Method, device and equipment for reconstructing topological structure of power distribution network and storage medium

Abstract

The invention discloses a method, device and equipment for reconstructing a topological structure of a power distribution network and a storage medium, and relates to the technical field of electric power, and the method comprises the steps: firstly obtaining multi-source heterogeneous data such as SCADA monitoring data and GIS spatial data, extracting node attribute features and edge attribute features from the multi-source heterogeneous data, and generating a neighbor matrix based on the edge attribute features; and judging the node connection probability, if the node connection probability is smaller than a threshold value, generating current layer features by using the neighbor matrix and previous layer node attribute features, calculating the connection probability through cosine similarity, if the connection probability is larger than the threshold value, outputting target node attribute features, and combining edge attribute features to output a topological structure. Meanwhile, a target risk assessment function and a recovery efficiency function are constructed, a multi-objective function is determined, values of the multi-objective function are calculated to obtain a local solution set, and an optimal solution set is screened according to the degree of congestion. According to the method, the topological structure can be accurately reconstructed, the risk and the recovery efficiency are considered, and support is provided for operation optimization of the power distribution network.

Description

Technical Field

[0001] This application relates to the field of power technology, and in particular to a method, apparatus, equipment, and storage medium for reconstructing the topology of a power distribution network. Background Technology

[0002] In recent years, with the large-scale integration of distributed resources such as photovoltaics and energy storage, the structure of power distribution systems has become more dynamic and complex. The volatility and uncertainty of these resources expose traditional power distribution systems to greater risks during operation, especially given the frequent occurrence of external uncertainties such as extreme weather, equipment aging, and cyberattacks, highlighting the vulnerability of power distribution systems. At the same time, the power distribution system plays an increasingly important role in the power grid, becoming the most vulnerable and critical link in the power system that most needs enhanced resilience.

[0003] When faced with faults, extreme environments, and equipment aging, the topology data of the distribution network is incomplete, and a complete and accurate topology cannot be obtained. Summary of the Invention

[0004] This application provides a method for reconstructing the topology of a power distribution network, which can obtain a complete and accurate topology.

[0005] To achieve the above objectives, this application adopts the following technical solution: Firstly, this application provides a method for reconstructing the topology of a power distribution network, including: Acquire multi-source heterogeneous data from the power distribution network, including monitoring data, geographic location data, equipment asset data, and dispatch log data; Extract node attribute features based on monitoring data, geographic location data, equipment asset data, and scheduling log data; The attribute characteristics of the edges are calculated based on monitoring data, geographic location data, equipment asset data, and scheduling log data; Generate a neighbor matrix based on the attribute characteristics of the edges; Determine if the node connection probability is less than a threshold. If it is less than the threshold, generate the node attribute features of layer t based on the neighbor matrix and the node attribute features of layer t-1. The node connection probability of the node attribute features in the t-th layer is calculated using cosine similarity. If the value exceeds the threshold, the target node attribute features are output directly. Output the topology based on the attribute characteristics of the target node and the attribute characteristics of the edges.

[0006] Optionally, the neighbor matrix and the node attribute features of layer (t-1) are used to generate the node attribute features of layer t, which can be calculated using the following expression:

[0007] in, For the node attribute features of layer t, For activation function, For degree matrix, For the neighbor matrix, For the node attribute features of the (t-1)th layer, Let be the weight matrix of the (t-1)th layer.

[0008] Alternatively, the expression for cosine similarity is:

[0009] in, Represents a node and nodes Similarity between them Represents a node In features eigenvalues ​​on, Represents a node In features eigenvalues ​​on, Represents a node The set of attribute features, Represents a node The set of attribute features, Represents a set of nodes.

[0010] Optionally, the method further includes: The target risk assessment function is determined based on topological vulnerability, load non-recovery rate, equipment aging level, and risk propagation between nodes; The risk of the distribution network topology is determined based on the target risk assessment function.

[0011] Optionally, the method further includes: The target recovery efficiency function is determined based on the load recovery rate and path congestion. The recovery efficiency of the distribution network topology is determined based on the target recovery efficiency function.

[0012] Optionally, the method further includes: Determine a multi-objective function that includes a target risk assessment function and a target recovery efficiency function; A local solution set is obtained by calculating the multi-objective function value, and the optimal solution set is selected from the local solution set based on the crowding degree.

[0013] Optionally, the expression for the congestion degree is:

[0014] in, Indicates the degree of crowding. Indicates the number of objective functions. For the first The solution is at the th solution. The values ​​that can be taken on the objective function, For the first The solution is at the th solution. The values ​​that can be taken on the objective function, For the first The maximum value of each objective function at the current level. for The minimum value of each objective function at the current level.

