Defect identification-based self-healing method and system for power distribution network containing distributed power supply

Abstract

The invention discloses a distributed power supply-containing power distribution network self-healing method and system based on defect identification, and the method comprises the steps: inputting operation data into a defect identification model, obtaining a defect identification result, correcting a feeder load rate and a weighted structure entropy according to the defect identification result to generate a candidate solution set, inputting the candidate solution set into an upper layer optimization model, and obtaining a defect identification result; taking the maximization of the load recovery amount in the island as a target, combining with a line medium to guide a search direction and adopting a second-order cone relaxation method to carry out optimization solution to obtain a load recovery scheme in the island, and inputting the candidate solution set and the load recovery scheme into a lower-layer optimization model to obtain a load recovery scheme in the island; carrying out iterative solution under a constraint condition by combining an improved star-sparrow optimization algorithm which introduces an inertia weight and a learning factor, and obtaining a load recovery result after fault reconstruction; and determining a target switch operation sequence and a load recovery strategy based on the load recovery result to perform self-healing of the power distribution network so as to improve the self-healing capability and the operation reliability of the power distribution network.

Description

Technical Field

[0001] This invention relates to the field of power distribution networks, and in particular to a self-healing method and system for power distribution networks with distributed generation based on defect identification. Background Technology

[0002] With the increasing proportion of new energy sources in the power system, a large number of distributed new energy sources have been connected to the distribution network, significantly changing the structure and operating characteristics of the distribution network. The intermittency and uncertainty of new energy sources increase the probability of distribution network faults, and the scope of fault impact and recovery difficulty also increase accordingly. Traditional distribution network fault recovery methods often rely on manual operation, which is inefficient and time-consuming, making it difficult to meet the current requirements of the power system. Therefore, it is urgent to develop self-healing technologies that can autonomously and rapidly respond to and accurately correct faults, ensuring that the distribution network maintains high reliability and efficiency even in complex and ever-changing environments.

[0003] While existing distribution network self-healing technologies can achieve rapid fault detection, isolation, and recovery, reducing manual intervention and improving the speed and quality of fault recovery, they fail to dynamically and accurately consider real-time changes in source-load information when planning recovery paths and load transfer. This limits the practical effectiveness and reliability of distribution network self-healing technologies in scenarios with a high proportion of renewable energy. Especially in extreme scenarios, existing technologies may still face the risk of recovery failure or poor performance. Furthermore, existing technologies fail to fully consider real-time operational data and defect information when generating candidate solution sets, resulting in insufficient accuracy and reliability of candidate solution sets, further affecting the optimization effect of self-healing strategies. Summary of the Invention

[0004] This invention provides a self-healing method and system for distribution networks with distributed power sources based on defect identification, which can improve the self-healing capability and operational reliability of the distribution network.

[0005] One embodiment of the present invention provides a self-healing method for distribution networks with distributed power sources based on defect identification, comprising:

[0006] Obtain operational data of the power distribution network;

[0007] The operating data is input into a preset defect identification model to obtain defect identification results. The feeder load rate and weighted structural entropy are then corrected based on the defect identification results to generate a candidate solution set that reflects the actual operating state of the distribution network. The feeder load rate and the weighted structural entropy are calculated based on the operating data.

[0008] The candidate solution set is input into the upper-level optimization model. With the goal of maximizing the load recovery within the island, the search direction is guided by the line medium, and the second-order cone relaxation method is used for optimization to obtain the load recovery scheme within the island. The candidate solution set and the load recovery scheme are input into the lower-level optimization model. The total fault recovery target, the operation reliability target, and the economic cost target after the distribution network fault reconstruction are used as multi-objective optimization functions. The improved sparrow optimization algorithm with the introduction of inertia weight and learning factor is used to iteratively solve the problem under constraints to obtain the load recovery result after fault reconstruction.

[0009] Based on the load restoration results, a target switch operation sequence and load restoration strategy are determined, and the distribution network is controlled to self-heal according to the target switch operation sequence and the load restoration strategy.

[0010] This application embodiment, by acquiring operational data of the distribution network and performing defect identification, can quickly and accurately identify defect parameters in the distribution network, thereby generating a candidate solution set reflecting the actual operating state of the distribution network. This provides the optimization model with an initial solution that is closer to the actual operating conditions, thus improving the optimization effect of the self-healing strategy. Combining line betweenness to guide the search direction and employing a second-order cone relaxation method for optimization, it can quickly find the nearest accessible island to important load nodes, prioritizing the restoration of power supply to critical loads. This significantly improves the efficiency of island partitioning and the amount of load restoration, enhancing the power supply reliability of the distribution network under fault conditions. An improved sparrow optimization algorithm, incorporating inertia weights and learning factors, comprehensively considers multiple objective optimization functions such as total fault recovery, operational reliability, and economic cost to solve the lower-level optimization model. This generates optimal fault reconfiguration and load restoration strategies, improving the self-healing capability and economic efficiency of the distribution network. By controlling the distribution network to self-heal according to the target switching operation sequence and load restoration strategy, rapid isolation and power restoration of fault areas can be achieved. This significantly shortens the recovery time, improves the self-healing efficiency of the distribution network, and reduces power outage losses. Compared with existing technologies, this application can improve the self-healing capability and operational reliability of the distribution network.

[0011] Furthermore, the step of correcting the feeder load rate and weighted structural entropy based on the defect identification results to generate a candidate solution set reflecting the actual operating state of the distribution network specifically involves:

[0012] Based on the defect identification results, the defective feeder and the defect impact range are determined, wherein the defect impact range is the impact of the defect on the correlation of the feeder;

[0013] The actual active power value transmitted in the defective feeder is corrected by combining the maximum transmittable power value to obtain the corrected feeder load rate.

[0014] Based on the scope of the defect's influence, the line importance parameter in the weighted structural entropy is corrected to obtain the corrected weighted structural entropy.

[0015] Based on the corrected feeder load rate and the weighted structural entropy, a candidate solution set containing different feeder load allocation schemes is constructed, wherein the candidate solution set includes several candidate solutions that satisfy the actual operating state of the distribution network under the influence of defects.

[0016] By correcting the feeder load rate and weighted structural entropy based on the defect identification results, a candidate solution set reflecting the actual operating state of the distribution network is generated. This more accurately reflects the network's true condition under the influence of defects, providing a high-quality initial solution for subsequent optimization models. This correction method not only improves the accuracy and reliability of the candidate solution set but also enhances the adaptability and flexibility of the distribution network's self-healing strategy, thereby significantly improving the distribution network's self-healing capability and operational reliability under complex operating conditions.

[0017] Furthermore, the formula for correcting the feeder load rate and weighted structural entropy based on the defect identification results to generate a candidate solution set reflecting the actual operating state of the distribution network is as follows:

[0018]

[0019] ln(2 / N l )<σ≤ln N l ;

[0020]

[0021]

[0022] In the formula, δ represents the lines in the distribution network; α δ P represents the corrected line load factor. δ 、P δmax These are the actual transmitted active power value and the maximum transmittable power value, respectively; P DG N represents the remaining capacity of new energy equipment. l σ is the number of lines; W is the corrected weighted structure entropy; δ This refers to the importance parameter of the line.

[0023] Furthermore, the candidate solution set is input into the upper-level optimization model, with the objective of maximizing the load restoration within the island. The search direction is guided by the line medium, and a second-order cone relaxation method is used for optimization to obtain the load restoration scheme within the island. Specifically:

[0024] A higher-level optimization model is constructed using the maximization of load recovery within the isolated island as the objective function.

[0025] Based on the candidate solution set, the medium parameters of each line in the distribution network are extracted, and the initial direction of the optimization search and the boundary of the feasible region are determined according to the medium parameters.

[0026] For non-convex constraints in distribution networks, a second-order cone relaxation method is used to transform power balance constraints, network loss cost functions, and node voltage constraints into convex constraints to form a relaxation model.

[0027] The relaxation model is used to iteratively optimize the solutions in the candidate solution set in the initial direction and the boundary of the feasible region. The search direction is dynamically adjusted in combination with the medium parameters until the convergence condition is met, and the load recovery scheme in the island is output. The load recovery scheme includes the output adjustment value of the distributed power source and the access status of the recoverable load.

