Topology library generation method and device for power distribution network fault multiple transfer demand

Abstract

The application discloses a kind of topological graph library generation method and device for power distribution network fault multiple transfer demand, and relates to power distribution network fault recovery technical field.The method first abstracts power distribution network as undirected graph model, generates topological unique coding based on edge state vector for hash storage and deduplication;Then generate fault isolation topological and divide live / dead area;Then filter candidate tie line to generate first transfer topological, by constructing power supply electronic topological local correlation domain, execute "first combination then division" topological transformation recursion to generate secondary and multiple transfer topological;Finally, based on unique coding, realize global topological deduplication, with fault isolation topological as root node to build hierarchical topological tree.The application solves the problems of low efficiency, repeated calculation and other problems in the prior art of multiple transfer topological generation, and provides strong support for power distribution network fault rapid recovery.

Description

Technical Field

[0001] This invention belongs to the field of power distribution network fault recovery and dispatch control technology, specifically relating to a method and apparatus for generating a topology library for multiple power transfer needs in power distribution network faults. Background Technology

[0002] As a crucial link connecting the transmission network and end users in the power system, the distribution network directly determines the reliability and quality of power supply. Its safe and stable operation plays an irreplaceable role in industrial production, residential life, and socio-economic development. Distribution networks typically have a radial structure, with sectionalizing switches and tie lines installed on the lines. Their core function is to isolate faulty sections and quickly restore power to non-faulty areas in the event of a fault. A typical fault handling process involves first isolating the faulty section, and then using operations such as closing tie lines and opening / closing sectionalizing switches to transfer loads and adjust power supply boundaries, ensuring normal power supply to non-faulty areas.

[0003] Currently, existing distribution network reconfiguration methods can be broadly categorized into heuristic methods, graph theory methods, mathematical programming methods, and hybrid methods. The core logic is as follows: abstracting the distribution network into a graph model, using switch states as decision variables, and employing methods such as spanning tree search, branch switching, or heuristic search to form a feasible recovery topology after fault isolation. Then, through power flow calculations or constraint checks, schemes that meet constraints such as radial shape, capacity, voltage, and current are selected.

[0004] However, in practical engineering applications, existing technical solutions have significant limitations in meeting the needs of multiple power transfers following distribution network faults. These limitations manifest in three main ways: First, the state space explosion problem is prominent: with the increasing number of distribution network switches and the complex topology, the generation of candidate topologies covering multiple power transfers easily leads to state space explosion and an exponential increase in the number of candidate topologies, resulting in low generation efficiency and an inability to quickly respond to online dispatching needs. Second, a large number of duplicate topologies appear: existing methods lack effective cross-layer and intra-layer deduplication mechanisms, easily generating a large number of duplicate topologies during multiple power transfers, increasing the redundant workload of constraint verification and scheme selection, further reducing dispatching efficiency. Third, insufficient expression of process constraints: existing methods struggle to fully consider basic structural constraints such as radial operation and single-source power supply, as well as engineering operation constraints such as transformer capacity, line current, and node voltage. Some generated candidate topologies, while meeting structural requirements, lack engineering feasibility.

[0005] Therefore, there is an urgent need to develop a method that can quickly generate multiple power transfer topologies for distribution network fault states, solve the problems of low topology generation efficiency, excessive repetitive calculations, and poor engineering feasibility in existing technologies, and meet the need for rapid power restoration after distribution network faults. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of existing technologies and address the problems of state space explosion, redundant and repetitive topologies, insufficient constraint expression, and difficulty in meeting engineering feasibility in candidate topology generation under the multiple power transfer requirements of distribution network faults. This invention provides a method and apparatus for generating a topology library for distribution network faults with multiple power transfer requirements. Under the premise of strictly meeting basic structural constraints such as radial operation and single-source power supply, it performs hierarchical recursive search on the fault-isolated network to quickly generate and deduplicate multi-level candidate topologies. Simultaneously, it retains the evolutionary relationships between candidate topologies and corresponding switch operation information, thereby providing fundamental support for the generation, selection, and online auxiliary decision-making of distribution network fault recovery schemes, and improving the fault recovery capability and power supply reliability of the distribution network.

[0007] The technical solution adopted in this invention is as follows:

[0008] A method for generating a topology library to meet the needs of multiple power transfers during distribution network faults includes the following steps:

[0009] S1: Construct a distribution network graph model and implement a unique topology code: Abstract the distribution network into an undirected graph containing a set of nodes and a set of edges, divide the edge set and define edge state variables, arrange them in a fixed order to form an edge state vector, and construct a unique topology code based on the edge state vector for hash storage, fast comparison and deduplication of the topology;

[0010] S2: Generate the topology after fault isolation: Determine the fault location, disconnect the nearest circuit breakers at both ends of the fault point to form an isolation edge, construct the topology after fault isolation, and divide the set of energized and de-energized nodes and the de-energized area;

[0011] S3: Generate a primary power transfer topology: Based on the fault isolation topology, enumerate all disconnected tie lines that are related to the restoration of power supply to the power-loss area, screen tie lines with power transfer potential, perform a closure operation on each tie line that meets the conditions to generate candidate topologies, establish radial constraints and single power supply constraints to screen effective primary power transfer topologies and form a set.

[0012] S4: Recursively generate secondary transfer topology: Define the power supply scale of the power node, the set of power sources whose power supply scale expands relative to the fault isolation topology after transfer, locate the power supply topology with transfer potential, construct the power supply topology related domain and screen operable edges, the operable edges include closable edges and disconnectable edges, perform a two-step topology transformation of "combining and then separating", including: first closing the closable edges to build a temporary transfer channel, then disconnecting the disconnectable edges to eliminate the temporary loop, restore the topology to the loop-free running state, complete one topology transformation, generate candidate topologies, traverse all combinations of closed operable edges, and perform multi-class constraint checks to generate all transfer candidate topologies;

[0013] S5: Topology Deduplication and Topology Library Construction: Based on the unique topology code, all candidate topologies are globally deduplicated. The topology after fault isolation is used as the root node, and a topology tree is constructed according to the transfer level. The above steps are repeated for different fault scenarios to form a topology library forest composed of multiple topology trees, and the storage, updating and retrieval of the library are realized.

