A primary-secondary fusion ring network box operation topology identification method

Abstract

The present application relates to power distribution operation data processing, topology analysis and computer information processing technical field, specifically to a kind of primary and secondary fusion ring network box operation topology identification method.The method obtains switch opening and closing telesignaling, event sequence record, protection action information and the voltage, current and power direction data of feeder interval, forms secondary action event sequence and primary electrical response sequence;According to the physical constraint of action and response in sequence Correction data channel time shift, build unified causal time axis and cross-layer event causal diagram;Based on reference wiring relationship, generate candidate operation topology, deduce section response sequence and power supply path change, and implement counterfactual replacement to pending switch state, determine actual operation topology according to causal conflict.The present application can restore the corresponding relationship of action and electrical response under the condition of clock offset, telesignaling jitter, event out-of-order or partial event missing, reduce topology misupdate caused by abnormal data, and retain topology edge evidence and abnormal event record.

Description

Technical Field

[0001] This invention relates to the fields of power distribution operation data processing, topology analysis and computer information processing technology, specifically a method for identifying the operating topology of a primary and secondary integrated ring network box. Background Technology

[0002] The integrated primary and secondary ring main unit integrates primary switchgear with distribution terminals, protection and control, and communication functions. The operating topology is typically updated by the master station or edge terminals based on a pre-established wiring model and remote signaling for switch opening and closing. The standard processing flow involves receiving remote signaling changes, event sequence records, and protection start-up and reset information from each bay according to a point-to-point mapping relationship. Switches in the closed position are then mapped as conducting sides, and switches in the open position as disconnected sides. The connectivity between busbar sections, feeder sections, and the power supply side is calculated on a fixed wiring diagram. When switching operations, fault isolation, or power restoration occur, the system overwrites the original state with the latest remote signaling status, regenerating the current operating topology. This generated topology is then provided for fault section judgment, power outage range analysis, and power supply path calculation. Some schemes also use voltage presence, current amplitude, or power direction to check the consistency of remote signaling results. However, such checks often use single-moment comparisons or fixed logic rules, typically only determining whether measured values ​​meet preset conditions, without establishing a continuous correspondence between switch actions, changes in connection relationships, and subsequent electrical responses.

[0003] Existing topology identification schemes typically use device-sent timestamps as a unified sorting criterion when processing multi-source operational data, assuming that clock deviations between different terminals, intervals, and communication channels are negligible. When switch auxiliary contacts experience jitter or delay, the remote signaling change time may be earlier or later than an actual electrical state change. Communication congestion, event buffering, and batch transmission can also alter the recording order of protection actions, undervoltage, overvoltage recovery, and current surge events. To address these issues, conventional techniques often employ time window merging, duplicate event filtering, or fixed delay compensation to group events with similar times into the same operational process, and then determine the switch state based on the majority of remote signaling results. If remote signaling and telemetry are inconsistent, one type of data is usually selected according to a preset priority, or the corresponding device is marked as abnormal while maintaining the original topology. This approach does not perform relative time-shift correction for each data channel based on the physical sequence reflected by current surges, undervoltage, overvoltage recovery, and power direction reversals, nor does it deduce the appropriate segment response sequence for different candidate topologies, making it difficult to recover the true operational process from contradictory events.

[0004] When the integrated primary and secondary ring network enclosure suffers from terminal clock skew, remote signal jitter, auxiliary contact lag, missed event reporting, or out-of-order events, the method of directly updating the topology based on switch remote signaling cannot reliably determine the actual connection relationship. The fundamental reason is that existing processing treats the secondary action state and primary electrical response as independent judgment criteria, lacking causal constraints that connect switch actions, changes in connection relationships, and section electrical changes, and also lacking a mechanism for reverse verification of candidate switch states. When multiple switches operate consecutively within a similar timeframe, the event sequence recorded by different data channels may be inconsistent with the actual physical process. Simply correcting the timestamp may still yield multiple candidate topologies that satisfy local conditions; simply adding voltage, current, or power direction verification cannot determine whether the conflict originates from incorrect switch states, misaligned event times, or missing events. The resulting operating topology may be inconsistent with the actual power supply path and will continue to retain erroneous connection relationships in subsequent calculations. Summary of the Invention

[0005] The purpose of this invention is to provide a method for identifying the operating topology of a primary and secondary integrated ring network box, which can solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] A method for identifying the operating topology of a primary and secondary integrated ring network enclosure includes:

[0008] S1. Obtain remote signaling of switch opening and closing, event sequence records, protection action information, and voltage, current and power direction data of each feeder interval within the preset identification time window, and form a secondary action event sequence and a primary electrical response sequence respectively.

[0009] S2. Based on the physical sequence constraints between switching actions and electrical responses, relative time shift correction is performed on different data channels to construct a unified causal time axis;

[0010] S3. Based on the unified causal time axis, establish a cross-layer event causal graph between action events, changes in connection relationships, and electrical responses;

[0011] S4. Generate candidate operating topologies based on the basic wiring relationship of the ring network box, and deduce the electrical response sequence corresponding to each candidate operating topology;

[0012] S5. Match the deduction results with the cross-layer event causal graph, perform counterfactual replacement on the switch states in the candidate running topology, and determine the actual running topology based on the causal conflicts before and after the replacement.

[0013] Preferably, the relative time shift correction includes: extracting the switch opening and closing times from the secondary action event sequence, and extracting the current surge, undervoltage, voltage recovery, and power direction reversal times from the primary electrical response sequence;

[0014] Establish an event pairing set based on the baseline connection relationship between the switch and the feeder interval;

[0015] For each data channel, a time shift evaluation metric is constructed based on the number of event pairs that satisfy the physical sequence relationship in the event pairing set, the event interval dispersion, and the cross-interval sequence consistency.

[0016] The channel time shift is determined while keeping the relative order of events within the same data channel unchanged, and the channel time shift is used to rearrange the secondary action event sequence and the primary electrical response sequence.

[0017] Preferably, the cross-layer event cause-effect graph includes an action event layer, a connection relationship change layer, and an electrical response layer;

[0018] The action event layer uses switch opening and closing events and protection action events after relative time shift correction as nodes; the connection relationship change layer uses changes in the conduction state of candidate switches and changes in the reachability of corresponding feeder sections as nodes; the electrical response layer uses changes in voltage state, current amplitude, and power direction as nodes.

[0019] Based on the device correspondence, the corrected event sequence, and the segment reachability, directed causal edges are established between adjacent layer nodes, and directed causal edges that do not satisfy the physical propagation order are marked as conflict edges.