[0015] Secondly, this application provides a topology reconstruction device for a power distribution network, comprising: The acquisition module is used to acquire multi-source heterogeneous data of the power distribution network, including monitoring data, geographic location data, equipment asset data, and dispatch log data; extract node attribute features based on the monitoring data, geographic location data, equipment asset data, and dispatch log data; calculate edge attribute features based on the monitoring data, geographic location data, equipment asset data, and dispatch log data; and generate a neighbor matrix based on the edge attribute features. The optimization module is used to determine whether the node connection probability is less than a threshold. If it is less than the threshold, it generates the node attribute features of the t-th layer based on the neighbor matrix and the node attribute features of the (t-1)-th layer. It calculates the node connection probability of the node attribute features of the t-th layer using cosine similarity. If it is greater than the threshold, it directly outputs the target node attribute features. If the difference is less than the threshold, it outputs the predicted features of the t-th round. The output module outputs the topology structure based on the attribute characteristics of the target node and the attribute characteristics of the edges.

[0016] Thirdly, this application provides a computing device, including a memory and a processor; The memory stores one or more computer programs, the one or more computer programs including instructions; when the instructions are executed by the processor, the computing device performs the method as described in any one of the first aspects.

[0017] Fourthly, this application provides a computer-readable storage medium for storing a computer program for performing the method as described in any one of the first aspects.

[0018] As can be seen from the above technical solution, this application has at least the following beneficial effects: In this application, by integrating multi-source heterogeneous data such as monitoring data, geographic location data, equipment asset data, and scheduling logs, and combining the extraction of attribute features of nodes and edges, the limitations of a single data source can be overcome, and the physical and operational characteristics of the power distribution network can be captured more comprehensively. This provides a rich and three-dimensional information foundation for topology reconstruction, effectively improving the integrity and accuracy of the topology structure.

[0019] Secondly, by utilizing the concept of graph convolutional networks (generating node attribute features through iteration of neighbor matrices and multi-layer features) and cosine similarity to calculate node connection probabilities, data-driven topology inference is achieved. This approach can both optimize the model using known local connectivity information and make reasonable predictions about the connectivity status of unknown regions. It is particularly suitable for real-world scenarios where distribution network data is fragmented and incomplete, solving the bottleneck of traditional topology identification that relies on complete measured data.

[0020] Furthermore, by introducing a risk assessment function (integrating factors such as topological vulnerability and equipment aging) and a recovery efficiency function (combining indicators such as path congestion), and by selecting the optimal solution set based on multi-objective functions and congestion, topology reconstruction not only stays at the structural identification level, but also simultaneously supports risk prediction and recovery strategy optimization of the distribution network. This achieves integration from topology perception to decision support, providing strong support for improving the resilience and efficient operation of the distribution network.

[0021] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description

[0022] Figure 1 A flowchart illustrating a method for reconstructing a power distribution network topology, provided in an embodiment of this application; Figure 2 A schematic diagram of a device for reconstructing a power distribution network topology, provided as an embodiment of this application; Figure 3 This is a schematic diagram of a computing device provided in an embodiment of this application. Detailed Implementation

[0023] The terms "first," "second," and "third," etc., used in this application specification and accompanying drawings are used to distinguish different objects, not to limit a specific order.

[0024] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0025] To ensure clarity and conciseness in the description of the following embodiments, a brief introduction to the related technologies is given first: Distribution network topology refers to the physical connection relationship between nodes (such as substations, load centers, distributed power sources, etc.) and lines (such as feeders, switches, etc.) in the distribution network, and is the basis for power grid dispatching, fault recovery, and risk assessment.

[0026] Node attribute characteristics are parameters that describe the characteristics of a distribution network node itself, such as the node's voltage level, load type, spatial location, and the aging degree of the connected equipment.

[0027] Edge attributes are characteristic parameters that describe the connection lines between nodes, such as line impedance, current carrying capacity, operating status (on / off), and historical fault frequency.

[0028] Multi-source data in power distribution networks, including monitoring data (SCADA monitoring data), geographic location data (GIS spatial data), equipment asset data, and dispatch log data, often have conflicts (such as lines marked as "disconnected" in GIS actually being connected), missing data (such as 11% of lines having missing connection labels), or inconsistent formats (such as inconsistent equipment naming) due to independent system deployment and asynchronous maintenance and updates. Traditional methods that rely on a single data source or manual verification are difficult to integrate heterogeneous information, leading to deviations in topology inference.