[0014] S6: Multi-level topology library construction: Preset the maximum number of transfers K, repeat the recursive generation of secondary transfer topologies described in step S4, the topology deduplication and topology library construction process described in step S5, generate multiple transfer topologies, and terminate the recursion when the number of transfer topologies reaches K or a certain layer cannot generate a new candidate topology that meets the constraints, and finally obtain a multi-level topology library with fault isolation topology as the root node.

[0015] Furthermore, in S1, the construction of the distribution network graph model includes abstracting the distribution network into an undirected graph containing power source nodes and load nodes; the edge set is divided into three categories according to equipment attributes: ordinary lines, lines with sectionalizing switches, and tie lines; and edge state variables are defined using binary closed and open states for the edge set. All edge state variables are combined sequentially to form an edge state vector, and different edge state vectors correspond to different topological states, constituting a unique topological state code.

[0016] Furthermore, in S2, the set of charged and de-charged nodes is calculated using a breadth-first search algorithm. Starting from the power node, all nodes reachable by closed edges are traversed to form the set of charged nodes, and the remaining nodes form the set of de-charged nodes. The de-charged region is divided using a de-charged region partitioning function to provide a clear object for subsequent power transfer operations.

[0017] Furthermore, in S2, a fault location and isolation algorithm is used to determine the fault location, including the acquisition and completion of three-phase current and voltage data uploaded by FTUs at each node of the distribution network, the construction of a fault location mathematical model based on the minimum fault set theory and the determination of isolation edges, the location of the fault based on the minimum fault set theory combined with intelligent optimization algorithm, the selection of the isolation edge corresponding to the circuit breaker at both ends of the fault point, and the disconnection of the isolation edge to form a fault isolation topology.

[0018] Furthermore, in S3, the screening criteria for the tie line with the potential for power transfer are: the tie line is in an open state, and the two ends of the tie line belong to the energized area and the de-energized area respectively, so as to ensure that the tie line can connect the energized area and the de-energized area to realize power transfer.

[0019] Furthermore, in S4, the power supply scale refers to the total load power of the power supply nodes, the total restored load capacity, or a weighted restoration benefit index that takes into account the importance of the load.

[0020] Furthermore, in S4, the candidate topology must satisfy the following three types of constraints: structural constraints, global charged set invariance constraints, and engineering operation constraints.

[0021] Furthermore, in S4, the relevant domain is a local subgraph consisting of the set of operable edges related to the power supply boundary adjustment and their adjacent nodes.

[0022] Furthermore, in S5, the deduplication of the topology is achieved through a global hash set, which compares the unique code of the newly generated topology with the code in the hash set to eliminate duplicate topologies.

[0023] A topology map library generation device for the needs of multiple power transfers in distribution network faults includes: a topology coding module, a fault isolation topology generation module, a primary power transfer topology generation module, a secondary power transfer topology generation module, a topology map library construction module, and a multi-level topology map library construction module.

[0024] The topology coding module is used to abstract the distribution network into an undirected graph containing a set of nodes and a set of edges, divide the edge set and define edge state variables, arrange them in a fixed order to form an edge state vector, and construct a unique topology state code based on the edge state vector for hash storage, fast comparison and deduplication of the topology.

[0025] The fault isolation topology generation module is used to determine the fault location, disconnect the nearest circuit breakers at both ends of the fault point to form an isolation edge, construct the fault isolation topology, and divide the set of energized and de-energized nodes and the de-energized area.

[0026] The primary power transfer topology generation module is used to enumerate all disconnected tie lines related to power restoration in the power-loss area based on the fault isolation topology, screen tie lines with power transfer potential, perform a closure operation on each tie line that meets the conditions to generate candidate topologies, establish radial constraints and single power supply constraints to screen effective primary power transfer topologies and form a set.

[0027] The secondary power transfer topology generation module is used to define the power supply scale of power nodes, the power supply set whose power supply scale has expanded relative to the fault isolation topology after power transfer, locate power supply topologies with power transfer potential, construct power supply topology related domains and screen operable edges. The operable edges include closable edges and detachable edges. Perform a two-step topology transformation of "combining first and then separating", including: first closing the closable edges to build a temporary power transfer channel, then disconnecting the detachable edges to eliminate the temporary loop, restoring the topology to a loop-free operating state, completing one topology transformation, generating candidate topologies, traversing all combinations of closed operable edges, and verifying through multiple types of constraints to generate all power transfer candidate topologies.

[0028] The topology library construction module is used to globally deduplicate all candidate topologies based on the unique topology code, and construct a topology tree according to the transfer level with the fault-isolated topology as the root node. The above steps are repeated for different fault scenarios to form a topology library forest composed of multiple topology trees, and to realize the storage, updating and retrieval of the library.

[0029] The multi-level topology library construction module is used to preset the maximum number of transfers K, repeat the process of recursively generating secondary transfer topologies, deduplicating topologies, and constructing the topology library, and generate multiple transfer topologies. When the number of transfer topologies reaches K or a certain layer cannot generate a new candidate topology that satisfies the constraints, the recursion is terminated, and finally a multi-level topology library with fault isolation topology as the root node is obtained.

[0030] This invention abstracts the power distribution network into a graph model and uses a binary topology code with a fixed edge order as a unique topology identifier for fast comparison of candidate topologies, hash storage, and cross-layer deduplication, thereby reducing the overhead of repeated searches.

[0031] This invention establishes a rapid determination method for structural constraints such as radial constraints and single-power-supply constraints, ensuring the safety and feasibility of generating topology structures;

[0032] This invention proposes to take the power source whose power supply scale has expanded as the recursive object, and to perform a local topology transformation of "combining first and then splitting" only in the relevant domain of its power supply topology, instead of blindly searching throughout the entire network. This can effectively compress the search space and improve the generation efficiency of secondary and multiple transfer candidate topologies.

[0033] This invention sets an invariant constraint on the global charged set, so that subsequent power transfers focus on adjusting the power supply boundary, thus avoiding expanding the power outage area or introducing new recovery objects.

[0034] This invention constructs a multi-level topology tree with fault isolation topology as the root node, saves the evolutionary relationship and switching operation information between candidate topologies, and forms a topology library that can directly serve scheme generation, screening, sequence optimization and online auxiliary decision-making.