[0020] Preferably, the generation of the candidate operating topology includes: constructing topology nodes and topology edges based on the bus section, feeder section and switchgear in the reference wiring relationship;

[0021] The causal relationship between remote signaling status and the cross-layer event Figure 1 Set the switch topology edge that is consistent with the determined edge, and set the switch topology edge that conflicts with the remote signaling state and the cross-layer event causal graph as the undetermined edge;

[0022] The undetermined edges are assigned to either an on state or an off state to form candidate combinations;

[0023] Based on the mutual exclusion constraints of the same switch opening and closing states, the continuous constraints of energized sections, and the open-loop operation constraints of the ring network, candidate combinations that do not meet the constraints are deleted to obtain the candidate operating topology.

[0024] Preferably, the relative time shift correction further includes: dividing the event pairing set of the same data channel into multiple time segments according to a continuous identification time window;

[0025] Calculate the local time shift for each timing segment, and determine the channel clock drift direction based on the time difference of common event pairs in adjacent timing segments;

[0026] When adjacent local time shifts are discontinuous, the data channel is divided into multiple correction intervals by using events that occur simultaneously in multiple data channels during protection actions, switch changes, and undervoltage events as segmentation anchor points.

[0027] Within each correction interval, events are rearranged using independent time shifts, while maintaining the correspondence between the segmented anchor points in each data channel.

[0028] Preferably, the establishment of the cross-layer event cause-effect graph further includes: when there is no connection relationship change layer node that satisfies the timing constraints between the action event layer node and the electrical response layer node, searching along the reference wiring relationship for the shortest reachable path connecting the switch corresponding to the action event layer node and the feeder segment corresponding to the electrical response layer node;

[0029] The undetermined switch state combinations in the shortest reachable path are converted into candidate connection relationship change nodes;

[0030] The interpretable range of other electrical response layer nodes is filtered according to the candidate connection relationship change nodes;

[0031] Candidate connection relationship change nodes that can simultaneously connect multiple independent electrical response nodes are added to the cross-layer event causal graph, and their corresponding missing event identifiers are recorded.

[0032] Preferably, the deducing of the electrical response sequence corresponding to each candidate operating topology includes: for each action event in the cross-layer event causal graph, determining the starting segment adjacent to the action switch in the candidate operating topology;

[0033] The set of perturbation reachable segments is calculated based on the direction and connection level of the conductive topology edges in the candidate running topology;

[0034] Generate a sequence of responses based on the topological distances from the starting segment to each reachable segment;

[0035] The expected changes in energized state and power direction of each reachable section are determined based on the power supply connection relationship before and after the action.

[0036] The set of reachable disturbance segments, the sequence of responses, the expected change in charged state, and the change in power direction are combined to form a candidate electrical response sequence.

[0037] Preferably, the counterfactual replacement includes: sequentially selecting undetermined switches in the candidate running topology, replacing the on state of the selected undetermined switches with the off state or replacing the off state with the on state, while keeping the states of other topological edges unchanged;

[0038] The candidate electrical response sequence is regenerated based on the replaced candidate operating topology;

[0039] The regenerated candidate electrical response sequence is compared with the cross-layer event causal graph, and the electrical response nodes that cannot establish causal edges, the node pairs that violate the response order, the power supply path interruption locations, and the power direction discontinuity locations are counted respectively.

[0040] The statistical results are used to construct a counterfactual conflict vector corresponding to the undetermined switch.

[0041] Preferably, determining the actual operating topology based on the causal conflicts before and after the replacement includes: establishing an evidence matrix consisting of action event rows and electrical response event columns for each candidate operating topology, and recording the causal edge matching status, conflict edge position, and missing event identifier in the evidence matrix;

[0042] The counterfactual conflict vectors of each pending switch are mapped to the evidence matrix to determine the conflict increment caused by the state change of each pending switch.

[0043] Construct topological consistency constraints based on the set of causal edges maintained together across multiple independent action events;

[0044] Among the candidate running topologies that satisfy the topology consistency constraint, the candidate running topology with the smallest total conflict increment and the fewest number of missing event identifiers is selected.

[0045] Preferably, it further includes: associating the determined edges and undetermined edges in the selected candidate running topology with the causal edge matching records in the evidence matrix to form a topology edge evidence set;

[0046] When multiple consecutive recognition time windows obtain the same undetermined edge state, the corresponding undetermined edge will be updated to a determined edge;

[0047] When the change in the topological edge state in the subsequent identification time window is supported only by a single data channel event and conflicts with the set of topological edge evidence in the previous identification time window, the topological edge state of the previous identification time window is retained, and the single data channel event is written into the abnormal event set.

[0048] The output consists of the determined edge states, the reserved undetermined edge states, the set of topological edge evidence, and the set of abnormal events.

[0049] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0050] 1. The secondary action event sequence and the primary electrical response sequence are placed on the same causal time axis. Based on the physical sequence constraints between switching actions and current surges, undervoltage, voltage restoration, and power direction reversals, relative time-shift corrections are applied to different data channels. A directed causal relationship is then established between the action event layer, the connection relationship change layer, and the electrical response layer based on the corrected events. The candidate operating topology is no longer solely determined by the remote signaling state; it must simultaneously satisfy constraints such as mutually exclusive switching states, continuous energized sections, open-loop operation of the ring network, and the order of electrical response propagation. For conflicting pending switching states, counterfactual substitution is used to regenerate the candidate electrical response sequence, and the causal edge missing, response order conflict, power supply path interruption, and power direction discontinuity before and after the substitution are compared. This allows for the elimination of candidate topologies that can only explain partial data and not the complete action process under conditions of clock skew, remote signaling jitter, auxiliary contact lag, or event disorder, ensuring that the determined operating topology corresponds to changes in the primary electrical state.

[0051] 2. Continuously identifying local time shifts, segmented anchor points, and channel clock drift directions within a time window is used to correct time deviations of the same channel at different stages, avoiding subsequent events from being out of order again due to the use of a single fixed compensation amount. When a direct event is missing, the cross-layer event causal graph searches for reachable paths between the action switch and the response segment along the baseline wiring relationship, and supplements missing event identifiers with nodes that can explain changes in the connection relationships of multiple independent electrical response nodes, ensuring that some missed events do not directly interrupt topology reasoning. The evidence matrix records the matching status, conflict locations, and missing identifiers between each action event and electrical response event, and maps counterfactual conflict vectors to specific pending switches, maintaining the correspondence between topology edge states and their supporting events. When multiple consecutive identification time windows obtain the same state, the pending edge is updated to a determined edge; changes supported only by a single data channel and conflicting with the existing evidence set are written into the abnormal event set, limiting the propagation range of local abnormal data to the overall topology and preserving a traceable record of the topology state formation process. Attached Figure Description

[0052] Figure 1 This is a flowchart illustrating the overall process of topology identification for a primary and secondary integrated ring network enclosure.

[0053] Figure 2 This is a graph showing the relative time shift of the data channel.

[0054] Figure 3 This is a graph showing the matching strength of cross-layer directed causal edges.

