Power distribution network reconstruction method, system and device based on improved PSO algorithm and medium

Abstract

The invention relates to a power distribution network reconstruction method, system and device based on an improved PSO algorithm and a medium, and the method comprises the steps: carrying out the double-layer coding processing of a switching state and a particle swarm parameter, and generating an initial particle swarm meeting the radial constraint; performing classification processing in combination with real-time current data and historical fluctuation characteristics, dividing the power distribution network into a high-load key area, a medium-load transition area and a low-load elastic area, constructing a differential tabu search strategy, and dynamically adjusting tabu period parameters through regional importance weights and load rates; a dynamic inertia weight adjustment mechanism and differential tabu parameter collaborative optimization particle swarm optimization algorithm are adopted, and candidate reconstruction schemes meeting power grid operation constraints are screened; according to the method, the network loss, the voltage deviation and the load balancing indexes are quantified, an optimal scheme is output, a network topology reconstruction instruction is generated based on priority ranking, power distribution automation accurate regulation and control are achieved, and therefore the real-time performance, the economical efficiency and the operation reliability of power distribution network reconstruction are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power distribution network reconstruction, in particular to a power distribution network reconstruction method, system, device and medium based on an improved PSO algorithm. BACKGROUND

[0002] In modern power systems, power distribution network reconstruction is an important technical means to optimize power network operation efficiency, reduce line loss and improve power supply reliability. With the widespread access of distributed energy and the dynamic changes of load demand, the power distribution network needs to realize the reasonable distribution of power flow by flexibly adjusting the network topology structure under the premise of ensuring safe operation. Traditional reconstruction methods are usually based on static load prediction and offline calculation, and the switch state is combined and optimized through preset heuristic rules or mathematical optimization models, so as to reduce network loss, balance load and achieve other goals under specific operating scenarios. However, in actual power grid operation, the load fluctuates frequently and the network structure is complex, so the static optimization method is difficult to dynamically adapt to real-time changes, resulting in limitations in the timeliness and adaptability of the reconstruction scheme.

[0003] The existing power distribution network reconstruction technology has the following significant defects: first, most methods rely on fixed parameter settings and a single optimization target, making it difficult to coordinate conflicts between network loss minimization, voltage quality improvement and load balancing in dynamic scenarios; second, traditional optimization algorithms are inefficient in dealing with large-scale networks and are prone to local optimal solutions, making it difficult to quickly generate feasible solutions that meet real-time requirements; in addition, existing methods lack differentiated regulation capabilities for key areas, such as frequent operation in high-load areas which may shorten the service life of equipment, and over-optimization in low-load areas which may result in resource waste. These problems result in a large gap between the actual application effect and the theoretical expectation of power distribution network reconstruction. SUMMARY

[0004] Therefore, the purpose of the present application is to provide a power distribution network reconstruction method, system, device and medium based on an improved PSO algorithm, which can dynamically respond to load changes, accurately balance multi-objective conflicts and improve power grid operation reliability.

[0005] The purpose of the present application is achieved by the following scheme:

[0006] In a first aspect, the present application provides a power distribution network reconstruction method based on an improved PSO algorithm, comprising the following steps:

[0007] S1: Based on the topology structure of the power distribution network, real-time load data and historical load data, the switch state and PSO parameters are processed by double-layer encoding, all switch nodes are traversed and radial constraints are verified, an initial particle swarm is generated, and the initial particle swarm includes double-layer encoding of switch state and PSO parameters;

[0008] S2: According to the switch state corresponding to each particle in the initial particle group, the load characteristics of each region of the power distribution network are analyzed and processed, the load characteristic index is calculated in combination with the real-time current value and historical fluctuation data, and the classification result of the high-load key area, the medium-load transition area and the low-load elastic area is generated through the preset load classification threshold interval;

[0009] S3: Based on the classification result, the taboo search strategy is dynamically parameterized, and a differentiated taboo length parameter is generated, including high-load key area taboo length parameter, medium-load transition area taboo length parameter and low-load elastic area taboo length parameter;

[0010] S4: Based on the improved PSO algorithm, the cooperative optimization of the differentiated taboo length parameter is processed, the particle position is dynamically updated and the non-taboo candidate scheme is selected, and a candidate reconstruction scheme set satisfying the power grid operation constraint is generated, and the improved PSO algorithm is realized by introducing a dynamic inertia weight adjustment mechanism;

[0011] S5: Each scheme in the candidate reconstruction scheme set is subjected to multi-objective comprehensive evaluation processing, and the scheme with the optimal comprehensive evaluation index is selected, and the optimal reconstruction scheme is output;

[0012] S6: Based on the priority sorting result of the optimal reconstruction scheme, the control instruction sequence is generated according to the region importance and the load emergency degree, the network topology reconstruction instruction is generated, and the network topology reconstruction instruction is sent to the power distribution automation execution mechanism.

[0013] In the second aspect, the application provides a power distribution network reconstruction system based on an improved PSO algorithm, which is configured with the following modules:

[0014] The improved PSO and actual coding module is used for double-layer coding processing of switch state and PSO parameter based on the topology structure of the power distribution network, real-time load data and historical load data, traversing all switch nodes and verifying the radial constraint, generating an initial particle group, and the initial particle group including double-layer coding of switch state and PSO parameter;

[0015] The load characteristic analysis module is used for analyzing and processing the load characteristics of each region of the power distribution network according to the switch state corresponding to each particle in the initial particle group, calculating the load characteristic index in combination with the real-time current value and historical fluctuation data, and classifying through the preset load classification threshold interval, generating the classification result of the high-load key area, the medium-load transition area and the low-load elastic area;

[0016] The dynamic taboo strategy generation module is used for dynamically parameterizing the taboo search strategy based on the classification result, and generating a differentiated taboo length parameter, including high-load key area taboo length parameter, medium-load transition area taboo length parameter and low-load elastic area taboo length parameter.

[0017] The PSO-tabu cooperative optimization module is used for cooperative optimization processing of the differentiated tabu length parameters based on the improved PSO algorithm, dynamic updating of the particle position and screening of the non-tabu candidate scheme, and generation of a candidate reconstruction scheme set satisfying the power grid operation constraint, and the improved PSO algorithm is realized by introducing a dynamic inertia weight adjustment mechanism;

[0018] The multi-objective comprehensive evaluation module is used for multi-objective comprehensive evaluation processing of each scheme in the candidate reconstruction scheme set, selection of a scheme with optimal comprehensive evaluation indexes, and output of an optimal reconstruction scheme.

[0019] The control instruction generation and execution module is used for generating a control instruction sequence according to the regional importance and load emergency degree based on the priority sorting result of the optimal reconstruction scheme, generating a network topology reconstruction instruction, and sending the network topology reconstruction instruction to a power distribution automation execution mechanism.

[0020] In a third aspect, the present application provides a computer device, comprising a memory and a processor, the memory stores a computer program, and the processor realizes any one of the power distribution grid reconstruction methods based on the improved PSO algorithm when executing the computer program.

[0021] In a fourth aspect, the present application provides a computer readable storage medium, which stores a computer program, and the computer program realizes any one of the power distribution grid reconstruction methods based on the improved PSO algorithm when executed by a processor.

[0022] In summary, the power distribution grid reconstruction method based on the improved PSO algorithm provided by the present application solves the problems of low optimization efficiency, poor adaptability, and limited power grid operation reliability caused by the solidification of static parameters, the insufficient conflict regulation of multi-objective, and the lack of differentiated processing of key areas through a progressive technical chain of dynamic coding, regional classification, tabu cooperation, multi-objective optimization, and closed-loop instruction, significantly improves the dynamic response capability and comprehensive operation efficiency of power distribution grid reconstruction, and provides effective technical support for the safe and economic operation of power grids under high proportion of new energy access and complex load fluctuation scenarios.

[0023] In order to better understand and implement, the present application is described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0024] Figure 1 A flowchart of a power distribution grid reconstruction method based on an improved PSO algorithm provided by an embodiment of the present application is shown in the figure.

[0025] Figure 2 A flowchart of generating differentiated tabu length parameters provided by an embodiment of the present application is shown in the figure.

[0026] Figure 3A schematic diagram illustrating the process of generating a set of candidate reconstruction schemes provided in an embodiment of this application;

[0027] Figure 4 A schematic diagram illustrating the process of generating the optimal reconstruction scheme provided in an embodiment of this application;

[0028] Figure 5 This is a schematic diagram of a distribution network reconfiguration system based on an improved PSO algorithm, provided as another embodiment of this application. Detailed Implementation

[0029] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention.

[0030] 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 invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0031] In one embodiment, such as Figure 1 As shown, a distribution network reconfiguration method based on an improved PSO algorithm is provided. This embodiment illustrates the method by applying it to a terminal. It is understood that this method can also be applied to a server, and further to a system including both a terminal and a server, and is implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0032] S1: Based on the topology of the distribution network, real-time load data and historical load data, perform double-layer encoding processing on switch status and PSO parameters, traverse all switch nodes and verify radial constraints, and generate an initial particle swarm, which includes double-layer encoding of switch status and PSO parameters.

[0033] Specifically, the system comprehensively collects topology information of the distribution network, covering detailed data such as the connection relationships of each node, line parameters, and the attributes of switching equipment, while also acquiring real-time load data and historical load data. This data undergoes preprocessing operations such as denoising and normalization to eliminate differences in dimensions and data magnitudes, ensuring data accuracy and consistency. Simultaneously, the system employs a two-layer coding strategy to process switch states and Particle Swarm Optimization (PSO) parameters. The upper-layer coding is specifically used to represent switch states. Based on the distribution network topology, the state of each switch is precisely set to open or closed, and represented using binary encoding, where each bit corresponds to a switch state, ensuring a complete and accurate description of the distribution network topology. The lower-layer coding is used to represent PSO parameters, including key parameters such as particle position, velocity, and inertia weight. A reasonable coding design is implemented based on the specific needs of the algorithm and the parameter range; for example, the initial value of the inertia weight is set within a reasonable range to balance the algorithm's global exploration capability and local exploitation capability.

