Power distribution network active supply guarantee partition method and system facing multi-type subject support

Abstract

This invention relates to the field of distribution network control technology, and provides a method and system for proactive power supply protection zoning in distribution networks supported by multiple types of entities. During normal operation of the distribution network, the method acquires and updates the electrical parameters and connection relationships of nodes, branches, and switches; acquires source-grid-load-storage data; constructs a resource state mapping reflecting the real-time operating boundaries of the source-grid-load-storage system; characterizes branch availability / connectivity reliability; and establishes an proactive power supply protection zoning model with the objective function of maximizing the power supply level of critical loads and minimizing the cost of zoning structure adjustments, while introducing dynamic frequency security constraints. This invention generates and updates power supply protection zoning plans online during normal distribution network operation, and calls the latest plan to form power supply protection zones after a main grid fault is triggered. The introduction of linearized dynamic frequency security constraints improves the executability and operational safety of zoning power supply protection and recovery plans.

Description

Technical Field

[0001] This invention relates to the field of distribution network control technology, and in particular to a method and system for active power supply protection zoning of distribution networks oriented towards multiple types of supporting entities. Background Technology

[0002] The information disclosed in this background section is intended only to enhance understanding of the overall background of the invention and is not necessarily to be construed as an admission or in any way implying that such information constitutes prior art known to those skilled in the art.

[0003] In recent years, frequent extreme natural disasters such as typhoons, ice storms, and earthquakes have easily led to power outages in the main power grid, causing widespread blackouts and seriously threatening the continuous power supply security of critical loads such as hospitals, communication base stations, and government emergency centers. Meanwhile, with the widespread integration of distributed power sources, energy storage devices, and controllable loads into distribution networks, modern distribution networks are gradually transforming into proactive distribution networks with coordinated regulation capabilities across sources, grids, loads, and storage. These new entities give distribution networks the potential to independently guarantee power supply to critical loads through islanded operation modes after the main grid loses support. Against this backdrop, proactive power supply zoning has become a key technical means to enhance the resilience of distribution networks.

[0004] In the event of extreme disasters leading to power supply constraints, multiple network outages, and damage to localized equipment, ensuring emergency power supply after a disaster typically requires a regionalized power supply approach. This involves rapidly establishing several self-sufficient regional power zones based on distributed power sources, energy storage devices, and controllable loads within the network. Priority is given to maintaining continuous power supply to critical loads within these zones. As the network and resource status gradually stabilize, further coordination and scheduling between these zones can be used to provide phased support to neighboring areas using the surplus regulatory capacity of each zone, gradually expanding the available power supply range.

[0005] However, under the aforementioned extreme scenarios, the existing distribution network's zoning and operation for power supply protection still face numerous difficulties. These mainly manifest in the following ways: extreme disasters can damage communication links, cause terminal outages, and result in inconsistent equipment statuses, easily leading to situations such as missing remote signaling for switches and conflicts in line / switch statuses, making it difficult to accurately determine the actual topology connections. Under these conditions, existing zoning and reconfiguration decisions often rely on deterministic topology information, making them difficult to apply directly. In engineering practice, this typically requires manual inspection and verification or covering potential fault states across multiple scenarios, resulting in large computational scales and insufficient response time. Furthermore, it is difficult to quickly screen and prioritize candidate topologies based on limited real-time measurements. Existing power supply zoning and scheduling strategies focus primarily on steady-state indicators such as power flow, voltage, and line capacity, while neglecting the low inertia and weak damping characteristics of inverters within islanded zones and their frequency support capabilities. Dynamic safety indicators such as frequency deviation and rate of change of frequency (ROCOF) are not incorporated into decision constraints, leading to frequency overruns, protection malfunctions, or even zone instability during phased load recovery, source-load fluctuations, or zone switching. This restricts the ability to maintain continuous power supply and provide external support in extreme disaster scenarios. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the purpose of this invention is to provide a method and system for proactive power supply protection zoning in distribution networks that supports multiple types of entities. This invention generates and updates power supply protection zoning plans online during the normal operation of the distribution network, and calls the latest plan to form power supply protection zones after a main grid fault is triggered. It introduces linearized dynamic frequency security constraints, thereby improving the executability and operational safety of zoning power supply protection and recovery plans.

[0007] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0008] A method for proactive power supply protection zoning in distribution networks supported by multiple types of entities includes the following steps:

[0009] During the normal operation of the distribution network, the electrical parameters and connection relationships of nodes, branches and switches are acquired and updated, source-grid-load-storage data are acquired, and a resource status mapping reflecting the real-time operation boundary of source-grid-load-storage is constructed.

[0010] Construct a candidate branch set, and based on the acquired information, divide the candidate branch set into a set of confirmed closures, a set of confirmed disconnections, and a set of states to be verified.

[0011] For the set of states to be verified, the voltage measurement sequence of key nodes is obtained within the rolling observation window, preprocessed, and features reflecting the electrical correlation of nodes are extracted to characterize the availability / connectivity confidence of branches.

[0012] Based on the characterization results of branch availability / connectivity reliability, the topology graph formed by candidate branches is pruned and its connectivity is repaired.

[0013] Based on the resource state mapping and the processed topology, an active supply guarantee partitioning model is established. The objective function is to maximize the supply level of critical loads and minimize the cost of partitioning structure adjustment. A linear dynamic frequency security constraint is introduced into the active supply guarantee partitioning model. The frequency response process of the isolated partitions is discretized and frequency deviation and frequency change rate threshold constraints are applied.

[0014] The active supply guarantee zoning model is optimized and solved to obtain the optimal supply guarantee zoning target structure at the current moment, and the target switch state vector is obtained. The boundary differential adjustment amount is generated based on the difference between the target switch state vector and the current operating state vector.

[0015] In response to the triggering event, the optimal supply guarantee zone target structure is invoked, and the formation of the optimal supply guarantee zone target structure is performed based on the boundary differential adjustment amount and the preset switching operation sequence.