[0055] Figure 4 The graph shows the deviation of the response order of the candidate running topology. Detailed Implementation

[0056] In one embodiment, please refer to Figure 1 A method for identifying the operating topology of a ring main unit (RNB) with primary and secondary integration includes: S1, acquiring remote signaling of switch opening and closing, event sequence records, protection action information, and voltage, current, and power direction data of each feeder interval within a preset identification time window, forming a secondary action event sequence and a primary electrical response sequence respectively; S2, performing relative time shift correction on different data channels based on the physical sequence constraints between switch actions and electrical responses, and constructing a unified causal time axis; S3, establishing a cross-layer event causal graph based on the unified causal time axis, connecting relationship changes, and electrical responses; S4, generating candidate operating topologies based on the RNB's baseline wiring relationship, and deducing the electrical response sequence corresponding to each candidate operating topology; S5, matching the deduction results with the cross-layer event causal graph, performing counterfactual replacement on the switch states in the candidate operating topologies, and determining the actual operating topology based on the causal conflicts before and after the replacement.

[0057] In this embodiment, the preset identification time window is established in an event-driven manner. When any switch opening / closing remote signaling, protection start-up, protection reset, feeder interval undervoltage, undervoltage, current sudden change, or power direction reversal is recorded, the reference interval before the action is extracted with the corresponding event time as the center, and the response observation interval is extracted afterward. The reference interval before the action is used to determine the energized state, current stability range, and power direction of each feeder interval before the state change. The response observation interval is used to accommodate the time difference formed by switch mechanical action, auxiliary contact feedback, terminal event buffering, and communication transmission. If the response observation intervals of adjacent events overlap, they are merged into the same identification time window to avoid the continuous switching process being split into independent identification tasks.

[0058] Specifically, the switch opening and closing remote signaling forms a secondary action event record according to the equipment identifier, opening and closing status, source timestamp, receiving timestamp, and data channel identifier. The event sequence record and protection action information form a protection event record according to the protected object, action type, action time, and reset time. Records belonging to the same switchgear or the same protected object are temporarily sorted by source timestamp to form a secondary action event sequence. Voltage, current, and power direction data are first aligned according to feeder interval identifiers, and then state change events are extracted from continuous sampling values. Voltage events are formed by the transition between a voltage state and an unvoltage state, current events are formed by the transition between a stable range and a sudden change range, and power direction events are formed by the transition between inflow, outflow, or near-zero states. The obtained events are temporarily sorted by receiving timestamp to form a primary electrical response sequence.

[0059] In a preferred embodiment, current surge events are determined by the difference between the median of the reference interval before the action and the continuous observations within the response observation interval. The direction of the difference is used to distinguish between current increases and decreases. Loss of voltage events are determined by the phase voltage state changing from voltage to no voltage. Re-voltage events are determined by the phase voltage state changing from no voltage to voltage. Power direction reversal events are determined by a continuous change in the sign of active power in the same feeder interval. Isolated changes whose duration is insufficient to cover the continuous observation records in the same interval are only retained as events to be verified and do not directly participate in the establishment of causal edges. This allows a single electrical response sequence to retain state transitions that can reflect changes in connectivity, rather than interpreting single-point fluctuations as topological changes.

[0060] In this embodiment, relative time shift correction does not employ a single fixed delay compensation. Instead, it uses the permissible physical order between switching action and electrical response as a constraint to search for the relative shift of different data channels within the same identification time window. The permissible physical order includes that the switch change must not be later than the stable voltage loss or stable voltage recovery caused by it, the protection start must not be later than the disappearance of current after the corresponding fault section is isolated, the power direction reversal related to power supply path switching must not be earlier than the switching action that forms a new conduction path, and for the case where the same action causes multiple section responses, the section response closer to the operating switch must not be continuously later than the response of the far section that can only be reached through the aforementioned section.

[0061] The relative time shift is determined by the following formula:

[0062] ;

[0063] in, This represents the corrected time shift of data channel c. This represents the searchable time-shift set determined based on the length of the recognition time window. Indicates the use of candidate time shift The quantity of events that violate physical chronology. Indicates the degree of dispersion of similar events over time intervals. Indicates the number of response sequence conflicts across feeder intervals. , and To maintain the fixed meaning of non-negative constraint weights, argmin represents selecting the candidate time shift that minimizes the result within the parentheses. For example, for a certain channel... When three sets of events occur, the response precedes the action. If there are no reverse events during the sampling interval of each event, then Take two event sampling intervals and shift the entire event of that channel to the next interval.

[0064] Please refer to Figure 2 , Figure 2 The middle horizontal axis represents the candidate time shift. The unit is seconds, indicating the selection of the candidate time shift that minimizes the result within the parentheses. For example, if a channel takes... When three sets of events occur, the response precedes the action. If there are no reverse events during the sampling interval of each event, then Take two event sampling intervals and shift the entire channel event backward. This indicates selecting the candidate time shift that minimizes the result within the parentheses; for example, for a certain channel... When three sets of events occur, the response precedes the action. If there are no reverse events during the sampling interval of each event, then Two event sampling intervals are taken, and the entire event for that channel is shifted backward. The values ​​cover three correction scenarios: negative advance, zero offset, and positive lag. The vertical axis represents the comprehensive evaluation metric formed by the reverse logarithm, interval dispersion, and cross-layer conflict. The black solid line corresponds to the final evaluation metric, while the gray dashed, dotted, and dotted lines correspond to... , and The three components are illustrated in the diagram to distinguish the changes of each evaluation component under different candidate time shifts.

[0065] from Figure 2 It is evident that the comprehensive evaluation quantity is... A trough forms near 0 seconds, and gradually rises on both sides as the absolute value of the time displacement increases, indicating that when the physical sequence of the secondary action event and the primary electrical response event is basically restored... The corresponding candidate region can simultaneously reduce the number of reverse-order events and the conflict in the response order across feeders. When the negative direction is significantly advanced or the positive direction is significantly delayed, the black curve rises significantly, indicating that although the event rearrangement still maintains the order within the same channel, it will disrupt the consistency of the response across intervals.

[0066] After relative time shift correction is completed, the corrected event time is used as the position coordinate of the unified causal time axis. The original relative order of events within the same data channel is preserved, and the events are reordered according to the correction time between different data channels. If two events are at the same time position on the unified causal time axis, the switching action or protection start is placed before the change in connection relationship according to physical constraints, and the change in connection relationship is placed before the stable electrical response. If the order cannot be determined, the two events are placed in parallel time layers and multiple connection possibilities are preserved in subsequent causal matching to avoid eliminating real uncertainties by arbitrarily sorting them.

[0067] Furthermore, the cross-layer event cause-effect graph is represented by a directed hierarchical graph. Action events are upstream nodes, connection relationship changes are intermediate nodes, and electrical responses are downstream nodes. Action event nodes record switch identification, action direction, correction time, and data source. Connection relationship change nodes record the conduction changes of candidate switch edges, the resulting changes in segment reachability, and the corresponding candidate topology identification. Electrical response nodes record feeder interval identification, response type, response direction, and correction time. Directed edges between nodes are established only when the equipment correspondence, time sequence, and reference wiring reachability are simultaneously satisfied.