[0035] Advantages of this invention:

[0036] 1. The generated objectives are more aligned with actual scheduling practices. Existing similar solutions often focus on finding a feasible or optimal solution, while this invention is geared towards building a topology graph library, which can generate a multi-level, multi-candidate, and comparable set of recovery solutions at once, making it more suitable for online decision support.

[0037] 2. Higher search efficiency. This invention utilizes fixed edge order unique encoding and global hash deduplication to avoid repeated generation and verification within and across layers; at the same time, it only searches locally within the electron topology-related domain, significantly narrowing the search range.

[0038] 3. Enhanced traceability of results. This invention not only preserves the final candidate topology, but also retains the evolutionary relationships between nodes and corresponding switching operation information, facilitating subsequent switching sequence optimization, decision support, and knowledge reuse.

[0039] 4. It is suitable for the rapid generation, deduplication, storage and retrieval of candidate topologies for multiple power transfers after a distribution network fault, providing technical support for fault recovery decision-making, power transfer scheme optimization and online dispatch assistance. Attached Figure Description

[0040] Figure 1 This is a flowchart illustrating the process of generating the topology library that has been repeatedly provided by this invention. Detailed Implementation

[0041] See Figure 1 .

[0042] A method for generating a topology library to meet the needs of multiple power transfers during distribution network faults includes the following steps:

[0043] S1. Construct a distribution network graph model and implement a unique topology code: Abstract the distribution network into an undirected graph containing a set of nodes and a set of edges, divide the edge set and define edge state variables, arrange them in a fixed order to form an edge state vector, and construct a unique topology code based on the edge state vector for hash storage, fast comparison and deduplication of the topology.

[0044] Furthermore, in one embodiment of the present invention, the construction of the distribution network graph model includes abstracting the distribution network into an undirected graph containing power source nodes and load nodes; dividing the edge set into three categories according to equipment attributes: ordinary lines, lines with sectionalizing switches, and tie lines; and defining edge state variables using binary closed and open states for the edge set. All edge state variables are combined sequentially to form an edge state vector, with different edge state vectors corresponding to different topological states, thus forming a unique topological state code.

[0045] Furthermore, in one embodiment of the present invention, constructing the distribution network diagram model and implementing topological unique coding specifically includes:

[0046] 1.1 Abstraction of Undirected Graphs in Distribution Networks

[0047] Abstracting the power distribution network as an undirected graph ,in The set of nodes includes the set of power nodes S and the set of load nodes; Let this be the set of edges, representing the connectivity relationships between power distribution lines and switching equipment. To facilitate the description of power distribution network switching operations, the set of edges is divided according to equipment attributes: ordinary line set. Collection of circuits including sectionalizing switches With connecting lines ,Right now

[0048]

[0049] 1.2 Definition and Ordered Arrangement of Edge State Variables

[0050] For any edge Define a binary variable ,in Representing an edge Closed conduction, Representing an edge Disconnect. Set the edges together. Arranged into an ordered list according to a pre-agreed fixed order. ,in , represents the total number of edges in the power distribution network diagram model.

[0051] 1.3 Topological Unique Encoding Construction and Control Algorithm

[0052] For any topological state , can be represented as

[0053]

[0054] The corresponding edge state vector is denoted as . And thereby construct a topologically unique code.

[0055]

[0056] The encoding can serve as a unique identifier for the topology state and can be directly used for hash storage, fast comparison, and deduplication.

[0057] S2. Topology generation after fault isolation: Determine the fault location, disconnect the nearest circuit breakers at both ends of the fault point to form an isolation edge, construct the topology after fault isolation, and divide the sets of energized and de-energized nodes and the de-energized area.

[0058] Furthermore, in one embodiment of the present invention, the fault location is determined by a fault location and isolation algorithm in S2, including the acquisition and completion of three-phase current and voltage data uploaded by FTUs at each node of the distribution network, the construction of a fault location mathematical model and the determination of isolation edges based on the minimum fault set theory, the location of the fault based on the minimum fault set theory combined with an intelligent optimization algorithm, the selection of the isolation edge corresponding to the circuit breaker closest to both ends of the fault point, and the disconnection of the isolation edge to form a fault isolation topology.

[0059] Furthermore, in one embodiment of the present invention, the set of charged and de-charged nodes in S2 is calculated by breadth-first search (BFS) algorithm. Starting from the power node, all nodes reachable by closed edges are traversed to form the set of charged nodes, and the remaining nodes are the set of de-charged nodes. The de-charged region is divided by the de-charged region partitioning function to provide a clear object for subsequent power transfer operations.

[0060] Furthermore, in one embodiment of the present invention, the topology generation after fault isolation specifically includes:

[0061] 2.1 Fault Location and Isolation Control Algorithm

[0062] Given fault location Following commonly used fault isolation strategies in engineering, the circuit breakers closest to the fault point are disconnected to form a fault-isolated topology. .

[0063] The control algorithm flow is as follows:

[0064] Step 1. Collect three-phase current and voltage data uploaded by FTUs at each node of the distribution network, and use a self-correction method (based on interpolation of data from adjacent nodes) to fill in missing information to ensure data integrity;

[0065] Step 2. Construct a fault location mathematical model based on the minimum fault set theory, introduce the improved quantum ant colony algorithm (QACA), and accelerate the convergence speed of the fault segment and improve the location accuracy by dynamically adjusting the rotation angle of the quantum rotating gate (rotation angle adjustment formula);

[0066] Step 3. Output the fault location Determine the isolation side corresponding to the nearest circuit breaker at both ends of the fault point to ensure complete isolation between the fault area and the normal area.

[0067] 2.2 Fault Isolation Topology Construction

[0068] In a graph model, if the corresponding isolation edge is and Then for the initial topology ,make In

[0069]

[0070] Thus, the topology after fault isolation is obtained. , can be represented as

[0071]

[0072] in .

[0073] After obtaining fault isolation topology , That is, the isolated edge state vector ensures that the faulty area is no longer energized, thus guaranteeing the safety of subsequent power transfer operations.

