A method and system for automatic control of power distribution networks with fault self-identification

Abstract

This invention discloses a distribution network automation control method and system with self-identification of faults, specifically relating to the field of distribution network automation technology. It addresses the problems of misjudgment, missed judgment, and difficulty in implementing identification results caused by load disturbances and the inability to self-verify topology states under conditions of sparse measurement points, communication jitter, grid connection backflow, and inconsistencies between topology and timestamps in distribution networks. The method constructs a dual-anchor time base and aligns it using out-of-order buffering, water level advancement, and playback to generate idempotent keys. It then uses a triggered snapshot matrix, incremental impedance tomography, and directional consistency lattice to determine the primary candidate fault. Under homogenized gating, it extracts hysteresis area, amplitude-phase transition, and slope-curvature features, combining arrival time and attenuation law for localization. After residual verification, it generates an execution list containing sequence, interlocking, secondary confirmation, and rollback, and solidifies the evidence chain. Thus, it can reliably self-identify and execute faults even under conditions of sparse measurement points, communication jitter, and backflow.

Description

Technical Field

[0001] This invention relates to the field of power distribution network automation technology, specifically to a power distribution network automation control method and system with fault self-identification. Background Technology

[0002] In the field of distribution network automation, SCADA / FTU / DTU and fault indicators have been widely deployed, combined with distribution network self-healing / FLISR strategies to achieve fault location and isolation based on current, voltage over-limit, and remote signaling status. Some systems introduce synchronous phasor measurement, traveling wave method, or injected signal method to improve fault identification and location accuracy. However, the above paths are usually based on strong premises: relatively sufficient measurement point coverage, highly consistent timestamps of each source or reliance on unified time synchronization, real-time reflection of switch topology by ledgers and remote signaling, and stable communication links with negligible packet loss. In actual operation on the medium and low voltage side, these premises are difficult to maintain in the long term, causing traditional criteria based on fixed thresholds, rule trees, or simple statistics to be prone to misjudgment and omission when there are multiple asynchronous sources and mixed disturbances.

[0003] In more realistic field conditions, distribution networks face three prominent contradictions: First, asynchronous multi-source data and clock drift cause event disorder, making it difficult to restore the timing of reclosing and multi-point tripping. Second, delayed switch operation records, remote signaling errors or loss, coupled with the backflow of distributed power generation, distort the current direction and sequence component characteristics, increasing topological uncertainty. Third, feeders with sparse measurement points and low signal-to-noise ratios are more sensitive to difficult-to-detect faults such as high-resistance grounding and intermittent arcing, and conventional thresholds and single characteristics are insufficient to stably separate fault transients under load steps, reactive power compensation switching, and grid disturbances. Furthermore, existing solutions often use the determination of "whether it is a fault / approximate section" as the endpoint, lacking directly actionable sequences and traceable evidence for execution, making it difficult to convert identification results into actionable isolation and reconnection actions, affecting handling efficiency and control of safety boundaries.

[0004] In real-world distribution network scenarios characterized by sparse measurement points, communication jitter, significant grid backflow, and inconsistencies between switch topology and timestamps, how can a mechanism be established to address actual operational constraints? This mechanism would enable multi-source asynchronous measurements to converge to a unified time base under fault-tolerant conditions, providing quantifiable reliability of the topology state. Furthermore, it would distinguish between load and grid disturbances and fault transients, thereby providing directly executable and traceable control data for fault identification, section location, and isolation and restoration. This problem points to a single core link: unified time base—reliable topology—disturbance separation—location determination—executable and auditable control. This aims to overcome the common shortcomings of existing technologies under real-world operating conditions: inconsistent timing, lack of self-verification of topology, unauditable evidence, and difficulty in executing results. Summary of the Invention

[0005] (a) Technical problems to be solved

[0006] To address the shortcomings of existing technologies, this invention provides a distribution network automation control method and system with fault self-identification, thereby solving the problems mentioned in the background section.

[0007] (II) Technical Solution

[0008] To achieve the above objectives, the present invention provides the following technical solution: a distribution network automation control method with fault self-identification, characterized in that it includes:

[0009] S1. Extract intrinsic anchors and communication anchors to form a dual-anchor time base, set an out-of-order buffer to advance the water level and replay for correction, and generate idempotent keys according to device identifier, time anchor identifier, window sequence number, template version, and feature signature to complete alignment;

[0010] S2. Under the condition of anchoring and over-limit triggering, issue temporary additional sampling, collect zero sequence, negative sequence, voltage transient, phasor phase-locked results and spectrum segments around the trigger window, form an event snapshot matrix and bind idempotent keys;

[0011] S3. Calculate the zero-sequence and negative-sequence equivalent incremental impedance distribution on the candidate topology based on the snapshot, and score the consistency of each candidate topology according to the directional consistency grid to obtain the topology confidence of each candidate topology. The candidate topology whose topology confidence meets the preset conditions is determined as the main candidate topology.

[0012] S4. Implement homogenization gating on the main candidate to extract zero-sequence current-voltage hysteresis area transition, phasor phase-locked amplitude-phase transition, and slope-curvature mutation spectrum to generate fault evidence vector.

[0013] S5. Based on the arrival time sequence and amplitude decay law, search for the maximum consistent sub-path on the main candidate to complete the fault type determination and give the segment location confidence.

