A power distribution network fault research and judgment method based on multi-source data fusion

Abstract

The application discloses a power distribution network fault research and judgment method based on multi-source data fusion and relates to the technical field of power systems and power distribution automation.In the application, first, a topology version sequence sorted by time is generated according to switch action events and dispatching control events, then candidate topology versions are screened according to associated object identifiers and time identifiers, and state compatibility determination is performed on evidence fragments according to power supply connection relationships, switch states and outage and restoration states, and meanwhile, evidence fragments meeting version switching window and influence range constraints are classified into adjacent topology versions; switch actions, outage information, restoration power supply information and electrical quantity changes arriving at different disposal stages are respectively included in corresponding power supply relationships to participate in judgment, power loss evidence formed before fault isolation is no longer continuously pressed into the network state after transfer supply, and the restoration power supply influence formed after transfer supply is no longer explained as a new fault trigger.

Description

Technical Field

[0001] This invention relates to the field of power system and distribution automation technology, and in particular to a method for fault assessment of distribution networks based on multi-source data fusion. Background Technology

[0002] After a fault occurs in the distribution network, the main substation typically combines sectionalizing switches, tie switches, and feeder automation strategies to perform fault isolation and power restoration. Regarding fault assessment, existing methods usually utilize protection action information, switch position information, remote signaling and telemetry information, electrical quantity changes, and power outage / restoration information to determine the fault section and power supply range. In scenarios with a single fault and relatively stable power supply relationships, these methods can complete fault location and power outage range analysis.

[0003] However, in scenarios involving continuous execution of interconnection and self-healing reconfiguration, sectionalizing switches, interconnecting switches, and circuit breakers may operate consecutively within a short period, causing a switch in the power supply path. Furthermore, the arrival times of master station commands, terminal transmissions, electrical quantity changes, and power outage / restoration information may be inconsistent, with message delays, unstable connections, and partial omissions. If data from different batches is directly aggregated into the network status at the same moment for judgment, it is easy to include evidence of power loss formed before fault isolation and evidence of power restoration formed after power restoration in the same power supply relationship. This can lead to the faulty section being mistakenly expanded, the restored section being repeatedly judged as abnormal, or the power supply status of the same object repeatedly switching in different judgment results.

[0004] While existing methods can process some abnormal data through event sorting, anomaly filtering, alarm deduplication, or real-time topology analysis, they remain insufficient in distinguishing the relationships between multiple sources of evidence before and after topology switching. In particular, they lack mechanisms for stably collecting, resolving conflicts, and determining consistency of evidence at different stages during continuous changes in power supply relationships. Consequently, when fault isolation, interconnection and power restoration overlap, situations still arise where old evidence continues to be used in subsequent assessments, or restored evidence is interpreted as new anomalies, affecting the stability of fault assessment results and the verifiability of the basis for handling the situation.

[0005] Therefore, how to distinguish the multi-source evidence corresponding to different topology stages during the continuous switching of power supply relationships, and form a consistent fault judgment conclusion based on the corresponding relationship, has become an urgent technical problem to be solved. Summary of the Invention

[0006] This application provides a method for fault assessment of power distribution networks based on multi-source data fusion, which solves the problem of inconsistent fault assessment caused by the cross-stage mixing of multi-source evidence when power supply relationships are continuously switched.

[0007] This invention provides a method for fault assessment in distribution networks based on multi-source data fusion, applicable to at least one scenario among post-fault interconnection and self-healing reconfiguration scenarios, including: Acquire operational topology data, switch action event data, dispatch control event data, electrical quantity monitoring data, and power outage / restoration status data; Multi-source data is broken down into evidence fragments based on associated objects, event types, and time boundaries; Generate a topology version sequence ordered by time based on switching action events and scheduling control events that cause changes in the power supply path; Candidate topology versions are filtered by associated object identifier and time identifier, and state compatibility is determined by power supply connection relationship, switch status and power outage / restoration status. Compatibility evidence is assigned to the corresponding topology version. Evidence that is incompatible under the current candidate topology version but compatible under adjacent topology versions that meet the version switching window and influence range constraints is assigned to the adjacent topology version. The rest are written into the conflict evidence set. Evidence fragments assigned to the same topology version but with inconsistent representation results are sorted by source and checked by time sequence. In each topology version, candidate fault segments are generated based on power connection relationships, switch status, changes in power outage range, and changes in power restoration range. A candidate fault event chain is then constructed around the candidate fault segments according to the equipment connection sequence and the event occurrence sequence. Compare the version consistency and matching level of each candidate fault event chain to determine the target topology version and target fault segment, and output the fault assessment result including the evidence fragments used and the handling stage identifier.

[0008] In some embodiments, the operational topology data includes feeder connection relationships, sectionalizing switch status, tie switch status, and power access relationships; The switch action event data includes protection action events, circuit breaker action events, sectionalizing switch action events, and tie switch action events; The scheduling and control event data includes fault isolation records, communication transfer records, and self-healing reconfiguration records; The electrical quantity monitoring data includes voltage, current, power, and waveform summary information; The power outage and restoration status data includes terminal power outage information, user power outage and restoration information, and power restoration information.

[0009] In some embodiments, the evidence fragment is the smallest associatable data unit formed by splitting according to the associated object, event type and time boundary, and each evidence fragment records the source identifier, time identifier, status identifier and associated object identifier; The evidence fragments include at least a single switch status change record, a single dispatch control record, a single device electrical status change record, and a single area power outage / restoration status change record.

[0010] In some embodiments, when generating the topology version sequence, events that cause changes in the power supply path are used as version switching events, including the opening and closing of sectionalizing switches, the opening and closing of tie switches, the opening and closing of circuit breakers, and power supply path switching caused by scheduling control. The fault handling process is divided into multiple handling stages according to the occurrence time of the version switching event, and a corresponding topology version identifier is generated for each handling stage.

[0011] In some embodiments, when performing source sorting and timing verification on conflict evidence within the same topology version, the evidence adoption priority is determined according to the order of switch action event data, electrical quantity monitoring data, dispatch control event data and power outage / restoration status data, and the event sequence before and after the event is verified according to the time identifier to ensure that the event sequence of isolation before power transfer and power loss before power restoration is met. Evidence fragments that do not conform to the stated order of events are treated as conflicting evidence.

[0012] In some embodiments, when generating the candidate fault segments, the power outage boundary segment is determined based on the change in the power outage range, the power transfer boundary segment is determined based on the power restoration range, and the isolation boundary segment is determined based on the change in the switch state. The power outage boundary segment, the power transfer boundary segment, and the isolation boundary segment are then superimposed in the current topology version to obtain a set of candidate fault segments.

[0013] In some embodiments, when constructing the candidate fault event chain, the switching action events, scheduling control events, electrical quantity change records, and power outage / restoration status change records surrounding the same candidate fault segment are connected in the order of equipment connection from the power supply side to the candidate fault segment, and then from the candidate fault segment to the power outage area or power restoration area under the current topology version, as well as the order of occurrence of fault-related events, isolation-related events, power transfer-related events, and power restoration-related events on a unified time axis to form a candidate fault event chain. For each candidate fault event chain, a chain identifier, the topology version identifier, the associated candidate fault segment identifier, the sequence of evidence fragment identifiers within the chain, the first event time, the last event time, and the matching level are recorded.

[0014] In some embodiments, when determining the matching level of the candidate fault event chain, the determination is based on the matching results of the power outage range and the isolation range, the matching results of the power restoration range and the interconnection and transfer path, the matching results of the electrical quantity change sequence and the switch action sequence, and the evidence of conflict within the chain. When multiple candidate fault event chains have the same matching level, the candidate fault event chain with fewer conflicting evidence is selected first. When the number of conflicting evidence is the same, the candidate fault event chain with more supporting evidence fragments covering the upstream boundary equipment, downstream boundary equipment, tie switch and power restoration boundary equipment of the candidate fault section is selected first.

[0015] In some embodiments, the fault assessment result further includes version tracking records, which include the target topology version identifier, the triggering event for switching from the previous topology version to the target topology version, the target fault segment, the set of evidence fragments used, and the set of conflicting evidence.