[0074] 2.3 Definition of the set of charged / de-charged nodes

[0075] Define topology The set of charged nodes is calculated using the breadth-first search (BFS) algorithm, starting from the power node and traversing all nodes reachable through closed edges.

[0076] For ease of subsequent description, an arbitrary topology is defined. The set of charged nodes below is

[0077]

[0078] in, For topology From the closed edge of the power node The set of all nodes reachable from the starting point, S is the topology. The set of power supply nodes. Correspondingly, the set of power-off nodes is defined. For the complete set of nodes The middle does not belong to the set of charged nodes The node can be represented as

[0079]

[0080] In fault isolation topology In general, one or more power loss areas will be formed. By using the power loss area division function, the power loss area is divided, providing a clear target area for subsequent power transfer operations.

[0081] S3. Generate a primary power transfer topology: Based on the fault isolation topology, enumerate all disconnected tie lines that are related to the restoration of power supply to the power-loss area, screen tie lines with power transfer potential, perform a closure operation on each tie line that meets the conditions to generate candidate topologies, establish radial constraints and single power supply constraints to screen effective primary power transfer topologies and form a set.

[0082] Furthermore, in one embodiment of the present invention, the screening condition for the tie line with the potential for power transfer in S3 is: the tie line is in an open state, and the two end nodes belong to the energized area and the de-energized area respectively, ensuring that the tie line can connect the energized and de-energized areas to realize power transfer.

[0083] Furthermore, in one embodiment of the present invention, the generation of a primary transfer topology specifically includes:

[0084] 3.1 Screening of candidate contact lines

[0085] In fault isolation topology Based on this, enumerate all disconnected tie lines that are related to the restoration of power to the power-loss area, and screen tie lines that have the significance of being candidates for primary power transfer. The screening criteria are as follows:

[0086] For any connection line Where i and j represent the two endpoint nodes of the connection line e, if the following conditions are met...

[0087]

[0088] Or satisfy

[0089]

[0090] That is, in fault isolation topology The set of charged nodes that are in a disconnected state and whose one end is located in a fault isolation topology. The other end is located in the set of power-out nodes in the fault isolation topology. If the tie line is deemed suitable for a power transfer, it is considered a candidate for power transfer. Ensure the tie line can connect energized and de-energized areas, thus possessing power transfer potential.

[0091] 3.2 One-time transfer of candidate topology generation

[0092] Perform a closure operation on each tie line that meets the conditions to generate multiple candidate topologies for primary feed-in. Each topology can be represented as

[0093]

[0094] in ,remove The remaining amount and same.

[0095] For each candidate link, perform a closure operation, adjust the edge state variables, and generate a transfer candidate topology. Each candidate topology corresponds to a single link closure transfer scheme.

[0096] 3.3 Candidate Topology Constraint Verification Control Algorithm

[0097] Closing the tie line may introduce loops or multiple power sources, leading to instability in the distribution network. Therefore, a fast judgment module for radial constraints and single power supply constraints is established to verify candidate topologies. The control algorithm flow is as follows:

[0098] Step 1. Extract any charged connected component of the candidate topology. The connected components are extracted using the depth-first search (DFS) algorithm.

[0099] In a subgraph of a multi-source radial distribution network model, if the subgraph contains source nodes in the candidate topology, and all edges and nodes in the subgraph are closed and charged, then the subgraph is called a charged connected component of the candidate topology. Let any charged connected component in the candidate topology be...

[0100] ,

[0101] V q E represents the set of charged nodes contained in this connected component. qLet be the set of edges within the connected component that are in a closed conducting state.

[0102] Step 2. Radial Constraint Verification: A determination method based on the non-connectivity condition of power supply loops is adopted. The normal operation of the distribution network requires maintaining a radial structure (no loops). The constraint formula is as follows:

[0103]

[0104] in Represents the electrically connected components The number of middle edges, Represents the electrically connected components The number of nodes in the middle. The electrically connected components that satisfy this constraint are connected and acyclic.

[0105] Step 3. Single-source power supply constraint verification: Count the number of power nodes in the electrically connected component. The constraint formula is:

[0106]

[0107] This means that any charged connected component contains at most one power source node. This constraint ensures that any charged region is powered by only one power source, avoiding voltage and current conflicts caused by multiple power sources operating in parallel.

[0108] Step 4. For candidate topologies that satisfy the above constraints, add them to the set of primary transfer topologies. , recorded as

[0109]

[0110] Where n is the number of topologies that satisfy the constraints in a single transfer, and the corresponding operation "closing the connection" is recorded. For example, a single transfer topology. By fault isolation topology Closed tie line e1 is obtained, and "closed tie line e1" is recorded. This is a single topology transfer. By fault isolation topology A closed connection line e2 is obtained and recorded as "closed connection line e2", and so on. Connections that do not meet the constraints are removed to ensure the rationality and security of the topology.

[0111] S4: Recursively generate secondary transfer topology: Define the power supply scale of the power node, the set of power sources whose power supply scale expands relative to the fault isolation topology after transfer, locate the power supply topology with transfer potential, construct the power supply topology related domain and screen operable edges, the operable edges include closable edges and disconnectable edges, perform a two-step topology transformation of "combining and then separating", including: first closing the closable edges to build a temporary transfer channel, then disconnecting the disconnectable edges to eliminate the temporary loop, restore the topology to the loop-free running state, complete one topology transformation, generate candidate topologies, traverse all combinations of closed operable edges, and perform multi-class constraint checks to generate all transfer candidate topologies;

[0112] Furthermore, in one embodiment of the present invention, the relevant domain mentioned in S4 is a local subgraph consisting of the set of operable edges related to power supply boundary adjustment and its adjacent nodes.

[0113] Without altering the fundamental idea of ​​local search, the relevant domain can be constructed in different ways, such as based on the adjacent boundaries of the power supply topology, several layers of nodes near the power supply boundary, candidate tie lines, and their basic loop range. Essentially, it still belongs to a variation of the implementation of locally generating candidate operations around the power supply topology.

[0114] Furthermore, in one embodiment of the present invention, the power supply scale mentioned in S4 refers to the total load power of the power supply nodes, the total restored load capacity, or a weighted restoration benefit index that takes into account the importance of the load.