[0068] The matching strength of a directed causal edge is calculated according to the following formula:

[0069] ;

[0070] in, This represents the matching strength between upstream node u and downstream node v. This represents a discrete value indicating the correspondence between device identifiers or protected objects; it is 1 if the correspondence is correct, and 0 otherwise. This indicates the degree to which the corrected time order is satisfied. A value of 1 indicates that the order is acceptable; a value between 0 and 1 indicates that the order is pending; and a value of 0 indicates that the order is reversed. This indicates whether a reachable path exists in the reference wiring; a value of 1 indicates a path that exists, and a value of 0 indicates a path that does not exist. This indicates whether the direction of motion is compatible with the direction of electrical response. to They are respectively represented as , , and The corresponding fixed non-negative coefficients, for example, when the tripping action and downstream pressure loss simultaneously satisfy the requirements of equipment correspondence, time sequence, path reachability and directional compatibility, all four discrete quantities are taken as 1, and the corresponding directed causal edge is retained.

[0071] Please refer to Figure 3 , Figure 3 The horizontal axis represents the sequence number of the cross-layer directed causal edge, with each sequence number corresponding to a candidate association between node u in the action event layer and node v in the electrical response layer; the vertical axis represents the matching strength. The solid black line represents the overall matching strength, while the other line types represent... , , and The value changes on the same causal edge are used to show the differences in the contributions of four types of evidence: device correspondence, correction time, segment reachability, and direction of action.

[0072] Figure 3The overall matching strength remains high, indicating that most causal edges simultaneously satisfy the device object correspondence and the reachability condition of the baseline wiring path. The black curves at some edge numbers show localized declines, mainly related to a decrease in time consistency or path reachability components. This suggests that while these edges may have correspondences in device identification or action direction, further verification using the evidence matrix is ​​needed to unify the causal timeline or explain the reachable path.

[0073] From the component curves Most values ​​are close to stable high values, reflecting a relatively clear equipment relationship between switches, protected objects, and feeder bays; and The fluctuations are more pronounced, reflecting the impact of time-shift correction results and baseline wiring path search on the validity of causal edges; It is used to constrain whether the switching action, undervoltage and overvoltage, current change and power direction change are compatible. When it decreases, it usually indicates that there is a weak consistency or undetermined state between the response direction and the change in connection relationship.

[0074] In this embodiment, the reference wiring relationship of the ring main unit consists of static connection records between busbar sections, feeder sections, incoming line bays, outgoing line bays, and switchgear, and is converted into a basic diagram without operating status. In the basic diagram, switchgear corresponds to variable topology edges, and fixed conductive connections correspond to fixed topology edges. Candidate operating topologies are generated by assigning an on or off state to each variable topology edge. However, for switchgear where there is no conflict between the remote signaling state and the primary electrical response, the remote signaling state is directly used. For switchgear where there is a conflict or missing remote signaling, it is reserved as an undetermined edge, thus limiting the candidate combinations to the local area related to the current event.

[0075] Each candidate operating topology must satisfy the following constraints: switch open and closed positions are mutually exclusive; the same switch side can only have a single operating state; confirmed energized sections can be connected to the power supply through a conductive path; confirmed unenergized sections cannot be connected to the power supply through an unisolated conductive path; and the predetermined open-loop position of the ring network does not form a closed power supply loop during normal operation. If the identification time window includes a fault isolation process, a short-term incomplete power supply path is allowed in the action transition layer, but a connection relationship that contradicts the stable electrical response is not allowed to be retained in the action completion layer.

[0076] The constraint evaluation of the candidate running topology is formed according to the following formula:

[0077] ;

[0078] in, Indicates candidate running topology The constraint is violated. This indicates that the mutual exclusion relationship between the switches violates the quantity. This indicates the number of conflicts between the energized state and the availability of power. This indicates the number of disallowed closed loops. This represents the number of stable electrical responses that cannot be explained by the candidate topology. to They are respectively represented as , , and The corresponding fixed non-negative coefficients are retained only. Or a candidate operating topology within the permissible range of missing events, for example, a candidate topology that allows a confirmed undervoltage section to remain connected to the power supply. Add and delete corresponding candidate combinations accordingly.

[0079] For the retained candidate operating topologies, starting from the switch edge corresponding to each action event, the topology state before the action is compared with the candidate state after the action to determine the newly accessible sections, unaccessible sections, and sections that remain energized despite changes in power supply path caused by the action. The expected response for newly accessible sections is the appearance of overvoltage or current, and the expected response for unaccessible sections is the disappearance of voltage or current. For sections with changed power supply path, it is determined whether the power direction should be reversed based on the new and old power supply directions. Then, a candidate electrical response sequence is generated according to the topology distance from the action switch to each response section.

[0080] When matching candidate electrical response sequences with cross-layer event causal graphs, it is not required that all electrical responses have exactly the same time interval. Instead, it is required that the response type, response direction, set of reachable segments, and relative order remain consistent. Multiple responses within the same topological distance layer are allowed to be parallel. When the same far segment can only be reached through the near segment, the near response is kept no later than the far response. For intermediate events that are missing in the actual record but supported by multiple adjacent segment responses, a missing event label is set. For events that exist in the actual record but cannot be reached by any candidate propagation path, an unexplained response label is set.

[0081] In this embodiment, counterfactual substitution is performed on each pending switch in the candidate operating topology, changing the on state in the original candidate state to off state or vice versa, while keeping the other switch states unchanged. The reachability, expected energized state, power direction and response sequence of each segment after the action are recalculated. Counterfactual substitution does not directly use the remote signaling state as the correct answer, but compares the explanatory power of the candidate topology before and after the substitution for the complete event process, in order to identify switch states that superficially conform to the remote signaling record but cannot explain an electrical response.

[0082] In the comparison before and after replacement, causal conflicts include the inability to establish a directed causal edge from the action event to the electrical response event, the actual response sequence being opposite to the candidate propagation path, the power supply path being interrupted in the confirmed energized section, and the power direction of adjacent sections not forming a continuous flow direction. If the conflict concentration is reduced after replacing a certain pending switch, the state after replacement is retained as the preferred state in the candidate topology. If different types of conflicts exist before and after replacement, the judgment is continued by combining the set of causal edges formed by other independent action events, rather than ending the identification based on the local consistency of a single electrical quantity.

[0083] When determining the actual operating topology based on the causal conflicts before and after the replacement, the matching results of each candidate topology under each action event are summarized. The candidate topology that can maintain more complete causal chains, rely less on missing events for supplementation, and does not violate the power supply reachability constraint is selected. If multiple candidate topologies have the same matching results, the switching edges that are different from each other are retained as undetermined edges and their respective evidence is output. A unique state is not forced to be generated. The corresponding undetermined edges are updated after new independent action events appear in the subsequent identification time window.