[0016] In some embodiments, when a new evidence fragment related to the target fault section is received after the fault assessment result is output, and the new evidence fragment is incompatible with the output fault assessment result, a topology version remapping is performed on the new evidence fragment and the candidate fault event chain is reconstructed. When the matching level of the reconstructed candidate fault event chain is higher than that of the adopted candidate fault event chain, update the target topology version and the target fault segment, and record the version differences before and after the update.

[0017] Through the above technical solution, the present invention can achieve at least the following beneficial effects: In this invention, in post-fault communication and power transfer scenarios and self-healing reconfiguration scenarios, a topology version sequence ordered by time is first generated based on switch action events and scheduling control events. Then, candidate topology versions are filtered by associated object identifiers and time identifiers. Evidence fragments are then evaluated for compatibility based on power supply connectivity, switch status, and power outage / restoration status. Simultaneously, evidence fragments satisfying version switching window and impact range constraints are categorized into adjacent topology versions. Thus, switch actions, power outage information, power restoration information, and electrical quantity changes arriving at different handling stages are incorporated into the corresponding power supply relationships for judgment. Power loss evidence formed before fault isolation is no longer continuously pushed into the post-transfer network state, and the power restoration impact formed after transfer is no longer interpreted as a new fault trigger. This suppresses cross-stage mixing, erroneous expansion of fault sections, and repeated fault judgments in restored sections caused by delayed evidence arrival due to topology switching.

[0018] For evidence fragments belonging to the same topology version but with inconsistent characterization results, source sorting and timing verification are performed. Candidate fault segments are generated under each topology version based on power connection relationships, switch states, changes in power outage range, and changes in power restoration range. Then, a candidate fault event chain is constructed around the candidate fault segments according to the equipment connection sequence and event occurrence sequence. As a result, protection actions, sectionalizing switch opening and closing, tie switch closing, power outage range, and power restoration range form a physical correspondence under the same version. Old remote signaling, delayed power outage information, and local jitter codes no longer participate in the fault chain construction with equal weight as subsequent stable states. This ensures that candidate fault segments are within the range jointly defined by isolation boundaries, power outage boundaries, and power transfer boundaries, reducing the possibility of power outage propagation, load transfer, and power restoration being mixed into the same abnormal chain.

[0019] By comparing the version consistency and matching level of candidate fault event chains, the system outputs fault assessment results including the target topology version, target fault segment, evidence fragments used, and handling stage identifiers. Remapping and reconstruction are performed when new evidence arrives. Thus, the assessment results maintain a one-to-one correspondence with the corresponding topology stage, used evidence, and conflicting evidence. The main station can distinguish between handling results where the original conclusion remains valid, the topology version is corrected due to new evidence, and the target fault segment is corrected due to new evidence. This ensures that the assessment basis during communication transfer and self-healing reconstruction processes remains verifiable, traceable, and continuously updatable. Even in situations involving continuous switching, multiple arrivals, and partial missing data, a convergent fault assessment result can still be output along a unified handling chain. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation on the scope of this application.

[0021] Figure 1 This is a flowchart of the power distribution network fault assessment method based on multi-source data fusion in the embodiments. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0023] All terms used in this application (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0024] To facilitate understanding, the relevant terms and concepts involved in the embodiments are introduced below: A topology version refers to a snapshot of the power supply connectivity determined by the current combination of switch states; the processing phase refers to the time interval between two adjacent version switching events; an evidence fragment refers to the smallest associatable data unit formed by splitting it according to the associated object, event type, and time boundary; a candidate topology version refers to the set of topology versions that can be used for attribution judgment based on the associated object and time position of the evidence fragment; conflicting evidence refers to evidence fragments that cannot simultaneously satisfy the power supply connectivity, switch state, and power outage / restoration state under the candidate topology versions; and the version tracking record refers to the set of version switching, evidence adoption, and conflict elimination information saved corresponding to the target topology version. Furthermore, the associated object refers to the device or area corresponding to the evidence fragment, and the associated object identifier is a record field used to uniquely identify the device or area.

[0025] A candidate fault segment refers to the range of line segments used for fault diagnosis, obtained by overlaying power outage boundary segments, power transfer boundary segments, and isolation boundary segments under the corresponding topology version. A candidate fault event chain refers to a sequence of evidence fragments connected around the same candidate fault segment according to the equipment connectivity order and event occurrence order. A version switching window refers to the time range within which evidence is allowed to be attributed across versions around a version switching event. The scope of impact refers to the range of equipment or areas whose power supply relationships are altered by a version switching event.

[0026] A unified timeline refers to a time base formed by aligning the main station time, terminal time, scheduling control time, and power outage / restoration record time. This definition serves as a consistent standard for subsequent evidence fragment attribution, candidate fault segment generation, candidate fault event chain comparison, and version tracking record retention. Version consistency refers to the degree of consistency between the topology version to which a candidate fault event chain belongs and the topology version to which the evidence fragments within the chain belong.

[0027] In addition, state compatibility determination refers to the determination of whether the evidence fragments satisfy the consistency of power supply connection relationship, switch state and power outage / restoration state under the corresponding topology version.

[0028] In some embodiments, the master station organizes and operates topology data, switch action event data, dispatch control event data, electrical quantity monitoring data, and power outage / restoration status data according to a unified device identifier and a unified timeline. It also performs evidence fragment splitting, topology version generation, evidence attribution verification, candidate fault segment generation, candidate fault event chain comparison, and fault judgment result output in the order of fault occurrence, fault isolation, interconnection and power supply transfer, and power restoration, so as to avoid data from different handling stages being merged into the same power supply relationship for judgment.

[0029] Example 1: like Figure 1As shown in this embodiment, a method for distribution network fault assessment based on multi-source data fusion is provided. This method is used in at least one scenario, including interconnection and transfer scenarios and self-healing reconfiguration scenarios, to perform evidence fragment segmentation, topology version generation, evidence attribution verification, candidate fault segment generation, and candidate fault event chain comparison on multi-source data to determine the target topology version and target fault segment. The method includes: Step S1: Obtain running topology data, switch action event data, scheduling control event data, electrical quantity monitoring data, and power outage / restoration status data; Step S2: Segment the multi-source data into evidence fragments according to the associated object, event type, and time boundary; Step S3: Generate a topology version sequence ordered by time based on the switching action events and scheduling control events that cause changes in the power supply path; Step S4: Filter candidate topology versions by associated object identifier and time identifier, and determine state compatibility based on power supply connection relationship, switch status and power outage / restoration status. Assign compatibility evidence to the corresponding topology version. Assign evidence that is incompatible under the current candidate topology version but compatible under adjacent topology versions that meet the version switching window and influence range constraints to the adjacent topology version. Write the rest into the conflict evidence set. Perform source sorting and time sequence verification on evidence fragments assigned to the same topology version but with inconsistent representation results. Step S5: Under each topology version, candidate fault segments are generated based on power connection relationship, switch status, changes in power outage range and changes in power restoration range, and candidate fault event chains are constructed around the candidate fault segments according to the equipment connection sequence and the event occurrence sequence. Step S6: Compare the version consistency and matching level of each candidate fault event chain, determine the target topology version and the target fault segment, and output the fault assessment result containing the evidence fragments used and the handling stage identifier. In this embodiment, candidate topology versions are determined by filtering based on the associated object identifier and the time identifier. During filtering, topology versions containing the associated object are first selected as the spatial candidate set, and then topology versions whose time identifiers fall within the start and end time range of the corresponding topology version or fall within the version switching window of the corresponding version switching boundary are selected as the temporal candidate set. The intersection of the spatial candidate set and the temporal candidate set is taken as the candidate topology version set.

[0030] Specifically, the main station generates a time alignment record table based on a unified timeline. This table records the time correction results of various original records according to the original time, alignment time, time source, alignment deviation, and alignment status. During alignment, priority is given to using the switch action confirmation time, scheduling control execution feedback time, electrical quantity sampling time, and power outage / restoration time. If the original record lacks an occurrence time, the upload time and storage time are used as substitute times, and these are marked as substitute times in the alignment status. The current candidate topology version refers to the topology version that matches both the associated object and time position of the evidence fragment. After filtering, a version attribution index is formed. This index records the attribution candidate results according to the evidence fragment identifier, candidate topology version identifier, time distance, and consistency of the handling stage, and is used for subsequent state compatibility determination and parallel resolution.