[0029] In practice, only a small amount of local real topology (such as the connection status of some nodes) can be obtained through on-site inspections. However, the distribution network is large in scale and the topology changes dynamically (such as fault reconfiguration and distributed power source access). Existing methods are difficult to accurately infer the global topology based on local information, and misjudgments are prone to occur, especially in areas where data is missing.

[0030] Existing topology identification methods mostly focus only on structural accuracy, without simultaneously considering the impact of topology on power grid risks and recovery efficiency, making it difficult for reconstruction results to directly support dispatch decisions.

[0031] In view of this, embodiments of this application provide a method for reconstructing a power distribution network topology, which can be executed by a processing device. This processing device can be a terminal or a server. Terminals include, but are not limited to, smartphones, tablets, laptops, personal digital assistants, or smart wearable devices. Servers can be cloud servers, such as central servers in a central cloud computing cluster or edge servers in an edge cloud computing cluster. Alternatively, servers can be located in a local data center. A local data center refers to a data center directly controlled by the user.

[0032] This method first addresses the challenge of integrating multi-source heterogeneous data (SCADA, GIS, equipment asset data, etc.) by proposing the simultaneous extraction of node attribute features (such as voltage, spatial coordinates, and equipment aging degree) and edge attribute features (such as line impedance and operating status). Through complementary fusion at the feature level, it overcomes the information limitations of a single data source and lays a data foundation for accurate inference of topology.

[0033] Secondly, to address the bias in global topology inference caused by insufficient local information, the iterative learning concept of Graph Convolutional Networks (GCN) is introduced: a neighbor matrix is ​​generated based on the attribute features of edges, node attribute features are dynamically optimized through multi-layer feature propagation, and node connection probabilities are calculated by combining cosine similarity, thereby achieving adaptive learning of "local known information guiding global unknown inference", which can still ensure the continuity and accuracy of topology recognition in scenarios with missing data.

[0034] Furthermore, to bridge the gap between topology reconstruction and actual decision-making, the topology structure is innovatively linked with risk and recovery indicators: a risk assessment function is constructed by integrating factors such as topology vulnerability and equipment aging, a recovery efficiency function is designed by combining path congestion, and the optimal solution set is selected through multi-objective optimization and congestion screening. This ensures that the topology reconstruction results not only reflect physical connectivity but also directly support risk prediction and recovery strategy formulation for the distribution network, achieving a closed loop from "structure identification" to "resilience enhancement".

[0035] To make the technical solution of this application clearer and easier to understand, the following describes a method for reconstructing a power distribution network topology according to an embodiment of this application, in conjunction with the accompanying drawings. Figure 1 As shown in the figure, this is a flowchart of a method for reconstructing a power distribution network topology according to an embodiment of this application.

[0036] S201. The processing equipment acquires multi-source heterogeneous data from the power distribution network, including monitoring data, geographic location data, equipment asset data, and dispatch log data.

[0037] Monitoring data (SCADA monitoring data): This data is collected in real time by sensors, smart meters, and other devices installed at various nodes of the distribution network (such as substations and feeder terminals). It includes electrical quantity data such as voltage, current, power, and frequency, as well as operational status information such as switch status and protection action signals. This data is transmitted to the SCADA system via communication networks (such as fiber optics or dedicated wireless networks) to reflect the real-time operating conditions of the power grid and to provide a basis for judging electrical correlations between nodes (such as voltage correlation, which can help infer connection relationships).

[0038] Geographic location data (GIS spatial data): Acquired through geographic information acquisition equipment (such as surveying instruments and satellite positioning systems), and stored in the GIS system, it includes spatial coordinates of power distribution equipment (such as transformer locations and line routes), initial annotations of physical connections (such as the connection relationship between lines and nodes), equipment models, and other spatial attribute information, which is the basis for constructing the topological spatial framework.

[0039] Equipment asset data: mainly from the asset management system (AMS), is generated by records during the entire lifecycle management of equipment. It includes attribute data such as equipment commissioning time, manufacturer information, rated parameters (such as capacity and impedance), maintenance records, aging degree, and fault history. This information is used to assess the impact of equipment status on topology connection reliability (such as aging equipment may lead to increased connection vulnerability).

[0040] Dispatch log data: generated by the dispatch automation system or manual recording, covering text or structured data such as dispatch instructions, switch operation records, fault handling process, and abnormal event descriptions. It contains logical operation information of the equipment (such as the on / off status of the line after a certain operation) and can be used to verify or correct the connection status annotations in other data sources.

[0041] S202. The processing equipment extracts node attribute features based on monitoring data, geographical location data, equipment asset data, and scheduling log data.