[0016] As an alternative implementation, the process of acquiring and updating the electrical parameters and connection relationships of nodes, branches and switches includes: acquiring each node, as well as the node type, load requirement and importance level; acquiring each candidate branch, the endpoints, impedance parameters, switch positions and initial switching states of each branch; and constructing a node set and a candidate branch set based on the acquired information of nodes, branches and switches.

[0017] As an alternative implementation method, the process of acquiring source-grid-load-storage data and constructing a resource state mapping that reflects the real-time operating boundaries of source-grid-load-storage includes: acquiring the output limit of distributed power sources, the charging and discharging power and energy boundaries, efficiency, initial state of charge and real-time state of charge of energy storage devices, and the adjustable capacity of flexible load / demand response resources; establishing and dynamically updating a load importance classification list based on user-side demand; and acquiring voltage measurement sequences, remote signaling / alarms, inspection and on-site confirmation information.

[0018] As an alternative implementation method, the process of dividing the candidate branch set into a determined closed set, a determined open set, and a status-to-verify set based on the acquired information includes: dividing the candidate branches into a determined closed set according to the basic information of the nodes, branches, and switches of the active distribution network to be restored, the switch historical records, communication status, and on-site inspection / remote signaling information. 、 Determine the disconnected set and the set whose state needs to be verified. Determine that branches in the closed set remain closed during the recovery process. Determine that branches in the disconnected set remain disconnected. The three sets are mutually exclusive and constitute the entire candidate branch set.

[0019] As an alternative implementation, the process of extracting features reflecting the electrical correlation of nodes and characterizing the availability / connectivity reliability of branches includes: targeting candidate branches in the set of states to be verified. e Based on nodes i With nodes j Statistical correlation indices are calculated from the voltage characteristic sequences, and branch scores are generated. s e Specifically:

[0020] ;

[0021] in, X' i and X' j They represent respectively to X i ( t )and X j ( t The vector after mean removal and normalization. r e branch road e The correlation coefficient of the voltage sequences at both ends indicates that the closer the correlation value is to 1, the stronger the consistency of the voltage changes at both ends.

[0022] ;

[0023] in, I e As a proxy indicator of mutual information, it is used to enhance the characterization of nonlinear correlations;

[0024] ;

[0025] in, T meas The number of sampling points in the measurement window. T eff The number of valid samples, T min The minimum sample threshold, g Con is the adjustment coefficient. f e The branch score is used to reflect the impact of measurement quality on the reliability of the score.

[0026] ;

[0027] in, κ Branch scoring for smoothing parameters s e The larger the value, the more likely the branch is to be a true connecting branch.

[0028] As an alternative implementation, the process of pruning and connectivity repairing the topology graph formed by candidate branches based on the characterization results of branch availability / connectivity reliability includes: [the following steps are taken based on branch scores]. s e The topology graph is pruned and its connectivity is restored. Under the premise that branches in the closed set must be selected and branches in the open set must be discarded, several adjacent uncertain branches with the highest branch scores are retained for each node to form a sparse candidate edge set. If pruning causes the candidate graph to be disconnected, edges that connect different connected components and whose branch scores are higher than a set value are added first to restore connectivity.

[0029] As a further defined implementation, pruning employs a node-by-node retention of Top- K Two strategies are used: scoring threshold screening and other methods.

[0030] Top-retained node by node K The strategy is: for each node i After eliminating the definite disconnected set and forcibly retaining the definite closed set, the remaining adjacent uncertain branches are scored according to branch. s e Sort by highest to lowest and retain the top results. k keep strip;

[0031] The scoring threshold filtering strategy is: only retain s e ≥ s th Uncertain branches;

[0032] in k keep and s th The settings are determined based on network size, measurement coverage, and computing resources.

[0033] As a further defined implementation method, the connectivity repair process is as follows: the pruned candidate graph is divided into connected components; when there are multiple connected components, the branch that connects different connected components and has the highest score is selected and added to the candidate edge set in sequence until the candidate graph is connected or reaches the preset edge replenishment limit; if global connectivity cannot be achieved within the edge replenishment limit, a locally connected sparse candidate graph is output, and multiple isolated root nodes are allowed to be formed to supply power separately during the subsequent active power supply partitioning model establishment process.

[0034] As an alternative implementation method, the process of establishing a proactive load balancing zoning model includes: considering the uncertainty of the topology scenario, establishing a proactive load balancing zoning model with the objective of maximizing the critical load balancing level and minimizing the cost of zoning structure adjustment, the objective function being expressed as:

[0035] ;

[0036] in, For the set of topological scenes { ω k}, k For scene indexing, π k For the corresponding topology scene ω k The probability of the scenario; V For a set of nodes, i For node indexing; T For the number of discrete scheduling periods, t For time period index, t The length of a single time period; ω i For nodes i The load importance weighting coefficient; P serv i,k ( t ) for scene k Next period t node i The active power of the restored power supply; sw The switching action cost weighting coefficient, E sw For a set of operable branches / switches, e For branch / switch index, u e For branch circuits / switches e Action cost (by |) z e - z 0,e | linearized form representation). R For the set of root nodes of the partition, q For the root node index, root Enable a penalty coefficient for the root node. y q ∈{0,1} indicates whether the partition / root node is enabled;

[0037] The proactive power supply zoning model is used for joint decision-making on zoning boundaries, switch target states, distributed power generation and energy storage output / energy planning, and load recovery.

[0038] As an alternative implementation, the active power supply zoning model includes energy storage state evolution and boundary constraints, linearized power flow constraints, and linearized constraints for switching / boundary difference.

[0039] As an alternative implementation, the dynamic frequency security constraint includes setting upper and lower limit thresholds for frequency deviation and frequency change rate to impose a feasible domain constraint on the island net power deficit.

[0040] A distribution network active power supply protection zoning system oriented towards supporting multiple types of entities includes:

[0041] The data acquisition and resource status mapping unit is configured to acquire and update the electrical parameters and connection relationships of nodes, branches and switches during the normal operation phase of the distribution network, acquire source-grid-load-storage data, and construct a resource status mapping that reflects the real-time operating boundary of source-grid-load-storage.