[0034] After receiving data collection and preprocessing signals, the system encodes the switch states and PSO parameters according to preset encoding rules. Following the distribution network topology, it sequentially traverses all switch nodes, checking each node's radial constraints under its current switch state. Radial constraints are a crucial characteristic of distribution networks, ensuring the absence of loop networks and guaranteeing grid safety and reliability. During the traversal, the system executes a series of logical judgments and mathematical operations to verify that the connection relationships of each switch node conform to the requirements of a radial network. For cases where constraints are not met, the system automatically adjusts the switch state encoding until all switch nodes satisfy the radial constraints. During this process, the system records each switch state encoding that satisfies the constraints, serving as the basis for generating the initial particle swarm. Based on the encoded switch states and PSO parameters, the system randomly generates the initial particle swarm. The size of the initial particle swarm is reasonably determined based on the scale and complexity of the distribution network, ensuring that the particle swarm can cover all possible operating states of the distribution network. Each particle in the particle swarm contains a two-layer encoded information of complete switch state and PSO parameters, and these particles will serve as the initial search points for subsequent optimization algorithms.

[0035] S2: Based on the switching state of each particle in the initial particle swarm, analyze and process the load characteristics of each area of ​​the distribution network, calculate the load characteristic index by combining real-time current value and historical fluctuation data, and classify the loads through the preset load classification threshold range to generate classification results of high load critical area, medium load transition area and low load elastic area.

[0036] Specifically, for each region, the system calculates the current load level by combining real-time current values ​​and deeply mines historical fluctuation data to comprehensively analyze the historical trends and fluctuation amplitudes of the load. Based on this, the system constructs a comprehensive load characteristic index system, which includes, but is not limited to, key indicators such as load factor and fluctuation coefficient, to comprehensively reflect the load characteristics of each region. For example, the load factor measures the load level of a region by calculating the ratio of the actual load to the rated load, while the fluctuation coefficient assesses the stability of the load by analyzing the ratio of the standard deviation to the average value of historical load data.

[0037] Based on historical power grid operation data and expert experience, the system pre-defines classification threshold ranges for high-load critical areas, medium-load transition areas, and low-load flexible areas. For example, areas with a load factor greater than 80% and a fluctuation coefficient greater than 0.3 are classified as high-load critical areas; areas with a load factor between 50% and 80% and a fluctuation coefficient between 0.1 and 0.3 are classified as medium-load transition areas; and areas with a load factor less than 50% and a fluctuation coefficient less than 0.1 are classified as low-load flexible areas. These threshold settings fully consider the safe operation requirements of the power grid and the differences in load characteristics across different areas, aiming to provide accurate classification criteria for subsequent optimization strategies. After obtaining the load characteristic indicators for each area, the system compares and matches these indicators with the pre-defined classification threshold ranges one by one. Through complex logical judgments and mathematical calculations, the system accurately classifies the distribution network, ultimately generating detailed classification results for high-load critical areas, medium-load transition areas, and low-load flexible areas. This classification result is stored in the system's memory in the form of a data structure, providing an important basis for subsequent tabu search strategy parameterization, enabling the optimization algorithm to process and optimize different load areas more specifically.

[0038] Throughout the classification process, the system records and updates the classification information for each region in real time to ensure the accuracy and timeliness of the classification results, so that subsequent steps can be effectively optimized based on the latest classification information.

[0039] S3: Based on the classification results, the taboo search strategy is dynamically parameterized to generate differentiated taboo length parameters, including taboo length parameters for high-load critical areas, medium-load transition areas, and low-load elastic areas.

[0040] Specifically, based on the significant impact of high-load critical areas on grid security and reliability, the system assigns longer prohibitory length parameters to these areas. This longer prohibitory length means that during optimization, the system will limit frequent operations on high-load critical areas, effectively preventing shortened equipment lifespan due to excessive operation and ensuring that equipment in high-load critical areas operates in a relatively stable state. For medium-load transition areas, the system comprehensively considers its load characteristics and optimization requirements, setting a medium-length prohibitory length parameter. This setting aims to balance optimization effectiveness and operational rationality, allowing the system to adjust medium-load transition areas to optimize grid operation to a certain extent while preventing unnecessary wear and tear on equipment from excessively frequent operations. For low-load flexible areas, due to their relatively low sensitivity to operation, the system sets shorter prohibitory length parameters. This allows the system to perform more flexible optimization adjustments to low-load flexible areas without affecting grid security, fully utilizing the flexibility of this area to improve the overall operating efficiency of the grid.

[0041] During the optimization process, the system continuously monitors the load characteristics of each region, collecting real-time power grid operating data such as current, voltage, and power, and using a pre-defined algorithm model to analyze and predict load change trends. When a significant change in the load characteristics of a high-load critical area is detected, the system automatically increases or decreases the taboo length parameter for that area according to predefined adjustment rules. For example, if the load rate of a high-load critical area continues to rise and approaches the equipment's load limit, the system will appropriately extend the taboo length and reduce the frequency of operation in that area to ensure the safe operation of the equipment. Similarly, for medium-load transition areas and low-load elastic areas, the system will dynamically adjust the corresponding taboo length parameters based on real-time changes in their load characteristics, ensuring that the taboo search strategy always closely matches the actual operating state of the power grid, thereby improving the performance and effectiveness of the optimization algorithm and making the optimization process more adaptable and effective.

[0042] S4: Based on the improved PSO algorithm for the collaborative optimization of the differentiated taboo length parameter, the particle position is dynamically updated and non-taboo candidate schemes are screened to generate a set of candidate reconfiguration schemes that meet the power grid operation constraints. The improved PSO algorithm is achieved by introducing a dynamic inertia weight adjustment mechanism.

[0043] Specifically, the improved PSO algorithm enhances its performance in distribution network reconfiguration optimization by introducing a dynamic inertia weight adjustment mechanism. This mechanism intelligently adjusts the inertia weight based on the number of iterations and the current state of the particles. In the initial optimization phase, the inertia weight is set to a large value, allowing the particles to conduct a broad global search, effectively expanding the search scope and preventing the algorithm from prematurely getting trapped in local optima. As the number of iterations increases, the system gradually reduces the inertia weight value, shifting the particle's search behavior from global exploration to local exploration. This shift helps the particles perform refined searches within local regions, improving search accuracy and convergence speed, and ensuring the algorithm efficiently finds high-quality solutions.

[0044] During the optimization process, the system organically combines the tabu search strategy with the improved PSO algorithm. When the PSO algorithm iteratively updates particle positions, the system constrains the particle's movement direction and updated position based on the tabu length parameter. Specifically, if the particle's movement direction points to a tabu region, the system automatically prevents the particle from entering that region and adjusts the particle's speed and position according to preset rules, guiding the particle to move towards a non-tabu region. Simultaneously, the system verifies the adjusted particle positions against power grid operation constraints, including node voltage constraints and line power constraints. Only when a particle position satisfies all constraints will the system consider it a valid candidate reconfiguration scheme and include it in the candidate reconfiguration scheme set.

[0045] To further improve the quality of the candidate reconfiguration scheme set, the system performed a series of screening and optimization operations on each candidate scheme. First, the system conducted an initial screening, eliminating schemes that clearly did not meet the requirements of power grid operation. Then, for the remaining candidate schemes, the system used a preset optimization algorithm to further refine them, improving their feasibility and effectiveness. Throughout the process, the system continuously updated and optimized the candidate reconfiguration scheme set to ensure it provided a sufficiently rich and high-quality selection for subsequent multi-objective comprehensive evaluation.

[0046] S5: Perform multi-objective comprehensive evaluation on each scheme in the candidate reconstruction scheme set, select the scheme with the best comprehensive evaluation index, and output the optimal reconstruction scheme.

[0047] Specifically, the system constructs a comprehensive evaluation index system to process each candidate reconfiguration scheme. This system covers multiple key indicators such as network loss, voltage quality, and load balance. The network loss index accurately measures the operational economy of the power grid by calculating the active and reactive power losses under different operating conditions. The system utilizes efficient numerical calculation methods to accurately assess the network loss corresponding to each candidate reconfiguration scheme, providing a strong basis for subsequent optimization decisions. The voltage quality index comprehensively evaluates the power supply quality and stability of the power grid by analyzing the voltage amplitude deviation and voltage fluctuation at each node. Based on preset voltage level standards and allowable voltage deviation ranges, the system quantitatively analyzes the voltage quality of each node, ensuring that the power grid can provide users with charts, results, and specific numerical values, displaying the load characteristic index values ​​of each region and the corresponding classification results, intuitively presenting the load distribution of the distribution network, and providing a clear direction for subsequent optimization strategies.

[0048] S6: Based on the priority ranking results of the optimal reconfiguration scheme, generate a sequence of control instructions according to regional importance and load urgency, generate network topology reconfiguration instructions, and send the network topology reconfiguration instructions to the distribution automation actuator.

[0049] Specifically, based on key information such as the reconfiguration priority and load urgency of each region contained in the optimal reconfiguration scheme, this embodiment formulates a priority ranking rule. This rule comprehensively considers multiple factors, including the urgency of high-load critical areas, the potential risks of medium-load transition areas, and the adjustment flexibility of low-load flexible areas, to ensure that the order of control command transmission best matches the actual operational needs of the power grid. For example, for regions located in high-load critical areas with extremely high load urgency, the system assigns them the highest priority, prioritizing the generation and transmission of corresponding control commands to quickly alleviate their high-load operating state, promptly eliminate potential power grid operational risks, and ensure the safe and stable operation of the power grid. For medium-load transition areas and low-load flexible areas, they are ranked according to their respective load characteristics and importance, and the order of control command transmission is rationally arranged to ensure that the power grid can achieve optimized adjustment of its overall operating state in the shortest possible time.