[0028] By combining line betweenness to guide the search direction and using the second-order cone relaxation method for optimization, we can quickly find the nearest accessible island to the critical load node, prioritize the restoration of power supply to critical loads, significantly improve the efficiency of islanding and the amount of load restoration, and enhance the power supply reliability of the distribution network under fault conditions.

[0029] Furthermore, the expression for the relaxation model is specifically as follows:

[0030]

[0031] ξ ij，t =U i，t U j，i G ij cosθ ij ;

[0032] η ij，t =U i，t U j，t B ij cosθ ij ;

[0033]

[0034]

[0035]

[0036] In the formula, G ij 、B ij Let θ be the real and imaginary parts of the admittance matrix of the distribution network nodes; ij The phase angle difference; ξ ij,t η ij,t 、u i,j p i,j and q i,j λ is a slack variable; ijp represents the weighting coefficients of the slack variables. i，t and Q i，t These represent the actual active and reactive power of the distributed generation, respectively; Δt is the time consumed in adjusting the network structure; R i C is the branch resistance; Loss The economic cost caused by transmission power loss; p ij，t and Q ij，t These represent the active power and reactive power flowing from node i to node j during time period t.

[0037] Furthermore, the construction process of the lower-level optimization model is as follows:

[0038] The goal of maximizing the total fault recovery after distribution network fault reconfiguration is to maximize the total fault recovery. The goal of operational reliability is to minimize the system mean fault time, mean fault frequency, and power shortage. The goal of economic cost is to minimize the sum of switching action cost, power curtailment cost, and network loss cost. A multi-objective optimization function is constructed based on the goals of total fault recovery, operational reliability, and economic cost.

[0039] The lower-level optimization model is constructed by combining island power balance constraints, distribution network power balance constraints, node voltage constraints, branch current constraints, network topology constraints, and switching action count constraints.

[0040] By comprehensively considering maximizing the total fault recovery, improving operational reliability, and reducing economic costs, a lower-level optimization model that fully reflects the fault reconfiguration process of the distribution network can be obtained.

[0041] Furthermore, the expression for the lower-level optimization model is specifically as follows:

[0042] The multi-objective optimization function is:

[0043]

[0044]

[0045]

[0046] In the formula, f2 represents the total fault recovery amount; z dis For load nodes outside the island; N dis This represents the number of load nodes outside the island. These are the weighting coefficients; This represents the active power of the load node; f3 is a binary variable; f3 is the reliability optimization function. and These represent mean time between failures, failure frequency, and power shortage, respectively; ω1, ω2, and ω3 are weighting coefficients; Nsys This refers to the number of isolated islands. N represents the number of load nodes within the island. i T represents the total number of users; i Mean time between failures; f4 represents the total operating power of the distribution network; c represents the comprehensive cost. Swi Unit cost of switching action; Ω L Ω represents the number of network branches. New X represents the load number; i，t X i，t-1 t and t-1 represent the branch switch states during time periods t and t-1, respectively; Δt represents the time consumed for network structure adjustment; c Cur The penalty cost incurred for failing to fully utilize new energy sources; Loss These represent the economic costs caused by transmission power loss; R1 is the branch resistance; P i，t , Q i，t These represent the active and reactive power flowing into the branch, respectively; U i，t This refers to the voltage at the end node of the branch.

[0047] The island power balance constraint is:

[0048]

[0049] In the formula, The number of new energy equipment; For the number of loads; and These are the active power and reactive power of new energy equipment, respectively. and These are the active power and reactive power of the i-th node, respectively; and These represent the active and reactive power losses of the line:

[0050] The power balance constraint of the distribution network is:

[0051]

[0052] In the formula, b is the set of branches flowing into node i; P i,t and Q i,t These represent the actual active power and reactive power of the distributed power source, respectively; I ji,t 、P ji,t and Q ji,t For the current, active power, and reactive power transmitted in the branch; R ji X ji These are the resistance and reactance of the branch circuit, respectively.

[0053] The node voltage constraint is:

[0054]

[0055] In the formula, These are the upper and lower limits of the permissible voltage value, respectively.

[0056] The branch current constraint is:

[0057]

[0058] In the formula, The maximum allowable current;

[0059] The network topology constraints are:

[0060]

[0061] In the formula, X i,t L is the decision variable. Node , L Sub These represent the number of distribution network nodes and the number of substations, respectively; Ω g Ω is the set of disintegratable branches; c This is a collection of normally closed branches;

[0062] The constraint on the number of switching actions is:

[0063] k≤k max ;

[0064] In the formula, K max This represents the maximum number of actions allowed.

[0065] Furthermore, the candidate solution set and the load restoration scheme are input into the lower-level optimization model. The model uses the total fault recovery target, operational reliability target, and economic cost target after distribution network fault reconfiguration as multi-objective optimization functions. An improved sparrow optimization algorithm incorporating inertia weights and learning factors is used to iteratively solve the problem under constraints to obtain the load restoration result after fault reconfiguration. Specifically:

[0066] The population of the improved sparrow optimization algorithm is initialized based on the candidate solution set and the load recovery scheme, wherein each individual in the population corresponds to a fault reconstruction scheme;

[0067] During the iteration process, the search direction and step size of the starbird individual in the solution space are dynamically adjusted by introducing the inertial weight, and the learning ability of the starbird individual to the best position of the population is enhanced according to the adaptive learning factor to execute the food storage strategy and the cache recovery strategy respectively.

[0068] The position of the starbird is updated by the food storage strategy and the cache recovery strategy, and the fitness of the individual solutions generated in each iteration is evaluated under the constraints according to the multi-objective optimization function in order to update the Pareto optimal solution set.

[0069] When the preset termination conditions are met, the scheme with the best overall performance is selected from the final Pareto optimal solution set as the load recovery result after fault reconstruction.

[0070] By introducing an improved Pareto optimization algorithm, combined with a multi-objective optimization function considering total fault recovery, operational reliability, and economic cost, and performing iterative solutions under constraints, the search direction and step size can be dynamically adjusted, enhancing individual learning capabilities and thus quickly finding the optimal fault reconstruction scheme. Through updating and selecting the Pareto optimal solution set, it is ensured that the scheme with the best overall performance, while satisfying multiple constraints, is selected as the final load recovery result.

[0071] Furthermore, the formula for updating the starbird's position using the food storage strategy and the cache recovery strategy is as follows:

[0072]

[0073] In the formula, c is the adaptive learning factor; up and c low represents the upper and lower bounds of the learning factor, respectively; t is the number of iterations; MaxIt is the maximum number of iterations; rand is a random number in the interval [0,1].

[0074] The location update formula for the food storage strategy is:

[0075]

[0076]

[0077] τ1≥τ2 and τ1≥τ3;

[0078] In the formula, Let i be the new position of the i-th starbird at time t+1; Let i be the new position of the i-th starbird at time t; λ represents the optimal position for the individual Star Sparrow; λ is a parameter generated based on Levi's flight; A and B are randomly selected indices; ι is a linear decreasing factor; τ1, τ2, τ3, and r1 are random numbers within the interval [0,1]; μ is a randomly selected parameter.

[0079] The location update formula for the cache recovery strategy is:

[0080]

[0081]

[0082] In the formula, Reference point information for storage location; r1 and r2 are random numbers within the interval [0,1]; τ3, τ4, and τ5 are random numbers within the interval [0,1]; C is a randomly selected index; This represents the j-th position of the i-th star before and after the update.

[0083] Another embodiment of the present invention provides a self-healing device for a distribution network with distributed power sources based on defect identification, comprising: an acquisition module, an identification module, an optimization module, and a self-healing module;

[0084] The acquisition module is used to acquire the operating data of the power distribution network;

[0085] The identification module is used to input the operating data into a preset defect identification model to obtain defect identification results, and to correct the feeder load rate and weighted structural entropy based on the defect identification results to generate a candidate solution set that reflects the actual operating state of the distribution network. The feeder load rate and the weighted structural entropy are calculated based on the operating data.