[0042] The branch status assessment unit is configured to construct a candidate branch set. Based on the acquired information, the candidate branch set is divided into a confirmed closed set, a confirmed open set, and a status to be verified set. For the status to be verified set, the voltage measurement sequence of key nodes is acquired within the rolling observation window, preprocessed, and features reflecting the electrical correlation of nodes are extracted to characterize the branch availability / connectivity confidence.

[0043] The pruning and connectivity repair unit is configured to perform pruning and connectivity repair on the topology graph formed by candidate branches based on the characterization results of branch availability / connectivity credibility.

[0044] The supply guarantee partitioning and recovery plan generation unit is configured to establish an active supply guarantee partitioning model based on the resource status mapping and the processed topology map. The objective function is to maximize the supply level of critical loads and minimize the cost of partitioning structure adjustment. Linear dynamic frequency security constraints are introduced into the active supply guarantee partitioning model, the frequency response process of the isolated partitions is discretized and frequency deviation and frequency change rate threshold constraints are applied.

[0045] The target structure and switching strategy output unit is configured to perform optimization solution of the active supply guarantee partition model, obtain the optimal supply guarantee partition target structure at the current time, obtain the target switching state vector, and generate boundary differential adjustment amount based on the difference between the target switching state vector and the current running state vector.

[0046] The trigger response and contingency plan execution unit is configured to respond to a trigger event, invoke the optimal supply guarantee zone target structure, and execute the formation of the optimal supply guarantee zone target structure according to the boundary differential adjustment amount and the preset switch operation sequence.

[0047] One or more technical solutions provided in the embodiments of the present invention have at least the following technical effects or advantages:

[0048] In cases where communication failures or incomplete switch states lead to topology uncertainty, this invention utilizes readily available node voltage measurements to construct branch confidence scores and generate candidate topologies, reducing reliance on manual verification and improving the ability to make rapid decisions after a fault.

[0049] This invention utilizes candidate graph pruning and Top- K Candidate radial topology generation can significantly compress the scene size while ensuring feasibility, avoid full enumeration of uncertain branches, and improve the efficiency of robust optimization solutions.

[0050] This invention quantifies topological uncertainty through scenario probability, and achieves a "probability-weighted / risk-controllable" solution output at the recovery decision level, taking into account both power restoration effectiveness and execution reliability.

[0051] This invention combines optimized power supply zones, switch switching, and coordinated scheduling of power supply, energy storage, and load to improve the power supply level and continuous power supply capability of critical loads while meeting voltage, power flow, and energy storage constraints.

[0052] This invention introduces dynamic frequency safety constraints, which couple the frequency deviation and rate of change limits of islanded zones with load recovery timing and energy storage regulation capabilities to reduce the risk of zone instability and protection actions.

[0053] This invention employs a rolling update mechanism, which can correct scores, scenarios, and scheduling schemes in real time as new measurement and topology confirmation information is added. Attached Figure Description

[0054] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0055] Figure 1 A flowchart illustrating an exemplary embodiment of the present invention for a distribution network active power supply protection zoning method for multiple types of novel subjects;

[0056] Figure 2 A schematic diagram of an active power supply protection zoning system for distribution networks oriented towards multiple types of novel subjects, provided as an exemplary embodiment of the present invention;

[0057] Figure 3 A schematic diagram of the candidate edge set and pruning provided for an exemplary embodiment of the present invention;

[0058] Figure 4 A schematic diagram of a supply guarantee partition (partition + root node) provided for an exemplary embodiment of the present invention;

[0059] Figure 5 A schematic diagram of islanded operation frequency security constraints provided for an exemplary embodiment of the present invention.

[0060] Figure 6 This is a schematic diagram comparing the active power supply protection zoning results of the distribution network without and with the introduction of dynamic frequency security constraints, provided as an exemplary embodiment of the present invention. (a) shows the active power supply protection zoning results of the distribution network without the introduction of dynamic frequency security constraints, and (b) shows the active power supply protection zoning results of the distribution network with the introduction of dynamic frequency security constraints. Detailed Implementation

[0061] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0062] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this disclosure. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms “comprising” and / or “including” are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0063] Example 1

[0064] A proactive power supply protection zoning method for distribution networks with multiple stakeholders. This method addresses scenarios where power supply capacity is insufficient due to extreme disasters or upstream grid failures. During normal distribution network operation, it generates and updates power supply protection zones and recovery plans online on a rolling basis. When triggered events such as grid fault warnings, grid power outages, or upstream power source disconnection are detected, the latest plan is directly invoked to form power supply protection zones. Zone formation is executed based on boundary differential adjustment amounts and preset switching operation sequences, prioritizing power supply to critical loads. Dynamic frequency safety constraints are embedded in a linearized form within the plan generation and rolling correction process to suppress power surge risks during load recovery, thereby improving the executability and operational safety of power supply protection zones and recovery plans. More specifically, such as... Figure 1 As shown, the process includes the following:

[0065] Step S101: Source-Grid-Load-Storage Data Acquisition and Resource Status Mapping Update. Acquire basic information about the active distribution network to be guaranteed and form a structured data interface, including at least:

[0066] Node information: Node set v Node type (critical load / general load / distributed generation (DG) / energy storage (ESS) / controllable load, etc.), load demand and importance level;

[0067] Branch and switch information: Candidate branch set EEach branch endpoint, impedance parameter, switch position, and initial switching state;

[0068] Resource boundaries: DG output limit, ESS charging and discharging power and energy boundary, efficiency, initial SOC;

[0069] Available measurement and communication functions include: voltage measurement sequences, remote signaling / alarms, inspection and on-site confirmation information, etc.

[0070] Obtain basic information on nodes, branches, and switches of the active distribution network to be restored, and construct a node set. V With candidate branch set E By combining switch history records, communication status, and on-site inspection / remote signaling information, candidate branches are divided into a definite closed set. E fix1 Determine the disconnect set E fix0 and the set of statuses to be verified E unc Branches in the closed set should remain closed during the recovery process; branches in the open set should remain open; and the states of branches in the unverified state set need to be determined through a combination of measurement inference and optimization decision-making.