[0050] During the generation of network topology reconfiguration instructions, the system performs detailed analysis of the switching operation sequence in the optimal reconfiguration scheme, clarifying the specific operating state and sequence of each switching device. These instructions are encapsulated in a standardized instruction format to ensure their accuracy and executability. The system performs multiple checks and verifications on the generated control instruction sequence, including instruction integrity checks, logical consistency checks, and matching checks with the current operating state of the power grid. Only after the control instruction sequence passes all verification stages will the system send it to the distribution automation actuator. Upon receiving the instruction, the distribution automation actuator responds quickly and performs precise switching operations on the switching devices in the distribution network according to the instruction requirements, thereby achieving network topology reconfiguration.

[0051] In summary, the distribution network reconfiguration method based on the improved PSO algorithm provided by this invention fuses the distribution network topology, real-time load data, and historical load data through a two-layer encoding process to construct an initial particle swarm containing switch states and PSO parameters. This enables dynamic optimization starting point generation in complex power grid scenarios. Based on load characteristic analysis and classification threshold determination using real-time current values ​​and historical fluctuation data, high, medium, and low load regions are dynamically divided to accurately identify key control areas and adapt to differentiated operational needs. By dynamically parameterizing the tabu search strategy in conjunction with the region classification results, long-term operation in high-load areas can be achieved. The PSO algorithm is improved by incorporating periodic taboo constraints and flexible short-cycle control in low-load areas, effectively balancing global search efficiency and local convergence accuracy. A dynamic inertia weight adjustment mechanism is introduced to improve the PSO algorithm, and combined with differentiated taboo parameters to collaboratively optimize the particle update process. This overcomes the limitation of traditional algorithms easily getting trapped in local optima, significantly improving the feasibility and diversity of candidate solutions. A multi-objective comprehensive evaluation model is used to quantify network loss, voltage deviation, and load balancing indicators to achieve Pareto optimal solution selection for reconfiguration schemes under multiple constraints. A security verification mechanism based on priority sorting and instruction compilation enables accurate generation and reliable issuance of network topology reconfiguration instructions.

[0052] The aforementioned distribution network reconfiguration method based on the improved PSO algorithm systematically solves the problems of low optimization efficiency, poor adaptability, and limited grid operation reliability caused by the existing methods due to the rigidity of static parameters, insufficient multi-objective conflict control, and lack of differentiated processing of key areas. It significantly improves the dynamic response capability and comprehensive operation efficiency of distribution network reconfiguration, and provides effective technical support for the safe and economical operation of the power grid under scenarios of high proportion of new energy access and complex load fluctuations.

[0053] In one embodiment, S1 of the distribution network reconfiguration method based on the improved PSO algorithm provided by the present invention specifically includes the following steps:

[0054] S11: Based on the topology of the distribution network, the switch states are processed by binary encoding. All switch nodes are traversed and mapped to a 0 / 1 state sequence to generate a binary switch code that satisfies the radial constraint.

[0055] Specifically, the system conducts a detailed analysis of the distribution network topology to determine the location and connection relationships of each switch node within the network. Based on the physical location and electrical connections of the switches, the system constructs a node-branch model of the distribution network, which is stored in the system's database in matrix form. Based on this model, the system initiates a binary encoding process for switch states. For each switch in the distribution network, the system maps its open / closed state to a binary "0" or "1," where "1" represents a closed switch, indicating that the corresponding branch is in a conducting state; and "0" represents an open switch, indicating that the corresponding branch is in a disconnected state. This binary encoding method is concise and efficient, comprehensively covering all possible combinations of switch states in the distribution network.

[0056] The system visits each switch node in the distribution network one by one according to a predetermined traversal algorithm. During the traversal, the system reads the current state of each switch in real time and converts it into corresponding binary bits. These binary bits are arranged sequentially according to the switch node number, forming a complete binary state sequence. This sequence fully reflects the topological connection status of the distribution network at the current moment.

[0057] Preferably, the system employs a topology verification algorithm to convert the binary state sequence into a corresponding graph structure, where nodes represent switches and load points, and edges represent closed branches. Using a depth-first search or breadth-first search algorithm, the system checks the connectivity of each node, ensuring that all nodes can be connected via only one path, thus forming a radial network. If loops or isolated nodes are detected, the system automatically triggers an adjustment mechanism. This mechanism, based on preset priority rules, prioritizes adjusting switch states located on critical paths or those significantly impacting network connectivity. Through continuous adjustment and verification cycles, the system ultimately generates a switch state encoding sequence that conforms to both binary encoding rules and radial constraints.

[0058] S12: Based on the real-time load data and historical load data of the distribution network, the inertia weight, individual learning factor and social learning factor of the PSO algorithm are encoded with real numbers and randomly initialized within the preset parameter range to generate an algorithm parameter vector containing the inertia weight, individual learning factor and social learning factor.

[0059] Specifically, during the preparation phase of distribution network reconfiguration, the computer system processes key parameters of the Particle Swarm Optimization (PSO) algorithm in parallel. The system acquires current load data from monitoring equipment in real time and retrieves past load records from historical databases. This data includes key indicators such as active power, reactive power, and load change rate at various nodes in the distribution network over different time periods. Through in-depth analysis of this data, the system can comprehensively understand the current operating status of the distribution network and its potential future trends.

[0060] Based on these analytical results, the system began to encode the three core parameters of the PSO algorithm—inertia weight, individual learning factor, and social learning factor—with real numbers. Inertia weight determines the tendency of a particle to maintain its original direction of motion during the search process; a larger inertia weight is beneficial for global search, while a smaller inertia weight is beneficial for local search. The individual learning factor reflects the particle's emphasis on its own historical best position, while the social learning factor reflects the particle's tendency to follow the group's best position. The system presets a reasonable value range for each parameter based on the scale, complexity, and load characteristics of the distribution network. For example, inertia weight is typically between 0.4 and 0.9, while the individual learning factor and social learning factor are in the range of 1.5 to 2.5.

[0061] To ensure initial diversity in the PSO algorithm, the system employs a random initialization strategy. During initialization, random values ​​are generated within a preset range for each parameter. These random values ​​are then standardized and combined into a three-element real-number vector, which serves as the algorithm parameter vector. The system generates multiple such vectors, forming an initial parameter set. Each parameter vector represents a unique PSO algorithm configuration, guiding the particle swarm's search behavior during subsequent optimization.

[0062] After generating these parameter vectors, the system performs a preliminary performance evaluation. By simulating the PSO algorithm, the system calculates the convergence rate and solution quality of each parameter vector on a simple test function. Based on the evaluation results, the system selects the parameter vectors with better performance, forming an optimal set of initial parameters. These parameter vectors not only possess good initial diversity but also have certain optimization potential, providing strong support for the subsequent distribution network reconfiguration process. In this way, the system ensures that the parameter configuration of the PSO algorithm can achieve efficient optimization capabilities while meeting the operational requirements of the distribution network.

[0063] S13: Process the binary switch code and algorithm parameter vector, and generate the initial particle swarm through topological verification and parameter combination processing.

[0064] Specifically, the system associates and combines each binary switch code with its corresponding algorithm parameter vector to form a complete particle representation. Each particle not only contains the topology information of the distribution network but also carries the control parameters of the PSO algorithm. This allows the particle to simultaneously optimize and adjust the network topology and algorithm parameters as it moves through the search space. The system performs topology verification on each particle, rigorously checking whether it meets the radial constraints of the distribution network, ensuring that there are no loops in the network to guarantee the safety and reliability of the power grid. For particles that do not meet the constraints, the system initiates a correction mechanism, adjusting some binary bits in the switch code to make it conform to the radial constraints. For example, the system checks whether the network topology corresponding to each particle is connected, whether there are loops, and whether it meets constraints such as voltage level and line capacity. If a situation where constraints are not met is found, the system will randomly select one or more switch states to toggle according to predefined rules until all constraints are met.

[0065] Simultaneously, the system verifies the validity of the algorithm parameter vectors within the particles, ensuring that the values ​​of inertia weight, individual learning factor, and social learning factor are within a reasonable range and can coordinate with the switching coding. After topology verification and parameter combination processing, the system generates an initial particle swarm. Each particle in this swarm represents a possible distribution network operating state and its corresponding optimized parameter configuration, laying a solid foundation for subsequent optimization iterations. The system rationally determines the size of the initial particle swarm based on the scale and complexity of the distribution network, ensuring that the particle swarm can cover all possible operating states of the distribution network and possess sufficient diversity in algorithm parameters. This improves the global search capability and convergence speed of the optimization algorithm, providing a strong guarantee for ultimately finding the optimal distribution network reconfiguration scheme.

[0066] In one embodiment, step S2 of the distribution network reconfiguration method based on the improved PSO algorithm provided by the present invention specifically includes the following steps:

[0067] S21: Analyze the switching state of each particle in the initial particle swarm, extract the switching on / off state of each area of ​​the distribution network, and generate a state mapping table of the area switches.

[0068] Specifically, the system sequentially extracts the switch state code of each particle from the initial particle swarm. These codes, in the form of binary sequences, record the on / off status of each switch in the distribution network. The system parses the code of each particle, and based on the topology information of the distribution network, precisely matches each bit in the binary sequence with the corresponding switch node to identify whether each switch node is in a closed or open state. Building on this, the system further analyzes the connection relationships between the switch nodes to determine the boundaries and composition of each region.

[0069] For each area of ​​the distribution network, the system generates a detailed area switch status mapping table based on the on / off status of all switch nodes within that area. This mapping table, in the form of a data structure, clearly displays the on / off status of each switch node in each area, providing an accurate topological basis for subsequent load characteristic analysis. During the generation of the mapping table, the system strictly adheres to preset data formats and specifications to ensure the completeness and accuracy of the information. Simultaneously, the system also verifies the mapping table, checking for logical contradictions or omissions to ensure that the mapping table accurately reflects the operating status of the distribution network.

[0070] S22: Based on the state mapping table of the regional switch, combined with the collected real-time current values ​​of the distribution network and the fluctuation data in the historical load database, load characteristic indicators are calculated and processed to calculate the load rate, fluctuation rate and voltage deviation, and generate load characteristic indicators for each region.