[0115] Furthermore, in one embodiment of the present invention, the candidate topology mentioned in S4 needs to satisfy the following three types of constraints: structural constraints, global charged set invariance constraints, and engineering operation constraints.

[0116] The structural constraints require that the energized connected components meet the conditions of radial and single power supply. The global energized set remains unchanged constraint requires that subsequent power transfers do not reduce the restored energized range. The engineering operation constraints require that the transformer capacity, line current and node voltage meet the preset thresholds.

[0117] Furthermore, in one embodiment of the present invention, the recursive generation of the secondary transfer topology specifically includes:

[0118] 4.1 Definition of Power Supply Scale and Expanded Power Supply Set

[0119] For any topology A certain power node The set of power supply nodes is Topology Power supply scale at power node s Defined as the sum of the load power of the power supply nodes, the formula is:

[0120]

[0121] in Let be the load power of node v.

[0122] Preferably, in engineering applications, it can also be Replace it with the total restored load capacity or a weighted recovery benefit index that takes into account the importance of the load.

[0123] For the Topology obtained after secondary transfer ( ), can be represented as Define a relative fault isolation topology The power supply scale has expanded to include the following:

[0124]

[0125] For any Take its topological value The charged connected component where the current is located is denoted as the electron-supplying topology of the power source. This set is used to locate power sources with further transfer potential, narrowing down the subsequent search range.

[0126] 4.2 Constructing the electron donor topological correlation domain

[0127] Based on the current power supply topology, focusing on the power supply boundary adjustment needs, relevant domains are constructed, and the specific operations are as follows:

[0128] Step 1. Define the relevant domain: The relevant domain only includes nodes and edges that are directly related to the power supply boundary adjustment, excluding topological elements in the entire network that are not related to the boundary adjustment, thus achieving local focus.

[0129] Step 2. Filter the core operable edges within the relevant domain: The relevant domain must contain at least the following two types of edges, and must strictly meet the filtering conditions:

[0130] (1) Closeable edge: Both ends of the node are energized and currently in the open state, and may form a temporary loop after closing; during screening, the state of the edge (open) and the energization status of both ends of the node (both energized) need to be checked one by one to ensure that the conditions for forming a temporary loop are met after closing, so as to lay the foundation for the subsequent transfer channel construction.

[0131] (2) Disconnectable edge: It is currently in a closed state, located inside the electron supply topology, and can be used to eliminate the loop introduced by the closing operation after being disconnected; during screening, the state (closed) and location (inside the electron supply topology) of the edge need to be verified, and it needs to be confirmed that it can accurately eliminate the loop generated by the closing operation after being disconnected, so as to avoid topological disorder.

[0132] Step 3. Determine the scope of the relevant domain: The two types of operable edges selected, along with the two adjacent nodes at both ends of each type of edge, together form a complete local subgraph of the relevant domain, clarifying the boundary and topological structure of the relevant domain, and providing a fixed range for subsequent edge set labeling and topological transformation.

[0133] To avoid blindly searching across the entire network, this patent constructs candidate operations only within the relevant domain of the power supply topology. The relevant domain is a local subgraph consisting of a set of operable edges related to power supply boundary adjustment and their adjacent nodes, which includes at least the following two types of edges: one type is a closable edge where both ends of the node are energized but currently in an open state, and may form a temporary loop after closure; the other type is a detachable edge that is currently in a closed state, located inside the power supply topology, and can be used to open or close the loop after forming a temporary loop.

[0134] Therefore, the relevant domain can be formally described, denoted as the first... The layer topology is For power nodes The set of power supply nodes is The corresponding electron-donating topology is denoted as:

[0135]

[0136] in,

[0137]

[0138]

[0139] In the formula, Indicates the first Layer topology The following is from the power node The set of power supply nodes, Indicates connection Internal node and in topology The set of edges that are in a closed state. Representing an edge It is in a closed conduction state in the current topology. Then the... Layers, for power nodes The electron-donating topological correlation domain can be represented as:

[0140]

[0141] in, Represents the electron topology related domain. Represents the set of related domain nodes. This represents the set of edges in the relevant domain. The set of edges in the relevant domain consists of both the set of closable edges and the set of detachable edges, i.e.:

[0142]

[0143] 4.3 Label the set of operable edges within the relevant domain

[0144] For a specific power source s at the k-th layer of the electron topology, within the relevant domain constructed in the above steps, the sets of closable edges and the sets of detachable edges are marked respectively, as follows:

[0145] Recorded in the first Layer, for power supply The set of closable edges within the relevant domain is The set of edges that can be broken is .in,

[0146] Closeable edges satisfy: in the topology Its state is disconnected, and both ends of the edge are located within the current energized region;

[0147] A disconnectable edge satisfies: in the topology Its state is closed, and when opened, it can be used to eliminate loops introduced by the closing operation.

[0148] Edge set verification: for the marked edge set and Verification is performed to ensure that the two edge sets have no intersection and no omissions, and that each edge meets the corresponding screening conditions, so as to prevent invalid or erroneous edges from entering the subsequent topology transformation process.

[0149] 4.4 Perform a "merge-then-divide" topology transformation to generate candidate topologies for transfer.

[0150] For the k-th layer and the power source s, select a set of closable edges. Any closable edge in and the set of detachable edges Any breakable edge in Perform a two-step topological transformation: "combine first, then separate." The specific steps are as follows:

[0151] Step 1. Close the closable edge: In accordance with the power operation specifications, switch the closable edge from the open state to the closed state. At this time, a temporary loop is formed with other edges in the relevant domain to build a temporary transfer channel, providing a basis for subsequent load transfer.

[0152] Step 2. Disconnect the disconnectable edge: After a temporary loop is formed, immediately switch the disconnectable edge from the closed state to the open state to eliminate the temporary loop, restore the power supply topology to a loop-free operating state, and avoid power supply safety hazards caused by loops.

[0153] Step 3. Generate candidate topologies: Through the above two-step operation of "combining then separating" (denoted as the topology transformation operator, i.e., completing one topology transformation, we obtain the first candidate topology). Secondary transfer of candidate topology

[0154]

[0155] in, Indicates closing the edge first. Disconnect the edge again The topology transformation operator. The "merge first, then split" approach is applicable to operational scenarios that allow short-term loop merging; for operational scenarios that do not allow loop merging, the logical transformation corresponding to the candidate topology can be mapped to an equivalent safe operation sequence that satisfies the scheduling procedure.