[0042] Real-time electrical characteristics of nodes are extracted from SCADA monitoring data, such as voltage amplitude, current magnitude, active / reactive power, and frequency deviation. These characteristics reflect the real-time operating status of the nodes (e.g., whether they are overloaded or whether the voltage exceeds the limit).

[0043] Spatial location features of nodes are extracted from GIS spatial data, including latitude and longitude coordinates, distance from adjacent nodes, and geographical environment of the area (such as whether it is located in a densely populated area). These features reflect the physical layout attributes of the nodes and provide a basis for selecting reasonable connection ranges (such as nodes that are too far apart have a low probability of being connected).

[0044] Extract the attribute characteristics of the equipment associated with the node from the equipment asset data, such as the rated capacity, commissioning years, aging degree, and maintenance cycle of the transformer connected to the node, as well as the model and action threshold of the protection device. These characteristics reflect the health status and tolerance of the node's equipment and affect the vulnerability assessment of the node in the topology.

[0045] Extracting historical operation and event characteristics of nodes from scheduling log data, such as the number of operations of node-associated switches, the time of the most recent failure, the type of failure (such as short circuit, overload), and recovery time, these characteristics reflect the operational stability and historical reliability of the nodes and can help correct the inference of the node connection status (such as nodes with frequent failures may have implicit connection problems).

[0046] S203. The processing equipment calculates the attribute characteristics of the edges based on monitoring data, geographical location data, equipment asset data, and scheduling log data.

[0047] The active / reactive power flow and power factor of the line are calculated by using the power and current data of the nodes at both ends of the line, reflecting the real-time load status of the line.

[0048] Based on the node voltage amplitude and phase difference, the voltage drop rate of the line is calculated. Combined with the line impedance parameters (from equipment asset data), the rationality of the connection relationship can be verified (such as abnormal voltage drop, which may indicate connection error).

[0049] Extract the number of historical faults, fault types (such as short circuits and grounding), and fault durations of the line to assess line reliability, which can then be used as a correction factor for connection confidence.

[0050] Based on the geographical coordinates of the nodes at both ends of the line, calculate the line length, direction, and type of area it traverses (such as urban or suburban areas), and assess the feasibility of physical connections (for ultra-long-distance lines, it is necessary to verify whether there are intermediate equipment such as booster stations).

[0051] Calculate the Euclidean distance or path distance between nodes, and determine the rationality of the connection based on the line type (overhead line / cable). For example, the probability of connecting node pairs that are too far apart in space but have no special cross-regional line markings is low.

[0052] Extract line type, conductor material, cross-sectional area, rated current carrying capacity, impedance parameters, etc., to calculate theoretical power flow distribution, and compare with SCADA measured values ​​to verify connection status.

[0053] Based on factors such as the line's years of operation, degree of aging, and maintenance records, the health index of the line is assessed and used as a weighting factor for connection reliability (e.g., severely aged lines may have a hidden risk of disconnection).

[0054] By using scheduling instruction logs and switch operation records, the on / off status of the line can be directly obtained (e.g., "circuit breaker closed" indicates that the line is connected), which can be used as a direct label for edge attributes.

[0055] By statistically analyzing the historical number of operations and operation time intervals of the lines, the scheduling flexibility and stability of the lines can be assessed to evaluate the dynamic characteristics of the connections.

[0056] Generate a neighbor matrix based on the attribute characteristics of the edges.

[0057] The neighbor matrix is ​​an n×n square matrix if there are n nodes in a distribution network. The rows and columns of the matrix correspond to the individual nodes, and the elements in the matrix... (No. Line, number Columns are used to represent nodes. With nodes Is it a directly connected neighbor node?

[0058] The processing device integrates the above-mentioned multi-source features to construct the attribute vector of the edge (such as "[power flow size, voltage drop rate, line length, aging rate, number of operations]"), and forms a neighbor matrix with a unified dimension through standardization, normalization and other processing.

[0059] S204. The processing device determines whether the node connection probability is less than the threshold.

[0060] If the node connection probability is less than the threshold, execute S205; if the node connection probability is greater than the threshold, execute S206.

[0061] The probability of node connection is calculated using cosine similarity, expressed as follows:

[0062] in, Represents a node and nodes Similarity between them Represents a node In features eigenvalues ​​on, Represents a node In features eigenvalues ​​on, Represents a node The set of attribute features, Represents a node The set of attribute features, Represents a set of nodes.

[0063] S205. Generate the node attribute features of layer t based on the neighbor matrix and the node attribute features of layer t-1, and calculate the node connection probability of the node attribute features of layer t using cosine similarity.