[0031] In this embodiment, state compatibility refers to the state represented by the evidence fragment being valid under the corresponding topology version. State compatibility includes power supply connectivity compatibility, switch state compatibility, and power outage / restoration state compatibility. Power supply connectivity compatibility is used to determine whether the associated object has a valid power supply path to the reachable power source under the current topology version; switch state compatibility is used to determine whether the states of the upstream disconnecting switches, sectionalizing switches, tie switches, and circuit breakers directly related to the validity of the evidence fragment allow the evidence fragment to be valid; power outage / restoration state compatibility is used to determine whether the associated area should be in a de-energized state, a restored energized state, or a continuously energized state under the current topology version. It can be understood that power supply connectivity compatibility calculates the valid power supply path according to the closed switch connection relationship and power access relationship under the current topology version, and determines whether the associated object has the conditions for receiving power accordingly; switch state compatibility is mainly based on actual position change records, supplemented by dispatch control execution feedback, and when control has been issued but there is no execution feedback, it is only used as a record to be verified; power outage / restoration state compatibility is jointly determined according to terminal power outage information, user power outage / restoration information, restored energization information, and electrical quantity change records. The continuous energized state is confirmed using the in-service threshold, which refers to the continuous voltage and current observation boundaries used to confirm that the associated object is in a normal energized state. The in-service threshold is met when the voltage is continuously higher than 70%–85% of the rated phase voltage, and the current is continuously higher than 2%–10% of the rated current of the associated object or 5%–10% of the median load current during the same period in the past 30 days. If there is information about voltage loss, continuous current loss, or power outage, the continuous energized state is no longer assigned. For cases where only voltage records are available but continuous current records are lacking, power outage / restoration compatibility is only used as an auxiliary criterion and does not independently overturn boundary judgments already established by power supply connectivity compatibility or switch state compatibility.

[0032] When an evidence fragment is incompatible in the current candidate topology version but compatible in an adjacent topology version, and its timestamp falls within the version switching window of the corresponding version switching boundary and its associated object is within the corresponding influence range, the evidence fragment is assigned to the adjacent topology version. When an evidence fragment is compatible in multiple candidate topology versions, the topology version with the smaller timestamp distance and consistent handling stage is selected as the assignment result. When the timestamp distance is the same, the assignment result is determined according to the order of switch action event data, electrical quantity monitoring data, dispatch control event data, and power outage / restoration status data. When an evidence fragment is incompatible in all candidate topology versions, or is compatible in an adjacent topology version but does not meet the version switching window and influence range constraints, the evidence fragment is written into the conflict evidence set.

[0033] In this embodiment, the conflict evidence set records at least the evidence fragment identifier, source identifier, associated object identifier, conflict type, corresponding candidate topology version set, and conflict determination time. The conflict types include at least power supply connectivity conflicts, switch status conflicts, power outage / restoration status conflicts, and cross-version attribution conflicts. Specifically, each record in the conflict evidence set may also include an original record identifier, initial determination time, most recent review time, corresponding handling stage identifier, and review status identifier. The review status identifier indicates whether the conflict evidence has been re-involved in version attribution determination in subsequent handling stages. Evidence fragments in the conflict evidence set do not directly participate in matching level improvement in the current round of candidate fault event chain comparison, but maintain association with the original record identifier and version tracking record. When a switch action event, electrical quantity change record, or power outage / restoration status change record corresponding to the conflict evidence is subsequently received, and the newly added record can eliminate the original conflict type, the main station re-executes candidate topology version screening and status compatibility determination based on the original record identifier, while retaining the results of both determinations.

[0034] In an optional implementation of Example 1, the running topology data includes feeder connection relationships, sectionalizing switch status, tie switch status, and power access relationships. Switch action event data includes protection action events, circuit breaker action events, sectionalizing switch action events, and tie switch action events; Dispatch control event data includes fault isolation records, communication and supply transfer records, and self-healing and reconfiguration records; Electrical quantity monitoring data includes voltage, current, power, and waveform summary information; Power outage and restoration status data includes terminal power outage information, user power outage and restoration information, and power restoration information; The operational topology data should include at least the line segment identifier, node identifier, upstream equipment identifier, downstream equipment identifier, switch status, power supply identifier, and acquisition time; the switch action event data should include at least the event identifier, equipment identifier, action type, action result, action time, and source identifier; the dispatch control event data should include at least the control record identifier, controlled object identifier, control type, issuance time, execution feedback time, and execution status; the electrical quantity monitoring data should include at least the monitoring point identifier, sampling time, voltage value, current value, power value, and waveform summary identifier; and the power outage / restoration status data should include at least the area identifier, associated object identifier, status type, occurrence time, and recovery time.

[0035] For example, waveform summary information refers to the set of summary fields directly related to fault assessment extracted from waveform records. This set of summary fields includes at least the waveform start time, phase identifier, voltage surge, current surge, power direction change identifier, and corresponding monitoring point identifier. Control execution status refers to the execution result identifier of the dispatch control record on the master station side, including at least "issued," "executed," "execution failed," and "execution unknown." To ensure that data from different sources can be correlated and located within the same power supply network, all types of records are stored using unified equipment identifiers, unified line segment identifiers, and unified area identifiers, maintaining a one-to-one correspondence with the original message based on the original record identifiers.

[0036] The device identifiers in the aforementioned multi-source data are mapped to feeders, line segments, sectionalizing switches, tie switches, circuit breakers, and user areas using a unified mapping relationship, so that data from different sources can be associated and located within the same power supply network.

[0037] In one optional implementation of Example 1, the evidence fragment is the smallest associatable data unit formed by splitting according to the associated object, event type and time boundary, and each evidence fragment records the source identifier, time identifier, status identifier and associated object identifier; The evidence fragments should include at least a single switch status change record, a single dispatch control record, a single equipment electrical status change record, and a single area power outage / restoration status change record; The time boundary refers to the time cutoff point that triggers the splitting of evidence segments. These time cutoff points include the moment of switch state change, the moment of dispatch control issuance, the moment of dispatch control execution feedback, the moment of electrical quantity abrupt change, and the moment of power outage or power restoration state change. For a single message containing multiple device states, it is split into multiple evidence segments according to each device. For the same device maintaining the same state within a continuous time period, this continuous time period is recorded as a continuous state evidence segment. For the same device repeatedly sending the same state within a short period, the first sending time and the last confirmation time are retained to form one evidence segment.

[0038] In this embodiment, each evidence fragment records at least a fragment identifier, a source identifier, a data category, an associated object identifier, a time identifier, a status identifier, and an original record identifier. The time identifier includes the event occurrence time or the valid time interval of the status. The status identifier is used to characterize one or more of the following states: split, closed, energized, de-energized, restored to energized state, control issuance, and control execution result.

[0039] In an optional implementation of Example 1, when generating the topology version sequence, events that cause changes in the power supply path are used as version switching events. Version switching events include the opening and closing of sectionalizing switches, the opening and closing of tie switches, the opening and closing of circuit breakers, and power supply path switching caused by scheduling control. The fault handling process is divided into multiple handling stages according to the occurrence time of the version switching event, and a corresponding topology version identifier is generated for each handling stage. The master station uses the real-time power supply relationship before the fault occurred as the initial topology version, and generates subsequent topology versions sequentially based on version switching events ordered by time. If multiple version switching events are received at the same time, the system first confirms whether the power supply path has changed according to the actual switch status changes, and then writes the events that have indeed caused changes in the power supply path into the corresponding topology version.

[0040] Each topology version records at least the version identifier, the previous version identifier, the effective time, the failure time, the line segment connectivity, the switch status set, and the power supply connection relationship; each handling stage records at least the stage identifier, the start time, the end time, and the corresponding topology version identifier. This forms a topology version sequence arranged continuously along the time direction to characterize the evolution of the power supply relationship before fault isolation, during isolation execution, during interconnection and power transfer, and after power restoration.