[0071] Specifically, for each region, the system first calculates the current load factor based on the real-time current value and the region's rated load parameters. Specifically, the system retrieves equipment parameters stored in the database, such as the transformer's rated capacity and the line's rated current, and combines this with real-time monitored current and voltage data. Using a power calculation formula, it derives the actual current load power, which is then compared to the rated power to obtain the load factor value. This process requires the system to possess efficient computing power and accurate data processing capabilities to ensure that the calculated load factor accurately reflects the current load level of the region. The load factor calculation formula is as follows:

[0072]

[0073]

[0074] in, (Load Ratio) is the load factor. This is the actual load power of the area, expressed in kilowatts (kW). It is the real-time voltage value. This is the real-time current value. It is the power factor. This is the rated load power of the area, in kilowatts (kW), obtained from the equipment parameter database.

[0075] Next, the system delves into the fluctuation data in the historical load database and calculates the load volatility for the region using statistical analysis methods. The system filters historical data within specific time periods, such as the past 24 hours, week, or month, to ensure the calculation results fully reflect the short-term and long-term fluctuation characteristics of the load. The volatility calculation involves analyzing the standard deviation and mean of historical load data. The system processes a large amount of historical data by calling statistical analysis algorithms to obtain the standard deviation and mean of load fluctuations, and then calculates the volatility. Simultaneously, the system considers the load fluctuation characteristics of different time periods, such as the differences between weekdays and weekends, and daytime and nighttime, to perform a more detailed analysis of the volatility, thus more comprehensively reflecting the dynamic changes in the load. Preferably, volatility is used to quantify the stability of load changes over time, and the calculation formula is as follows:

[0076]

[0077]

[0078]

[0079] in, Fluctuation Ratio is the volatility. The standard deviation of the historical load data. This is the average of historical load data. Let N be the load power at the i-th time point, and N be the number of data points.

[0080] Simultaneously, the system analyzes voltage data within the area and calculates voltage deviation indicators. This requires comparing real-time voltage measurements with the system's rated voltage value, and using a preset deviation calculation formula to derive the specific value of the voltage deviation. The system retrieves the rated voltage value stored in the database and compares it point-by-point with the real-time collected voltage data to calculate the voltage deviation at each monitoring point. Then, statistical methods are used to derive indicators such as the average voltage deviation and maximum voltage deviation within the area. Voltage deviation is a crucial parameter for assessing power quality, reflecting the stability and reliability of power supply. The system's accurate calculation of voltage deviation provides a key basis for subsequent classification and processing. The voltage deviation is a key indicator for assessing power quality, and its calculation formula is as follows:

[0081]

[0082] in, (Voltage Deviation) refers to the voltage deviation. This is the real-time voltage value, in volts (V). This is the rated voltage value, and the unit is volts (V).

[0083] During the calculation process, the system strictly follows the preset algorithms and data processing procedures to ensure that the calculation results of each indicator have a high degree of accuracy and reliability. All calculated load characteristic indicators are organized and stored by the system to provide data support for subsequent classification processing. The system also performs preliminary analysis and screening of the calculation results to remove obviously abnormal data points, thereby improving the quality and reliability of the data.

[0084] S23: Classify the load characteristic indicators by using the preset load classification threshold range. Mark the areas with indicator values ​​higher than the upper limit of the threshold as high load critical areas, the areas between the upper and lower limits of the threshold as medium load transition areas, and the areas below the lower limit of the threshold as low load elastic areas, thus generating the classification results of the distribution network area.

[0085] Specifically, the system performs detailed classification processing on the calculated load characteristic indicators based on preset load classification threshold ranges. These thresholds are pre-set based on the operating experience and safety standards of the distribution network, aiming to distinguish areas with different load levels and operating states. For each area, the system first compares the three indicators—load factor, volatility, and voltage deviation—with the corresponding classification thresholds. The classification thresholds are set through rigorous analysis and verification. For example, the upper limit of the load factor threshold may be set at 80% of the rated load, and the lower limit at 30%; the volatility threshold may be determined based on statistical analysis of historical data at ±10%; and the voltage deviation threshold is set at ±5% according to power industry standards. These thresholds together constitute the basis for classification.

[0086] For areas with load rates exceeding the upper threshold, the system marks them as high-load critical areas. The load in these areas is approaching or exceeding the safe operating limit, and the system further analyzes their load growth trends and duration to assess the potential risk level. For example, if the load rate of an area exceeds the upper threshold for multiple consecutive time periods and shows a continuous upward trend, the system will mark it as a high-risk area requiring priority monitoring and handling to prevent potential overload situations from threatening power grid safety.

[0087] For areas where the load rate falls between the upper and lower limits of the threshold, the system marks them as medium-load transition zones. These zones have moderate load levels, currently within safe limits, but still offer room for further optimization. The system analyzes the load fluctuation characteristics and voltage stability of these zones, identifying potential problems such as increased equipment wear due to large load fluctuations, and potential power quality issues caused by voltage deviations approaching the upper limit. This provides targeted suggestions for subsequent optimization measures.

[0088] Areas with load rates below the lower threshold are designated as low-load flexibility zones. These areas have relatively low loads and some flexibility. The system analyzes their load characteristics and equipment operating status to identify potential resource waste or inefficiency. For example, some equipment operating under low load may lead to reduced efficiency and increased energy consumption. The system will then propose corresponding optimization suggestions, such as adjusting equipment operating parameters or consolidating equipment, to improve resource utilization efficiency.

[0089] During the classification process, the system comprehensively considers three load characteristic indicators to ensure the comprehensiveness and accuracy of the classification results. For example, even if the load factor of a certain area is below the lower threshold, if its voltage deviation exceeds the allowable range, the system will still mark it specially according to the specific circumstances and propose corresponding handling suggestions. The classification results are compiled into a detailed distribution network area classification report, which includes not only the classification category of each area, but also key load characteristic indicator values ​​and trend analysis, providing a clear basis for subsequent optimization strategy formulation. Through this rigorous classification process, the system can help power grid managers better understand the operating status of the distribution network, thereby formulating more targeted optimization measures and improving the overall operating efficiency and security of the distribution network.

[0090] In one embodiment, such as Figure 2 As shown, S3 of the distribution network reconfiguration method based on the improved PSO algorithm provided by this invention specifically includes the following steps:

[0091] S31: Based on the high-load critical area, medium-load transition area and low-load elastic area in the classification results, the number of important users, load density and power quality requirements are weighted and summed using a weighted summation algorithm to generate regional importance weights.

[0092] Specifically, the system quantitatively assesses the number of important users. Important users typically include locations with extremely high power supply reliability requirements, such as hospitals and data centers. The system assigns corresponding weight coefficients to important users based on pre-defined priorities. For example, a hospital, as a critical important user, might have a weight coefficient of 0.6, while a typical data center might have a weight coefficient of 0.4. The system determines the number of important users in each region by reading user category information stored in the database, and then multiplies this number by the corresponding weight coefficient to obtain the weighted value of the indicator.

[0093] Secondly, the system analyzes the load density of each area. Load density reflects the electrical load per unit area and is an important indicator for measuring the intensity of regional power demand. The system obtains regional area data from a Geographic Information System (GIS) database and combines it with real-time monitored total load data to calculate the load density of each area. Based on pre-defined load density levels, the system assigns weighting coefficients; for example, a weighting coefficient of 0.5 for high load density areas, 0.3 for medium load density areas, and 0.2 for low load density areas. The system multiplies the load density value by the weighting coefficient to obtain the weighted value of the indicator.

[0094] The system assesses power quality requirements. Power quality requirements involve multiple aspects, including voltage stability and frequency fluctuations, and different users have significantly different requirements. Based on industry standards and user agreements, the system assigns weighting coefficients to the power quality requirements for each region. For example, for users with extremely high power quality requirements, such as precision instrument manufacturing workshops, the weighting coefficient might be set to 0.7, while for ordinary commercial users it might be 0.3. The system determines the power quality requirement level for each region by analyzing historical power quality monitoring data and calculates its weighted value.

[0095] The system sums the weighted values ​​of the three indicators mentioned above to generate the importance weight for each region. This weight is presented numerically and comprehensively reflects the region's overall importance in the distribution network. For example, the importance weight of a high-load critical area, due to its large number of important users, high load density, and high power quality requirements, may reach 0.8 to 0.9; the medium-load transition area may be between 0.4 and 0.6; and the low-load flexibility area may be below 0.4. These weights provide a precise quantitative basis for subsequent tabu search strategy adjustments.

[0096] S32: Based on the regional importance weight and the load rate of high-load key areas in the classification results, high-load taboo processing is performed. The taboo period is linearly adjusted according to the product relationship between the load rate and the weight, and the taboo length parameter of high-load key areas is generated.

[0097] Specifically, the load factor reflects the ratio of a region's current load to its maximum load capacity, while the regional importance weight reflects the region's criticality in the power grid operation. The system calculates an adjustment coefficient by multiplying the load factor and the regional importance weight according to a preset linear adjustment formula. This adjustment coefficient is used to linearly adjust the taboo period to generate the taboo length parameter for high-load critical areas. For example, assuming the initial taboo period is T, after considering the load factor and weight, the adjusted taboo length parameter is: In this way, the system allows the taboo length parameter of high-load critical areas to dynamically change according to their load conditions and importance. For areas with high load rates and high weight, the taboo period will be extended accordingly, thereby reducing frequent operations on these critical areas, protecting equipment, and ensuring the stable operation of the power grid.

[0098] S33: Based on the regional importance weight and the load rate of the load transition zone in the classification results, a step-wise taboo period is set according to the ratio of load rate to weight, and a taboo length parameter for the medium load transition zone is generated.

[0099] Specifically, the system presets multiple tiered threshold ranges and sets corresponding taboo periods. For example, when the load factor to weight ratio is in a low range, the taboo period is shorter; as the ratio increases, the taboo period gradually increases until it reaches its maximum value. The system matches the corresponding tier based on the actually calculated ratio, thereby determining the taboo length parameter of the medium load transition zone. This tiered adjustment method can moderately optimize the medium load transition zone while ensuring the stability of the power grid operation. For areas with relatively low load factors and small weights, a shorter taboo period allows for more flexible operation; while for areas with high load factors and weights, a longer taboo period helps avoid the potential risks of frequent adjustments.