[0064] The node attribute characteristics of layer t can be calculated using the following expression:

[0065] in, For the node attribute features of layer t, For activation function, For degree matrix, For the neighbor matrix, For the node attribute features of the (t-1)th layer, Let be the weight matrix of the (t-1)th layer.

[0066] Then, the node connection probability of the node attribute features in the t-th layer is calculated using cosine similarity.

[0067] S206, Output the target node attribute characteristics.

[0068] The target node attribute features refer to the node attribute features that need to be output or used as the core reference in the entire iterative calculation or judgment process.

[0069] S207. Output the topology based on the attribute characteristics of the target node and the attribute characteristics of the edges.

[0070] By integrating the core characteristics of nodes (such as node voltage, spatial location, equipment status, etc.) after iterative optimization or screening, and the characteristics of the connecting lines between nodes (such as line impedance, on / off status, aging rate, etc.), the physical connection network between nodes and lines in the distribution network is finally constructed, that is, it is clear which nodes are directly connected through which lines to form a complete topology.

[0071] In the process of topology reconstruction of distribution networks, it is not enough to simply infer connectivity relationships from data characteristics. A series of constraints closely related to the physical laws, operational rules, and actual scenarios of the power grid must also be met. These constraints are crucial to ensuring that the reconstruction results accurately reflect the actual operating state of the distribution network. Specifically, these constraints can be expanded from the following dimensions: Node power flow balance constraint: Ensures power balance at each node, conforming to the law of conservation of electricity.

[0072]

[0073] in, For nodes The set, Indicates at time and scene The next node and nodes Connection weights between them Represents a node voltage amplitude, Represents a node voltage amplitude, Represents a node Angle, Represents a node Angle, Represents a node Power generation capacity, Represents a node The load power, Represents a node Reduce load power.

[0074] Maximum number of switching operations: Limit the number of switching operations within each time step to avoid system instability caused by frequent switching.

[0075]

[0076] in, For the set of branches, For binary activation indicator, For branches The switching state This indicates the maximum number of switching operations allowed by the system, representing the number of operations within a single time period. Within, the maximum total number of switching operations that the system can perform.

[0077] Consistent topology constraints: Ensure that the identified topology is consistent with known GIS data, and ensure the accuracy of the risk assessment model.

[0078]

[0079] in, Indicates time Scene Next, node and The strength of the connection between them Indicates structural vulnerability markers identified by GIS. Represents a node The structural vulnerability weight is used to measure the structural vulnerability of nodes. Vulnerability in the network.

[0080] Equipment failure probability constraint: Ensure that the failure probability of the equipment does not exceed the preset tolerance value, so as to take into account the aging and load conditions of the equipment in the risk assessment.

[0081]

[0082] in, Indicates the possibility of failure Less than or equal to The probability of the function. This is represented as a parameter indicating the degree of equipment aging. Load power parameters and structural vulnerability parameters A function jointly determined Indicates the tolerance level.

[0083] node The total input supply (generation and input) at the point must equal the sum of service demand, load shedding, and system losses to maintain power consistency.

[0084]

[0085] in, Represents a node Power generation capacity, Represents a node Input power, Represents a node The load power, Represents a node Reduce load power, Represents a node The power loss.

[0086] Switching operation restrictions: Limit the maximum number of switching operations allowed within each time step to ensure the rationality of switching operations during the refactoring process.

[0087]

[0088] in, Represents an indicator function, when The function value is 1 when the condition is met, and 1 when the condition is not met. When the function value is 0, It is a binary decision variable, representing the time... ,line Whether it is activated, This indicates that the line was reconfigured at that moment. Indicates time The maximum number of lines that can be activated is limited.

[0089] Line flow and topology consistency: Ensure that the power flow after topology switching is consistent with the line status to avoid invalid power flow transfer.

[0090]

[0091] in, Representing a path The actual power flow, Represents a binary decision variable. This represents the theoretical power flow.

[0092] S208. Determine the target risk assessment function based on topology vulnerability, load non-recovery rate, equipment aging degree, and risk propagation between nodes.

[0093] The calculation expression is:

[0094] in, This represents the target risk assessment function, which calculates the total risk. Indicates time, For the scene The set, Represents a set of nodes. Representing a scene The probability weights below, Indicates structural fragility. Represents the structural vulnerability weight. Indicates the load recovery efficiency weight. This indicates unrecovered load. For total load, Indicates error. Indicates the weight of equipment aging. Indicates the degree of equipment aging. Indicates the risk propagation weight. This represents a measure of risk propagation.