[0041] For example, the version switching window width, as an implementation parameter setting, is primarily determined based on the time resolution and transmission delay envelope of the primary data source corresponding to the version switching event. The default value is 120s, with an adjustable range of 30~300s. A smaller value is used when station-side sequential events are used as the primary data source, and a larger value is used when relying mainly on terminal uploads or user power outage / restoration information. The scope of impact is determined according to the power supply path changed by the version switching event: sectionalizing switch tripping corresponds to its downstream isolated section; tie switch closure corresponds to the tie transfer path and its restored power supply area; circuit breaker operation corresponds to the section between its downstream and the nearest isolation boundary; and power supply path switching caused by dispatch control corresponds to the isolated section and transfer section covered by the dispatch control record. The version switching window and scope of impact are frozen within the same topology version sequence and are only refreshed when the topology version sequence is regenerated to avoid drift in attribution criteria during the comparison of candidate topology versions in the same round.

[0042] In an optional implementation of Example 1, when performing source sorting and timing verification on conflict evidence within the same topology version, the evidence adoption priority is determined according to the order of switch action event data, electrical quantity monitoring data, dispatch control event data and power outage / restoration status data, and the event sequence before and after the event is verified according to the time identifier to ensure that the event sequence of isolation before power transfer and power loss before power restoration is met. Evidence fragments that do not conform to the order of events are treated as conflicting evidence. Switch action event data directly characterizes equipment status changes, electrical quantity monitoring data characterizes electrical status changes before and after a fault, dispatch control event data characterizes the master station's handling actions, and power outage / restoration status data characterizes the power supply results on the user side or regional side. Therefore, the evidence is executed in the order of switch action event data, electrical quantity monitoring data, dispatch control event data, and power outage / restoration status data, and priority is used for judgment.

[0043] When there is a discrepancy in the time signature, the master station time, terminal time, and power outage / restoration record times are first aligned according to a unified timeline. If, after alignment, contradictions still exist in the state sequence of the same device, the evidence fragment with the later time and compatible with the current topology version is retained, while the evidence fragment with the earlier time and incompatible is added to the conflict evidence set. For cases where the same device repeatedly transmits split and merge data within a short period, evidence fragments are formed according to the final stable state, and intermediate jitter records are treated as conflict evidence.

[0044] Specifically, when aligning the timeline, the master station time and scheduling control time are recorded at the millisecond level, while the terminal time and power outage / restore recording time are recorded at the second level. When the time granularity of different sources is inconsistent, the coarser time granularity is rounded up to the same second or 10-second time slice for alignment. The time slice width is a default implementation parameter of 1 second, which can be adjusted to 10 seconds when there is a large amount of retransmission from the terminal. Records with an alignment deviation exceeding half of the corresponding version switching window are marked as late records. Late records do not directly change the matching level of the current round of candidate fault event chains; they only participate in the comparison again when rebuilding the candidate fault event chain in subsequent rounds. When multiple state values ​​of the same associated object appear in the same time slice, the master station prioritizes retaining records that are compatible with the current topology version and have a more direct time source. The remaining records are retained as evidence of conflict to prevent the repeated absorption of contradictory state information in the same time slice.

[0045] In an optional implementation of Example 1, when generating candidate fault sections, the power outage boundary section is determined based on the change in the power outage range, the power transfer boundary section is determined based on the power restoration range, and the isolation boundary section is determined based on the change in the switch state. The power outage boundary section, the power transfer boundary section, and the isolation boundary section are then superimposed in the current topology version to obtain a set of candidate fault sections. The power loss boundary section refers to the line segment that transitions from a energized area to a power loss area in the current topology version. The power transfer boundary section refers to the line segment that is restored to power by the tie transfer path in the current topology version. The isolation boundary section refers to the line segment defined by the disconnected sectionalizing switch, tie switch or circuit breaker.

[0046] The candidate fault segment set includes at least the segment identifier, the topology version identifier, the upstream boundary device identifier, the downstream boundary device identifier, the associated power outage boundary segment identifier, the associated power transfer boundary segment identifier, the associated isolation boundary segment identifier, and a set of supporting evidence fragment identifiers. For multi-branch feeder scenarios, the power supply connectivity of each branch is first calculated according to the current topology version. Then, the power outage boundary segments, power transfer boundary segments, and isolation boundary segments on each branch are identified separately. Finally, the segments are overlaid under the same topology version to avoid erroneous cross-branch connections.

[0047] Based on the above analysis, when power outage boundary sections, power transfer boundary sections, and isolation boundary sections overlap, the upstream and downstream boundaries of candidate fault sections are first defined by the isolation boundary section, and then the section range is corrected by the power outage boundary section and the power transfer boundary section. When only a power outage boundary section exists and the power transfer boundary section is missing, the candidate fault section remains the intersection of the isolation boundary section and the power outage boundary section; when only a power transfer boundary section exists and the power outage boundary section is missing, the candidate fault section remains the compensation range of the isolation boundary section and the power transfer boundary section; when all three types of boundary sections are missing, no new candidate fault sections are generated, and the candidate fault sections generated in the previous handling stage that have not been eliminated are retained for subsequent verification. To ensure feasibility in multi-branch feeder scenarios, each boundary section records its upstream boundary equipment identifier, downstream boundary equipment identifier, and branch identifier according to the line segment identifier. The branch identifier is a record field used to distinguish different power supply branches under the same topology version.

[0048] In an optional embodiment of Example 1, the fault assessment result further includes a version tracking record, which includes the target topology version identifier, the triggering event for switching from the previous topology version to the target topology version, the target fault segment, the set of evidence fragments used, and the set of conflicting evidence. Version tracking records include at least the record identifier, target topology version identifier, previous topology version identifier, version switch event identifier, target fault segment identifier, set of adopted evidence fragment identifiers, set of conflict evidence identifiers, handling stage identifier, analysis output time, and analysis result status. Version tracking records are associated with the original record identifiers to maintain the correspondence between the fault analysis result and the original message and event. Furthermore, version tracking records can establish a bidirectional index relationship based on the record identifier, the original record identifier, the evidence fragment identifier, and the topology version identifier to support tracing back from the fault analysis result to the original message, from the original message to the adopted result, and from the conflict elimination result. To ensure the update process is verifiable, version tracking records can also store the current round of candidate fault event chain identifiers, the previous round of candidate fault event chain identifiers, the update trigger time, and the update reason description; the update reason description is used to record changes in version attribution, candidate fault segment changes, or matching level changes triggered by newly added evidence fragments. In cases where the same target fault segment is updated multiple times, the main station retains all version tracking records in the order of analysis output time, without overwriting historical records.

[0049] When the main station subsequently receives new evidence fragments related to the target fault section, and the new evidence fragments are incompatible with the output fault assessment results, it first performs candidate topology version screening and version attribution verification on the new evidence fragments, and then replaces or supplements the corresponding positions in the original candidate fault event chain with the new evidence fragments, reconstructs the candidate fault event chain, and recalculates the matching level; when the matching level of the reconstructed candidate fault event chain is higher than the matching level of the original candidate fault event chain, it updates the target topology version, target fault section, and version tracking record, and writes the differences in topology version, evidence fragments, and assessment results before and after the update into the version tracking record.

[0050] In an optional implementation of Example 1, when a new evidence fragment related to the target fault section is received after the fault assessment result is output, and the new evidence fragment is incompatible with the output fault assessment result, topology version remapping is performed on the new evidence fragment and the candidate fault event chain is reconstructed. When the matching level of the reconstructed candidate fault event chain is higher than that of the adopted candidate fault event chain, update the target topology version and the target fault segment, and record the version differences before and after the update. Version differences include at least the identifiers for newly added evidence fragments, replaced evidence fragments, the original topology version, the updated topology version, the original target fault segment, the updated target fault segment, and the time the update was triggered. For cases where the target topology version changes but the adopted evidence fragments do not, only the evidence fragment differences in the version tracking record are updated. For cases where both the target topology version and the target fault segment change simultaneously, both the version differences and the result differences in the version tracking record are updated concurrently.

[0051] The main site saves version tracking records before and after the update to form a historical sequence of analysis that unfolds continuously along the fault handling process.

[0052] In a preferred embodiment of Example 1, during the process of mapping evidence fragments to the corresponding topology version, state compatibility is determined sequentially in three layers: power supply connectivity compatibility, switch state compatibility, and power outage / restoration state compatibility.