[0084] In a preferred embodiment, the relative time shift correction includes: extracting the switch opening and closing times from the secondary action event sequence, and extracting the current surge, undervoltage, voltage recovery, and power direction reversal times from the primary electrical response sequence; establishing an event pairing set according to the reference connection relationship between the switch and the feeder interval; for each data channel, constructing a time shift evaluation quantity based on the number of event pairs satisfying the physical sequence relationship in the event pairing set, the event interval dispersion, and the consistency of the cross-interval order; determining the channel time shift while keeping the relative order of events within the same data channel unchanged, and rearranging the secondary action event sequence and the primary electrical response sequence using the channel time shift quantity.

[0085] In this embodiment, the establishment of the event pairing set is limited to the electrical adjacency range in the reference connection relationship. The opening event is paired with the current reduction, undervoltage or power direction change that may occur downstream. The closing event is paired with the current increase, overvoltage or power direction change that may occur downstream. The protection start event is paired with the current change of the section where the protected object is located and the adjacent isolation section. For the case where the same response event may be interpreted by the operation of multiple adjacent switches, multiple pairing branches are retained and branches that do not meet the propagation path are eliminated in the candidate topology deduction stage to avoid pre-fixing the topology state that has not yet been determined in the time shift correction stage.

[0086] Preferably, the relative time shift correction further includes: dividing the event pairing set of the same data channel into multiple time segments according to the continuous identification time window; calculating the local time shift corresponding to each time segment, and determining the channel clock drift direction according to the time difference of common event pairs in adjacent time segments; when adjacent local time shifts are discontinuous, using events that simultaneously appear in multiple data channels in protection actions, switch changes, and undervoltage events as segmentation anchor points to divide the data channel into multiple correction intervals; rearranging events using independent time shifts in each correction interval, while maintaining the correspondence between segmentation anchor points in each data channel.

[0087] Specifically, the continuous identification time window is divided into time segments according to the changes in event density, the communication batch transmission boundary, and the distribution of common event pairs. Each time segment contains at least a pair of records that can connect the secondary action event and the primary electrical response event. The local time shift is obtained independently using the aforementioned time shift evaluation quantity. When adjacent local time shifts show a continuous increase or decrease, it is determined to be clock drift. When adjacent local time shifts undergo a jump without a common trend and the jump position coincides with the arrival position of the event batch, it is determined to be a segmented offset caused by the release of the channel buffer, and a new correction interval is established at the corresponding position.

[0088] Segmented anchor points require that the same physical event can be identified through different data sources. For example, protection start-up records and fault section current surges both point to the same fault occurrence process, and switch position change records and adjacent section voltage loss both point to the same isolation process. Segmented anchor points do not require that each data channel has the same source timestamp, but require that the event correspondence be stable after relative correction. If a candidate segmentation method causes the same anchor point to appear repeatedly or in reverse order in adjacent correction intervals, the corresponding segmentation method is discarded.

[0089] In a preferred embodiment, the cross-layer event causal graph includes an action event layer, a connection relationship change layer, and an electrical response layer. The action event layer uses switch opening and closing events and protection action events corrected for relative time shift as nodes. The connection relationship change layer uses changes in the conduction state of candidate switches and changes in the reachability of corresponding feeder sections as nodes. The electrical response layer uses changes in voltage state, current amplitude, and power direction as nodes. Directed causal edges are established between adjacent layer nodes based on device correspondence, corrected event sequence, and section reachability. Directed causal edges that do not satisfy the physical propagation order are marked as conflict edges.

[0090] In this embodiment, switch opening and closing events in the action event layer are deduplicated according to the mutual exclusion rule of opening and closing of the same device. Change records that occur repeatedly in a short period of time without corresponding electrical response are merged into state jitter clusters. State jitter clusters only retain the earliest record, the latest record, and the final stable state. Protection action events are associated with switch events that may perform isolation according to the protected object and action type. The connection relationship change layer does not directly copy the remote signaling state, but records the difference caused by each candidate switch state to the reachable set of the section. The electrical response layer merges continuous changes in the same direction of the same section into a single state transition node.

[0091] Furthermore, the establishment of the cross-layer event cause-effect graph also includes: when there is no connection relationship change layer node that satisfies the timing constraints between the action event layer node and the electrical response layer node, searching along the baseline wiring relationship for the shortest reachable path connecting the switch corresponding to the action event layer node and the feeder section corresponding to the electrical response layer node; converting the undetermined switch state combinations in the shortest reachable path into candidate connection relationship change nodes; filtering the interpretable range of other electrical response layer nodes according to the candidate connection relationship change nodes; adding candidate connection relationship change nodes that can simultaneously connect multiple independent electrical response nodes into the cross-layer event cause-effect graph, and recording their corresponding missing event identifiers.

[0092] The shortest reachable path search uses the fixed topology edges and candidate conducting edges in the baseline wiring as the search range. The path stops expanding when it passes a disconnected edge and generates a new state branch when it passes a pending edge. Paths that pass through fewer pending edges, can cover more actual response nodes, and do not form prohibited closed loops are retained as retention conditions. If a candidate connection change node can only explain a single isolated response and contradicts other responses in the same time layer, it will not be added to the cross-layer event cause-effect graph. If it can explain the undervoltage section, the current disappearance section, and the adjacent power direction change at the same time, it will be added and the missing switch change or intermediate section response will be recorded as a missing event identifier.

[0093] The addition of candidate connection relationship change nodes does not change the original event record, but adds a presumed node with source markers to the graph. The source markers include the corresponding shortest reachable path, the set of associated undetermined edges, the set of electrical response nodes that can be explained, and the required missing event type. If the corresponding shortest reachable path becomes invalid due to subsequent counterfactual replacement, the presumed node and its causal edges are deleted simultaneously. If a real record corresponding to the missing event appears in a subsequent time window, the presumed node is replaced with the real event node while the original node association is retained.

[0094] In a preferred embodiment, the generation of the candidate operating topology includes: constructing topology nodes and topology edges based on the bus sections, feeder sections, and switchgear in the baseline wiring relationship; and causally linking the remote signaling status with the cross-layer event. Figure 1 The corresponding switch topology edges are set as determined edges, and the switch topology edges that conflict with the remote signaling state and the cross-layer event causal graph are set as undetermined edges. The undetermined edges are assigned to conduction state and disconnection state respectively to form candidate combinations. Based on the mutual exclusion constraint of the same switch opening and closing state, the continuous constraint of the energized section and the open-loop operation constraint of the ring network, the candidate combinations that do not meet the constraints are deleted to obtain the candidate operating topology.