[0100] S34: Based on the regional importance weight and the load redundancy of the low-load elastic zone in the classification results, the taboo period is shortened according to the nonlinear function of the load redundancy to generate the taboo length parameter of the low-load elastic zone.

[0101] Specifically, load redundancy refers to the portion of a region's current load capacity that exceeds its actual load, reflecting the region's reserve load capacity. The system adjusts the taboo period based on a preset nonlinear function, comprehensively considering load redundancy and regional importance weights. For example, an exponential decay function can be used, where the taboo period gradually shortens as load redundancy increases, and the rate of shortening gradually slows down. This nonlinear relationship ensures that for regions with high load redundancy, the taboo period can be shortened rapidly, allowing for more frequent operations and optimizations. In this way, the system generates taboo length parameters suitable for low-load flexibility zones. In low-load flexibility zones, due to relatively low load pressure, appropriately shortening the taboo period can improve the grid's flexibility and resource utilization efficiency, while the consideration of regional importance weights ensures the rationality and targeted nature of the adjustment.

[0102] S35: Integrate the forbidden length parameters of the high-load critical zone, the forbidden length parameters of the medium-load transition zone, and the forbidden length parameters of the low-load elastic zone to generate a differentiated set of forbidden length parameters.

[0103] Specifically, the system categorizes and stores these parameters according to regional identifiers, constructing a comprehensive set of differentiated taboo length parameters. This parameter set not only includes the taboo length parameters for each region but also retains related regional classification information and key indicators, such as load factor and regional importance weight. Ultimately, the system generates a structured dataset, providing crucial parameter support for subsequent distribution network optimization strategies. This ensures that the optimization process fully considers the characteristics and needs of different regions, achieving precise control of the distribution network.

[0104] The aforementioned distribution network reconfiguration method based on an improved PSO algorithm dynamically generates differentiated taboo parameters based on regional classification results, enabling precise adaptation and taboo strategy optimization for key control areas during distribution network reconfiguration. For high-load critical areas, the taboo period is linearly adjusted by combining the product of regional importance weight and load rate, effectively extending the taboo duration of switch operations in critical areas and avoiding frequent disturbances to important load nodes. For medium-load transition areas, a stepped taboo period is set using the ratio of load rate to weight, achieving reasonable hierarchical control of operation frequency in transition areas while balancing search efficiency and convergence stability. For low-load elastic areas, the taboo period is shortened using a nonlinear function of load redundancy, overcoming the limitations of traditional fixed taboo strategies and fully releasing the topology adjustment potential of low-load areas. Through the differentiated generation and integration of taboo parameters for these three types of areas, a taboo search strategy system that dynamically adapts to the power grid operating state can be constructed, effectively balancing global exploration and local development capabilities, preventing the algorithm from getting trapped in local optima, and reducing operational losses on high-load critical equipment. This process, through a multi-level parameter coordination mechanism, systematically solves the problems of insufficient protection of key areas, inefficient search, and reduced equipment lifespan caused by the rigidity of tabu strategies in existing reconstruction methods. It significantly improves the adaptability and optimization effect of tabu search algorithms in complex distribution network scenarios, and provides reliable support for the safe and economical operation of the power grid under dynamic load conditions.

[0105] In one embodiment, such as Figure 3 As shown, S4 of the distribution network reconfiguration method based on the improved PSO algorithm provided by this invention specifically includes the following steps:

[0106] S41: The standard particle swarm optimization algorithm is improved based on the dynamic inertia weight adjustment mechanism. During the iteration process, the inertia weight is updated according to the deviation between the current particle fitness and the optimal fitness of the swarm, and the improved PSO algorithm is generated.

[0107] Specifically, the system improves the standard Particle Swarm Optimization (PSO) algorithm by introducing a dynamic inertia weight adjustment mechanism to enhance the balance between the algorithm's global and local search capabilities. The system first initializes the particle swarm, with each particle representing a potential distribution network topology, including switch states and PSO algorithm parameters. During the iteration process, the system calculates the fitness value of each particle in real time. This fitness value is derived through a multi-objective comprehensive evaluation function, comprehensively considering multiple key indicators such as network loss, voltage quality, and load balancing.

[0108] The system simultaneously tracks the optimal fitness value of the swarm, which is the best fitness value among all particles in the current iteration. In each iteration, the system calculates the deviation between the current particle fitness and the swarm's optimal fitness. This deviation reflects the difference between the particle's current state and the swarm's optimal state. The system adjusts the inertia weight based on a preset dynamic weight adjustment coefficient and the magnitude of the deviation. For example, a larger deviation indicates a greater gap between the particle and the swarm's optimal state, and the system increases the inertia weight, making the particle more likely to maintain its original direction of motion, expanding the search range, and enhancing global search capabilities. Conversely, a smaller deviation indicates that the particle is close to the swarm's optimal state, and the system decreases the inertia weight, making the particle more likely to follow the swarm's optimal particle, enhancing local search capabilities, and improving search accuracy.

[0109] The system dynamically adjusts the inertia weights in this way, enabling the particle swarm to achieve a good balance between global and local search. During the iteration process, the system continuously updates the velocity and position of the particles; the velocity update formula is:

[0110]

[0111]

[0112] in, For inertial weights, and These are individual learning factors and social learning factors, respectively. It is a random number. This represents the historical best position of particle i. The optimal position for the group. Let be the position of particle i in the t-th iteration.

[0113] Through a dynamic inertia weight adjustment mechanism, the system can adaptively adjust the inertia weight based on the deviation between the current particle's fitness and the swarm's optimal fitness, thereby generating an improved PSO algorithm. This improved PSO algorithm can explore the solution space more effectively during the optimization process, improving search efficiency and optimization results.

[0114] S42: The improved PSO algorithm and the differentiated taboo length parameter are co-optimized. The particle velocity and position are adjusted based on the dynamic inertia weight in the improved PSO algorithm. A dynamic taboo table is constructed in combination with the differentiated taboo length parameter. A long taboo period is set for switching operations involving high-load critical areas, and a short taboo period is set for operations in low-load elastic areas to generate an updated particle swarm.

[0115] Specifically, the system utilizes the dynamic inertia weight adjustment mechanism in the improved PSO algorithm to update the particle velocity and position, and constructs a dynamic tabu table by combining it with a differentiated tabu length parameter. The dynamic tabu table records solutions that have already been searched or the characteristics of solutions, avoiding repeated searches and improving search efficiency. Based on the tabu length parameters for high-load critical areas, medium-load transition areas, and low-load flexible areas, the system sets different tabu periods for switching operations in different regions. For switching operations in high-load critical areas, due to their significant impact on the distribution network operation, the system sets a longer tabu period, such as 15 to 20 iterations, to expand the search range and avoid getting trapped in local optima. For switching operations in low-load flexible areas, due to their smaller impact on the distribution network operation, the system sets a shorter tabu period, such as 5 to 10 iterations, to improve search efficiency.

[0116] In each iteration, the system checks whether the corresponding switch operation of the updated particle position is in the taboo list. If the switch operation involved in the particle's new position is in the taboo list and the taboo period has not exceeded, the position is marked as taboo, and the particle will adjust its flight direction according to the taboo list to select a new position. If the switch operation involved in the particle's new position is not in the taboo list or has exceeded the taboo period, the particle's position is updated, the new switch operation is recorded in the taboo list, and the taboo period is updated simultaneously.

[0117] Through this collaborative optimization process, the system generates an updated particle swarm. Each particle in the updated swarm undergoes not only speed and position updates based on the improved PSO algorithm but also filtering through a dynamic tabu list, ensuring both diversity and search efficiency. This collaborative optimization mechanism effectively combines the global search capability of the improved PSO algorithm with the local search capability of the tabu search algorithm, enhancing optimization performance and providing a high-quality particle swarm for subsequent candidate solution selection.

[0118] S43: Based on the updated particle swarm and the initial particle swarm, candidate schemes that do not violate taboo rules are screened, and based on the topology connectivity verification of the distribution network, the maximum current limit of real-time load data, and the node voltage deviation threshold, schemes that violate constraints are eliminated, and a set of candidate reconfiguration schemes is generated.

[0119] Specifically, the system checks each particle in the updated particle swarm to ensure that its switching operation sequence does not violate taboo rules, i.e., it is not marked as a taboo state in the dynamic taboo table. Next, the system further rigorously screens candidate schemes based on the topology connectivity verification of the distribution network, the maximum current limit of real-time load data, and node voltage deviation thresholds. Specifically, the system verifies whether each candidate scheme meets the connectivity requirements of the distribution network, ensuring that there are no islands or breakpoints in the network; checks whether the line current exceeds the maximum current limit to ensure the safe operation of equipment; and verifies whether the node voltage is within the allowable deviation range to ensure power quality. Schemes that violate the above constraints are eliminated by the system, ultimately generating a candidate reconfiguration scheme set. Each scheme in this set satisfies the operating constraints of the distribution network and does not violate taboo rules, providing feasible options for subsequent optimization decisions. In this way, the system ensures the quality and reliability of the candidate reconfiguration scheme set, laying a solid foundation for finally determining the optimal reconfiguration scheme.

[0120] This embodiment provides a distribution network reconfiguration method based on an improved PSO algorithm. Through dynamic inertia weight adjustment mechanism and differentiated taboo parameters, it can improve the adaptive search capability of the particle swarm optimization algorithm and efficiently screen candidate schemes during the distribution network reconfiguration process. Based on the dynamic adjustment of inertia weights according to particle fitness bias, it can overcome the limitation of the imbalance between global exploration and local development capabilities in the standard particle swarm optimization algorithm, adaptively balancing the search breadth and convergence accuracy during iteration. By constructing a dynamic taboo table through differentiated taboo length parameters, it can realize long-cycle taboo restrictions for switching operations in high-load critical areas and short-cycle flexible de-tabulation for operations in low-load flexible areas, fully releasing the topology adjustment potential of low-load areas while avoiding frequent actions of critical equipment. Based on multi-constraint verification of grid topology connectivity, real-time current limits, and voltage deviation thresholds, it can ensure the operational safety and physical feasibility of candidate reconfiguration schemes.