[0053] For any piece of evidence, first locate the corresponding device or region based on its associated object identifier, and then locate the current candidate topology version based on the time identifier. Under the current candidate topology version, if all three layers of verification are valid, the evidence piece is included in the current candidate topology version. If there is incompatibility under the current candidate topology version, it is then determined whether it meets the transfer conditions of adjacent topology versions. In the three-layer verification, the previous layer is used to limit the judgment scope of the next layer. If a clear conflict has occurred in the previous layer, the subsequent layer verification can still be performed to record the source of the conflict, but the candidate topology version is not used as the adopted version for the evidence piece.

[0054] Power connectivity compatibility is used to determine whether the associated objects of the evidence fragment have a valid power supply path to a reachable power source under this topology version. The connectivity state represented by the evidence fragment can be denoted as... The number of valid power supply paths from associated objects to reachable power sources under the topology version is denoted as . The method for determining the compatibility of power supply connectivity is as follows: , in, Indicates the first The evidence fragment in the first Compatibility determination results of power supply connectivity under each topology version; Indicates the first The connectivity state represented by each piece of evidence is 1, which indicates that the piece of evidence points to the existence of a power supply path, and 0, which indicates that the piece of evidence points to the non-existence of a power supply path. Indicates the first Under the first topology version, the first The number of valid power supply paths from the associated objects of each piece of evidence to at least one reachable power source; the reachable power source is the number of valid power supply paths from the associated objects of each piece of evidence to at least one reachable power source. A power source that can supply power to the associated object along a closed power supply path in each topology version.

[0055] right The extraction of data can be performed according to the source of evidence: switch action event data and dispatch control event data are directly extracted based on their status identifiers; electrical quantity monitoring data is extracted jointly based on the direction of voltage, current, and power. When the phase voltage is 15%–30% higher than the rated phase voltage and the current is 5%–10% higher than the median load current of the equipment in the same period over the past 30 days, a power supply path is considered to exist; when the phase voltage is 10%–20% lower than the rated phase voltage and the current is 3%–8% lower than the aforementioned median load current and remains so for at least two sampling cycles or two upload cycles, a power supply path is considered to not exist. In power outage and restoration status data, power outage information corresponds to no power supply path, while power restoration information corresponds to a power supply path. When there are contradictions in the directions of voltage, current, and power, switch action event data and dispatch control event data are used as primary evidence, with electrical quantity monitoring data serving as supplementary evidence, without independently overturning established action boundaries. The median load current in the same period over the past 30 days is only considered valid if the number of valid historical samples of the equipment in the corresponding period over the past 30 days is not less than [a certain percentage]. When used, A value of 7-15 is acceptable; when the number of valid historical samples is less than... At this time, the median load current will no longer be used as the judgment benchmark. Instead, the lower limit of the current reference will be determined according to 2% to 10% of the rated current of the equipment, while the original judgment rules for no voltage, undervoltage, and switch action boundaries will remain unchanged. During the handling phase of load migration caused by interconnection transfer and self-healing reconfiguration, the current samples newly added during the current fault handling process will not be fed back into the median load current of the same period in the past 30 days; the median load current will be frozen during this fault handling process until the topology version sequence is regenerated and then updated.

[0056] Furthermore, the historical samples of the median load current for the same period in the past 30 days only use valid historical records corresponding to the same measurement point, sampling granularity, and operating mode as the currently associated object. Valid historical records refer to records in which no power outage information, voltage loss information, continuous current loss information, or significant jitter codes occurred within the historical period. Abnormal current samples that significantly deviate from the historical percentile range are processed according to the abnormal sample rejection criteria; the abnormal sample rejection criteria, as an implementation parameter, can be determined according to the historical sample percentile, with the default being the upper 5% and lower 5% sample rejection range, and an adjustable range of 1%~10%. When the valid historical records for the past 30 days are insufficient, the main station maintains a fallback criterion of 2%~10% of the rated current as the current reference lower limit, and writes a historical sample insufficiency mark in the corresponding evidence segment to avoid threshold drift during the handling process due to load migration.

[0057] Switch state compatibility is used to determine whether the open / closed states of upstream disconnecting switches, sectionalizing switches, tie switches, and circuit breakers allow the validity of a given evidence segment. Based on the evidence segment type, a necessary set of switches corresponding to that evidence segment is generated. And provide evidence of the required state for each switch in the necessary switch set. The method for determining the compatibility of switch states is as follows: , in, Indicates the first The evidence fragment in the first Switch state compatibility determination results under each topology version; Indicates the first The first topology version The status of each switch is recorded as follows: open is 0, closed is 1. Indicates by the first The constraint obtained from the first piece of evidence fragment The required state of each switch; Indicates the switch index; Indicates the first The set of necessary switches corresponding to each piece of evidence.

[0058] Necessary switch set Instead of considering all network switches, only boundary switches directly related to the establishment of evidence should be selected. For protection operation events or circuit breaker operation events, the necessary switch set must include at least the operated circuit breaker and the sectionalizing switches on its downstream fault isolation boundary; for tie-line transfer records, the necessary switch set must include at least the tie-line switch and the sectionalizing switches that cooperate with it to form the isolation boundary; for self-healing reconfiguration records, the necessary switch set must include at least the switch set pointed to by the dispatch control command; for power outage / restoration status data and electrical quantity monitoring data, hard constraints at the switch level are only introduced when a corresponding operating switch or dispatch command already exists. If the specific break point cannot be uniquely located, the switch level is only recorded as not forming counter-evidence, and incompatibility is not directly determined based on fuzzy boundaries. For evidence of power restoration, the tie-line switch on the tie path should be in the closed state, and the switch on the fault isolation boundary should remain in the open state; for evidence of power loss, the key upstream switch on the power supply path should at least have a break boundary, and should not have formed an effective transfer path.

[0059] Power outage / restoration compatibility is used to determine whether the region corresponding to an evidence fragment should be in a power-off state, a restored power state, or a continuously energized state under this topology version. The region associated with the evidence fragment can be denoted as... The power supply status of this area under the corresponding topology version is recorded as follows: The state represented by the evidence fragment is denoted as The compatibility determination method for power outage and restoration states is as follows: , in, Indicates the first The evidence fragment in the first Compatibility determination results of power outage and restoration states under each topology version; Indicates the first The power supply area associated with each piece of evidence; Indicates the first The power supply status of the associated power supply area under each topology version is recorded as follows: power outage is recorded as 0, power restoration is recorded as 1, and continuous power supply is recorded as 2. Indicates the first The power supply status is represented by each piece of evidence, with 0 for power loss, 1 for power restoration, and 2 for continuous power supply.

[0060] The generation of the status can be determined by combining the power supply path and the handling stage: when there is no valid power supply path to any power source in the current topology version, the area is marked as power outage; when the area was power outage in the previous topology version and a power supply path is re-established in the current topology version, it is marked as power restored; when the area maintains a valid power supply path in both the previous and current topology versions, it is marked as continuously energized. In the extraction process, user power outage information, terminal voltage loss information, and continuous current loss information correspond to the power outage state; power restoration information, voltage restoration information, and current restoration information correspond to the power restoration state; and electrical quantity change records with no power outage records and voltage and current continuously meeting the in-service threshold correspond to the continuous energized state. A higher threshold is used for confirming power restoration; the voltage confirmation threshold can be 70%–85% of the rated phase voltage, and it must be continuous for at least two sampling periods. In cases of communication anomalies, missing waveforms, or a large number of delayed user reports, the power outage / restoration layer adopts a conservative strategy, using it only as a basis for adoption when it does not contradict the connectivity layer and the switching layer. The continuous energized state is only considered when continuous voltage and current observations are simultaneously available in the same associated area, and no power outage, voltage loss, or continuous current loss information appears within the start and end time range corresponding to the current candidate topology version. The value is assigned to 2. If there is a lack of continuous voltage or current observations based solely on the absence of power outage reports, the value is not directly assigned to a continuously energized state. Instead, the power outage / restoration layer is marked as inapplicable to this evidence fragment, and the original judgment results of the connectivity layer and the switching layer are maintained.

[0061] After the three-layer verification is completed, the state compatibility results can be combined into a single decision value: , in, Indicates the first The evidence fragment in the first The overall state compatibility determination results under each topology version; This indicates the compatibility determination result of the power supply connectivity relationship; This indicates the result of the switch state compatibility determination; This indicates the compatibility determination result for power outage and restoration states.