[0086] The optimization module is used to input the candidate solution set into the upper-level optimization model, with the goal of maximizing the load recovery within the island. It combines the line medium to guide the search direction and uses the second-order cone relaxation method to optimize and solve the problem, thereby obtaining the load recovery scheme within the island. The candidate solution set and the load recovery scheme are then input into the lower-level optimization model, with the total fault recovery target, operational reliability target, and economic cost target after the distribution network fault reconstruction as the multi-objective optimization function. The improved sparrow optimization algorithm, which introduces inertia weights and learning factors, is used to iteratively solve the problem under constraints to obtain the load recovery result after the fault reconstruction.

[0087] The self-healing module is used to determine the target switch operation sequence and load restoration strategy based on the load restoration result, and control the distribution network to perform self-healing according to the target switch operation sequence and the load restoration strategy.

[0088] This application embodiment, by acquiring operational data of the distribution network and performing defect identification, can quickly and accurately identify defect parameters in the distribution network, thereby generating a candidate solution set reflecting the actual operating state of the distribution network. This provides the optimization model with an initial solution that is closer to the actual operating conditions, thus improving the optimization effect of the self-healing strategy. Combining line betweenness to guide the search direction and employing a second-order cone relaxation method for optimization, it can quickly find the nearest accessible island to important load nodes, prioritizing the restoration of power supply to critical loads. This significantly improves the efficiency of island partitioning and the amount of load restoration, enhancing the power supply reliability of the distribution network under fault conditions. An improved sparrow optimization algorithm, incorporating inertia weights and learning factors, comprehensively considers multiple objective optimization functions such as total fault recovery, operational reliability, and economic cost to solve the lower-level optimization model. This generates optimal fault reconfiguration and load restoration strategies, improving the self-healing capability and economic efficiency of the distribution network. By controlling the distribution network to self-heal according to the target switching operation sequence and load restoration strategy, rapid isolation and power restoration of fault areas can be achieved. This significantly shortens the recovery time, improves the self-healing efficiency of the distribution network, and reduces power outage losses. Compared with existing technologies, this application can improve the self-healing capability and operational reliability of the distribution network. Attached Figure Description

[0089] To more clearly illustrate the technical solution of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0090] Figure 1 This is a flowchart illustrating an embodiment of the self-healing method for distribution networks with distributed power sources based on defect identification provided in this application.

[0091] Figure 2 This is a flowchart illustrating steps S201 to S204 provided in this application;

[0092] Figure 3 This is a flowchart illustrating steps S301 to S304 provided in this application;

[0093] Figure 4 This is a flowchart illustrating steps S401 to S404 provided in this application;

[0094] Figure 5 This is a flowchart illustrating another embodiment of the self-healing method for distribution networks with distributed power sources based on defect identification provided in this application.

[0095] Figure 6 This is a schematic diagram of the IEEE 69-node model provided in this application;

[0096] Figure 7 This is a diagram showing the comparison between fault and load recovery probability provided in this application;

[0097] Figure 8 This is a schematic diagram of an embodiment of the self-healing system for a distribution network with distributed power sources based on defect identification provided in this application. Detailed Implementation

[0098] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0099] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0100] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly defined.

[0101] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0102] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0103] In the description of the embodiments of this application, the term "multiple" refers to two or more (including two), similarly, "multiple sets" refers to two or more (including two sets), and "multiple pieces" refers to two or more (including two pieces).

[0104] In the description of the embodiments of this application, unless otherwise expressly specified and limited, technical terms such as "installation," "connection," "joining," and "fixing" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. For those skilled in the art, the specific meaning of the above terms in the embodiments of this application can be understood according to the specific circumstances.

[0105] With the increasing proportion of renewable energy sources, the integration of distributed power sources into the distribution network has led to changes in its structure and operating characteristics, increasing the probability of failures and the difficulty of recovery. Traditional manual recovery methods are inefficient and time-consuming, failing to meet the demands. While existing self-healing technologies can quickly detect, isolate, and recover from faults, they fail to dynamically consider real-time source-load information when planning recovery paths and load transfer, limiting their effectiveness and reliability in scenarios with a high proportion of renewable energy, and potentially leading to recovery failures in extreme scenarios. Furthermore, the candidate solution sets generated by existing technologies do not fully consider real-time operational data and defect information, resulting in insufficient accuracy and reliability, affecting the optimization effect of self-healing strategies.

[0106] See Figure 1 To improve the self-healing capability and operational reliability of the distribution network, an embodiment of the present invention provides a self-healing method for a distribution network with distributed power sources based on defect identification, including steps S101 to S104.

[0107] Step S101: Obtain the operating data of the power distribution network;

[0108] In some embodiments, acquiring operational data involves several key aspects. Specifically, the acquisition methods are as follows: First, voltage and current values ​​at each node in the distribution network are collected using devices such as smart meters, sensor networks, and monitoring systems. This data directly reflects the electrical state of the grid and potential anomalies. Simultaneously, the real-time output power of distributed power sources (such as photovoltaic and wind power equipment) is acquired, which is crucial for assessing the impact of renewable energy integration on grid stability. Furthermore, active and reactive power data of the loads need to be collected to understand the electricity demand and power balance in different areas. Acquiring switch status data helps determine the network topology and connectivity, which is essential for subsequent fault isolation and network reconfiguration strategies. Finally, the collation and analysis of historical fault data can provide valuable experience and reference for fault mode identification and the optimization of self-healing strategies.

[0109] In some embodiments, after acquiring the operational data, it is necessary to preprocess the collected operational data, including data cleaning, standardization, and feature engineering, to ensure data quality and consistency. For example, this includes removing outliers, filling in missing values, and standardizing the data to a uniform range.

[0110] Step S102: Input the operating data into a preset defect identification model to obtain defect identification results, and correct the feeder load rate and weighted structural entropy according to the defect identification results to generate a candidate solution set that reflects the actual operating state of the distribution network. The feeder load rate and the weighted structural entropy are calculated based on the operating data.

[0111] In some embodiments, the running data is input into a preset defect identification model to obtain defect identification results. Specifically, the defect identification model is constructed using the XGBoost algorithm. Then, the processed running data is assigned to the defect identification model to determine the defect identification results, which include the defect location, defect type, and defect duration.

[0112] It should be noted that XGBoost is an ensemble learning algorithm based on gradient boosting, which has the ability to process large-scale data and performs well in classification and regression tasks.

[0113] It should be noted that historical fault data is used to train the XGBoost model. During training, historical fault data is used as input, and the corresponding fault type and fault location are used as output labels. Features related to fault identification are extracted from the historical fault data, such as line load rate, weighted structural entropy, voltage deviation, current change rate, and power fluctuation of distributed power sources. The importance of each feature is analyzed, and features that have a greater impact on fault identification, such as feeder load rate and weighted structural entropy, are identified. This helps to understand which factors are more likely to cause faults, and then adjust the model parameters so that the model can accurately identify different types of faults.

[0114] Please refer to Figure 2 In some embodiments, the step of correcting the feeder load rate and weighted structural entropy based on the defect identification results to generate a candidate solution set that reflects the actual operating state of the distribution network includes steps S201 to S204.

[0115] Step S201: Determine the defective feeder and the defect impact range based on the defect identification result, wherein the defect impact range is the impact of the defect on the correlation of the feeder;

[0116] In some embodiments, defects in the distribution network may be hidden within complex network structures, significantly impacting the accuracy of feeder load rates and the calculation results of weighted structural entropy. Therefore, it is necessary to correct the feeder load rates and weighted structural entropy based on the defect identification results. Specifically, based on the defect identification results, the defective feeder is identified, and the impact of the defect on feeder correlation is analyzed to determine the set of load nodes affected by the defect and the set of lines affected. For example, if a single-phase ground fault occurs on a feeder, it may affect the load distribution and power supply reliability of that feeder and other connected feeders.

[0117] Step S202: Correct the actual active power value transmitted in the defective feeder by combining the maximum transmittable power value to obtain the corrected feeder load rate.

[0118] In some embodiments, firstly, for each feeder, the actual active power value P delivered is measured in real time. δ Then, obtain the maximum transmittable power value P for each feeder from the operating data. δmax Finally, based on the defect identification results, the actual active power value transmitted by the defective feeder is corrected.

[0119] It should be noted that if a feeder has defects, its load rate should be appropriately reduced to reflect its actual load level. For example, for a defective feeder, the load rate can be multiplied by a correction factor (such as 0.8 or 0.9) to indicate a reduction in its actual available capacity.