[0095] The determination of edges requires that the remote signaling state, the node with the change in connection relationship, and at least one type of primary electrical response are consistent, and there is no stable event opposite to it on the unified causal time axis. Undetermined edges include edges with missing remote signaling, edges with conflict between remote signaling and electrical response, switch edges with state jitter clusters and whose final state cannot explain the response, and switch edges located at the intersection of multiple candidate propagation paths. When assigning state values ​​to undetermined edges, a local combination generation method is adopted, and only undetermined edges that can change the reachability relationship of the observed response segment are combined. For undetermined edges not located on the relevant power supply path, the state of the previous identification time window is used.

[0096] In a preferred embodiment, the deducing of the electrical response sequence corresponding to each candidate operating topology includes: for each action event in the cross-layer event causal graph, determining the starting segment adjacent to the action switch in the candidate operating topology; calculating the set of disturbance reachable segments based on the direction and connection level of the conducting topology edges in the candidate operating topology; generating a response sequence according to the topological distance from the starting segment to each reachable segment; determining the expected change in energized state and power direction of each reachable segment based on the power supply connection relationship before and after the action; and combining the set of disturbance reachable segments, the response sequence, the expected change in energized state, and the power direction into a candidate electrical response sequence.

[0097] The order deviation between the candidate electrical response sequence and the actual response sequence is calculated according to the following formula:

[0098] ;

[0099] in, Indicates candidate running topology Response order deviation, This represents the set of response node pairs in the candidate topology that have a definite sequential relationship. This represents the fixed non-negative weight of the node for both p and q. This indicates that the value is 1 if the condition is true and 0 otherwise. and These represent the topological distances from the corresponding segments of response nodes p and q to the starting segment, respectively. and These represent the actual response times on the unified causal timeline. This represents the set of actual response nodes that cannot be reached by the candidate topology. This indicates a fixed number of collisions that cannot reach node r. For example, when the topological distance of the near segment is less than that of the far segment but the actual response is later than that of the far segment, the indicator function is set to 1 and an order collision is formed.

[0100] Please refer to Figure 4 , Figure 4 The horizontal axis represents the candidate running topology number. The vertical axis represents the order deviation between the candidate electrical response sequence and the actual response sequence. The black solid line represents the overall deviation, and the gray auxiliary curves represent the response sequence conflict items and unreachable response items, respectively. The response sequence conflict originates from... The node pair comparison in the data indicates that the unreachable response originates from... The actual set of response nodes in the system.

[0101] Figure 4 This shows the differences between different candidate running topologies. The differences are quite obvious. If the black curve corresponding to a certain candidate number is close to a low value, it means that the order of responses obtained by the initial segment propagating along the connected topology edge under the candidate topology is basically compatible with the actual response time recorded in the unified causal time axis. If the black curve forms a peak at individual candidates, it means that although the candidate may satisfy some static connectivity constraints, it cannot explain the physical propagation order of the responses of the near-end segment before the far-end segment.

[0102] In this embodiment, the topological distance is calculated based on the number of conducting switch edges that the action effect needs to pass through for propagation. Fixed conductive connections do not increase the topological distance. Response nodes within the same topological distance layer are not forced to establish a sequential relationship. Different distance layers with a single series path establish a definite sequential relationship. When there are multiple parallel paths and the propagation order cannot be determined from the reference wiring, only the reachable range and response direction are compared. This avoids identifying normal time fluctuations of parallel branches as causal conflicts.

[0103] In a preferred embodiment, the counterfactual replacement includes: sequentially selecting undetermined switches in the candidate operating topology, replacing the on state of the selected undetermined switches with an off state or vice versa, while keeping the states of other topological edges unchanged; regenerating a candidate electrical response sequence based on the replaced candidate operating topology; comparing the regenerated candidate electrical response sequence with the cross-layer event causal graph, and respectively counting electrical response nodes that cannot establish causal edges, node pairs that violate the order of responses, power supply path interruption locations, and power direction discontinuities; and forming the counterfactual conflict vectors of the corresponding undetermined switches using the statistical results.

[0104] The counterfactual conflict vector is represented by the following formula:

[0105] ;

[0106] in, Indicates candidate running topology The collision vector after the pending switch j completes its state replacement. This indicates the number of electrical response nodes for which causal edges cannot be established. This indicates the number of node pairs that violated the order of responses. Indicates the number of locations where the power supply path is interrupted. The number of discontinuous positions in the power direction is indicated. The four components are used only to describe different conflict types and do not replace each other. For example, when the vector changes from (2,1,0,0) to (0,0,0,0) after replacement, the replacement state is retained. If it changes to (0,0,1,0), it means that the local event is explained but the power supply reachability relationship is broken, and the corresponding replacement state is still not accepted.

[0107] When performing counterfactual replacement, pending switches are sorted according to their coverage in the candidate propagation path. Pending switches that cover multiple actual response segments are replaced first. After replacement, only the local connectivity components affected by the state of the switch and the corresponding electrical response sequence are recalculated. Unaffected topological edges, causal nodes and evidence records remain unchanged. If two pending switches are located on the same series path and individual replacement cannot eliminate the conflict, a binary replacement branch is established and the joint replacement results are compared. However, joint replacement is limited to the set of pending switches associated with the same missing event identifier or the same conflict edge.

[0108] In a preferred embodiment, determining the actual operating topology based on causal conflicts before and after replacement includes: establishing an evidence matrix for each candidate operating topology, consisting of rows of action events and columns of electrical response events, and recording the causal edge matching status, conflict edge positions, and missing event identifiers in the evidence matrix; mapping the counterfactual conflict vectors of each pending switch to the evidence matrix to determine the conflict increment caused by the state change of each pending switch; constructing topology consistency constraints according to the set of causal edges maintained by multiple independent action events; and selecting the candidate operating topology with the smallest total conflict increment and the fewest missing event identifiers among the candidate operating topologies that satisfy the topology consistency constraints.

[0109] Each row in the evidence matrix corresponds to one switch action event or protection action event, and each column corresponds to one voltage state change, current amplitude change, or power direction change event. Matrix elements record direct matching, indirect matching through connection relationship change nodes, missing event supplementation matching, timing conflict, path conflict, or unrelated states. When mapping counterfactual conflict vectors, the matrix element that caused the change due to the replacement of the pending switch is marked as the corresponding switch associated element, thereby determining the source of conflict change. Independent action events are distinguished according to the non-overlapping action time windows or different disturbance source segments.

[0110] Topology consistency constraints require that the same pending switch maintains the same stable state in multiple independent action events, unless there is a new change event for the corresponding switch in the unified causal time axis. They also require that the determined edges on the same power supply path not be changed by irrelevant action events, and that multiple causal edges supported by the same connection relationship be retained. If a candidate topology can only be interpreted by repeatedly changing the state of the same switch between different action events without corresponding change records, then the corresponding candidate topology is deleted.