[0071] In one implementation, the basic information of the nodes, branches, and switches may include: node type (load node, distributed generation node, energy storage node, tie switch node, etc.), load level and importance weight, upper limit of available output of distributed generation, energy storage capacity and initial SOC, and branch electrical parameters. R e , X e This includes information such as the line's thermal stability capacity and the allowed number of switch operations / operation costs. By structuring and organizing this basic information, a unified data interface can be provided for subsequent candidate topology generation and recovery plan optimization.

[0072] When grouping candidate branches, information such as switch remote signaling, protection action information, fault indicators, historical operation records, on-site inspection results, and disaster impact assessments can be comprehensively utilized. Preferably, branches that have been confirmed to be damaged, isolated, or prohibited from closure are included. E fix0 Branches that have been confirmed as usable and must be closed to maintain trunk connectivity will be included. E fix1 Other branches with unclear status or questionable availability are included. E unc .

[0073] when E fix1When the large number of branches in a network could potentially create a ring network structure, some branches can be rerouted based on their confidence level, operational feasibility, or impact on the power supply to critical loads. E fix1 Downgraded to E unc This ensures that a feasible solution exists for subsequent radial candidate topology generation;

[0074] Step S102: Continuously acquire the per-unit voltage measurement sequence of key nodes within the rolling observation window. V pu ( i , t The measurement sequence undergoes preprocessing such as time alignment, noise reduction, anomaly removal, and missing data completion to form a voltage feature sequence. X i ( t Optionally, the voltage series can be detrended, differentially filtered, or bandpass filtered to reduce the impact of common voltage source fluctuations on correlation calculations.

[0075] The voltage measurement sequence can be obtained from distribution automation terminals, micro PMUs, smart meters, or edge computing devices. When communication conditions are limited, key nodes (such as partition root nodes, tie points, and important load nodes) can be selected first for measurement and collection to reduce data requirements while maintaining the ability to identify topological relationships.

[0076] In the preprocessing stage, median filtering, Kalman filtering, or moving average methods can be used to suppress measurement noise; for missing data, interpolation or collaborative completion based on adjacent nodes can be used; for outliers, the 3σ criterion, outlier detection based on sliding windows, or removal rules based on physical boundaries can be used for processing.

[0077] To further mitigate the impact of common voltage source fluctuations on correlation calculations, voltage sequences can be detrended, differentially processed, or bandpass filtered to make the characteristic sequences better reflect the propagation characteristics of local disturbances, thereby improving the robustness of identifying real electrical connection relationships.

[0078] Step S103: For the set of states to be verified E unc Candidate branches in e = ( i , j ), based on nodes i With nodes j Statistical correlation indices are calculated from the voltage characteristic sequences, and branch scores are generated. s e The following is a feasible method for constructing the score:

[0079] ;

[0080] in, X' i and X' j They represent respectively to X i ( t )and X j ( t The vector after mean removal and normalization. r e branch road e The correlation coefficient of the voltage sequences at both ends. When the correlation value is close to 1, it indicates that the voltage changes at both ends have a stronger consistency.

[0081] ;

[0082] in, I e It can be used as a proxy indicator for mutual information to enhance the characterization of nonlinear correlations.

[0083] ;

[0084] in, T meas The number of sampling points in the measurement window. T eff The number of valid samples, T min The minimum sample threshold, g This is the adjustment coefficient. f e It is used to reflect the impact of measurement quality on the reliability of the score.

[0085] ;

[0086] in, κ For smoothing parameters, used to avoid I e The score is overly sensitive when the score is too small. Branch score s e The larger the value, the more likely the branch is to be a true connecting branch. Optionally, multiple time delay scenarios can be considered, and the maximum / weighted average score under different time delays can be taken.

[0087] To account for the time lag in measurements, a finite lag set can be used. The correlation index is calculated on each of the ∈{0,…,L}, and the maximum value or weighted average is taken as the final score of the branch. Correspondingly, the weight coefficient can be set in descending order of lag to reflect the timeliness of disturbance propagation.

[0088] When the number of valid samplesT eff When the value is small or the measurement quality is poor, a conservative strategy can be set: for example, setting the value of con... f e Limit the score to the [0,1] range and set an upper limit on the score, or T eff < T min The branch score is set to 0 directly to reduce the risk of misjudgment caused by false high correlation;

[0089] Step S104: Scoring based on branch paths s e Pruning and connectivity restoration are performed on the candidate graph. For example... Figure 3 As shown, it is possible to satisfy a definite closed set. E fix1 Middle branch path must be selected, and the set of paths to be disconnected must be determined. E fix0 Under the premise that the middle branch must be discarded, several adjacent uncertain branches with the highest scores are retained for each node to form a sparse candidate edge set; if pruning causes the candidate graph to be disconnected, edges with higher scores that connect different connected components can be added first to repair connectivity, thereby reducing the scale while ensuring the existence of feasible candidate topologies.

[0090] In one implementation, pruning can be done by retaining the "Top-" node by node. K Two strategies: "or scoring threshold screening"; for each node i In picking E fix0 And forcibly retain E fix1 Based on this, the remaining adjacent uncertain branches are divided according to their scores. s e Sort by highest to lowest and retain the top results. k keep Article; or only retain s e ≥ s th Uncertain branches, among which k keep and s th It can be configured according to network size, measurement coverage, and computing resources.

[0091] Connectivity restoration can be achieved through the following process: The pruned candidate graph is divided into connected components; when multiple connected components exist, the branch with the highest score connecting different connected components is sequentially added to the candidate edge set until the candidate graph is connected or reaches a preset edge replenishment limit. If global connectivity cannot be achieved within the edge replenishment limit, a sparse candidate graph with "local connectivity" can be output, allowing multiple isolated root nodes to be formed and powered separately in subsequent partitioning models.