[0120] Step S203: Based on the scope of the defect's influence, correct the line importance parameter in the weighted structural entropy to obtain the corrected weighted structural entropy;

[0121] In some embodiments, an initial weighted structural entropy σ is calculated for each line based on operational data. Then, based on the defect impact range, the line importance parameter W in the weighted structural entropy is adjusted. δ The correction is performed as follows: For affected lines, the line importance parameter is adjusted according to the degree of impact of the defect on line correlation. For example, if the correlation of a line is significantly affected, its line importance parameter can be appropriately reduced. Finally, the weighted structural entropy is recalculated using the corrected line importance parameters.

[0122] Step S204: Based on the corrected feeder load rate and the weighted structural entropy, construct a candidate solution set containing different feeder load allocation schemes, wherein the candidate solution set includes several candidate solutions that satisfy the actual operating state of the distribution network under the influence of defects.

[0123] In some embodiments, the correlation between feeders and the complexity of load distribution are taken into account when generating candidate solution sets. For example, for a defective feeder, priority is given to distributing its load to an adjacent, unaffected feeder.

[0124] It should be noted that the feeder load rate can intuitively reflect the current load level of the feeder, while the weighted structural entropy can comprehensively consider the correlation between feeders and the complexity of load distribution, thus providing a more reliable basis for network reconfiguration.

[0125] In some embodiments, the formula for correcting the feeder load rate and weighted structural entropy based on the defect identification results to generate a candidate solution set reflecting the actual operating state of the distribution network is specifically as follows:

[0126]

[0127] ln(2 / N l )<σ≤ln N l ;

[0128]

[0129]

[0130] In the formula, δ represents the lines in the distribution network; α δ P represents the corrected line load factor. δ 、P δmax These are the actual transmitted active power value and the maximum transmittable power value, respectively; P DG N represents the remaining capacity of new energy equipment. l σ is the number of lines; W is the corrected weighted structure entropy; δ This refers to the importance parameter of the line.

[0131] By correcting the feeder load rate and weighted structural entropy based on the defect identification results, a candidate solution set reflecting the actual operating state of the distribution network is generated. This more accurately reflects the network's true condition under the influence of defects, providing a high-quality initial solution for subsequent optimization models. This correction method not only improves the accuracy and reliability of the candidate solution set but also enhances the adaptability and flexibility of the distribution network's self-healing strategy, thereby significantly improving the distribution network's self-healing capability and operational reliability under complex operating conditions.

[0132] Step S103: Input the candidate solution set into the upper-level optimization model. With the goal of maximizing the load recovery amount within the island, combine the line medium to guide the search direction and use the second-order cone relaxation method for optimization to obtain the load recovery scheme within the island. Input the candidate solution set and the load recovery scheme into the lower-level optimization model. With the total fault recovery amount target, operation reliability target and economic cost target after distribution network fault reconstruction as the multi-objective optimization function, and combine the improved sparrow optimization algorithm with inertia weight and learning factor to iteratively solve under constraints to obtain the load recovery result after fault reconstruction.

[0133] Please refer to Figure 3 In some embodiments, the process of inputting the candidate solution set into the upper-level optimization model, taking the maximization of load recovery within the island as the objective, combining the line medium to guide the search direction and using the second-order cone relaxation method for optimization, and obtaining the load recovery scheme within the island, includes steps S301 to S304.

[0134] Step S301: Construct an upper-level optimization model with the objective function of maximizing the load recovery within the island;

[0135] In some embodiments, the upper-level optimization model is constructed using maximizing the load restoration within the island as the objective function. This objective function aims to maximize the load restoration within the island, prioritizing the power supply to critical loads. For islands k formed within the distribution network... is Its optimization model is shown in the following equation: In the formula, f1 is the load for which power is restored; N D ψ is the number of load nodes within the island; ψ is the set of islands; ω i P represents the weighting coefficient. i The active power of the load; α i It is a binary variable.

[0136] It should be noted that, to ensure the stable operation of the independent power supply areas formed after islanding, priority must be given to ensuring the power supply to critical loads when dividing them. Furthermore, the division scheme must ensure that power generation and consumption within the island are balanced, while simultaneously meeting the voltage requirements of all loads and the current constraints of each line. This invention uses a random search to determine the maximum power supply area contained within each island.

[0137] Step S302: Based on the candidate solution set, extract the medium parameters of each line in the distribution network, and determine the initial direction of the optimization search and the boundary of the feasible region according to the medium parameters.

[0138] In some embodiments, based on the candidate solution set, the dielectric parameters (such as line betweenness numbers) of each line in the distribution network are extracted, and the relevant formula is: In the formula, b lThe line medium is T; the set of important nodes is T; i and j are distinct load nodes; N j N represents the number of shortest paths. i (l) represents the number of branches that overlap with the shortest path; I z→DG →N z The equivalent impedance is determined by the electrical characteristics between the load and the new energy equipment, based on these dielectric parameters b. l Determine the initial direction and feasible region boundary for the optimization search. Specifically, prioritize searching along lines with lower betweenness to find the nearest accessible islands to critical load nodes. This method effectively avoids frequent operations on critical lines, reducing interference with system stability.

[0139] It is important to note that in power system islanding, the line betweenness factor (LBD) is a key indicator of a line's importance within the network. Its core function is to reflect the proportion of the shortest paths between all pairs of nodes that pass through that line; a higher LBD value indicates a more important role for the line in network connectivity. A failure or disconnection of this line will significantly impact the overall system operation. Therefore, in the search process of upper-level islanding optimization strategies, the LBD is introduced as an important guiding parameter to improve the scientific rigor and efficiency of islanding. By prioritizing the search of lines with lower LBD values, frequent operations on critical lines can be effectively avoided, reducing interference with system stability. This ensures that the upper-level islanding optimization model can quickly identify the nearest accessible island to important load nodes, thereby achieving priority power supply for critical loads.

[0140] Step S303: For the non-convex constraints in the distribution network, the second-order cone relaxation method is used to transform the power balance constraint, network loss cost function and node voltage constraint into convex constraints to form a relaxation model.

[0141] In some embodiments, the upper-level model may contain non-convex or difficult-to-solve constraints. Therefore, for nonlinear constraints in the distribution network (such as power balance constraints and network loss cost functions), relaxation variables are introduced to transform them into the standard form of second-order cone programming. For example, for power balance constraints, second-order cone relaxation is used to transform them into the standard form of second-order cone programming, i.e., to obtain a relaxation model, thereby reducing the difficulty of solving the problem.

[0142] In some embodiments, the non-convex components in the original problem are transformed into the standard form of a second-order cone programming problem, specifically:

[0143]

[0144] ξ ij，t =U i，t U j，t G ij cosθij ;

[0145] η ij，t =U i，t U j，t B ij cosθ ij ;

[0146] The network loss cost and constraints containing square and trigonometric terms in the model, after second-order cone relaxation, are as follows:

[0147]

[0148]

[0149]

[0150] In the formula, G ij 、B ij Let θ be the real and imaginary parts of the admittance matrix of the distribution network nodes; ij The phase angle difference; ξ ij，t η ij，t 、u i，j p i，j and q i，j λ is a slack variable; ij p represents the weighting coefficients of the slack variables. i，t and Q i，t These represent the actual active and reactive power of the distributed generation, respectively; Δt is the time consumed in adjusting the network structure; R i C is the branch resistance; Loss The economic cost caused by transmission power loss; p ij，t and Q ij，t These represent the active power and reactive power flowing from node i to node j during time period t.

[0151] Step S304: The relaxation model is used to iteratively optimize the solutions in the candidate solution set in the initial direction and the boundary of the feasible region. The search direction is dynamically adjusted in combination with the medium parameters until the convergence condition is met. The load recovery scheme in the island is output, wherein the load recovery scheme includes the output adjustment value of the distributed power source and the access status of the recoverable load.

[0152] In some embodiments, firstly, the parameters of the optimization algorithm are initialized based on the candidate solution set and the initial direction. Then, in each iteration, the search direction is dynamically adjusted based on the fitness value of the current solution and the medium parameters; for example, variables corresponding to lines with lower betweenness factors are adjusted first to optimize islanding. When the change in the objective function value is less than a set threshold or the maximum number of iterations is reached, the iteration stops. The final output load recovery scheme includes the output adjustment value of the distributed power source and the access status of recoverable loads.