[0095] S209. Determine the target recovery efficiency function based on the load recovery rate and path congestion.

[0096] The calculation expression is:

[0097] in, This represents the target recovery efficiency function, used to calculate the effectiveness of the entire recovery process. Representing a path The set, Indicates the recovered load. Representing a path In time The recovery weight represents the degree of activation or participation of that path. Indicates redundancy weight. Indicates path congestion level.

[0098] S2010, Determine the multi-objective function of the target risk assessment function and the target recovery efficiency function.

[0099] The calculation expression is:

[0100] in, This indicates a weighting or balancing process. This represents the target risk assessment function. Target recovery efficiency function.

[0101] S2011, Output the optimal solution set of the multi-objective function.

[0102] A local solution set is obtained by calculating the multi-objective function value, and the optimal solution set is selected from the local solution set based on the crowding degree.

[0103] A multi-objective function refers to optimizing multiple objectives simultaneously (such as "lowest risk assessment value" or "highest recovery efficiency"). These objectives often conflict (for example, reducing risk may require adding redundant lines, leading to increased topology complexity). Each objective corresponds to a quantification function (such as a risk assessment function or a recovery efficiency function), and the merits of the solution can be measured by calculating the values ​​of these functions.

[0104] For a specific region or problem scenario, a set of candidate solutions (each solution corresponding to a topology) is obtained through algorithmic search (such as genetic algorithm or particle swarm optimization). Each solution has a corresponding multi-objective function value (e.g., solution A has a risk value of 0.3 and an efficiency of 0.8). These solutions have their own advantages and disadvantages in multiple objectives, and it is impossible to judge their absolute good or bad based on a single objective, thus forming a "local solution set".

[0105] Each solution is treated as a "particle" and randomly distributed in the solution space. Each particle has its own position (corresponding to a solution) and velocity (which determines the next search direction).

[0106] Each particle calculates a multi-objective function value based on its own position, and records its own "historical best position" (individual best) and the "global best position" (group best) of the entire swarm.

[0107] Each particle adjusts its speed and position based on its own experience (individual optimality) and the collective experience (global optimality), moving towards a better region (such as making the solution more balanced between the goals of "risk" and "efficiency").

[0108] After multiple iterations, the particle swarm gradually gathers in a better region of the solution space, eventually forming a set of widely distributed solutions that perform well across multiple objectives, known as the "local solution set".

[0109] Crowding degree is used to measure the "density" of different solutions in the solution set, reflecting the diversity among solutions. The lower the crowding degree, the greater the difference between the solution and other solutions, and the more unique optimization information is retained; the higher the crowding degree, the more similar the solution is to surrounding solutions, and can be regarded as redundant.

[0110] (For example, if two options have almost the same risk and efficiency values, with only slight differences in error, then the crowding level is high, and one can be retained; if an option differs significantly from other options in both risk and efficiency, then the crowding level is low, and it should be retained first.) The expression for congestion is:

[0111] in, Indicates the degree of crowding. Indicates the number of objective functions. For the first The solution is at the th solution. The values ​​that can be taken on the objective function For the first The solution is at the th solution. The values ​​that can be taken on the objective function For the first The maximum value of each objective function at the current level. for The minimum value of each objective function at the current level.

[0112] From the local solution set, by calculating the congestion of each scheme, the scheme with low congestion is retained first. The final "optimal solution set" is not a single scheme, but a set of high-quality schemes that perform in a balanced manner across multiple objectives, which allows decision-makers to select the most suitable topology scheme according to the actual scenario (such as emergency response and daily scheduling).

[0113] Based on the above description, this application has the following beneficial effects: In this application, by integrating multi-source heterogeneous data such as SCADA monitoring data, GIS spatial data, equipment asset data, and dispatch logs, and combining the extraction of attribute features of nodes and edges, the limitations of a single data source can be overcome, and the physical and operational characteristics of the power distribution network can be captured more comprehensively. This provides a rich and three-dimensional information foundation for topology reconstruction, effectively improving the integrity and accuracy of the topology structure.

[0114] Secondly, by utilizing the concept of graph convolutional networks (generating node attribute features through iteration of neighbor matrices and multi-layer features) and cosine similarity to calculate node connection probabilities, data-driven topology inference is achieved. This approach can both optimize the model using known local connectivity information and make reasonable predictions about the connectivity status of unknown regions. It is particularly suitable for real-world scenarios where distribution network data is fragmented and incomplete, solving the bottleneck of traditional topology identification that relies on complete measured data.