[0156] 4.5 Iteratively generate all candidate topologies

[0157] Traverse all combinations of the closable edge set and the detachable edge set, repeat the steps to generate all candidate topologies for power transfer corresponding to this layer and this power source, ensuring that no effective power transfer scheme is missed.

[0158] The candidate topology must satisfy the following three types of constraints:

[0159] (1) Structural constraints: For any charged connected component, radial operation constraints and single power supply constraints are satisfied;

[0160] (2) Global charged set invariance constraint:

[0161]

[0162] In other words, subsequent power transfers are only used to adjust power supply boundaries and power sharing, and cannot reduce the currently restored power supply range;

[0163] (3) Engineering operation constraints: The transformer capacity meets the requirements.

[0164]

[0165] The line current satisfies

[0166]

[0167] Node voltages satisfy

[0168]

[0169] in, For the distribution set, For the distribution transformer r in the topology The load capacity that it can bear. Its maximum capacity; Let l be the current in line l. This is the upper limit of the allowable current for the line. For topology Set of closed conducting edges; Let be the voltage at node i. , These represent the lower and upper limits of the node voltage, respectively. Topologies satisfying the above constraints are added to the [number]th [node]. Layer transfer to topology set And record the corresponding operation "close". ,disconnect Following the same recursive method, the above process can be repeated to generate three, four, or even the preset maximum number of iterations. Multi-level candidate topology set

[0170]

[0171] If a certain layer cannot generate a new candidate topology that satisfies the constraints, the recursive generation process is terminated.

[0172] S5. Topology Deduplication and Topology Library Construction: Based on the unique topology code, all candidate topologies are globally deduplicated. The topology after fault isolation is used as the root node, and a topology tree is constructed according to the transfer level. The above steps are repeated for different fault scenarios to form a topology library forest composed of multiple topology trees, and the storage, updating and retrieval of the library are realized.

[0173] Furthermore, in one embodiment of the present invention, the topology deduplication described in S5 is achieved through a global hash set, which compares the unique code of the newly generated topology with the code in the hash set to eliminate duplicate topologies.

[0174] Furthermore, in one embodiment of the present invention, the topology deduplication and topology library construction specifically include:

[0175] 5.1 Topology Deduplication Control Algorithm

[0176] Maintain a global hash set throughout the entire topology generation process. A unique code used to store the topology that has appeared. The deduplication control algorithm flow is as follows, ensuring topological uniqueness:

[0177] For any newly generated candidate topology ,like

[0178]

[0179] If it is a duplicate topology, it will be removed; otherwise, it will be...

[0180]

[0181] The topology is added to the hash set and retained as a valid candidate topology. Through the above mechanism, intra-layer deduplication and cross-layer deduplication can be achieved, avoiding repeated searches and repeated checks.

[0182] 5.2 Construction of Topology Library

[0183] For a given fault scenario, the topology after fault isolation is used. Using the root node, a hierarchical tree-building algorithm is used to construct a topology tree according to the progressive hierarchical relationship. The rules for constructing a topology tree are as follows:

[0184] (1) The nodes in the tree represent valid candidate topologies. Each node stores topology information (edge ​​state vector, unique code, power supply scale) and corresponding switching operations, which facilitates subsequent querying and reuse.

[0185] (2) The directed edges between nodes represent one or a group of switching operations corresponding to the evolution from the previous topology to the next topology. The edge weight is the power supply scale increase, which is used to prioritize the topology with high power supply efficiency when selecting subsequent schemes.

[0186] Steps 2 to 4 are repeated for different fault scenarios (different fault locations, different load distributions, and a total of N preset fault scenarios) to obtain a topology map forest composed of multiple topology trees. A map management and control module is constructed to realize the storage, update, and retrieval functions of the map library. The topology tree of the corresponding fault scenario can be quickly called for fault recovery scheme generation, candidate scheme screening, switch sequence optimization, online auxiliary decision-making, and subsequent knowledge accumulation and experience reuse.

[0187] S6: Multi-level topology library construction: Preset the maximum number of transfers K, repeat the recursive generation of secondary transfer topologies described in step S4, the topology deduplication and topology library construction process described in step S5, generate multiple transfer topologies, and terminate the recursion when the number of transfer topologies reaches K or a certain layer cannot generate a new candidate topology that meets the constraints, and finally obtain a multi-level topology library with fault isolation topology as the root node.

[0188] Furthermore, in one embodiment of the present invention, in S6: a rapid generation control algorithm for the topology library is constructed to improve the algorithm's operating efficiency and adaptability. A parallel computing mechanism is introduced to process the constraint verification of candidate topologies for transfer and the recursive generation of multiple transfer topologies in parallel, utilizing multi-threaded resources to reduce generation time. At the same time, an adaptive adjustment module for algorithm termination conditions is set up to dynamically adjust the preset maximum number of transfers K based on the topology generation efficiency and constraint satisfaction rate, and to preset the maximum generation time (set according to actual engineering needs, usually 1~5s), ensuring that the algorithm achieves a balance between efficiency and effectiveness, and ultimately realizing the rapid and accurate generation of the multiple transfer topology library for distribution network fault states.

[0189] Specifically, to improve the efficiency of topology selection for power transfer, the radial constraint verification, single-power supply constraint verification, and engineering operation constraint verification of different candidate topologies are divided into independent parallel tasks, which are processed synchronously by multi-threading or thread pools. Each thread is responsible for the constraint verification of one or more candidate topologies. In the generation of multiple power transfer topologies, due to the high independence between different topologies, different extended power sources, and different combinations of "closable edges - detachable edges", the recursive generation process of multiple power transfer topologies is divided into multi-level parallel tasks: the first level of parallelism targets different parent topologies in the same power transfer level; the second level of parallelism targets different extended power sources in the same parent topology; and the third level of parallelism targets different combinations of operable edges in the same related domain. Each parallel task performs a "combination-separation" topology transformation to generate the next level of candidate topologies, and performs structural constraints, global charged set invariant constraints, and engineering operation constraint verification on the generated candidate topologies.