[0153] By combining line betweenness to guide the search direction and using the second-order cone relaxation method for optimization, we can quickly find the nearest accessible island to the critical load node, prioritize the restoration of power supply to critical loads, significantly improve the efficiency of islanding and the amount of load restoration, and enhance the power supply reliability of the distribution network under fault conditions.

[0154] In some embodiments, the construction process of the lower-level optimization model specifically involves: maximizing the total fault recovery amount after distribution network fault reconfiguration as the total fault recovery amount objective; minimizing the system mean fault time, mean fault frequency, and power shortage as the operational reliability objective; and minimizing the sum of switching action cost, power curtailment cost, and network loss cost as the economic cost objective. A multi-objective optimization function is then constructed based on the total fault recovery amount objective, the operational reliability objective, and the economic cost objective. The lower-level optimization model is then constructed by combining islanded power balance constraints, distribution network power balance constraints, node voltage constraints, branch current constraints, network topology constraints, and switching action count constraints.

[0155] In some embodiments, the expression of the lower-level optimization model is specifically:

[0156] The multi-objective optimization function is:

[0157]

[0158]

[0159]

[0160] In the formula, f2 represents the total fault recovery amount; z dis For load nodes outside the island; N dis This represents the number of load nodes outside the island. These are the weighting coefficients; This represents the active power of the load node; f3 is a binary variable; f3 is the reliability optimization function. and These represent mean time between failures, failure frequency, and power shortage, respectively; ω1, ω2, and ω3 are weighting coefficients; N sys This refers to the number of isolated islands. N represents the number of load nodes within the island. i T represents the total number of users; i Mean time between failures; f4 represents the total operating power of the distribution network; c represents the comprehensive cost. Swi Unit cost of switching action; Ω L Ω represents the number of network branches. New X represents the load number;i,t X i,t-1 t and t-1 represent the branch switch states during time periods t and t-1, respectively; Δt represents the time consumed for network structure adjustment; c Cur The penalty cost incurred for failing to fully utilize new energy sources; Loss These represent the economic costs caused by transmission power loss; R1 is the branch resistance; P i,t , Q i，t These represent the active and reactive power flowing into the branch, respectively; U i，t This refers to the voltage at the end node of the branch.

[0161] The island power balance constraint is:

[0162]

[0163] In the formula, The number of new energy equipment; For the number of loads; and These are the active power and reactive power of new energy equipment, respectively. and These are the active power and reactive power of the i-th node, respectively; and These represent the active and reactive power losses of the line:

[0164] The power balance constraint of the distribution network is:

[0165]

[0166] In the formula, b is the set of branches flowing into node i; P i,t and Q i,t These represent the actual active power and reactive power of the distributed power source, respectively; I ji,t 、P ji,t and Q ji,t For the current, active power, and reactive power transmitted in the branch; R ji X ji These are the resistance and reactance of the branch circuit, respectively.

[0167] The node voltage constraint is:

[0168]

[0169] In the formula, These are the upper and lower limits of the permissible voltage value, respectively.

[0170] The branch current constraint is:

[0171]

[0172] In the formula, The maximum allowable current;

[0173] The network topology constraints are:

[0174]

[0175] In the formula, X i,t L is the decision variable. Node , L Sub These represent the number of distribution network nodes and the number of substations, respectively; Ω g Ω is the set of disintegratable branches; c This is a collection of normally closed branches;

[0176] The constraint on the number of switching actions is:

[0177] k≤k max ;

[0178] In the formula, K max This represents the maximum number of actions allowed.

[0179] It should be noted that the total fault recovery target comprehensively reflects the system's ability to restore power supply after a fault. This indicator includes two parts: first, islanded load recovery, which is the load recovered by forming independent islanded areas after a fault in a part of the power grid; second, reconfigured load recovery, which is the load restored by reconnecting loads that were originally de-energized due to the fault to the main grid or a new power supply network through methods such as adjusting the grid topology and reconnecting lines. The operational reliability target needs to comprehensively consider the overall characteristics of multi-island operation and reconfigured networks, and uses the mean time between faults, the system's mean fault frequency, and the system's power shortage as assessment criteria to quantitatively evaluate the power supply reliability level of the distribution network system under different operating modes. The main components of the economic cost target include three parts: first, switching operation cost, which refers to the cost incurred when performing line switching operations to form islands or reconfigure networks; second, curtailment cost, which represents the economic loss caused by abandoning renewable energy generation or reducing load due to system constraints; and third, network loss cost, which reflects the economic cost of power loss during line transmission.

[0180] It should be noted that island power balance constraints refer to the total power generated by available power generation resources within a region being equal to the total power consumed by the load. This real-time dynamic balance between supply and demand directly affects the stability of island operation. Distribution network power balance constraints: After refactoring, the total power generation of each node in the distribution network is equal to the sum of the total load consumption and network losses, in order to maintain and ensure that the global power reaches a dynamic balance. Node voltage constraints ensure that the voltage amplitude of all nodes must be maintained within the set upper and lower limits to ensure power quality and safe operation of equipment. Branch current constraints: The current flowing through each line or branch does not exceed the set maximum limit to prevent line overload. Network topology constraints: The refactored distribution network topology must maintain connectivity, and each island or main network section should have a radial structure to avoid forming loops. Switching operation number constraints: During the refactoring or islanding process, the allowed number of switching operations does not exceed the set maximum value to reduce operating costs and the impact on system stability.

[0181] By comprehensively considering maximizing the total fault recovery, improving operational reliability, and reducing economic costs, a lower-level optimization model that fully reflects the fault reconfiguration process of the distribution network can be obtained.

[0182] Please refer to Figure 4 In some embodiments, the candidate solution set and the load recovery scheme are input into the lower-level optimization model, and the total fault recovery target, the operation reliability target and the economic cost target after the distribution network fault reconstruction are used as multi-objective optimization functions. The improved sparrow optimization algorithm with inertia weight and learning factor is combined to iteratively solve under constraints to obtain the load recovery result after fault reconstruction, including steps S401 to S404.

[0183] Step S401: Initialize the population of the improved star sparrow optimization algorithm based on the candidate solution set and the load recovery scheme, wherein each individual in the population corresponds to a fault reconstruction scheme;

[0184] In some embodiments, firstly, an initial population is randomly generated based on the candidate solution set and load restoration schemes. Each individual represents a possible fault reconfiguration scheme, and its position can be represented as a combination of switch states in the distribution network. Then, each fault reconfiguration scheme is encoded as an individual, with the individual's position vector representing the switch state (e.g., 1 for switch closed, 0 for switch open). Next, relevant parameters of the improved sparrow optimization algorithm are set, such as population size, maximum number of iterations, initial values ​​of inertia weights, and upper and lower bounds of the learning factor.

[0185] Step S402: During the iteration process, the search direction and step size of the starbird individual in the solution space are dynamically adjusted by introducing the inertial weight, and the learning ability of the starbird individual to the best position of the population is enhanced according to the adaptive learning factor so as to execute the food storage strategy and the cache recovery strategy respectively.

[0186] In some embodiments, firstly, in each iteration, the inertia weight is dynamically adjusted according to the current iteration number, gradually decreasing from its initial value to near zero. This helps enhance global search capability in the early stages and local search capability in the later stages. Then, the search direction and step size of the starling individual in the solution space are dynamically adjusted by introducing the inertia weight. Next, an adaptive learning factor is used to enhance the starling individual's ability to learn towards the population's optimal position. This learning factor can be dynamically adjusted according to the current iteration number and the individual's fitness, enabling the individual to better utilize global optimum information during the search process. Finally, based on the updated search direction and step size, a food storage strategy and a cache recovery strategy are executed, respectively. The food storage strategy is used for local search around the current solution, while the cache recovery strategy is used to obtain information from the global optimum to avoid getting trapped in local optima.

[0187] In some embodiments, the expression for the inertia weight W is:

[0188]

[0189]

[0190] In the formula, The minimum fitness in the t-th iteration; α represents the fitness of the i-th starbird in the t-th iteration; α is the sigmoid function parameter, whose value affects the algorithm's global and local search capabilities; MaxIt is the maximum number of iterations.