[0092] By pruning and repairing, the topology of candidate radial structures can be significantly compressed while ensuring that they can be generated. E unc Scale, thereby reducing Top- K The complexity of candidate topology generation and subsequent robust optimization solution;

[0093] Step S105: Generate Top- on the sparse candidate edge set K A set of candidate radial topologies. Candidate radial topologies can be constructed using a constrained minimum spanning tree approach, with edge weights as follows:

[0094] ;

[0095] in, s e ∈[0,1] is a branch e The rating, w e The edge weights are used in candidate topology generation. The higher the score, the smaller the edge weight, so branches with higher scores are preferred when generating candidate radial topologies.

[0096] Sort and select uncertain branches, and force inclusion. E fix1 ,exclude E fix0 To obtain diverse candidate topologies, a penalty term can be introduced into the selected edges after each generation, or a multi-starting point / multi-weight perturbation method can be used to obtain... K A differentiated candidate topology.

[0097] Top- K Candidate radial topologies can be generated using a constrained minimum spanning tree (MST) or an equivalent mixed integer programming approach: closed branches are designated as mandatory edges, and disconnected branches are designated as prohibited edges; loops are avoided during generation to ensure radiality. For candidate graphs containing tie switches, a radial topology rooted at the power / energy storage node is preferred.

[0098] To obtain differentiated candidate topologies, a penalty is introduced on the selected branches and the edge weights are updated after each candidate topology is generated:

[0099] ;

[0100] in, r The iteration number generated for the candidate topology. wr e For the first r Edge weights during secondary generation, cnt r e up to the r Secondary generation branch e The cumulative number of times a branch is selected; by iteratively increasing the edge weights of high-frequency branches, subsequent candidate topologies can be differentiated in their selection of key branches. Alternatively, small random perturbations can be applied to the edge weights to influence the selection of key branches in subsequent topologies; or a K-best spanning tree algorithm can be used to generate candidate topologies in ascending order of cost, in order to achieve a balance within a limited number of branches. K It balances optimality and diversity.

[0101] In practical applications, K The value of can be determined based on the network size and computing resources, for example... K A value of 5 to 20 can be used to strike a balance between scene coverage and solution efficiency.

[0102] Step S106: Construct a topology scene based on candidate topologies ω k And calculate the scene probability. π k Candidate topology costs can be defined:

[0103] ;

[0104] The scene probability is obtained by using softmax mapping:

[0105] ;

[0106] in, c min Let t be the minimum cost of the candidate topology, and t be the temperature parameter. The scenario probability can be further used to calculate the existence probability of uncertain branches:

[0107] ;

[0108] Where 1{·} is an indicator function: when e ∈ T k Select 1 if the value is 1, otherwise select 0. p ( e The larger the number, the more likely it is to be a branch. e The higher the probability of it appearing in the candidate topology.

[0109] The temperature parameter t is used to adjust the dispersion of scene probabilities: when t is small, scene probabilities are more concentrated in low-cost candidate topologies; when t is large, scene probabilities tend to be more uniform. To avoid excessive concentration or dispersion of probabilities, a minimum probability lower bound can be set. π floor and to π k After truncation, normalize again.

[0110] Probability of branch existence p ( e This can serve as a risk assessment indicator for subsequent recovery decisions: for example p ( e For branches with lower critical power supply paths, conservative margins can be added to the optimization model, alternative connection paths can be introduced, or dependence on that branch can be reduced; p ( e For branches with higher success rates, closure operations can be prioritized to improve the success rate.

[0111] When new measurements or on-site verification information arrive, it can be updated. E fix1 , E fix0 and E unc And recalculate the score. s e With scene probability π k This enables rolling correction of topological uncertainties.

[0112] like Figure 4 As shown, this invention maps Top-K candidate radial topologies to a set of topological scenarios, and obtains scenario probabilities through softmax based on candidate topology costs. This allows for the calculation of posterior connection probabilities of uncertain branches, which is used to characterize topological uncertainty and provide probability weights and screening criteria for robust supply partition optimization.

[0113] Step S107: Considering the uncertainty of the topology scenario, establish a unified optimization model for proactive load balancing and recovery, with the objective of maximizing the critical load supply level and minimizing the cost of partition structure adjustment. The objective function can be expressed as:

[0114] ;

[0115] in, u e For linearization | z e - z 0,e |(Cost of switching action); y q∈{0,1} indicates whether partition / root node is enabled (used to suppress too many partitions and overly complex structures);

[0116] The model integrates decision-making on zoning boundaries, switch target states, distributed power generation and energy storage output / energy plans, and load recovery. Simultaneously, it introduces linearized dynamic frequency safety constraints into the model to limit frequency deviation and rate of change of frequency (ROCOF), avoiding power imbalance caused by step-by-step recovery that could lead to frequency exceeding limits.

[0117] The linearization constraint for the switch / boundary difference can be written as:

[0118] ;

[0119] in, z e ∈{0,1} represents the target open / closed state (1 closed, 0 open). z 0,e This indicates the current running status.

[0120] In one implementation, the power supply zoning model can be constructed as a scenario-based mixed-integer linear programming (MILP) model. To improve the solution efficiency of large-scale problems, a linearized power flow model can be preferred, and the nonlinear terms can be linearized so that the model can be solved quickly by a general-purpose MILP solver.

[0121] Constraint forms such as linearized power flow (LinDistFlow):

[0122] ;

[0123] ;

[0124] Among them, P ij,k ( t ), Q ij,k ( t Branch power; V i,k ( t () represents the node voltage; R ij , X ij Line parameters; S max ij This represents the maximum line capacity. w k ( e This is used to characterize the availability of branch paths in a scene (if a branch is unavailable, its effective capacity is 0).

[0125] Energy storage energy state evolution and boundary constraints:

[0126] ;

[0127] ;

[0128] in, E s,k ( t ) for energy storage s During the period t The energy state; P ch , P dis This refers to the charging and discharging power. h ch , h dis For charge and discharge efficiency; E min s , E max s This is the energy boundary.