[0111] In another preferred embodiment, the method further includes: associating the determined edges and undetermined edges in the selected candidate running topology with the causal edge matching records in the evidence matrix to form a topology edge evidence set; when multiple consecutive identification time windows obtain the same undetermined edge state, updating the corresponding undetermined edge to a determined edge; when the topology edge state change in the later identification time window is supported only by a single data channel event and conflicts with the topology edge evidence set of the previous identification time window, retaining the topology edge state of the previous identification time window and writing the single data channel event into the abnormal event set; and outputting the running topology identification result consisting of the determined edge state, the retained undetermined edge state, the topology edge evidence set, and the abnormal event set.

[0112] The topological edge evidence set must record at least the associated action events, nodes with changes in associated connection relationships, associated electrical response nodes, candidate propagation paths, conflict vectors before and after counterfactual substitution, and missing event identifiers. The evidence set for determining an edge must contain compatible records from different data types. The evidence set for undetermined edges retains records that support both the on-state and off-state. When the next identification time window arrives, the previous evidence set is read using the topological edge identifier as an index, and new records are appended without overwriting the original evidence source.

[0113] The state of topological edges in the continuous identification time window is updated according to the following formula:

[0114] ;

[0115] in, This represents the state support quantity of topological edge e in the nth recognition time window. The value is 1 if the current time window is the same as the previous time window, and 0 otherwise. This indicates whether a new causal chain exists, formed by independent action events; a value of 1 indicates existence, otherwise a value of 0. This indicates whether there is cross-source support between a primary electrical response and a secondary action event; a value of 1 indicates support if it exists, and a value of 0 indicates otherwise. This indicates whether a single data channel conflicts with the existing evidence set; a value of 1 indicates conflict, otherwise a value of 0. to They are respectively represented as , , and The corresponding fixed non-negative coefficients, for example, when the state is continuous and consistent and has cross-source common support, the corresponding value of the first 3 terms is 1. If a single channel conflict occurs at the same time, the 4th term is deducted. Only when the state support quantity continuously meets the solidification condition will the undetermined edge be updated to a determined edge.

[0116] The set of abnormal events is stored according to data channel identifier, event type, event time, associated topology edge, and conflict evidence. A single data channel event includes isolated remote signaling changes, protection records without corresponding electrical responses, electrical changes that occur only in a single interval and cannot propagate along the candidate topology, and time records that conflict with multiple anchor points in the unified causal time axis. If an abnormal event is supported by other data sources in subsequent time windows, it will change from an isolated abnormal state to a verifiable event and re-participate in candidate topology generation. If it continues to conflict with the set of topology edge evidence, it will only be used for channel anomaly tracking without changing the actual operating topology.

[0117] During actual operation, after receiving the incoming switch closed position record, tie switch open position record, adjacent feeder revoltage record, and power direction reversal record within a certain identification time window, the system first determines the relative time shift between the incoming switch action and the revoltage event through the event pairing set. Then, it establishes a causal chain between the incoming action, the restoration of power supply reachability, and the feeder revoltage. At the same time, it generates a candidate topology to maintain an open loop based on the tie switch open position status. If, counterfactually, setting the tie switch to the closed position would form an unacceptable closed loop and cause a conflict with the continuity of power direction, then the open position status is retained. If a feeder revoltage record is earlier than the incoming action but the physical order is restored after channel time shift correction, then it is not treated as a switch state error.

[0118] In another operation, after the system receives records of protection activation, outgoing switch isolation remote signaling, downstream current disappearance, and unchanged power direction in adjacent sections, the cross-layer event cause-effect graph associates protection activation with outgoing switch isolation action, limits the change in connection relationship to the loss of power supply reachability in the fault section, and takes the disappearance of downstream current as the electrical response. If the outgoing switch remote signaling is delayed by the auxiliary contact and lags behind the current disappearance, the relative time shift correction adjusts the corresponding channel according to the common anchor point of protection activation and current change. If the remote signaling continuously shows closed, the closed state in the candidate topology will generate a power supply reachability conflict with the current disappearance in the counterfactual comparison. The outgoing switch isolation state will be retained as a topology edge to be confirmed and the remote signaling conflict event will be recorded.

[0119] During operation with event omissions, the system only receives records of protection actions, multiple downstream section undervoltage, and upstream current reduction, but does not receive records of isolating switch changes. The cross-layer event cause-effect graph searches along the baseline wiring for the shortest reachable path between the protected object and the undervoltage section. The undetermined switch state combinations in the path are converted into candidate connection relationship change nodes. The combination that can explain multiple undervoltage nodes and upstream current reduction nodes is used as the supplementary node. The supplementary node carries a missing switch change identifier. If subsequent counterfactual replacement proves that other switch state combinations require fewer missing events and do not have path conflicts, the new connection relationship change node replaces the original supplementary node.

[0120] During the operation of multiple switches operating continuously, each action event enters the same identification time window according to a unified causal time axis. The candidate operating topology starts from the basic state before the action and applies the verifiable switch state changes in sequence. For actions with parallel time and uncertain order, multiple transition topology branches are established, and the electrical response sequence is deduced for each. The transition topology is only used to explain the short-time response during the action. The actual operating topology is determined by the branch that can maintain a stable energized state, power direction and open-loop constraints after the action is completed, so as to avoid mistaking the intermediate transition state as the final operating state.

[0121] In optional operation where data point mapping is abnormal, if the voltage change of a certain feeder interval continuously conflicts with the reference path corresponding to its equipment identifier, but maintains a stable causal relationship with the action events and power direction changes of adjacent intervals, the corresponding electrical response node will be retained as an unexplained response and associated with the abnormal event set. The data point mapping relationship will not be directly exchanged. Only when the same group of response nodes continuously shows cross-correspondence in multiple independent identification time windows and the counterfactual switch state replacement cannot eliminate the conflict will the data point mapping abnormality be output as external verification information. The actual operating topology is still determined based on the current reference wiring, event causal chain, and candidate topology constraints.

[0122] In this embodiment, data processing can be performed by a distribution master station, an edge computing terminal, or a software processing environment connected to the operation data of the integrated primary and secondary ring network box. The input data all come from the existing switch remote signaling, event sequence records, protection action information, and feeder interval voltage, current, and power direction records of the ring network box. The processing uses event records, graph structures, candidate state sets, evidence matrices, and abnormal event sets as data objects. It does not require changes to the primary equipment structure of the ring network box. The output results include the current actual operating topology, the determined or pending state of each topology edge, event evidence supporting the corresponding state, and abnormal event records that did not participate in the topology update.