[0191] This introduces an inertia weight parameter to precisely control the search direction and step size of the individual starbird in the solution space. This mechanism allows the starbird to dynamically adjust its exploration and development behavior based on its current optimization stage and environmental feedback, thereby improving the probability and efficiency of the algorithm in finding the global optimum.

[0192] It should be noted that the core idea of ​​the adaptive learning factor c is to enhance information sharing and experience transfer among individuals, thereby accelerating the convergence of the population towards the optimal solution region. The magnitude of the adaptive learning factor directly affects the degree to which individuals follow the optimal position of the population. In the later stages of the algorithm, setting a larger learning factor can encourage individuals to move more quickly toward the currently known optimal position, thus accelerating the convergence speed of the algorithm.

[0193] Step S403: Update the position of the starbird through the food storage strategy and the cache recovery strategy, and evaluate the fitness of the individual solutions generated in each iteration under the constraints according to the multi-objective optimization function, so as to update the Pareto optimal solution set;

[0194] In some embodiments, firstly, the food storage strategy updates the individual positions by performing a local search around the current solution; the cache recovery strategy updates the individual positions by obtaining information from the global optimum. Then, based on a multi-objective optimization function (such as total fault recovery, operational reliability, and economic cost), the fitness of the individual solutions generated in each iteration is evaluated under constraints. The fitness evaluation can employ Pareto dominance relations to divide the individual solutions into different Pareto layers. Finally, based on the fitness evaluation results, the Pareto optimal solution set is updated, retaining individual solutions with higher fitness and eliminating those with lower fitness. Through multiple iterations, the Pareto optimal solution set is gradually optimized.

[0195] In some embodiments, the formula for updating the stargazer's position using the food storage strategy and the cache recovery strategy is specifically as follows:

[0196]

[0197] In the formula, c is the adaptive learning factor; up and c low represents the upper and lower bounds of the learning factor, respectively; t is the number of iterations; MaxIt is the maximum number of iterations; rand is a random number in the interval [0,1].

[0198] The location update formula for the food storage strategy is:

[0199]

[0200]

[0201] τ1≥τ2 and τ1≥τ3;

[0202] In the formula, Let i be the new position of the i-th starbird at time t+1; Let i be the new position of the i-th starbird at time t; λ represents the optimal position for the individual Star Sparrow; λ is a parameter generated based on Levi's flight; A and B are randomly selected indices; ι is a linear decreasing factor; τ1, τ2, τ3, and r1 are random numbers within the interval [0,1]; μ is a randomly selected parameter.

[0203] The location update formula for the cache recovery strategy is:

[0204]

[0205]

[0206] In the formula, Reference point information for storage location; r1 and r2 are random numbers within the interval [0,1]; τ3, τ4, and τ5 are random numbers within the interval [0,1]; C is a randomly selected index; This represents the j-th position of the i-th star before and after the update.

[0207] Step S404: When the preset termination condition is met, select the scheme with the best overall performance from the final Pareto optimal solution set as the load recovery result after fault reconstruction.

[0208] In some embodiments, it is determined whether a preset termination condition is met, such as reaching the maximum number of iterations or fitness convergence. From the final Pareto optimal solution set, the optimal solution is selected based on comprehensive performance indicators (such as a weighted sum of total fault recovery, operational reliability, and economic cost). Finally, the selected optimal solution is used as the load recovery result after fault reconstruction, and relevant parameters, such as the switching operation sequence and load recovery strategy, are output.

[0209] It should be noted that the improved algorithm can effectively avoid getting trapped in local optima and ensure that the recovery strategy found is optimal overall, thereby improving the efficiency and reliability of distribution network fault recovery.

[0210] By introducing an improved Pareto optimization algorithm, combined with a multi-objective optimization function considering total fault recovery, operational reliability, and economic cost, and performing iterative solutions under constraints, the search direction and step size can be dynamically adjusted, enhancing individual learning capabilities and thus quickly finding the optimal fault reconstruction scheme. Through updating and selecting the Pareto optimal solution set, it is ensured that the scheme with the best overall performance, while satisfying multiple constraints, is selected as the final load recovery result.

[0211] Step S104: Determine the target switch operation sequence and load restoration strategy based on the load restoration results, and control the distribution network to perform self-healing according to the target switch operation sequence and the load restoration strategy.

[0212] In some embodiments, firstly, based on the load restoration results obtained in step S403, these results include the optimal solution of a multi-objective optimization function, considering factors such as the load restoration amount after fault reconfiguration, operational reliability, and economic cost. Next, based on the load restoration results, a target switch operation sequence is generated. This step involves determining which switches need to be operated (e.g., opened or closed) and the order of operation based on the optimization results. The generation of the switch operation sequence aims to achieve load transfer from the fault area to the normal area, as well as islanding and reconfiguration. Simultaneously, a load restoration strategy is formulated, including determining which loads should be restored first and how to allocate remaining power resources. This strategy is based on factors such as operational reliability and economic cost considered in the optimization model. Then, the generated target switch operation sequence is input into the distribution network automation system. The automation system controls the corresponding switches to operate according to the operation sequence, thereby achieving distribution network reconfiguration. Finally, the distribution network automation system simultaneously controls the load restoration process according to the formulated load restoration strategy. This may involve load redistribution, priority settings, etc., to ensure the rapid restoration of critical and high-priority loads.

[0213] This application embodiment, by acquiring operational data of the distribution network and performing defect identification, can quickly and accurately identify defect parameters in the distribution network, thereby generating a candidate solution set reflecting the actual operating state of the distribution network. This provides the optimization model with an initial solution that is closer to the actual operating conditions, thus improving the optimization effect of the self-healing strategy. Combining line betweenness to guide the search direction and employing a second-order cone relaxation method for optimization, it can quickly find the nearest accessible island to important load nodes, prioritizing the restoration of power supply to critical loads. This significantly improves the efficiency of island partitioning and the amount of load restoration, enhancing the power supply reliability of the distribution network under fault conditions. An improved sparrow optimization algorithm, incorporating inertia weights and learning factors, comprehensively considers multiple objective optimization functions such as total fault recovery, operational reliability, and economic cost to solve the lower-level optimization model. This generates optimal fault reconfiguration and load restoration strategies, improving the self-healing capability and economic efficiency of the distribution network. By controlling the distribution network to self-heal according to the target switching operation sequence and load restoration strategy, rapid isolation and power restoration of fault areas can be achieved. This significantly shortens the recovery time, improves the self-healing efficiency of the distribution network, and reduces power outage losses. Compared with existing technologies, this application can improve the self-healing capability and operational reliability of the distribution network.

[0214] This application provides Figure 5 To make it easier to understand, specifically, it includes three steps:

[0215] Step 1: Model Transformation: Based on the power balance and other constraints in the constructed optimization model, a non-negative variable is introduced to relax it. The new mathematical model will no longer require the power to be strictly balanced at every node, but will allow it to fluctuate within a certain preset relaxation range, reducing the difficulty and complexity of subsequent optimization solutions.

[0216] Step Two: Upper-Level Model Based on Mathematical Optimization: The system uses real-time power grid status information (active power, line load rate, actual transmitted active power and maximum transmittable power, etc.), historical fault data (mean fault time, fault frequency, and power shortage, etc.), and predicted information on renewable energy sources (remaining capacity of renewable energy equipment) to determine the most likely fault areas. Simultaneously, based on the determined network topology, the betweenness number of each line is calculated. Based on this, and considering factors such as line betweenness number and load importance level, several potential island sets that can form independent operations are initially selected. Finally, the optimal island operation scheme is determined by solving the upper-level optimization problem.

[0217] Step 3: Lower-level model based on the improved star-sparrow algorithm: Generate the initial population required by the algorithm and initialize the algorithm parameters based on the optimal island solution output by the upper-level model. In each iteration, calculate and evaluate based on the specific objective of the lower-level multi-objective optimization to obtain the fitness value or objective function value of the candidate solution. Simultaneously, use improved inertia weights and learning factors to dynamically adjust the flight speed and direction of each candidate solution, enabling it to better explore the search space and utilize discovered advantageous information. At the end of each iteration, update the individual optimal position of each star-sparrow in the population and the overall population's optimal position information. Repeat the iteration process until the preset termination condition is met, and output the found optimal recovery strategy and the related multi-objective optimization results.

