Monitor the actual execution process of the device in real time and obtain the execution status; Check whether the instructions are executed as planned and determine whether the plan needs to be adjusted; The collection device uploads the execution feedback data and compares it with the planned results. If there is any deviation, the rescheduling logic is triggered in real time, and the adjusted scheduling results are reissued to form a closed-loop control.

10. A distribution network resource optimization allocation system, characterized in that: The method comprises a processor, a memory and a computer program stored in the memory. When the processor executes the computer program, the method specifically executes the steps in the method for optimizing allocation of distribution network resources as described in any one of claims 1 to 9.

Images