[0062] When a certain layer is inherently inapplicable to a particular piece of evidence, the result of that layer is set to 1, and it only participates in the comprehensive judgment for applicable layers. When any applicable layer has a value of 0, the piece of evidence is considered incompatible under the current candidate topology version. This process maintains a consistent comprehensive judgment standard and avoids mistaking inapplicable layers as sources of conflict due to differences in evidence types.

[0063] When the When a piece of evidence is incompatible in the current candidate topology version but potentially compatible in an adjacent topology version, cross-version transfer is allowed only if both the version switching window and the scope of influence constraints are satisfied. A transfer permission can be represented as: , in, Indicates the first The evidence fragment is relative to the first Cross-version transfer license results at each version switch boundary; Indicates the first The time stamp of each piece of evidence; Indicates the first The time of occurrence of the version switch event corresponding to each version switch boundary; Indicates the first The width of the version switching window corresponding to each version switching boundary; Indicates the first The associated object identifier of the first piece of evidence; when the first When the first piece of evidence corresponds to equipment-related evidence, the associated object is the equipment; when the first piece of evidence corresponds to the second piece of evidence, the associated object is the equipment. When a piece of evidence corresponds to a region-type piece of evidence, the associated object is the region. Indicates the first The scope of impact corresponding to each version switch boundary.

[0064] The timescale resolution and transmission delay are jointly calibrated. A window of 30s to 180s before and after the version switching event can be used. When the resolution of the station-end sequential event is high, a smaller value is used. When the main data depends on the terminal upload and the user's power outage and restoration status data, a larger value is used, and the upper limit is not more than 300s. , in, Indicates the first The width of the version switching window corresponding to each version switching boundary; This indicates the lower limit of the version switching window, which can be set to 30 seconds. This indicates the maximum window size for version switching, which can be set to 300 seconds. Indicates the first The time resolution of the primary data source used for each version switching boundary; Indicates the first The transmission delay envelope of the primary data source used at each version switching boundary; and Indicates the calibration coefficient. One to two can be selected. A value of 1 to 3 is acceptable; when the same version switch boundary involves multiple data sources simultaneously, and Take the larger, more conservative value; the first Once the version switch boundary is generated... It remains unchanged during the comparison of the topology version sequence and is only refreshed when the topology version sequence is regenerated.

[0065] When a clock drift alarm, communication congestion, or bulk retransmission occurs, the algorithm can be extended upwards once, but only a single-step transfer is still allowed between the current candidate topology version and its directly adjacent topology version. Scope of impact. The power supply path is determined based on the version switching event: when a sectionalizing switch is opened, its downstream isolated section is taken; when a tie switch is closed, the tie transfer path and its restored power supply area are taken; when a circuit breaker operates, the section from its downstream to the nearest isolation boundary is taken; when a path switch is caused by dispatch control, the isolated section and transfer section covered by the dispatch command are taken. Only when the associated object of the evidence fragment falls within the scope of influence, and its timestamp falls within the version switching window, is a transition from the current candidate topology version to the adjacent topology version allowed; evidence outside the window or outside the scope, even if interpretable under the adjacent topology version, is retained as conflicting evidence and not absorbed across versions. The current candidate topology version is incompatible, the adjacent topology version is compatible, and the following conditions are met: When the evidence fragment is transferred to an adjacent topology version, if both the previous and next adjacent topology versions of the current candidate topology version meet the transfer conditions, it is preferentially assigned to the topology version with a smaller time interval than the evidence fragment. If the time intervals are the same, its affiliation is determined according to the order of switch action event data, electrical quantity monitoring data, dispatch control event data, and power outage / restoration status data.

[0066] When the current candidate topology version and its adjacent topology versions are incompatible, or although they are compatible under adjacent topology versions but do not meet the cross-version transfer permission, the evidence fragment is marked as conflict evidence.

[0067] Example 2: Based on Example 1, this example further provides a method for determining the matching level of candidate fault event chains. This method compares candidate fault event chains under different topology versions based on the generated candidate fault segments and candidate fault event chains, and determines the target topology version and target fault segment, as follows: When constructing candidate fault event chains, the switching action events, scheduling control events, electrical quantity change records, and power outage / restoration status change records around the same candidate fault section are connected according to the equipment connection sequence from the power supply side to the candidate fault section, and then from the candidate fault section to the power loss area or power restoration area under the current topology version, as well as the event occurrence sequence of fault-related events, isolation-related events, power transfer-related events, and power restoration-related events on a unified time axis. Each candidate fault event chain is recorded with a chain identifier, the topology version identifier, the associated candidate fault section identifier, the sequence of evidence fragment identifiers within the chain, the first event time, the last event time, and the matching level. In this embodiment, the matching level of the candidate fault event chain is determined based on the matching results of the power outage range and the isolation range, the matching results of the power restoration range and the interconnection and transfer path, the matching results of the electrical quantity change sequence and the switch action sequence, and the evidence of conflict within the chain. When multiple candidate fault event chains have the same matching level, the candidate fault event chain with fewer conflicting evidence is selected first. When the number of conflicting evidence is the same, the candidate fault event chain with more supporting evidence fragments covering the upstream boundary equipment, downstream boundary equipment, tie switch and power restoration boundary equipment of the candidate fault section is selected first. In this embodiment, a candidate fault event chain refers to a sequence of evidence formed by connecting devices around the same candidate fault segment according to the device connectivity order and the event occurrence order. The device connectivity order refers to the sequence of line segments and devices connected from the power supply side to the candidate fault segment, and then from the candidate fault segment to the power outage area or the power restoration area, under the current topology version. The event occurrence order refers to the arrangement of fault-related events, isolation-related events, power transfer-related events, and power restoration-related events on a unified timeline.

[0068] In this embodiment, each candidate fault event chain records at least the chain identifier, the topology version identifier, the associated candidate fault segment identifier, the sequence of evidence fragment identifiers within the chain, the time of the first event, the time of the last event, and the matching level. The matching level is determined based on at least the matching results of the power outage range and the isolation range, the matching results of the power restoration range and the interconnection and transfer path, the matching results of the electrical quantity change sequence and the switch action sequence, and the conflict evidence within the chain; wherein, the conflict evidence within the chain includes at least the number of conflicting pieces of evidence and the proportion of conflicting evidence. To ensure consistent engineering standards for matching levels across different data completeness levels, the reference threshold for evidence coverage can be 0.65~0.85, with a default value of 0.75; the reference threshold for critical equipment coverage can be 0.75~1.00, with a default value of 0.85; the reference threshold for conflicting evidence ratio can be 0.10~0.25, with a default value of 0.20; and the reference threshold for time deviation can be 10s~120s, with the default value primarily determined by the primary evidence data source. Specifically, 10s~30s is used when station-side sequential events dominate, and 30s~120s is used when terminal transmission and power outage / restoration records dominate. These thresholds can be adjusted based on historical sample quantiles, validation set tuning results, or industry limits, and remain unchanged during the same round of candidate fault event chain comparison. When multiple candidate fault event chains have identical counting and ranking results, the primary station performs parallel resolution in the order of earlier target topology version effective time, more complete evidence for candidate fault segment boundary equipment, and more continuous version tracking records.

[0069] Version consistency is determined based on the correspondence between the topology version to which the candidate fault event chain belongs and the topology version to which the evidence fragments used in the chain belong.

[0070] When all the evidence fragments used in the chain belong to the same topology version as the candidate fault event chain, or when the inconsistent evidence fragments are only marked conflict evidence and do not participate in the matching level calculation, the candidate fault event chain is determined to meet the version consistency requirement; when there are evidence fragments used to participate in the matching level calculation whose topology version belongs to the candidate fault event chain is inconsistent with the topology version of the candidate fault event chain, the candidate fault event chain is determined to not meet the version consistency requirement.

[0071] In a preferred embodiment of Example 2, during the construction and comparison of candidate fault event chains, the matching level is determined in the order of strong conflict elimination, hard constraint counting, soft constraint counting, and tie-breaker.