[0121] In one embodiment, such as Figure 4 As shown, step S5 of the distribution network reconfiguration method based on the improved PSO algorithm provided by this invention specifically includes the following steps:

[0122] S51: Based on the distribution network simulation model, perform multi-objective index quantification on each scheme in the candidate reconfiguration scheme set, calculate the network loss value, the average node voltage deviation and the load balance degree of each scheme, and generate a scheme index dataset.

[0123] Specifically, the system utilizes a detailed distribution network simulation model to perform comprehensive multi-objective index quantification on each scheme in the candidate reconfiguration scheme set. The system first initializes the simulation environment, loading the topology and switch state configuration of each candidate scheme into the simulation model. Based on detailed parameters of the distribution network, including line impedance, transformer parameters, and node load characteristics, the simulation model can accurately simulate the electrical behavior of the distribution network under different operating conditions.

[0124] For calculating network loss, the system uses Kirchhoff's laws and nodal power balance equations to solve for the active and reactive power losses of the distribution network under the current topology and load conditions. Preferably, the system can use the Newton-Raphson iterative method for power flow calculation to ensure high accuracy of the results. The formula for calculating network loss is:

[0125]

[0126] in, and Let be the active power loss and reactive power loss of branch ij, respectively, and n and m be the number of nodes and branches, respectively. The calculation of the average node voltage deviation is based on the node voltage data obtained from simulation. The system calculates the voltage deviation of each node, that is, the difference between the node voltage and the rated voltage, and calculates the average of their absolute values. The calculation formula is:

[0127]

[0128] Where N is the number of nodes. Let be the voltage at node i. The rated voltage is used. Load balance is assessed by analyzing the load distribution in each area. The system calculates the ratio of the load in each area to the maximum load capacity of that area and then calculates its standard deviation. The formula for calculating load balance is:

[0129]

[0130] in, The standard deviation of the regional load ratio, This represents the average load ratio of the regions. The load balance score ranges from 0 to 1, with a higher value indicating a more balanced load distribution.

[0131] The system compiles the network loss value, average node voltage deviation, and load balancing degree of each scheme into a dataset, namely the scheme index dataset. This dataset is stored in tabular form, with each scheme corresponding to one row, containing four fields: scheme number, network loss value, average node voltage deviation, and load balancing degree. The system employs efficient data processing algorithms to ensure the completeness and accuracy of the dataset, providing reliable data support for subsequent comprehensive evaluation.

[0132] S52: Based on the dynamic weight allocation mechanism, the dataset of scheme indicators is comprehensively evaluated and processed to generate a comprehensive score for each scheme.

[0133] Specifically, the system calculates a comprehensive score by weighting and summing the network loss value, average node voltage deviation, and load balance degree of each scheme according to a preset dynamic weight allocation mechanism. The dynamic weight allocation mechanism aims to dynamically adjust the weight coefficients of each indicator based on the current operating status and optimization objectives of the distribution network, so as to achieve a reasonable trade-off between multiple objectives.

[0134] Dynamic weighting coefficients , and The calculations are based on the current operational priorities of the distribution network. The system monitors the operational status of the distribution network in real time, analyzing the current main challenges and optimization needs. For example, during peak grid load periods, reducing network losses may be the primary task. The weight of [something] will increase relatively; while during periods when voltage quality problems are prominent, the importance of improving voltage quality becomes more apparent. The weight will increase accordingly; when there is a significant load imbalance in the power grid, The weight will increase. The system dynamically calculates the weight coefficient of each indicator through preset weight adjustment rules and algorithms.

[0135] The formula for calculating the overall score is as follows:

[0136]

[0137] Where S represents the overall score. , , The dynamic weighting coefficients satisfy the following conditions: , The network loss value of the solution. As the baseline network loss, The mean of node voltage deviation. To allow the maximum voltage deviation, For load balancing, and The system sets the minimum and maximum load balancing values ​​for candidate solutions. It normalizes the indicator data for each solution to ensure that indicators of different dimensions and orders of magnitude can be compared and synthesized on the same scale. The normalized indicator values ​​are then substituted into the comprehensive scoring formula to calculate the comprehensive score for each solution. The comprehensive score reflects the overall performance of the solution in multi-objective optimization, providing a quantitative basis for subsequent solution selection.

[0138] S53: Sort the candidate reconstruction scheme set based on the comprehensive score of each scheme, and select the scheme with the best comprehensive score as the optimal reconstruction scheme.

[0139] Specifically, the system sorts all solutions in descending order based on their comprehensive scores. The sorting algorithm used is quicksort, which has a time complexity of O(nlogn), enabling efficient handling of large-scale candidate solution sets. Specifically, the system first initializes an empty sorted list, then reads each solution and its comprehensive score into memory one by one. Using the quicksort algorithm, the system sorts the solutions by their comprehensive score. During the sorting process, the system compares the comprehensive scores of adjacent solutions, gradually moving solutions with higher scores to the front of the list and solutions with lower scores to the back. Finally, the system generates a list of solutions sorted from highest to lowest comprehensive score.

[0140] After ranking, the system selects the scheme with the highest comprehensive score as the optimal reconfiguration scheme. This scheme achieves the best balance in multi-objective optimization, reducing network losses while improving voltage quality and load balance. The system extracts detailed information about the optimal reconfiguration scheme, including key indicators such as the switching operation sequence, expected network loss value, average node voltage deviation, and load balance. This information will be compiled into a report to provide decision support for power grid dispatchers.

[0141] To verify the effectiveness and feasibility of the optimal reconfiguration scheme, the system also performs a series of post-verification steps. These steps include re-simulating and verifying the scheme, comparing and analyzing it with historical data, and assessing potential risks. Through these verification steps, the system ensures that the optimal reconfiguration scheme can achieve the expected optimization effect in actual power grid operation, improving the overall operating efficiency and reliability of the distribution network.

[0142] In summary, the distribution network reconfiguration method based on the improved PSO algorithm provided in this embodiment achieves refined evaluation of distribution network reconfiguration schemes and accurate selection of optimal solutions through the comprehensive application of multi-objective index quantification and dynamic weight allocation mechanism. Based on the distribution network simulation model, candidate schemes are quantified from multiple dimensions, including network loss, voltage deviation, and load balance, covering core indicators such as grid operation economy, power quality, and resource utilization, to construct a comprehensive index dataset reflecting scheme performance. The dynamic weight allocation mechanism adaptively evaluates multi-objective indicators, dynamically adjusting the weight allocation strategy according to the real-time grid operation status, effectively coordinating conflicts between multiple objectives. The Pareto front screening mechanism based on comprehensive scoring and ranking overcomes the limitations of traditional single-objective or fixed-weight evaluation, ensuring that the final selected reconfiguration scheme achieves a balance between global optimality and scenario adaptability.

[0143] In one embodiment, step S6 of the distribution network reconfiguration method based on the improved PSO algorithm provided by the present invention specifically includes the following steps:

[0144] S61: The priority ranking results of the optimal reconfiguration scheme are processed by combining regional importance and load urgency. The execution priority score of each region is calculated based on the differentiated taboo length parameter and the real-time load change rate, and a priority ranking sequence is generated.

[0145] Specifically, the system integrates regional importance and load urgency in the priority ranking results of the optimal reconfiguration scheme. Based on the differentiated taboo length parameters of high-load critical areas, medium-load transition areas, and low-load elastic areas, and combined with the real-time monitored load change rate, the system calculates the execution priority score for each area. Specifically, the system extracts the taboo length parameter and real-time load change rate for each area. For high-load critical areas, due to their significant impact on grid operation, the system assigns a higher base weight; medium-load transition areas and low-load elastic areas are assigned medium and lower base weights respectively based on their characteristics. Then, the system standardizes the real-time load change rate to fall within the range of [0, 1]. The standardized load change rate is multiplied by the base weight and then weighted and summed with the taboo length parameter to obtain the priority score for each area. Finally, the system sorts the areas according to their priority scores, generating a priority ranking sequence.

[0146] S62: Based on the priority sorting sequence, control instructions are generated and processed. The switch operation instructions, action timestamps and area labels are encapsulated into a structured instruction set according to the execution order to generate a control instruction sequence.

[0147] Specifically, the system reads a priority sequence from the memory buffer and uses it as the basis for the execution order of control instructions. For each region in the sequence, the system retrieves the corresponding switch operation instruction from the switch operation instruction library. These instructions are stored in a standardized format and contain detailed information such as the operation type (closed or open), the target switch number, and the expected result of the operation.

[0148] To ensure orderly operation and stable power grid operation, the system assigns precise action timestamps to each switching operation command according to preset timing rules. The generation of these timestamps comprehensively considers the time required for the switching operation, the dynamic response characteristics of the power grid, and the dependencies between different operations. The system uses a complex timing algorithm to calculate the optimal execution time for each operation and embeds it into the command.

[0149] Simultaneously, the system extracts corresponding area tag information from the area tag database and associates it with the switch operation command. The area tag contains key information such as the area's geographical location, functional attributes, and topological location in the network, which helps operators quickly identify and trace the area where the command is executed.

[0150] The system integrates and encapsulates switch operation commands, action timestamps, and region tags to form a structured instruction set. Each instruction set is organized according to a unified data structure to ensure the integrity and consistency of the commands. Finally, the system arranges all generated structured instruction sets in execution order to form a complete control command sequence, which is then stored in an instruction buffer for further processing and distribution.

[0151] S63: Perform topology reconfiguration instruction compilation processing on the control instruction sequence, convert the instructions into binary code stream according to the communication protocol of the distribution automation terminal, and generate network topology reconfiguration instructions.