[0129] Step S108: Dynamic Frequency Safety Constraint Modeling (Linearized Embedding) To limit the risk of frequency overruns caused by phased load deployment, simplified frequency response constraints are established for each power supply zone (island), and these constraints are embedded into the unified optimization model of Step S108 in the form of linear inequalities. This constraint affects the frequency deviation D. f And set a threshold limit for the rate of change of frequency (ROCOF) to limit the net power deficit (D) of the island. P Apply feasible region constraints.

[0130] First, define the partition net power deficit:

[0131] ;

[0132] And define the partition equivalent inertia / damping as the sum of the parameters that can participate in the frequency modulation unit within the partition:

[0133] ;

[0134] in, q For partition numbering; V q The set of nodes within the partition; DG q , ESS q This refers to the combination of DG and energy storage within the zone; u For participation in the frequency modulation unit assembly; H u , D u These are the unit inertia and damping parameters, respectively. v u,q ∈{0,1} represents a unitu Does it belong to a partition? q .

[0135] Under the simplified model, the frequency deviation and ROCOF can be constrained using a linear threshold:

[0136] ;

[0137] ;

[0138] Among them, D f max ROCOF is the frequency deviation threshold. max This is the frequency change rate threshold. The above constraint is equivalent to limiting the net power imbalance caused by phased load connection, thereby improving the frequency security of islanding.

[0139] like Figure 5 As shown, during the islanding process and load recovery, the islanding frequency deviation D... f It should be constrained within the preset safety boundary ±D f max Within this range, a threshold upper limit is set for the rate of change of frequency (ROCOF). max This is to suppress the risk of low-frequency over-limits caused by load step input and distributed power source tracking errors, thereby improving the operational safety of the power supply zone.

[0140] Step S109: Establish the target network structure and adjust the boundary difference D. E Generation and contingency plan updates:

[0141] Solving the optimization model in steps S108 to S109 yields the optimal supply guarantee partition target structure at the current rolling moment, and also obtains the target switch state vector. z .like Figure 6 As shown, dynamic frequency safety constraints link power imbalances at the zonal level with the zonal equivalent inertia / damping and available frequency regulation reserves, ensuring that the frequency deviation and rate of change of each power supply zone meet threshold limits under load recovery or output fluctuation disturbances. Therefore, optimization solutions tend to configure or retain regulation units with rapid power support capabilities (such as energy storage or distributed power sources with frequency regulation capabilities) within the zonal area and adjust the zonal boundary / tether switch states accordingly to improve the zonal frequency safety margin and reduce the risk of low-frequency over-limit violations.

[0142] based on z Current running status z 0 difference generation boundary differential adjustment amount D E(Represents the set of switches that need to be changed). When a main grid fault warning, main grid power failure, or upstream power supply disconnection is detected, the latest contingency plan is directly invoked to form a power supply protection zone; the zone formation is executed according to the boundary differential adjustment amount and the preset switch operation sequence, and priority is given to ensuring power supply to critical loads;

[0143] Example 2

[0144] A distribution network active supply protection zoning system for multiple new types of entities, such as Figure 2 As shown, it includes:

[0145] The data acquisition and resource status mapping unit is configured to: acquire and update the electrical parameters and connection relationships of nodes, branches and switches in real time during normal operation; dynamically monitor the predicted output curve / available output limit of distributed generation (DG), the state of charge (SOC) and charge / discharge power / energy boundary of energy storage devices (ESS), and the adjustable capacity of flexible loads / demand response resources; establish and dynamically update the load importance classification list and critical load list based on user-side demand, and construct a resource status mapping that reflects the real-time operation boundary of "source-grid-load-storage", providing an input data foundation for power supply zoning and contingency plan generation.

[0146] The branch state evaluation and constraint update unit is configured to: construct a set of candidate branches. E Based on the status of switch remote signaling / telemetry and communication, protection and automation action records, online monitoring alarms, and manual inspection information, candidate branches are divided into defined closed sets. E fix1 Determine the disconnect set E fix0 Status pending verification set E unc The sets of feet are pairwise mutually exclusive and satisfy the following conditions: E fix1 ∪ E fix0 ∪ E unc = E ;against E unc In the middle branch, the measurement sequences and alarm information of key nodes are fused and evaluated within the rolling observation window to form the rolling update results of branch connectivity confidence score and connectivity constraints, which are used for subsequent plan optimization and invocation; when new measurement or on-site confirmation information arrives, the above set division and confidence / constraint input are updated.

[0147] The dynamic frequency safety constraint unit is configured to: discretize and linearize the equivalent frequency response process of the island partition before solving the pre-plan optimization, forming a dynamic frequency safety constraint form that can be embedded in mixed integer optimization; the dynamic frequency safety constraint includes at least the partition frequency deviation D.f The amplitude constraint and the threshold constraint of the rate of change of frequency (ROCOF) are combined with the inverter power supply / energy storage droop control, reserve capacity and ramping capability to form constraint parameters or constraint sets; among them, the dynamic frequency safety constraint serves as a supporting input for the power supply zoning and recovery plan generation unit, and is used to suppress the low-frequency risk caused by power jumps during the load recovery process at the decision level.

[0148] The power supply zoning and recovery plan generation unit is configured to: establish and solve the active power supply zoning model based on the resource status mapping, branch status assessment and constraint update results, and dynamic frequency security constraint input; jointly determine the selection of the zoning root node, the division of the zoning boundary, the source-load-storage output / energy plan and load recovery of each zoning; and satisfy constraints such as power flow balance, node voltage and line capacity, energy storage energy boundary and equipment operation; and generate a power supply plan that can be triggered for execution.

[0149] The target structure and switching strategy output unit is configured to output the optimal supply guarantee zone target structure and branch / switch target on / off state variables. z And based on z Current running status z 0 difference generation boundary differential adjustment amount D E (Representing the set of switches that need to be changed); Optionally, a set of switch control actions or a sequence of actions can be further output and the intermediate states can be checked for safety; wherein, the output is the target state and change set at the plan level, and the present invention does not limit the specific execution sequence of the switch actions.