[0190] Furthermore, when the topology generation efficiency of a certain transfer level is high, the constraint satisfaction rate is high, and the cumulative running time has not reached the preset maximum generation time, it indicates that continued recursion may still generate an effective transfer topology with engineering significance. In this case, the number of transfers can be limited to no more than K, which is the maximum allowable number of transfers in the project. max Within a certain range, K can be appropriately increased to expand the coverage of the topology library. When the topology generation efficiency of a certain transfer level decreases significantly, the constraint satisfaction rate is lower than the preset threshold, or the number of newly generated valid topologies in several consecutive layers is lower than the preset number, it indicates that the probability of continuing to generate valid topologies is low. At this time, K can be reduced or the recursion can be terminated in advance to avoid invalid searches and repeated verifications.

[0191] Furthermore, the termination conditions in S6 include at least one of the following: First, the current transfer level k reaches the current maximum effective transfer count K; second, the current level cannot generate a new candidate topology that satisfies the constraints; third, all candidate topologies generated by the current level are duplicate topologies after topology unique coding comparison; fourth, the cumulative running time reaches the preset maximum generation time T. max If any of the above conditions are met, the recursive generation process is stopped, and the generated multi-level topology set and its parent-child evolution relationship are output as a fault state for multiple transfer to the topology library.

[0192] A comparison in a 94-node distribution network shows that, under the same fault scenario, the traditional global random graph search generation method generates 198 topologies in 3.00 seconds, but many of these topologies lack practical power transfer significance. The method of this invention generates 10 topologies, all of which are valid and have practical power transfer significance, in 0.039 seconds, improving generation efficiency while ensuring topology validity. Furthermore, the traditional method only provides the final topology result without describing the power transfer hierarchy between topologies; the method of this invention constructs a topology tree according to the power transfer hierarchy and records the switching operation information between nodes, making it more suitable for selecting appropriate power transfer schemes based on actual needs and facilitating tasks such as switching sequence optimization.

[0193] Example 2

[0194] A topology map library generation device for the needs of multiple power transfers in distribution network faults includes: a topology coding module, a fault isolation topology generation module, a primary power transfer topology generation module, a secondary power transfer topology generation module, a topology map library construction module, and a multi-level topology map library construction module.

[0195] The topology coding module is used to abstract the distribution network into an undirected graph containing a set of nodes and a set of edges, divide the edge set and define edge state variables, arrange them in a fixed order to form an edge state vector, and construct a unique topology state code based on the edge state vector for hash storage, fast comparison and deduplication of the topology.

[0196] The fault isolation topology generation module is used to determine the fault location, disconnect the nearest circuit breakers at both ends of the fault point to form an isolation edge, construct the fault isolation topology, and divide the set of energized and de-energized nodes and the de-energized area.

[0197] The primary power transfer topology generation module is used to enumerate all disconnected tie lines related to power restoration in the power-loss area based on the fault isolation topology, screen tie lines with power transfer potential, perform a closure operation on each tie line that meets the conditions to generate candidate topologies, establish radial constraints and single power supply constraints to screen effective primary power transfer topologies and form a set.

[0198] The secondary power transfer topology generation module is used to define the power supply scale of power nodes, the power supply set whose power supply scale has expanded relative to the fault isolation topology after power transfer, locate power supply topologies with power transfer potential, construct power supply topology related domains and screen operable edges. The operable edges include closable edges and detachable edges. Perform a two-step topology transformation of "combining first and then separating", including: first closing the closable edges to build a temporary power transfer channel, then disconnecting the detachable edges to eliminate the temporary loop, restoring the topology to a loop-free operating state, completing one topology transformation, generating candidate topologies, traversing all combinations of closed operable edges, and verifying through multiple types of constraints to generate all power transfer candidate topologies.

[0199] The topology library construction module is used to globally deduplicate all candidate topologies based on the unique topology code, and construct a topology tree according to the transfer level with the fault-isolated topology as the root node. The above steps are repeated for different fault scenarios to form a topology library forest composed of multiple topology trees, and to realize the storage, updating and retrieval of the library.

[0200] The multi-level topology library construction module is used to preset the maximum number of transfers K, repeat the process of recursively generating secondary transfer topologies, deduplicating topologies, and constructing the topology library, and generate multiple transfer topologies. When the number of transfer topologies reaches K or a certain layer cannot generate a new candidate topology that satisfies the constraints, the recursion is terminated, and finally a multi-level topology library with fault isolation topology as the root node is obtained.

Claims

1. A method for generating a topology library to meet the needs of multiple power transfers during distribution network faults, characterized in that, Includes the following steps: S1: Construct a distribution network graph model and implement a unique topology code: Abstract the distribution network into an undirected graph containing a set of nodes and a set of edges, divide the edge set and define edge state variables, arrange them in a fixed order to form an edge state vector, and construct a unique topology code based on the edge state vector for hash storage, fast comparison and deduplication of the topology; S2: Generate the topology after fault isolation: Determine the fault location, disconnect the nearest circuit breakers at both ends of the fault point to form an isolation edge, construct the topology after fault isolation, and divide the set of energized and de-energized nodes and the de-energized area; S3: Generate a primary power transfer topology: Based on the fault isolation topology, enumerate all disconnected tie lines that are related to the restoration of power supply to the power-loss area, screen tie lines with power transfer potential, perform a closure operation on each tie line that meets the conditions to generate candidate topologies, establish radial constraints and single power supply constraints to screen effective primary power transfer topologies and form a set. S4: Recursively generate secondary transfer topology: Define the power supply scale of the power node, the set of power sources whose power supply scale expands relative to the fault isolation topology after transfer, locate the power supply topology with transfer potential, construct the power supply topology related domain and screen operable edges, the operable edges include closable edges and detachable edges, perform a two-step topology transformation of "combining and then separating", including: first closing the closable edges to build a temporary transfer channel, then disconnecting the detachable edges to eliminate the temporary loop, restore the topology to the loop-free running state, complete one topology transformation, generate candidate topologies, traverse all combinations of closed operable edges, and perform multi-constraint checks to generate all transfer candidate topologies; S5: Topology Deduplication and Topology Library Construction: Based on the unique topology code, all candidate topologies are globally deduplicated. The topology after fault isolation is used as the root node, and a topology tree is constructed according to the transfer level. The steps of S2-S4 above are repeated for different fault scenarios to form a topology library forest composed of multiple topology trees, and the storage, updating and retrieval of the library are realized. S6: Multi-level topology library construction: Preset the maximum number of transfers K, repeat the recursive generation of secondary transfer topologies described in step S4, the topology deduplication and topology library construction process described in step S5, generate multiple transfer topologies, and terminate the recursion when the number of transfer topologies reaches K or a certain layer cannot generate a new candidate topology that meets the constraints, and finally obtain a multi-level topology library with fault isolation topology as the root node.