[0072] For each candidate fault event chain, three types of relationships are first examined: the power outage range and the isolation range, the power restoration range and the interconnection and transfer path, and the sequence of electrical quantity changes and the sequence of switch actions. The action anchor point is the event moment used for timing alignment in the candidate fault event chain, including the isolation action confirmation moment, the interconnection switch closing moment, the circuit breaker reclosing moment, and the dispatch control event activation moment. A candidate fault event chain is directly excluded from subsequent sorting if any of the following conditions occur: First, the power outage range continuously crosses the isolation boundary without a corresponding upstream circuit breaker action or sectionalizing switch opening as an explanation; second, the power restoration range crosses the disconnecting switch that is kept open, or exceeds the actual reachable transfer path after the interconnection switch is closed; third, voltage recovery, current recovery, or power recovery occurs before the corresponding closing action, interconnection and transfer action, or dispatch control event, and this reverse sequence continuously exceeds the version switching window corresponding to the version switching boundary of the action anchor point. Candidate fault event chains that are not excluded then proceed to the matching level calculation.

[0073] For those not excluded For each candidate fault event chain, first count the number of hard constraints satisfied and the number of applicable hard constraints: , , in, Indicates the first The number of hard constraints satisfied in a candidate fault event chain; Indicates the first The number of applicable hard constraints for each candidate fault event chain; Indicates the first The relationship between the power outage range and isolation range of each candidate fault event chain is indicated by an applicable flag: 1 for applicable and 0 for inapplicable. Indicates the first The relationship between the power restoration range and the interconnection and transfer path of each candidate fault event chain is indicated by an flag: 1 for applicable and 0 for non-applicable. Indicates the first The relationship between the electrical quantity change sequence and the switch action sequence of a candidate fault event chain is indicated by an applicable flag: 1 for applicable, 0 for non-applicable. , and These represent the results of satisfying the corresponding hard constraints, with 1 indicating satisfaction and 0 indicating non-satisfaction.

[0074] When a certain processing step is in If a candidate fault event chain does not exist in the corresponding topology version and handling stage, the applicable flag is set to 0 and is not counted. ; When this handling step should exist in the topology version and handling phase, but the first When a candidate fault event chain does not form a corresponding evidence chain or there is contrary evidence, the corresponding applicable flag is set to 1, and the result is set to 0.

[0075] The determination is based on the isolation boundaries on both sides of the candidate fault section: power outage terminals, voltage loss terminals, or current loss terminals should be concentrated downstream of the isolation boundary, and equipment that remains energized outside the boundary should not be largely enclosed within the isolation section; when the power outage range only has one additional terminal branch near the boundary, and there is a timescale delay between this branch and the main trunk that does not exceed one sampling period in the power outage report, it can still be considered satisfactory. When the power outage range spans two or more segmented sections on both sides of the isolation boundary, Take 0. The determination is based on the reachable area of ​​the tie-and-transfer path: the power restoration terminal should be located in the section that can be reached from the backup power source after the tie switch is closed, and should not cross the disconnect switch that is still open; the restoration range can be slightly smaller than the theoretical power transfer path, and complete coverage is not required, but it should not be larger than the path. The determination is based on the action anchor point in the event chain: loss of voltage, loss of current, and power drop should not be later than the confirmation time of the isolation action; the restoration of voltage, current, and power recovery caused by power restoration should not be earlier than the closing of the tie switch, the reclosing of the circuit breaker, or the effective time of the dispatch transfer instruction; when the time scale resolution of the electrical quantity record and the action record are not synchronized, they are allowed to be regarded as consistent in sequence within the same version switching window.

[0076] In addition to hard constraints, soft constraints are applied to the coverage of evidence and the degree of conflict. Evidence coverage can be expressed as: , in, Indicates the first Evidence coverage of each candidate failure event chain; Indicates the first The number of evidence fragments used in each candidate failure event chain; This indicates the number of target evidence fragments that should appear according to the link structure of the candidate failure event chain.

[0077] The coverage of key equipment can be expressed as: , in, Indicates the first Key device coverage of each candidate failure event chain; Indicates the first The number of critical devices in a candidate failure event chain that have been supported by valid evidence; Indicates the first The total number of critical devices that should be covered in the candidate fault event chain.

[0078] Prioritize fault isolation points, interconnection and power transfer equipment, upstream and downstream sectionalizing switches of candidate fault sections, and equipment at the power restoration boundary. For simple fault isolation scenarios, This includes at least upstream operating switches, downstream isolating boundary switches, and outage boundary equipment; for interconnection and transfer or self-healing reconfiguration scenarios, It also includes tie switches and power restoration boundary equipment. The larger the value, the more complete the coverage of the power outage, isolation, power transfer, and restoration processes of the candidate fault event chain, and the direction of its change is monotonically increasing; The larger the value, the more complete the evidence at the key location. The direction of change is also monotonically increasing, and the upper limit of both is 1. The reference criterion can be taken as 0.65 to 0.85, with a higher value taken when the communication integrity is high and a lower value taken when there are intermittent terminal disconnections; The reference criterion can be taken as 0.75 to 1.00, and should be no less than 0.85 when it involves communication transfer and self-healing reconstruction. In cases where there is a lack of evidence for critical equipment but a large amount of evidence for ordinary equipment, the number of ordinary evidence should not be used to replace the coverage of critical equipment.

[0079] and Only statistics in the first The candidate fault event chain corresponds to the location of target evidence and key equipment that are configured, collectable, and time-aligned under the topology version. Objects that are not connected to monitoring, have been offline for a long time, have no corresponding data source configuration, or do not belong to this handling stage are not included in the denominator. Once a candidate fault event chain is generated... and The comparison remains unchanged in this round; newly added evidence fragments are only recalculated in the denominator when reconstructing the candidate failure event chain.

[0080] After the candidate failure event chain is constructed, and All integers are greater than 0; when the first... If a candidate fault event chain does not contain any adopted evidence fragments, the candidate fault event chain is directly excluded, and evidence coverage, critical equipment coverage, conflict evidence ratio, and time deviation are no longer calculated.

[0081] The number of soft constraints satisfied can be expressed as: , in, Indicates the first The number of soft constraints satisfied in a candidate fault event chain; This represents an indicative function; it takes the value 1 if the condition is true and 0 if the condition is false. Indicates the first Evidence coverage of each candidate failure event chain; Indicates the reference threshold for evidence coverage; Indicates the first Key device coverage of each candidate failure event chain; This indicates a reference threshold for the coverage of critical equipment; Indicates the first The percentage of conflicting evidence in a candidate fault event chain is the ratio of the number of conflicting evidence fragments in that candidate fault event chain to the number of adopted evidence fragments. Indicates the reference threshold for the proportion of conflicting evidence; Indicates the first The average absolute deviation between the time of each evidence fragment and the time of the corresponding action anchor point in each candidate fault event chain; This indicates the reference threshold for time deviation.

[0082] , in, Indicates the first The percentage of conflicting evidence in each candidate failure event chain; Indicates the first The number of conflicting evidence fragments in each candidate failure event chain; Indicates the first The number of evidence fragments used in each candidate failure event chain.

[0083] , in, Indicates the first The average absolute deviation between the time of each evidence fragment and the time of the corresponding action anchor point in each candidate fault event chain; Indicates the first The number of evidence fragments used in each candidate failure event chain; Indicates the first The sequence number of the evidence fragment used in the chain of candidate failure events; Indicates the first In the candidate fault event chain, the first The time of each piece of evidence used; Indicates the first In the candidate fault event chain, the first The time of the action anchor point corresponding to each piece of evidence used; Indicates and The width of the version switching window corresponding to the version switching boundary of the action anchor point.

[0084] when At that time, the evidence fragment was... Contribution by Include it, and keep its original conflict evidence markers unchanged. Once a candidate fault event chain is generated, each It is bound to the corresponding version switch boundary and will not be refreshed in the current comparison; it is only recalculated when rebuilding the candidate fault event chain.

[0085] The smaller the value, the fewer contradictions there are within the same candidate fault event chain. Its direction of change is monotonically decreasing, with a lower limit of 0. The smaller the value, the closer the evidence time is to the version switch, isolation action, communication and transfer action, or power restoration action. Its change direction is monotonically decreasing, with a lower limit of 0. The value can be between 0.10 and 0.25. A smaller value should be used for scenarios involving relatively complete waveform summaries and sequential events, while a larger value should be used when only relying on terminal power outage and restoration status data. The time stamp can be determined based on the source of the time stamp. When station-end sequential events are the primary data type, 10s to 30s can be used; when terminal-uploaded data and user power outage / restoration status data account for a high proportion, 30s to 120s can be used. When clock drift alarms, communication congestion, or batch retransmissions occur, the time deviation term is handled using a conservative strategy, meaning it is only used as a sorting criterion among parallel candidate fault event chains and not as a separate exclusion condition. The tolerance can be relaxed to within 1.5 times the original calibration value, but not exceeding 180 seconds.