[0218] To verify the effectiveness of the present invention, the following example illustrates the advancement of the proposed method for optimizing the self-healing performance of distribution networks with distributed power sources based on defect identification.

[0219] Please refer to Figure 6 , Figure 6 This is a schematic diagram of the IEEE 69-node model. The simulation examples of this invention are based on the IEEE 69-node model, with the addition of distributed power sources. In the diagram, solid lines represent branches, and dashed lines represent tie switches. An equivalent model is built in Matlab / Simulink.

[0220] Assuming any branch can be disconnected, faults occur at switches 45-46 and 62-63, as shown in Table 3. Fault types include single-phase grounding, two-phase short circuit, three-phase short circuit, overload, open circuit, and undervoltage. Node load levels and weights are shown in Table 1, and new energy equipment access parameters are shown in Table 2. The improved Star Sparrow algorithm uses an initial population size of 50; a lower bound of -100 for variables; an upper bound of 100 for variables; an iteration dimension of 10; and a maximum number of iterations of 500.

[0221] Table 1 Distribution Network Fault Settings

[0222]

[0223] Table 2 Classification of Node Load Importance

[0224]

[0225] Table 3 Distributed Power Supply Access Information

[0226]

[0227] To verify the effectiveness of the self-healing defect correction strategy for distribution networks proposed in this invention, the following comparative schemes are shown. Table 4 compares the optimized performance of different schemes. Figure 7 This is a diagram comparing the probability of fault and load recovery.

[0228] Table 4 Comparison of optimization performance of different algorithms

[0229]

[0230] Among them, Scheme A generates candidate solutions based on random variables; Scheme B generates candidate solutions based on distribution network parameters; and Scheme C generates candidate solutions based on defect identification results.

[0231] As can be seen from the above data, the method proposed in this invention exhibits significant superiority in several key indicators. Regarding node correction, the method can more accurately identify and correct uncorrected nodes, effectively reducing the number of uncorrected nodes compared to other schemes, thus ensuring the stability and reliability of the distribution network node status. In terms of load recovery, the method achieves a higher load recovery rate, indicating that after a distribution network fault, it can restore users' normal power demand more quickly and comprehensively, improving the continuity and stability of power supply. Regarding fault handling efficiency, the method significantly shortens the mean time between failures, indicating that the strategy can respond quickly to faults, rapidly formulate and execute correction plans, and improve the response speed and effectiveness of the distribution network in dealing with faults. In terms of fault recovery rate and load recovery rate at different levels, both the overall fault recovery rate and the recovery rate of different load levels are higher than the other two schemes. This fully verifies that the distribution network self-healing effectiveness optimization strategy proposed in this invention can more comprehensively and deeply solve distribution network fault problems and ensure stable power supply to loads at all levels.

[0232] like Figure 8 As shown, based on the above method embodiments, corresponding apparatus embodiments are provided;

[0233] An embodiment of the present invention provides a self-healing system for a distribution network with distributed power sources based on defect identification, comprising: an acquisition module 100, an identification module 200, an optimization module 300, and a self-healing module 400;

[0234] The acquisition module 100 is used to acquire the operating data of the power distribution network;

[0235] The identification module 200 is used to input the operating data into a preset defect identification model to obtain defect identification results, and to correct the feeder load rate and weighted structural entropy based on the defect identification results to generate a candidate solution set that reflects the actual operating state of the distribution network. The feeder load rate and the weighted structural entropy are calculated based on the operating data.

[0236] The optimization module 300 is used to input the candidate solution set into the upper-level optimization model, with the goal of maximizing the load recovery amount within the island, combined with the line medium to guide the search direction and using the second-order cone relaxation method for optimization solution, to obtain the load recovery scheme within the island. The candidate solution set and the load recovery scheme are then input into the lower-level optimization model, with the total fault recovery amount target, operation reliability target and economic cost target after distribution network fault reconstruction as multi-objective optimization functions, and combined with the improved sparrow optimization algorithm that introduces inertia weight and learning factor to perform iterative solution under constraints, to obtain the load recovery result after fault reconstruction.

[0237] The self-healing module 400 is used to determine the target switch operation sequence and load restoration strategy based on the load restoration result, and control the distribution network to perform self-healing according to the target switch operation sequence and the load restoration strategy.

[0238] It is understood that the above-described device embodiments correspond to the method embodiments of the present invention, and can realize the self-healing method for distribution networks with distributed power sources based on defect identification provided by any of the above-described method embodiments of the present invention.

[0239] It should be noted that the device embodiments described above are merely illustrative, and some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can specifically be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0240] Based on the above embodiments of the self-healing method for distribution networks with distributed power sources based on defect identification, another embodiment of the present invention provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the self-healing method for distribution networks with distributed power sources based on defect identification according to any embodiment of the present invention.

[0241] For example, in this embodiment, the computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the terminal device.

[0242] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0243] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal device, connecting all parts of the terminal device via various interfaces and lines.

[0244] Based on the above-described method embodiments, another embodiment of the present invention provides a computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the self-healing method for a distribution network with distributed power sources based on defect identification as described in any of the above-described method embodiments of the present invention.

[0245] The modules / units integrated in the device / terminal equipment, if implemented as software functional units and sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0246] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A self-healing method for distribution networks with distributed power sources based on defect identification, characterized in that, include: Obtain operational data of the power distribution network; The operating data is input into a preset defect identification model to obtain defect identification results. The feeder load rate and weighted structural entropy are then corrected based on the defect identification results to generate a candidate solution set that reflects the actual operating state of the distribution network. The feeder load rate and the weighted structural entropy are calculated based on the operating data. The candidate solution set is input into the upper-level optimization model. With the goal of maximizing the load recovery within the island, the search direction is guided by the line medium, and the second-order cone relaxation method is used for optimization to obtain the load recovery scheme within the island. The candidate solution set and the load recovery scheme are input into the lower-level optimization model. The total fault recovery target, the operation reliability target, and the economic cost target after the distribution network fault reconstruction are used as multi-objective optimization functions. The improved sparrow optimization algorithm with the introduction of inertia weight and learning factor is used to iteratively solve the problem under constraints to obtain the load recovery result after fault reconstruction. Based on the load restoration results, a target switch operation sequence and load restoration strategy are determined, and the distribution network is controlled to self-heal according to the target switch operation sequence and the load restoration strategy.

2. The self-healing method for distribution networks with distributed power sources based on defect identification according to claim 1, characterized in that, The step of correcting the feeder load rate and weighted structural entropy based on the defect identification results to generate a candidate solution set reflecting the actual operating state of the distribution network is as follows: Based on the defect identification results, the defective feeder and the defect impact range are determined, wherein the defect impact range is the impact of the defect on the correlation of the feeder; The actual active power value transmitted in the defective feeder is corrected by combining the maximum transmittable power value to obtain the corrected feeder load rate. Based on the scope of the defect's influence, the line importance parameter in the weighted structural entropy is corrected to obtain the corrected weighted structural entropy. Based on the corrected feeder load rate and the weighted structural entropy, a candidate solution set containing different feeder load allocation schemes is constructed, wherein the candidate solution set includes several candidate solutions that satisfy the actual operating state of the distribution network under the influence of defects.

3. The self-healing method for distribution networks with distributed power sources based on defect identification according to claim 2, characterized in that, The formula for correcting the feeder load rate and weighted structural entropy based on the defect identification results to generate a candidate solution set reflecting the actual operating state of the distribution network is as follows: ln(2 / N l )<σ≤ln N l ; In the formula, δ represents the lines in the distribution network; α δ P represents the corrected line load factor. δ 、P δmax These are the actual transmitted active power value and the maximum transmittable power value, respectively; P DG N represents the remaining capacity of new energy equipment. l σ is the number of lines; W is the corrected weighted structure entropy; δ This refers to the importance parameter of the line.