[0115] Furthermore, by introducing a risk assessment function (integrating factors such as topological vulnerability and equipment aging) and a recovery efficiency function (combining indicators such as path redundancy and congestion), and by selecting the optimal solution set based on multi-objective functions and congestion, topology reconstruction not only stays at the structural identification level, but also simultaneously supports risk prediction and recovery strategy optimization of the distribution network. This achieves integration from topology perception to decision support, providing strong support for improving the resilience and efficient operation of the distribution network.

[0116] The above text combined Figure 1 The method for reconstructing the power distribution network topology provided in the embodiments of this application has been described in detail. The apparatus and equipment provided in the embodiments of this application will be described below with reference to the accompanying drawings.

[0117] like Figure 2 As shown in the figure, this is a schematic diagram of a distribution network topology reconstruction device provided in an embodiment of this application. The device includes: The acquisition module 301 is used to acquire multi-source heterogeneous data of the power distribution network, including monitoring data, geographical location data, equipment asset data, and dispatch log data; extract node attribute features based on the monitoring data, geographical location data, equipment asset data, and dispatch log data; calculate edge attribute features based on the monitoring data, geographical location data, equipment asset data, and dispatch log data; and generate a neighbor matrix based on the edge attribute features. Optimization module 302 is used to determine whether the node connection probability is less than a threshold. If it is less than the threshold, it generates the node attribute features of layer t based on the neighbor matrix and the node attribute features of layer t-1. It calculates the node connection probability of the node attribute features of layer t using cosine similarity. If it is greater than the threshold, it directly outputs the target node attribute features. If the difference is less than the threshold, it outputs the predicted features of round t. Output module 303 outputs the topology structure based on the attribute characteristics of the target node and the attribute characteristics of the edge.

[0118] Optionally, optimization module 302 is specifically used to calculate the following expression:

[0119] in, For the node attribute features of layer t, For activation function, For degree matrix, For the neighbor matrix, For the node attribute features of the (t-1)th layer, Let be the weight matrix of the (t-1)th layer.

[0120] Optionally, the output module 303 is also used to determine the target risk assessment function based on topology vulnerability, load non-recovery rate, equipment aging degree and risk propagation between nodes; The risk of the distribution network topology is determined based on the target risk assessment function.

[0121] Optionally, the output module 303 is also used to determine the target recovery efficiency function based on the load recovery rate and path congestion. The recovery efficiency of the distribution network topology is determined based on the target recovery efficiency function.

[0122] Optionally, the optimization module 303 is also used to determine a multi-objective function that includes the target risk assessment function and the target recovery efficiency function; A local solution set is obtained by calculating the multi-objective function value, and the optimal solution set is selected from the local solution set based on the crowding degree.

[0123] The apparatus for reconstructing the distribution network topology according to the embodiments of this application can correspond to performing the method described in the embodiments of this application, and the other operations and / or functions of each module / unit of the apparatus for reconstructing the distribution network topology are respectively for implementing Figure 1 For the sake of brevity, the corresponding processes of each method in the illustrated embodiments will not be described in detail here.

[0124] This application also provides a computing device. For example... Figure 3 As shown in the figure, this is a schematic diagram of a computing device provided in an embodiment of this application. The computing device 700 includes a bus 701, a processor 702, a communication interface 703, and a memory 704. The processor 702, the memory 704, and the communication interface 703 communicate with each other via the bus 701.

[0125] The 701 bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0126] The processor 702 can be any one or more of the following processors: central processing unit (CPU), graphics processing unit (GPU), microprocessor (MP), or digital signal processor (DSP).

[0127] The communication interface 703 is used for communication with external devices.

[0128] Memory 704 may include volatile memory, such as random access memory (RAM). Memory 704 may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid state drive (SSD).

[0129] The memory 704 stores executable code, and the processor 702 executes the executable code to perform the aforementioned method for reconstructing the power distribution network topology.

[0130] Specifically, in achieving Figure 2 In the case of the illustrated embodiment, and Figure 2 When the modules or units of the distribution network topology reconstruction device described in the embodiment are implemented by software, the execution... Figure 2 The software or program code required for the functions of each module / unit can be partially or entirely stored in the memory 704. The processor 702 executes the program code corresponding to each unit stored in the memory 704 to perform the aforementioned method for reconstructing the power distribution network topology.

[0131] This application also provides a computer-readable storage medium. The computer-readable storage medium can be any available medium that a computing device can store, or a data storage device such as a data center containing one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive). The computer-readable storage medium includes instructions that instruct a computing device to perform the aforementioned method for reconstructing the power distribution network topology.