2. The method for generating a topology library for multiple power transfer needs in a distribution network based on claim 1, characterized in that, In S1, the construction of the distribution network diagram model includes abstracting the distribution network into an undirected graph containing power supply nodes and load nodes; the edge set is divided into three categories according to equipment attributes: ordinary lines, lines with sectionalizing switches, and tie lines; the edge set is defined by binary closed and open states to define edge state variables; all edge state variables are combined in sequence to form an edge state vector; different edge state vectors correspond to different topological states, forming a unique topological state code.

3. The method for generating a topology library for multiple power transfer needs in a distribution network based on claim 1, characterized in that, In S2, the sets of charged and de-charged nodes are calculated using a breadth-first search algorithm. Starting from the power node, all nodes reachable by closed edges are traversed to form the set of charged nodes, and the remaining nodes form the set of de-charged nodes. The de-charged region is divided using a de-charged region partitioning function to provide a clear object for subsequent power transfer operations.

4. The method for generating a topology library for multiple power transfer needs in a distribution network based on claim 1, characterized in that, In S2, fault location and isolation algorithms are used to determine the fault location, including the acquisition and completion of three-phase current and voltage data uploaded by FTUs at each node of the distribution network, the construction of a fault location mathematical model based on the minimum fault set theory and the determination of isolation edges, the location of the fault based on the minimum fault set theory combined with intelligent optimization algorithms, the selection of the isolation edge corresponding to the circuit breaker at both ends of the fault point, and the disconnection of the isolation edge to form a fault isolation topology.

5. The method for generating a topology library for multiple power transfer needs in a distribution network based on claim 1, characterized in that, In S3, the screening criteria for the tie line with the potential for power transfer are: the tie line is in an open state, and the two ends of the tie line belong to the energized area and the de-energized area respectively, so as to ensure that the tie line can connect the energized area and the de-energized area to realize power transfer.

6. The method for generating a topology library for multiple power transfer needs in a distribution network based on claim 1, characterized in that, In S4, the power supply scale refers to the total load power of the power supply nodes, the total restored load capacity, or a weighted restoration benefit index that takes into account the importance of the load.

7. The method for generating a topology library for multiple power transfer needs in a distribution network based on claim 1, characterized in that, In S4, the candidate topology must satisfy the following three types of constraints: structural constraints, global charged set invariance constraints, and engineering operation constraints.

8. The method for generating a topology library for multiple power transfer needs in a distribution network based on claim 1, characterized in that, In S4, the relevant domain is a local subgraph consisting of the set of operable edges related to the power supply boundary adjustment and its adjacent nodes.

9. The method for generating a topology library for multiple power transfer needs in a distribution network based on claim 1, characterized in that, In S5, the deduplication of topology is achieved through a global hash set. The unique code of the newly generated topology is compared with the code in the hash set to eliminate duplicate topologies.

10. A topology map generation device for multiple power transfer needs in distribution network faults, characterized in that, include: Topology coding module, fault isolation topology generation module, primary transfer topology generation module, secondary transfer topology generation module, topology map library construction module, and multi-level topology map library construction module; The topology coding module is used to construct a distribution network graph model and implement a unique topology coding: the distribution network is abstracted as an undirected graph containing a set of nodes and a set of edges, the edge set is divided and edge state variables are defined, and the edge state vectors are arranged in a fixed order to form an edge state vector. A unique topology state coding is constructed based on the edge state vectors for hash storage, fast comparison and deduplication of the topology. The fault isolation topology generation module is used to generate the fault isolation topology: determine the fault location, disconnect the nearest circuit breakers at both ends of the fault point to form an isolation edge, construct the fault isolation topology, and divide the set of energized and de-energized nodes and the de-energized area. The primary power transfer topology generation module is used to generate a primary power transfer topology: based on the fault isolation topology, it enumerates all disconnected tie lines that are related to the restoration of power supply to the power-loss area, filters tie lines with power transfer potential, performs a closure operation on each tie line that meets the conditions to generate a candidate topology, establishes radial constraints and single power supply constraints to filter effective primary power transfer topologies and form a set. The secondary power transfer topology generation module is used to recursively generate secondary power transfer topologies: defining the power supply scale of power nodes, the set of power sources whose power supply scale expands relative to the fault isolation topology after power transfer, locating power supply topologies with power transfer potential, constructing power supply topology related domains and screening operable edges, the operable edges including closable edges and detachable edges, performing a two-step topology transformation of "combining and then separating", including: first closing closable edges to build temporary power transfer channels, then disconnecting detachable edges to eliminate temporary loops, restoring the topology to a loop-free operating state, completing one topology transformation, generating candidate topologies, traversing all combinations of closed operable edges, and verifying through multiple types of constraints to generate all power transfer candidate topologies; The topology library construction module is used for topology deduplication and topology library construction: based on the unique topology code, all candidate topologies are globally deduplicated. Taking the fault-isolated topology as the root node, a topology tree is constructed according to the transfer level. For different fault scenarios, the steps of generating the fault-isolated topology, generating the first transfer topology, and recursively generating the second transfer topology are repeatedly executed to form a topology library forest composed of multiple topology trees, and the storage, updating and retrieval of the library are realized. The multi-level topology library construction module is used to preset the maximum number of transfers K, repeatedly generate secondary transfer topologies, deduplicate topologies, and construct the topology library, generating multiple transfer topologies. When the number of transfer topologies reaches K or a certain layer cannot generate a new candidate topology that meets the constraints, the recursion is terminated, and finally a multi-level topology library with fault isolation topology as the root node is obtained.

Images