4. The self-healing method for distribution networks with distributed power sources based on defect identification according to claim 1, characterized in that, The candidate solution set is input into the upper-level optimization model. With the objective of maximizing load restoration within the island, the search direction is guided by the line medium, and a second-order cone relaxation method is used for optimization to obtain the load restoration scheme within the island. Specifically: A higher-level optimization model is constructed using the maximization of load recovery within the isolated island as the objective function. Based on the candidate solution set, the medium parameters of each line in the distribution network are extracted, and the initial direction of the optimization search and the boundary of the feasible region are determined according to the medium parameters. For non-convex constraints in distribution networks, a second-order cone relaxation method is used to transform power balance constraints, network loss cost functions, and node voltage constraints into convex constraints to form a relaxation model. The relaxation model is used to iteratively optimize the solutions in the candidate solution set in the initial direction and the boundary of the feasible region. The search direction is dynamically adjusted in combination with the medium parameters until the convergence condition is met, and the load recovery scheme in the island is output. The load recovery scheme includes the output adjustment value of the distributed power source and the access status of the recoverable load.

5. The self-healing method for distribution networks with distributed power sources based on defect identification according to claim 4, characterized in that, The expression for the relaxation model is as follows: ξ ij，t =U i，t ,U j,t G ij cosθ ij ; or ij，t =U i,t U j，t B ij cosθ ij ; In the formula, G ij 、B ij Let θ be the real and imaginary parts of the admittance matrix of the distribution network nodes; ij The phase angle difference; ξ ij，t η ij，t 、u i，j p i，j and q i，j λ is a slack variable; ij p represents the weighting coefficients of the slack variables. i，t and Q i，t These represent the actual active and reactive power of the distributed generation, respectively; Δt is the time consumed in adjusting the network structure; R i C is the branch resistance; Loss The economic cost caused by transmission power loss; p ij，t and Q ij，t These represent the active power and reactive power flowing from node i to node j during time period t.

6. The self-healing method for distribution networks with distributed power sources based on defect identification according to claim 1, characterized in that, The construction process of the lower-level optimization model is as follows: The goal of maximizing the total fault recovery after distribution network fault reconfiguration is to maximize the total fault recovery. The goal of operational reliability is to minimize the system mean fault time, mean fault frequency, and power shortage. The goal of economic cost is to minimize the sum of switching action cost, power curtailment cost, and network loss cost. A multi-objective optimization function is constructed based on the goals of total fault recovery, operational reliability, and economic cost. The lower-level optimization model is constructed by combining island power balance constraints, distribution network power balance constraints, node voltage constraints, branch current constraints, network topology constraints, and switching action count constraints.

7. The self-healing method for distribution networks with distributed power sources based on defect identification according to claim 6, characterized in that, The expression for the lower-level optimization model is as follows: The multi-objective optimization function is: In the formula, f2 represents the total fault recovery amount; z dis For load nodes outside the island; N dis This represents the number of load nodes outside the island. These are the weighting coefficients; This represents the active power of the load node; f3 is a binary variable; f3 is the reliability optimization function. and These represent mean time between failures, failure frequency, and power shortage, respectively; ω1, ω2, and ω3 are weighting coefficients; N sys This refers to the number of isolated islands. N represents the number of load nodes within the island. i T represents the total number of users; i Mean time between failures; f4 represents the total operating power of the distribution network; c represents the comprehensive cost. Swi Unit cost of switching action; Ω L Ω represents the number of network branches. New X represents the load number; i，t X i，t-1 The branch switch statuses for time periods t and t-1 are respectively. Δt is the time consumed in adjusting the network structure; c Cur The penalty cost incurred for failing to fully utilize new energy sources; Loss These represent the economic costs caused by transmission power loss; R1 is the branch resistance; P i，t , Q i，t These represent the active and reactive power flowing into the branch, respectively; U i，t This refers to the voltage at the end node of the branch. The island power balance constraint is: In the formula, The number of new energy equipment; For the number of loads; and These are the active power and reactive power of new energy equipment, respectively. and These are the active power and reactive power of the i-th node, respectively; and These represent the active and reactive power losses of the line: The power balance constraint of the distribution network is: In the formula, b is the set of branches flowing into node i; P i，t and Q i，t These represent the actual active power and reactive power of the distributed power source, respectively; I ji,t 、P ji,t and Q ji,t For the current, active power, and reactive power transmitted in the branch; R ji X ji These are the resistance and reactance of the branch circuit, respectively. The node voltage constraint is: In the formula, These are the upper and lower limits of the permissible voltage value, respectively. The branch current constraint is: In the formula, The maximum allowable current; The network topology constraints are: In the formula, X i,t L is the decision variable. Node , L Sub These represent the number of distribution network nodes and the number of substations, respectively; Ω g Ω is the set of disintegratable branches; c This is a collection of normally closed branches; The constraint on the number of switching actions is: k≤k max ; In the formula, K max This represents the maximum number of actions allowed.

8. The self-healing method for distribution networks with distributed power sources based on defect identification according to claim 1, characterized in that, The candidate solution set and the load restoration scheme are input into the lower-level optimization model. The model uses the total fault recovery target, operational reliability target, and economic cost target after distribution network fault reconfiguration as multi-objective optimization functions. An improved sparrow optimization algorithm incorporating inertia weights and learning factors is used to iteratively solve the problem under constraints, yielding the load restoration result after fault reconfiguration. Specifically: The population of the improved sparrow optimization algorithm is initialized based on the candidate solution set and the load recovery scheme, wherein each individual in the population corresponds to a fault reconstruction scheme; During the iteration process, the search direction and step size of the starbird individual in the solution space are dynamically adjusted by introducing the inertial weight, and the learning ability of the starbird individual to the best position of the population is enhanced according to the adaptive learning factor to execute the food storage strategy and the cache recovery strategy respectively. The position of the starbird is updated by the food storage strategy and the cache recovery strategy, and the fitness of the individual solutions generated in each iteration is evaluated under the constraints according to the multi-objective optimization function in order to update the Pareto optimal solution set. When the preset termination conditions are met, the scheme with the best overall performance is selected from the final Pareto optimal solution set as the load recovery result after fault reconstruction.

9. The self-healing method for distribution networks with distributed power sources based on defect identification according to claim 8, characterized in that, The formula for updating the starbird's position using the food storage strategy and the cache recovery strategy is as follows: In the formula, c is the adaptive learning factor; up and c low represents the upper and lower bounds of the learning factor, respectively; t is the number of iterations; MaxIt is the maximum number of iterations; rand is a random number in the interval [0,1]. The location update formula for the food storage strategy is: τ1≥τ2 and τ1≥τ3; In the formula, Let i be the new position of the i-th starbird at time t+1; Let i be the new position of the i-th starbird at time t; λ represents the optimal position for the individual Star Sparrow; λ is a parameter generated based on Levi's flight; A and B are randomly selected indices; ι is a linear decreasing factor; τ1, τ2, τ3, and r1 are random numbers within the interval [0,1]; μ is a randomly selected parameter. The location update formula for the cache recovery strategy is: In the formula, Reference point information for storage location; r1 and r2 are random numbers within the interval [0,1]; τ3, τ4, and τ5 are random numbers within the interval [0,1]; C is a randomly selected index; This represents the j-th position of the i-th star before and after the update.

10. A self-healing system for a distribution network with distributed power sources based on defect identification, characterized in that, include: Acquisition module, identification module, optimization module, and self-healing module; The acquisition module is used to acquire the operating data of the power distribution network; The identification module is used to input the operating data into a preset defect identification model to obtain defect identification results, and to correct the feeder load rate and weighted structural entropy based on the defect identification results to generate a candidate solution set that reflects the actual operating state of the distribution network. The feeder load rate and the weighted structural entropy are calculated based on the operating data. The optimization module is used to input the candidate solution set into the upper-level optimization model, with the goal of maximizing the load recovery within the island. It combines the line medium to guide the search direction and uses the second-order cone relaxation method to optimize and solve the problem, thereby obtaining the load recovery scheme within the island. The candidate solution set and the load recovery scheme are then input into the lower-level optimization model, with the total fault recovery target, operational reliability target, and economic cost target after the distribution network fault reconstruction as the multi-objective optimization function. The improved sparrow optimization algorithm, which introduces inertia weights and learning factors, is used to iteratively solve the problem under constraints to obtain the load recovery result after the fault reconstruction. The self-healing module is used to determine the target switch operation sequence and load restoration strategy based on the load restoration result, and control the distribution network to perform self-healing according to the target switch operation sequence and the load restoration strategy.

Images