[0132] This application also provides a computer program product comprising one or more computer instructions. When the computer instructions are loaded and executed on a computing device, all or part of the processes or functions described in this application are generated.

[0133] The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, or data center to another website, computer, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means.

[0134] When the computer program product is executed by a computer, the computer performs any of the aforementioned methods for reconstructing the distribution network topology. The computer program product can be a software installation package; when any of the aforementioned methods for reconstructing the distribution network topology needs to be used, the computer program product can be downloaded and executed on the computer.

[0135] The descriptions of the processes or structures corresponding to the above figures each have their own emphasis. For parts of a process or structure that are not described in detail, please refer to the relevant descriptions of other processes or structures.

[0136] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be covered within the scope of protection of this application.

Claims

1. A method for reconstructing the topology of a power distribution network, characterized in that, The method includes: Acquire multi-source heterogeneous data from the power distribution network, including monitoring data, geographic location data, equipment asset data, and dispatch log data; Extract node attribute features based on monitoring data, geographic location data, equipment asset data, and scheduling log data; The attribute characteristics of the edges are calculated based on monitoring data, geographic location data, equipment asset data, and scheduling log data; Generate a neighbor matrix based on the attribute characteristics of the edges; Determine if the node connection probability is less than a threshold. If it is less than the threshold, generate the node attribute features of layer t based on the neighbor matrix and the node attribute features of layer t-1. The node connection probability of the node attribute features in the t-th layer is calculated using cosine similarity. If the value exceeds the threshold, the target node attribute features are output directly. Output the topology based on the attribute characteristics of the target node and the attribute characteristics of the edges.

2. The method according to claim 1, characterized in that, The neighbor matrix and the node attribute features of layer (t-1) generate the node attribute features of layer t, which can be calculated using the following expression: in, For the node attribute features of layer t, For activation function, For degree matrix, For the neighbor matrix, For the node attribute features of the (t-1)th layer, Let be the weight matrix of the (t-1)th layer.

3. The method according to claim 1, characterized in that, The expression for the cosine similarity is: in, Represents a node and nodes Similarity between them Represents a node In features eigenvalues ​​on, Represents a node In features eigenvalues ​​on, Represents a node The set of attribute features, Represents a node The set of attribute features, Represents a set of nodes.

4. The method according to claim 1, characterized in that, The method further includes: The target risk assessment function is determined based on topological vulnerability, load non-recovery rate, equipment aging level, and risk propagation between nodes; The risk of the distribution network topology is determined based on the target risk assessment function.

5. The method according to claim 1, characterized in that, The method further includes: The target recovery efficiency function is determined based on the load recovery rate and path congestion. The recovery efficiency of the distribution network topology is determined based on the target recovery efficiency function.

6. The method according to claim 1, characterized in that, The method further includes: Determine a multi-objective function that includes a target risk assessment function and a target recovery efficiency function; A local solution set is obtained by calculating the multi-objective function value, and the optimal solution set is selected from the local solution set based on the crowding degree.

7. The method according to claim 1, characterized in that, The expression for the congestion level is: in, Indicates the degree of crowding. Indicates the number of objective functions. For the first The solution is at the th solution. The values ​​that can be taken on the objective function, For the first The solution is at the th solution. The values ​​that can be taken on the objective function, For the first The maximum value of each objective function at the current level. for The minimum value of each objective function at the current level.

8. A topology reconstruction device for a power distribution network, characterized in that, The device includes: The acquisition module is used to acquire multi-source heterogeneous data of the power distribution network, including monitoring data, geographic location data, equipment asset data, and dispatch log data; extract node attribute features based on the monitoring data, geographic location data, equipment asset data, and dispatch log data; calculate edge attribute features based on the monitoring data, geographic location data, equipment asset data, and dispatch log data; and generate a neighbor matrix based on the edge attribute features. The optimization module is used to determine whether the node connection probability is less than a threshold. If it is less than the threshold, it generates the node attribute features of the t-th layer based on the neighbor matrix and the node attribute features of the (t-1)-th layer. It calculates the node connection probability of the node attribute features of the t-th layer using cosine similarity. If it is greater than the threshold, it directly outputs the target node attribute features. If the difference is less than the threshold, it outputs the predicted features of the t-th round. The output module outputs the topology structure based on the attribute characteristics of the target node and the attribute characteristics of the edges.

9. A computing device, characterized in that, Including memory and processor; The memory stores one or more computer programs, the one or more computer programs including instructions; when the instructions are executed by the processor, the computing device performs the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a computer program for performing the method as described in any one of claims 1 to 7.

Images