[0150] The trigger response and contingency plan execution unit is configured to: establish an active triggering mechanism; when a main grid fault warning, main grid power failure, or upstream power supply disconnection is detected, directly call the latest contingency plan to form a power supply protection zone; execute the zone formation according to the boundary differential adjustment amount and the preset switch operation sequence, and prioritize the power supply of critical loads; when the new plan differs significantly from the current structure and meets the change cost constraint, update the zone boundary and target switch status; otherwise, keep the zone structure unchanged and only modify the scheduling plan to achieve dynamic optimization and consistency control.

[0151] Example 3

[0152] A distribution network active power supply protection zoning system oriented towards supporting multiple types of entities includes:

[0153] The data acquisition and resource status mapping unit is configured to acquire and update the electrical parameters and connection relationships of nodes, branches and switches during the normal operation phase of the distribution network, acquire source-grid-load-storage data, and construct a resource status mapping that reflects the real-time operating boundary of source-grid-load-storage.

[0154] The branch status assessment unit is configured to construct a candidate branch set. Based on the acquired information, the candidate branch set is divided into a confirmed closed set, a confirmed open set, and a status to be verified set. For the status to be verified set, the voltage measurement sequence of key nodes is acquired within the rolling observation window, preprocessed, and features reflecting the electrical correlation of nodes are extracted to characterize the branch availability / connectivity confidence.

[0155] The pruning and connectivity repair unit is configured to perform pruning and connectivity repair on the topology graph formed by candidate branches based on the characterization results of branch availability / connectivity credibility.

[0156] The supply guarantee partitioning and recovery plan generation unit is configured to establish an active supply guarantee partitioning model based on the resource status mapping and the processed topology map. The objective function is to maximize the supply level of critical loads and minimize the cost of partitioning structure adjustment. Linear dynamic frequency security constraints are introduced into the active supply guarantee partitioning model, the frequency response process of the isolated partitions is discretized and frequency deviation and frequency change rate threshold constraints are applied.

[0157] The target structure and switching strategy output unit is configured to perform optimization solution of the active supply guarantee partition model, obtain the optimal supply guarantee partition target structure at the current time, obtain the target switching state vector, and generate boundary differential adjustment amount based on the difference between the target switching state vector and the current running state vector.

[0158] The trigger response and contingency plan execution unit is configured to respond to a trigger event, invoke the optimal supply guarantee zone target structure, and execute the formation of the optimal supply guarantee zone target structure according to the boundary differential adjustment amount and the preset switch operation sequence.

[0159] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0160] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0161] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0162] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0163] While the specific embodiments of this disclosure have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of this disclosure. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of this disclosure are still within the scope of protection of this disclosure.

Claims

1. A method for active power supply protection zoning in distribution networks supported by multiple types of entities, characterized in that, Includes the following steps: During the normal operation of the distribution network, the electrical parameters and connection relationships of nodes, branches and switches are acquired and updated, source-grid-load-storage data are acquired, and a resource status mapping reflecting the real-time operation boundary of source-grid-load-storage is constructed. Construct a candidate branch set, and based on the acquired information, divide the candidate branch set into a set of confirmed closures, a set of confirmed disconnections, and a set of states to be verified. For the set of states to be verified, the voltage measurement sequence of key nodes is obtained within the rolling observation window, preprocessed, and features reflecting the electrical correlation of nodes are extracted to characterize the availability / connectivity confidence of branches. Based on the characterization results of branch availability / connectivity reliability, the topology graph formed by candidate branches is pruned and its connectivity is repaired. Based on the resource state mapping and the processed topology, an active supply guarantee partitioning model is established. The objective function is to maximize the supply level of critical loads and minimize the cost of partitioning structure adjustment. A linear dynamic frequency security constraint is introduced into the active supply guarantee partitioning model. The frequency response process of the isolated partitions is discretized and frequency deviation and frequency change rate threshold constraints are applied. The active supply guarantee zoning model is optimized and solved to obtain the optimal supply guarantee zoning target structure at the current moment, and the target switch state vector is obtained. The boundary differential adjustment amount is generated based on the difference between the target switch state vector and the current operating state vector. In response to the triggering event, the optimal supply guarantee zone target structure is invoked, and the formation of the optimal supply guarantee zone target structure is performed based on the boundary differential adjustment amount and the preset switching operation sequence.

2. The active power supply guarantee zoning method for distribution networks supported by multiple types of entities as described in claim 1, characterized in that, The process of acquiring and updating the electrical parameters and connection relationships of nodes, branches, and switches includes: acquiring each node, as well as its type, load requirements, and importance level; acquiring each candidate branch, including its endpoints, impedance parameters, switch positions, and initial switching states; and constructing a node set and a candidate branch set based on the acquired node, branch, and switch information.

3. The active power supply guarantee zoning method for distribution networks supported by multiple types of entities as described in claim 1, characterized in that, The process of acquiring source-grid-load-storage data and constructing a resource status mapping that reflects the real-time operating boundaries of source-grid-load-storage includes: acquiring the output limit of distributed power sources, the charging and discharging power and energy boundaries, efficiency, initial state of charge and real-time state of charge of energy storage devices, and the adjustable capacity of flexible load / demand response resources; establishing and dynamically updating a load importance classification list based on user-side demand; and acquiring voltage measurement sequences, remote signaling / alarms, inspection and on-site confirmation information.

4. The active power supply guarantee zoning method for distribution networks supported by multiple types of entities as described in claim 1, characterized in that, Based on the acquired information, the process of dividing the candidate branch set into a determined closed set, a determined open set, and a status-to-verify set includes: dividing the candidate branches into a determined closed set according to the basic information of the nodes, branches, and switches of the active distribution network to be restored, the switch historical records, communication status, and on-site inspection / remote signaling information. 、 Determine the disconnected set and the set whose state needs to be verified. Determine that branches in the closed set remain closed during the recovery process. Determine that branches in the disconnected set remain disconnected. The three sets are mutually exclusive and constitute the entire candidate branch set.