[0086] The candidate fault event chain with the highest matching level is selected by hierarchical comparison, and its sorting vector can be represented as: , in, Indicates the first A hierarchical sorting vector of candidate fault event chains; Indicates the first The number of hard constraints satisfied in a candidate fault event chain; Indicates the first The number of applicable hard constraints for each candidate fault event chain; Indicates the first The number of soft constraints satisfied in a candidate fault event chain; Indicates the first The number of evidence fragments used in each candidate failure event chain; Indicates the first Key device coverage of each candidate failure event chain; Indicates the first The percentage of conflicting evidence in each candidate failure event chain; Indicates the first The average absolute deviation between the time of each evidence fragment and the time of the corresponding action anchor point in each candidate fault event chain.

[0087] Comparison of candidate fault event chains by The components are selected sequentially: priority is given to... A large chain of candidate failure events; When they are the same, choose A large chain of candidate failure events; If they are still the same, choose A large chain of candidate failure events; If they are still the same, choose Larger candidate fault event chains; if they are the same, select... If a large chain of candidate failure events still cannot be distinguished, select them sequentially. Smaller Smaller candidate fault event chains. This approach ensures that hard conflicts are not masked by high coverage, and that ordinary evidence does not replace the role of critical equipment evidence, thus maintaining consistent physical interpretability between the topology version corresponding to the highest matching level and the candidate fault segment. The topology version corresponding to this candidate fault event chain is determined as the target topology version, and the candidate fault segment corresponding to this candidate fault event chain is determined as the target fault segment. If a distinction still cannot be made, the candidate fault event chain with fewer conflicting evidence fragments is prioritized; if the number of conflicting evidence fragments is the same, the candidate fault event chain with the earlier occurrence time is selected based on the order of occurrence of the version switching event corresponding to the topology version.

[0088] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0089] Furthermore, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, combinations of features from different embodiments are meant to be within the scope of this application and form different embodiments. For example, all the embodiments above can be used in any combination. The information disclosed in this background section is intended only to enhance the understanding of the general background of this application and should not be construed as an admission or in any way implying that such information constitutes prior art known to those skilled in the art.

Claims

1. A method for fault assessment in distribution networks based on multi-source data fusion, characterized in that, Applied to at least one of the following scenarios: post-failure communication and supply transfer scenarios and self-healing reconfiguration scenarios, including: Acquire operational topology data, switch action event data, dispatch control event data, electrical quantity monitoring data, and power outage / restoration status data; Multi-source data is broken down into evidence fragments based on associated objects, event types, and time boundaries; Generate a topology version sequence ordered by time based on switching action events and scheduling control events that cause changes in the power supply path; Candidate topology versions are filtered by associated object identifier and time identifier, and state compatibility is determined by power supply connection relationship, switch status and power outage / restoration status. Compatibility evidence is assigned to the corresponding topology version. Evidence that is incompatible under the current candidate topology version but compatible under adjacent topology versions that meet the version switching window and influence range constraints is assigned to the adjacent topology version. The rest are written into the conflict evidence set. Evidence fragments assigned to the same topology version but with inconsistent representation results are sorted by source and checked by time sequence. In each topology version, candidate fault segments are generated based on power connection relationships, switch status, changes in power outage range, and changes in power restoration range. A candidate fault event chain is then constructed around the candidate fault segments according to the equipment connection sequence and the event occurrence sequence. Compare the version consistency and matching level of each candidate fault event chain to determine the target topology version and target fault segment, and output the fault assessment result including the evidence fragments used and the handling stage identifier.

2. The method for distribution network fault assessment based on multi-source data fusion according to claim 1, characterized in that, The operational topology data includes feeder connection relationships, sectionalizing switch status, tie switch status, and power supply access relationships; The switch action event data includes protection action events, circuit breaker action events, sectionalizing switch action events, and tie switch action events; The scheduling and control event data includes fault isolation records, communication transfer records, and self-healing reconfiguration records; The electrical quantity monitoring data includes voltage, current, power, and waveform summary information; The power outage and restoration status data includes terminal power outage information, user power outage and restoration information, and power restoration information.

3. The method for distribution network fault assessment based on multi-source data fusion according to claim 1, characterized in that, The evidence fragments are the smallest associatable data units formed by splitting them according to the associated object, event type and time boundary, and each evidence fragment records the source identifier, time identifier, status identifier and associated object identifier; The evidence fragments include at least a single switch status change record, a single dispatch control record, a single device electrical status change record, and a single area power outage / restoration status change record.

4. The method for distribution network fault assessment based on multi-source data fusion according to claim 1, characterized in that, When generating the topology version sequence, events that cause changes in the power supply path are used as version switching events. These version switching events include the opening and closing of sectionalizing switches, the opening and closing of tie switches, the opening and closing of circuit breakers, and power supply path switching caused by scheduling control. The fault handling process is divided into multiple handling stages according to the occurrence time of the version switching event, and a corresponding topology version identifier is generated for each handling stage.

5. The method for distribution network fault assessment based on multi-source data fusion according to claim 1, characterized in that, When performing source sorting and timing verification on conflict evidence within the same topology version, the priority of evidence adoption is determined according to the order of switch action event data, electrical quantity monitoring data, dispatch control event data, and power outage / restoration status data. The event sequence of isolation before power supply and power loss before power restoration is verified according to the time identifier. Evidence fragments that do not conform to the stated order of events are treated as conflicting evidence.

6. The method for distribution network fault assessment based on multi-source data fusion according to claim 1, characterized in that, When generating the candidate fault sections, the power outage boundary section is determined based on the change in the power outage range, the power transfer boundary section is determined based on the power restoration range, and the isolation boundary section is determined based on the change in the switch status. The power outage boundary section, the power transfer boundary section, and the isolation boundary section are then superimposed in the current topology version to obtain the candidate fault section set.

7. The method for distribution network fault assessment based on multi-source data fusion according to claim 1, characterized in that, When constructing the candidate fault event chain, the switching action events, scheduling control events, electrical quantity change records, and power outage / restoration status change records around the same candidate fault section are connected according to the equipment connection sequence from the power supply side to the candidate fault section, and then from the candidate fault section to the power loss area or power restoration area under the current topology version, as well as the event occurrence sequence of fault-related events, isolation-related events, power transfer-related events, and power restoration-related events on a unified time axis to form a candidate fault event chain. For each candidate fault event chain, the chain identifier, the topology version identifier, the associated candidate fault section identifier, the sequence of evidence fragment identifiers within the chain, the first event time, the last event time, and the matching level are recorded.

8. The method for distribution network fault assessment based on multi-source data fusion according to claim 1, characterized in that, When determining the matching level of the candidate fault event chain, it is based on the matching results of the power outage range and the isolation range, the matching results of the power restoration range and the interconnection and transfer path, the matching results of the electrical quantity change sequence and the switch action sequence, and the evidence of conflict within the chain. When multiple candidate fault event chains have the same matching level, the candidate fault event chain with fewer conflicting evidence is selected first. When the number of conflicting evidence is the same, the candidate fault event chain with more supporting evidence fragments covering the upstream boundary equipment, downstream boundary equipment, tie switch and power restoration boundary equipment of the candidate fault section is selected first.

9. The method for distribution network fault assessment based on multi-source data fusion according to claim 1, characterized in that, The fault assessment results also include version tracking records, which include the target topology version identifier, the triggering event for switching from the previous topology version to the target topology version, the target fault section, the set of evidence fragments used, and the set of conflicting evidence.

10. The method for distribution network fault assessment based on multi-source data fusion according to claim 1, characterized in that, When a new evidence fragment related to the target fault section is received after the fault assessment result is output, and the new evidence fragment is incompatible with the output fault assessment result, a topology version remapping is performed on the new evidence fragment and the candidate fault event chain is reconstructed. When the matching level of the reconstructed candidate fault event chain is higher than that of the adopted candidate fault event chain, update the target topology version and the target fault segment, and record the version differences before and after the update.

Images