[0152] Specifically, the system converts each instruction in the control command sequence into a binary code stream according to the communication protocol of the distribution automation terminal, generating a network topology reconfiguration instruction. The system acquires the communication protocol of the distribution automation terminal and parses its specified binary code stream format, including start bits, data bits, parity bits, and stop bits. For each switch operation instruction in the control command sequence, the system encodes its operation type (closed or open), switch number, and other information, converting it into binary data conforming to the communication protocol. During the encoding process, the system adds necessary verification information to ensure the integrity and accuracy of the instruction. Finally, the system sequentially integrates all binary code streams to generate a complete network topology reconfiguration instruction, preparing for subsequent instruction distribution and execution.

[0153] S64: Performs security verification and distribution processing on network topology reconfiguration commands, verifies the integrity of the commands through digital signatures, and sends them to the distribution automation actuators according to the timing requirements of the SCADA system.

[0154] Specifically, the system invokes the digital signature verification module, using a pre-stored public key to perform digital signature verification on the network topology reconstruction command, checking whether the command has been tampered with during transmission or storage. During the verification process, the system strictly follows the steps of the digital signature algorithm for calculation and comparison to ensure the accuracy of the verification. Simultaneously, the system checks whether the format and content of the command conform to the requirements of the distribution automation terminal, including the command length and encoding method. If the command passes the security verification, the system stores it in the command buffer, awaiting transmission.

[0155] The system acquires the current system time and status information of each power distribution automation actuator according to the timing requirements of the SCADA system. Based on the actuator's priority and task scheduling rules, the system determines the order and time interval for sending instructions. During transmission, the system monitors the communication link status in real time to ensure that instructions are reliably transmitted to the target actuator. If a communication failure or timeout occurs during transmission, the system will automatically initiate a retransmission mechanism until the instruction is confirmed to have been successfully delivered. Through these rigorous processing steps, the computer system ensures the secure transmission and accurate execution of network topology reconfiguration instructions.

[0156] Preferably, such as Figure 5 As shown, this invention provides a distribution network reconfiguration system 700 based on an improved PSO algorithm, which is configured with the following modules:

[0157] The PSO and actual coding module 710 is improved to perform double-layer coding of switch status and PSO parameters based on the topology of the distribution network, real-time load data and historical load data, traverse all switch nodes and verify radial constraints, and generate an initial particle swarm, which includes double-layer coding of switch status and PSO parameters.

[0158] The load characteristic analysis module 720 is used to analyze and process the load characteristics of each area of ​​the distribution network according to the switching state of each particle in the initial particle swarm. It calculates the load characteristic index by combining real-time current value and historical fluctuation data, and classifies the loads through a preset load classification threshold range to generate classification results for high load critical area, medium load transition area and low load elastic area.

[0159] The dynamic taboo strategy generation module 730 is used to dynamically parameterize the taboo search strategy based on the classification results and generate differentiated taboo length parameters. The differentiated taboo length parameters include taboo length parameters for high-load critical areas, medium-load transition areas, and low-load elastic areas.

[0160] The PSO-Taboo Cooperative Optimization Module 740 is used for cooperative optimization of differentiated taboo length parameters based on the improved PSO algorithm, dynamically updating particle positions and screening non-taboo candidate schemes to generate a set of candidate reconfiguration schemes that meet the power grid operation constraints. The improved PSO algorithm is achieved by introducing a dynamic inertia weight adjustment mechanism.

[0161] The multi-objective comprehensive evaluation module 750 is used to perform multi-objective comprehensive evaluation on each scheme in the candidate reconstruction scheme set, select the scheme with the best comprehensive evaluation index, and output the optimal reconstruction scheme.

[0162] The control instruction generation and execution module 760 is used to generate a sequence of control instructions based on the priority ranking results of the optimal reconfiguration scheme, according to regional importance and load urgency, generate network topology reconfiguration instructions, and send the network topology reconfiguration instructions to the distribution automation execution mechanism.

[0163] In summary, the distribution network reconfiguration system based on the improved PSO algorithm provided by this invention fuses the distribution network topology, real-time load data, and historical load data through a two-layer encoding process to construct an initial particle swarm containing switch states and PSO parameters. This enables dynamic optimization starting point generation in complex power grid scenarios. Based on load characteristic analysis and classification threshold determination using real-time current values ​​and historical fluctuation data, it dynamically divides high, medium, and low load areas to accurately identify key control areas and adapt to differentiated operational needs. By dynamically parameterizing the tabu search strategy in conjunction with the area classification results, it can achieve long-term operation in high-load areas. The PSO algorithm is improved by incorporating periodic taboo constraints and flexible short-cycle control in low-load areas, effectively balancing global search efficiency and local convergence accuracy. A dynamic inertia weight adjustment mechanism is introduced to improve the PSO algorithm, and combined with differentiated taboo parameters to collaboratively optimize the particle update process. This overcomes the limitation of traditional algorithms easily getting trapped in local optima, significantly improving the feasibility and diversity of candidate solutions. A multi-objective comprehensive evaluation model is used to quantify network loss, voltage deviation, and load balancing indicators to achieve Pareto optimal solution selection for reconfiguration schemes under multiple constraints. A security verification mechanism based on priority sorting and instruction compilation enables accurate generation and reliable issuance of network topology reconfiguration instructions.

[0164] Preferably, the improved PSO and the actual encoding module 710 provided in this embodiment are configured with the following units:

[0165] The switch coding generation unit is used to perform binary coding processing on the switch state based on the topology of the distribution network, traverse all switch nodes and map them to a 0 / 1 state sequence, and generate binary switch codes that satisfy radial constraints.

[0166] The parameter vector generation unit is used to encode the inertia weight, individual learning factor and social learning factor of the PSO algorithm with real numbers based on the real-time load data and historical load data of the distribution network. It is randomly initialized within the preset parameter range to generate an algorithm parameter vector containing the inertia weight, individual learning factor and social learning factor.

[0167] The initial particle swarm generation unit is used to process the binary switch code and algorithm parameter vector, and generate the initial particle swarm through topology verification and parameter combination processing.

[0168] Preferably, the load characteristic analysis module 720 provided in this embodiment is configured with the following units:

[0169] The regional switch state mapping unit is used to analyze the switch state of each particle in the initial particle swarm, extract the switch on / off state corresponding to each region of the distribution network, and generate a regional switch state mapping table.

[0170] The load characteristic index calculation unit is used to calculate the load rate, volatility and voltage deviation based on the status mapping table of the area switch, combined with real-time current value and historical load fluctuation data, and generate the load characteristic index of each area.

[0171] The regional classification unit is used to classify load characteristic indicators by a preset load classification threshold range, marking high load critical areas, medium load transition areas, and low load elastic areas, and generating regional classification results.

[0172] Preferably, the dynamic tabu policy generation module 730 provided in this embodiment is configured with the following units:

[0173] The regional importance weight calculation unit is used to calculate the number of important users, load density, and power quality requirements based on the regional type in the classification results, and generate regional importance weights through a weighted summation algorithm.

[0174] The high-load taboo parameter generation unit is used to generate the taboo length parameter of the high-load critical area by linearly adjusting the taboo period based on the product relationship between the load rate and the weight, according to the regional importance weight and the load rate of the high-load critical area.

[0175] The medium load taboo parameter generation unit is used to generate the medium load transition zone taboo length parameter by setting a step taboo period based on the regional importance weight and the load rate of the medium load transition zone according to the ratio of load rate to weight.

[0176] The low-load taboo parameter generation unit is used to generate the low-load elastic zone taboo length parameter by shortening the taboo period through a nonlinear function of load redundancy based on the regional importance weight and the load redundancy of the low-load elastic zone.

[0177] Differentiated taboo parameter integration unit: Integrates taboo length parameters for high-load critical zone, medium-load transition zone, and low-load elastic zone to generate a set of differentiated taboo length parameters.

[0178] Preferably, the PSO-Taboo collaborative optimization module 740 provided in this embodiment is configured with the following units:

[0179] The PSO algorithm improvement unit is used to improve the standard particle swarm algorithm based on the dynamic inertia weight adjustment mechanism. During the iteration process, the inertia weight is updated according to the deviation between the current particle fitness and the optimal fitness of the swarm, and the inertia weight is updated according to the preset dynamic weight adjustment coefficient to generate the improved PSO algorithm.

[0180] The particle swarm generation unit is updated to perform collaborative optimization of the improved PSO algorithm and the differentiated taboo length parameter. Based on the dynamic inertia weight in the improved PSO algorithm, the particle velocity and position are adjusted. A dynamic taboo table is constructed in combination with the differentiated taboo length parameter. A long taboo period is set for switching operations involving high-load critical areas, and a short taboo period is set for operations in low-load elastic areas to generate an updated particle swarm.

[0181] The candidate reconfiguration scheme set generation unit is used to screen candidate schemes that do not violate taboo rules based on the updated particle swarm and the initial particle swarm, and to eliminate schemes that violate constraints based on the topology connectivity verification of the distribution network, the maximum current limit of real-time load data and the node voltage deviation threshold, thereby generating a candidate reconfiguration scheme set.

[0182] Preferably, the multi-objective comprehensive evaluation module 750 provided in this embodiment is configured with the following units:

[0183] The scheme index quantification unit is used to perform multi-objective index quantification processing on each scheme in the candidate reconfiguration scheme set based on the distribution network simulation model, calculate the network loss value, the average node voltage deviation and the load balance degree, and generate a scheme index dataset.

[0184] The comprehensive score generation unit is used to perform comprehensive evaluation processing on the scheme indicator dataset based on the dynamic weight allocation mechanism, and generate a comprehensive score for each scheme.

[0185] The optimal solution selection unit is used to sort the candidate reconstruction solution set based on the comprehensive score and select the solution with the best comprehensive score as the optimal reconstruction solution.

[0186] Preferably, the control instruction generation and execution module 760 provided in this embodiment is configured with the following units:

[0187] The priority score calculation unit is used to integrate the regional importance and load urgency of the priority ranking results of the optimal reconstruction scheme. Based on the differentiated taboo length parameter and the real-time load change rate, it calculates the execution priority score of each region and generates a priority ranking sequence.