5. The active power supply zoning method for distribution networks supported by multiple types of entities as described in claim 1, characterized in that, The process of extracting features reflecting the electrical correlation of nodes and characterizing the availability / connectivity reliability of branches includes: targeting candidate branches in the set of states to be verified. e Based on nodes i With nodes j Statistical correlation indices are calculated from the voltage characteristic sequences, and branch scores are generated. s e Specifically: ； in, X' i and X' j They represent respectively to X i ( t )and X j ( t The vector after mean removal and normalization. r e branch road e The correlation coefficient of the voltage sequences at both ends indicates that the closer the correlation value is to 1, the stronger the consistency of the voltage changes at both ends. ； in, I e As a proxy indicator of mutual information, it is used to enhance the characterization of nonlinear correlations; ； in, T meas The number of sampling points in the measurement window. T eff The number of valid samples, T min The minimum sample threshold, g Con is the adjustment coefficient. f e The branch score is used to reflect the impact of measurement quality on the reliability of the score. ； in, κ Branch scoring for smoothing parameters s e The larger the value, the more likely the branch is to be a true connecting branch.

6. The active power supply zoning method for distribution networks supported by multiple types of entities as described in claim 1, characterized in that, Based on the characterization results of branch availability / connectivity reliability, the process of pruning and connectivity repairing the topology graph formed by candidate branches includes: based on branch scores... s e The topology graph is pruned and its connectivity is restored. Under the premise that branches in the closed set must be selected and branches in the open set must be discarded, several adjacent uncertain branches with the highest branch scores are retained for each node to form a sparse candidate edge set. If pruning causes the candidate graph to be disconnected, edges that connect different connected components and whose branch scores are higher than a set value are added first to restore connectivity.

7. The active power supply guarantee zoning method for distribution networks supported by multiple types of entities as described in claim 6, characterized in that, Pruning uses a node-by-node retention method. K Two strategies are used: scoring threshold screening and other methods. Top-retained node by node K The strategy is: for each node i After eliminating the definite disconnected set and forcibly retaining the definite closed set, the remaining adjacent uncertain branches are scored according to branch. s e Sort by highest to lowest and retain the top results. k keep strip; The scoring threshold filtering strategy is: only retain s e ≥ s th Uncertain branches; in k keep and s th The settings are based on network size, measurement coverage, and computing resources; The connectivity repair process is as follows: divide the pruned candidate graph into connected components; when there are multiple connected components, select the branch that connects different connected components and has the highest score and add it to the candidate edge set in turn, until the candidate graph is connected or reaches the preset edge filling limit; If global connectivity cannot be achieved within the edge-filling limit, a sparse candidate graph with local connectivity is output, and multiple isolated root nodes are allowed to supply power separately during the subsequent establishment of the active power supply partitioning model.

8. The active power supply guarantee zoning method for distribution networks supported by multiple types of entities as described in claim 1, characterized in that, The process of establishing a proactive load balancing zoning model includes: considering the uncertainty of the topology scenario, establishing a proactive load balancing zoning model with the objective of maximizing the critical load balancing level and minimizing the cost of zoning structure adjustment. The objective function is expressed as: ； in, For the set of topological scenes { ω k }, k For scene indexing, π k For the corresponding topology scene ω k The probability of the scenario; V For a set of nodes, i For node indexing; T For the number of discrete scheduling periods, t For time period index, t The length of a single time period; ω i For nodes i The load importance weighting coefficient; P serv i,k ( t ) for scene k Next period t node i The active power of the restored power supply; sw The switching action cost weighting coefficient, E sw For a set of operable branches / switches, e For branch / switch index, u e For branch circuits / switches e Action cost, according to | z e - z 0,e | is represented by a linearized form; R For the set of root nodes of the partition, q For the root node index, root Enable a penalty coefficient for the root node. y q ∈{0,1} indicates whether the partition / root node is enabled; The proactive power supply zoning model is used for joint decision-making on zoning boundaries, switch target states, distributed power generation and energy storage output / energy planning, and load recovery.

9. The active power supply zoning method for distribution networks supported by multiple types of entities as described in claim 1, characterized in that, The active power supply zoning model includes energy storage energy state evolution and boundary constraints, linearized power flow constraints, and linearized constraints of on / off / boundary difference. The dynamic frequency safety constraints include setting upper and lower limit thresholds for frequency deviation and frequency change rate to impose feasible domain constraints on the island net power deficit.

10. A distribution network active supply guarantee zoning system oriented towards multiple types of entities, characterized in that, include: The data acquisition and resource status mapping unit is configured to acquire and update the electrical parameters and connection relationships of nodes, branches and switches during the normal operation phase of the distribution network, acquire source-grid-load-storage data, and construct a resource status mapping that reflects the real-time operating boundary of source-grid-load-storage. The branch status assessment unit is configured to construct a candidate branch set, and based on the acquired information, divide the candidate branch set into a confirmed closed set, a confirmed open set, and a status to be verified set. For the set of states to be verified, the voltage measurement sequence of key nodes is obtained within the rolling observation window, preprocessed, and features reflecting the electrical correlation of nodes are extracted to characterize the availability / connectivity confidence of branches. The pruning and connectivity repair unit is configured to perform pruning and connectivity repair on the topology graph formed by candidate branches based on the characterization results of branch availability / connectivity credibility. The supply guarantee partitioning and recovery plan generation unit is configured to establish an active supply guarantee partitioning model based on the resource status mapping and the processed topology map. The objective function is to maximize the supply level of critical loads and minimize the cost of partitioning structure adjustment. Linear dynamic frequency security constraints are introduced into the active supply guarantee partitioning model, the frequency response process of the isolated partitions is discretized and frequency deviation and frequency change rate threshold constraints are applied. The target structure and switching strategy output unit is configured to perform optimization solution of the active supply guarantee partition model, obtain the optimal supply guarantee partition target structure at the current time, obtain the target switching state vector, and generate boundary differential adjustment amount based on the difference between the target switching state vector and the current running state vector. The trigger response and contingency plan execution unit is configured to respond to a trigger event, invoke the optimal supply guarantee zone target structure, and execute the formation of the optimal supply guarantee zone target structure according to the boundary differential adjustment amount and the preset switch operation sequence.

Images