Claims

1. A method for identifying the operating topology of a primary and secondary integrated ring network box, characterized in that, include: S1. Obtain remote signaling of switch opening and closing, event sequence records, protection action information, and voltage, current and power direction data of each feeder interval within the preset identification time window, and form a secondary action event sequence and a primary electrical response sequence respectively. S2. Based on the physical sequence constraints between switching actions and electrical responses, relative time shift correction is performed on different data channels to construct a unified causal time axis; S3. Based on the unified causal time axis, establish a cross-layer event causal graph between action events, changes in connection relationships, and electrical responses; S4. Generate candidate operating topologies based on the basic wiring relationship of the ring network box, and deduce the electrical response sequence corresponding to each candidate operating topology; S5. Match the deduction results with the cross-layer event causal graph, perform counterfactual replacement on the switch states in the candidate running topology, and determine the actual running topology based on the causal conflicts before and after the replacement.

2. The method for identifying the operating topology of a primary and secondary integrated ring network box according to claim 1, characterized in that, The relative time shift correction includes: extracting the switch opening and closing times from the secondary action event sequence, and extracting the current surge, undervoltage, voltage recovery, and power direction reversal times from the primary electrical response sequence; Establish an event pairing set based on the baseline connection relationship between the switch and the feeder interval; For each data channel, a time shift evaluation metric is constructed based on the number of event pairs that satisfy the physical sequence relationship in the event pairing set, the event interval dispersion, and the cross-interval sequence consistency. The channel time shift is determined while keeping the relative order of events within the same data channel unchanged, and the channel time shift is used to rearrange the secondary action event sequence and the primary electrical response sequence.

3. The method for identifying the operating topology of a primary and secondary integrated ring network box according to claim 1, characterized in that, The cross-layer event cause-effect graph includes an action event layer, a connection relationship change layer, and an electrical response layer; The action event layer uses switch opening and closing events and protection action events after relative time shift correction as nodes; the connection relationship change layer uses changes in the conduction state of candidate switches and changes in the reachability of corresponding feeder sections as nodes; the electrical response layer uses changes in voltage state, current amplitude, and power direction as nodes. Based on the device correspondence, the corrected event sequence, and the segment reachability, directed causal edges are established between adjacent layer nodes, and directed causal edges that do not satisfy the physical propagation order are marked as conflict edges.

4. The method for identifying the operating topology of a primary and secondary integrated ring network box according to claim 1, characterized in that, The generation of the candidate operating topology includes: constructing topology nodes and topology edges based on the bus section, feeder section and switchgear in the baseline wiring relationship; Set the switch topology edge whose remote signaling state is consistent with the cross-layer event causal graph as a determined edge, and set the switch topology edge whose remote signaling state conflicts with the cross-layer event causal graph as an undetermined edge. The undetermined edges are assigned to either an on state or an off state to form candidate combinations; Based on the mutual exclusion constraints of the same switch opening and closing states, the continuous constraints of energized sections, and the open-loop operation constraints of the ring network, candidate combinations that do not meet the constraints are deleted to obtain the candidate operating topology.

5. The method for identifying the operating topology of a primary and secondary integrated ring network box according to claim 2, characterized in that, The relative time shift correction further includes: dividing the event pairing set of the same data channel into multiple time segments according to the continuous identification time window; Calculate the local time shift for each timing segment, and determine the channel clock drift direction based on the time difference of common event pairs in adjacent timing segments; When adjacent local time shifts are discontinuous, the data channel is divided into multiple correction intervals by using events that occur simultaneously in multiple data channels during protection actions, switch changes, and undervoltage events as segmentation anchor points. Within each correction interval, events are rearranged using independent time shifts, while maintaining the correspondence between the segmented anchor points in each data channel.

6. The method for identifying the operating topology of a primary and secondary integrated ring network box according to claim 1, characterized in that, The establishment of the cross-layer event cause-effect graph also includes: when there is no connection relationship change layer node that satisfies the timing constraints between the action event layer node and the electrical response layer node, searching along the reference wiring relationship for the shortest reachable path connecting the switch corresponding to the action event layer node and the feeder section corresponding to the electrical response layer node; The undetermined switch state combinations in the shortest reachable path are converted into candidate connection relationship change nodes; The interpretable range of other electrical response layer nodes is filtered according to the candidate connection relationship change nodes; Candidate connection relationship change nodes that can simultaneously connect multiple independent electrical response nodes are added to the cross-layer event causal graph, and their corresponding missing event identifiers are recorded.

7. The method for identifying the operating topology of a primary and secondary integrated ring network box according to claim 4, characterized in that, The deduction of the electrical response sequence corresponding to each candidate operating topology includes: for each action event in the cross-layer event causal graph, determining the starting segment adjacent to the action switch in the candidate operating topology; The set of perturbation reachable segments is calculated based on the direction and connection level of the conductive topology edges in the candidate running topology; Generate a sequence of responses based on the topological distances from the starting segment to each reachable segment; The expected changes in energized state and power direction of each reachable section are determined based on the power supply connection relationship before and after the action. The set of disturbance reachable segments, the sequence of responses, the expected change in charged state, and the change in power direction are combined to form a candidate electrical response sequence.

8. The method for identifying the operating topology of a primary and secondary integrated ring network box according to claim 7, characterized in that, The counterfactual replacement includes: sequentially selecting undetermined switches in the candidate running topology, replacing the on state of the selected undetermined switches with the off state or the off state with the on state, while keeping the states of other topological edges unchanged; The candidate electrical response sequence is regenerated based on the replaced candidate operating topology; The regenerated candidate electrical response sequence is compared with the cross-layer event causal graph, and the electrical response nodes that cannot establish causal edges, the node pairs that violate the response order, the power supply path interruption locations, and the power direction discontinuity locations are counted respectively. The statistical results are used to construct a counterfactual conflict vector corresponding to the undetermined switch.

9. The method for identifying the operating topology of a primary and secondary integrated ring network box according to claim 1, characterized in that, Determining the actual operating topology based on causal conflicts before and after replacement includes: establishing an evidence matrix consisting of action event rows and electrical response event columns for each candidate operating topology, and recording the causal edge matching status, conflict edge position, and missing event identifier in the evidence matrix; The counterfactual conflict vectors of each pending switch are mapped to the evidence matrix to determine the conflict increment caused by the state change of each pending switch. Construct topological consistency constraints based on the set of causal edges maintained together across multiple independent action events; Among the candidate running topologies that satisfy the topology consistency constraint, the candidate running topology with the smallest total conflict increment and the fewest number of missing event identifiers is selected.

10. The method for identifying the operating topology of a primary and secondary integrated ring network box according to claim 9, characterized in that, Also includes: The determined edges and undetermined edges in the selected candidate running topology are respectively associated with the causal edge matching records in the evidence matrix to form a topology edge evidence set; When multiple consecutive recognition time windows obtain the same undetermined edge state, the corresponding undetermined edge will be updated to a determined edge; When the change in the topological edge state in the subsequent identification time window is supported only by a single data channel event and conflicts with the set of topological edge evidence in the previous identification time window, the topological edge state of the previous identification time window is retained, and the single data channel event is written into the abnormal event set. The output consists of the determined edge states, the reserved undetermined edge states, the set of topological edge evidence, and the set of abnormal events.

Images