[0188] The control instruction sequence generation unit is used to generate control instructions based on a priority-ordered sequence. It encapsulates the switch operation instructions, action timestamps, and area labels into a structured instruction set according to the execution order to generate a control instruction sequence.

[0189] The topology reconfiguration instruction compilation unit is used to compile and process the control instruction sequence into topology reconfiguration instructions. It converts the instructions into binary code streams according to the communication protocol of the distribution automation terminal and generates network topology reconfiguration instructions.

[0190] The instruction security verification and distribution unit is used to perform security verification and distribution processing on network topology reconstruction instructions. It verifies the integrity of the instructions through digital signatures and sends them to the distribution automation actuators according to the timing requirements of the SCADA system.

[0191] In one embodiment, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described distribution network reconfiguration method based on the improved PSO algorithm.

[0192] In one embodiment, this application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described distribution network reconfiguration method based on the improved PSO algorithm.

[0193] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0194] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0195] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope disclosed in this application, and these should all be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A power distribution network reconfiguration method based on an improved PSO algorithm, characterized in that, The method comprises the following steps: S1: based on the topology structure of the power distribution network, real-time load data and historical load data, the switch state and PSO parameters are processed by double-layer coding, all switch nodes are traversed and the radial constraint is verified, the initial particle swarm is generated, and the initial particle swarm comprises double-layer coding of the switch state and the PSO parameters; S2: according to the switch state corresponding to each particle in the initial particle swarm, the load characteristics of each region of the power distribution network are analyzed and processed, the load characteristic index is calculated in combination with the real-time current value and the historical fluctuation data, and the classification result of the high-load key region, the medium-load transition region and the low-load elastic region is generated through the preset load classification threshold interval; S3: based on the classification result, the taboo search strategy is processed by dynamic parameterization, and a differentiated taboo length parameter is generated, the differentiated taboo length parameter comprises a high-load key region taboo length parameter, a medium-load transition region taboo length parameter and a low-load elastic region taboo length parameter; S4: based on the improved PSO, the differentiated taboo length parameter is processed by cooperative optimization, the particle position is dynamically updated and the non-taboo candidate scheme is selected, and a candidate reconstruction scheme set satisfying the power grid operation constraint is generated, and the improved PSO algorithm is realized by introducing a dynamic inertia weight adjustment mechanism; S5: each scheme in the candidate reconstruction scheme set is processed by multi-objective comprehensive evaluation, the scheme with the optimal comprehensive evaluation index is selected, and the optimal reconstruction scheme is output; S6: based on the priority sorting result of the optimal reconstruction scheme, the control instruction sequence is generated according to the region importance and the load emergency degree, the network topology reconstruction instruction is generated, and the network topology reconstruction instruction is sent to the power distribution automation execution mechanism.

2. The method of claim 1, wherein, The S1 comprises: S11: based on the topology structure of the power distribution network, the switch state is processed by binary coding, all switch nodes are traversed and mapped to 0 / 1 state sequence, and binary switch coding satisfying the radial constraint is generated; S12: based on the real-time load data and the historical load data of the power distribution network, the inertia weight, the individual learning factor and the social learning factor of the PSO algorithm are processed by real number coding, and are randomly initialized in the preset parameter range to generate an algorithm parameter vector comprising the inertia weight, the individual learning factor and the social learning factor; S13: the binary switch coding and the algorithm parameter vector are processed, the topology verification and parameter combination processing are performed, and the initial particle swarm is generated.

3. The method of claim 1, wherein, The S2 comprises: S21: the switch state of each particle in the initial particle swarm is analyzed and processed, the switch on-off state corresponding to each region of the power distribution network is extracted, and a state mapping table of the regional switch is generated; S22: based on the state mapping table of the regional switch, in combination with the fluctuation data in the historical load database and the real-time current value of the power distribution network, the load characteristic index calculation processing is performed, the load rate, the fluctuation rate and the voltage deviation are calculated, and the load characteristic index of each region is generated; S23: Classify the load characteristic indicators by a preset load classification threshold interval, mark the area with an indicator value higher than the upper threshold as a high-load critical area, mark the area between the upper and lower thresholds as a medium-load transition area, and mark the area lower than the lower threshold as a low-load flexible area, to generate a classification result of the power distribution network area.

4. The method of claim 1, wherein, The S3 includes: S31: Based on the high-load critical area, medium-load transition area, and low-load flexible area in the classification result, perform weighted summation on the number of important users, load density, and power quality requirements by a weighted summation algorithm, to generate an area importance weight; S32: Perform high-load taboo processing based on the area importance weight and the load rate of the high-load critical area in the classification result, linearly adjust the taboo period according to the product relationship between the load rate and the weight, to generate a high-load critical area taboo length parameter; S33: Perform processing based on the area importance weight and the load rate of the load transition area in the classification result, set a stepped taboo period according to the proportional relationship between the load rate and the weight, to generate a medium-load transition area taboo length parameter; S34: Perform processing based on the area importance weight and the load redundancy of the low-load flexible area in the classification result, shorten the taboo period according to the nonlinear function of the load redundancy, to generate a low-load flexible area taboo length parameter; S35: Integrate the high-load critical area taboo length parameter, the medium-load transition area taboo length parameter, and the low-load flexible area taboo length parameter, to generate a set of differentiated taboo length parameters.

5. The method of claim 1, wherein, The S4 includes: S41: Improve the standard PSO algorithm based on a dynamic inertia weight adjustment mechanism, update the inertia weight according to the deviation value of the current particle fitness and the group optimal fitness in the iteration process, and generate an improved PSO algorithm according to a preset dynamic weight adjustment coefficient; S42: Perform collaborative optimization processing on the improved PSO algorithm and the differentiated taboo length parameters, adjust the particle speed and position based on the dynamic inertia weight in the improved PSO algorithm, and construct a dynamic taboo table by combining the differentiated taboo length parameters, to set a long taboo period for the switch operation involving the high-load critical area and a short taboo period for the low-load flexible area operation, to generate an updated particle swarm; S43: Based on the updated particle swarm and the initial particle swarm, screen candidate schemes that do not violate the taboo rule, and based on the topological connectivity verification of the power distribution network, the maximum current limit of the real-time load data, and the node voltage deviation threshold, eliminate schemes that violate the constraints, to generate a set of candidate reconstruction schemes.

6. The method of claim 1, wherein, The S5 includes: S51: Based on the power distribution network simulation model, perform multi-objective indicator quantification processing on each scheme in the set of candidate reconstruction schemes, calculate the network loss value, node voltage deviation mean value, and load balancing degree of each scheme, to generate a scheme indicator data set; S52: Based on a dynamic weight allocation mechanism, perform comprehensive evaluation processing on the scheme indicator data set, to generate a comprehensive score of each scheme, and the calculation formula of the comprehensive score is: wherein S is a comprehensive score, , , is a dynamic weight coefficient and satisfies , is a scheme network loss value, is a reference network loss, is a node voltage deviation average, is an allowed maximum voltage deviation, is a load balance degree, and are a minimum value and a maximum value of the load balance degree in the candidate scheme set. S53: Perform sorting processing on the candidate reconstruction scheme set based on the comprehensive scores of each scheme, and select the scheme with the optimal comprehensive score as the optimal reconstruction scheme.

7. The method according to any one of claims 1 to 6, characterized in that, The S6 includes: S61: Perform regional importance and load urgency fusion processing on the priority sorting result of the optimal reconstruction scheme, calculate the execution priority score of each region based on the differentiated taboo length parameter and the real-time load change rate, and generate a priority sorting sequence; S62: Perform control instruction generation processing based on the priority sorting sequence, encapsulate switch operation instructions, action timestamps, and region labels into a structured instruction set in execution order, and generate a control instruction sequence; S63: Perform topology reconstruction instruction compilation processing on the control instruction sequence, convert the instructions into binary code streams according to the communication protocol of the power distribution automation terminal, and generate network topology reconstruction instructions; S64: Perform security check and distribution processing on the network topology reconstruction instructions, verify the integrity of the instructions through digital signature, and send them to the power distribution automation execution mechanism according to the SCADA system timing requirements.

8. A power distribution network reconfiguration system based on improved PSO algorithm, characterized in that, The system includes: An improved PSO and actual coding module is configured to perform double-layer coding processing on switch states and PSO parameters based on the topology structure of a power distribution network, real-time load data, and historical load data, traverse all switch nodes and verify radial constraints, and generate an initial particle swarm, which includes double-layer coding of switch states and PSO parameters; A load characteristic analysis module is configured to analyze the load characteristics of each region of the power distribution network according to the switch state corresponding to each particle in the initial particle swarm, calculate load characteristic indicators in combination with real-time current values and historical fluctuation data, and classify them through a preset load classification threshold interval to generate classification results of high-load key regions, medium-load transition regions, and low-load flexible regions; A dynamic taboo strategy generation module is configured to perform dynamic parameterization processing on the taboo search strategy based on the classification results to generate differentiated taboo length parameters, which include high-load key region taboo length parameters, medium-load transition region taboo length parameters, and low-load flexible region taboo length parameters; A PSO-taboo collaborative optimization module is configured to perform collaborative optimization processing on the differentiated taboo length parameters based on an improved PSO algorithm to dynamically update particle positions and select non-taboo candidate schemes, and generate a candidate reconstruction scheme set that meets power grid operation constraints, wherein the improved PSO algorithm is realized by introducing a dynamic inertia weight adjustment mechanism; A multi-objective comprehensive evaluation module is configured to perform multi-objective comprehensive evaluation processing on each scheme in the candidate reconstruction scheme set, select a scheme with the optimal comprehensive evaluation indicator, and output an optimal reconstruction scheme; A control instruction generation and execution module is configured to generate a control instruction sequence based on the priority sorting result of the optimal reconstruction scheme according to regional importance and load urgency, generate network topology reconstruction instructions, and send the network topology reconstruction instructions to a power distribution automation execution mechanism. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8 when the computer program is executed by the processor. The processor implements the method of any one of claims 1-7 when executing the computer program.

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

Images