[0014] S6. Perform linear power flow residual verification on the primary candidate, generate an execution list consisting of four items: sequence, interlock, secondary confirmation, and rollback, solidify the evidence index, and record the execution log.

[0015] In a preferred embodiment, S1 includes:

[0016] A dual-anchor time base is constructed in parallel with intrinsic anchors and communication anchors. Anchoring is determined when both anchors simultaneously satisfy the requirements and the anchoring strength reaches the threshold within the observation window.

[0017] The measurements and signaling are written to the out-of-order buffer, the water level is advanced according to the time anchor and window number, and the corrections are replayed in the same observation window;

[0018] Generate idempotent keys based on device identifier, time anchor identifier, window sequence number, template version, and feature signature to form aligned window records, write them to the time series warehouse, and generate evidence indexes.

[0019] When the two anchors are inconsistent, the deviation is determined and recorded based on the communication anchor time stamp.

[0020] In a preferred embodiment, S2 includes:

[0021] When the anchor is established and the over-limit conditions are met, a burst frame acquisition command for temporary additional acquisition is issued within the alignment window;

[0022] Zero-sequence current, negative-sequence current, voltage transient, phasor phase-locked loop results and spectral energy segments are placed into the disordered buffer, aligned by advancing and replaying according to the water level line, bandwidth limiting and noise suppression are performed, missing measurement points are continuously supplemented and their credibility is marked.

[0023] The snapshot is retained based on the threshold and duration, an event snapshot matrix is ​​generated and bound with an idempotent key. The idempotent key consists of device identifier, time anchor identifier, window sequence number, template version, and feature signature. The same idempotent key is valid only once.

[0024] Write the snapshot matrix into the snapshot repository and generate an evidence index that is then stored in the audit repository.

[0025] One playback correction is allowed if the water level line is not crossed; if the retention conditions are still not met, it is marked as a candidate snapshot and recorded in the audit repository.

[0026] A sudden frame failure is registered as an incomplete snapshot and recorded in the audit repository.

[0027] In a preferred embodiment, S3 includes:

[0028] Within the alignment window, based on the event snapshot matrix bound by the dual-anchor time base, the zero-sequence equivalent incremental impedance distribution and the negative-sequence equivalent incremental impedance distribution are estimated on each candidate topology.

[0029] Using the alignment window baseline as a reference, a directional consistency grid is constructed, consisting of active power increment symbols, reactive power increment symbols, and phasor sequence relations.

[0030] A consistency score is generated according to a preset weight; and the consistency score is associated with the window identifier, time anchor identifier, template version, and threshold version and stored in the topology repository using an idempotent key.

[0031] In a preferred embodiment, a topological credibility is generated based on the generated consistency score, combined with evidence completeness, evidence freshness, and consistency of statements, according to a preset priority.

[0032] The primary candidate topology is determined when the topology credibility reaches a threshold.

[0033] When scores are tied, the following criteria are used in order of importance: consistency of direction, freshness of evidence, and consistency of interpretation.

[0034] Generate topology records containing candidate number, topology credibility, consistency score, primary candidate identifier, time anchor identifier, window identifier, template version, threshold version, and fingerprint value, and store them in the topology warehouse with idempotent key association. At the same time, solidify the shadow record and the fingerprint value of the audit warehouse.

[0035] In a preferred embodiment, S4 includes:

[0036] Within the primary candidate topology and limited to the alignment window, when the confidence of the primary candidate topology and the completeness of the snapshot reach the threshold, the snapshot is called from the snapshot repository according to the idempotent key, and bandwidth limiting and baseline correction are performed sequentially according to the template version and the threshold version. Zero-sequence current voltage hysteresis area transition, phase-locked amplitude-phase transition, slope mutation, and curvature mutation are extracted to generate fault evidence vectors.

[0037] Detect signs of stable reflux, and if the criteria are met, execute the reflux immunization strategy and generate a downgrade flag;

[0038] The fault evidence vector, along with the window identifier, main candidate identifier, time anchor identifier, idempotent key, template version, and threshold version, is solidified into the evidence repository, and arrival confirmation and secondary confirmation are completed through the publish-subscribe channel;

[0039] Each idempotent key is valid only once. Subsequent records are overwritten based on the freshness of the time anchor and are permanently recorded in the audit repository.

[0040] In a preferred embodiment, S5 includes:

[0041] Within the alignment window, based on the path length and equivalent impedance of the primary candidate topology, the evidence vector, which is fixed with idempotent keys and time anchors, is invoked to perform geometric consistency screening according to arrival time sequence and amplitude attenuation law.

[0042] When the evidence coverage, consistency residual, and gating satisfaction meet the criteria, the maximum consistent sub-path is searched and a location record is formed. The location record includes the segment identifier, fault type, location confidence, time anchor identifier, idempotent key, primary candidate identifier, template version, threshold version, weight version, and evidence index.

[0043] Evidence that extends beyond the reachable segments of the primary candidate topology is not included in the consensus calculation.

[0044] In a preferred embodiment, S6 includes:

[0045] Within the alignment window, based on the time anchor identifier and associated with an idempotent key, the positioning record is obtained from the positioning warehouse, the main candidate topology is obtained from the topology warehouse, and the current measurement is collected. A conservative coefficient is applied according to the measurement freshness to participate in the linear power flow residual verification.

[0046] Under the interlock table, interlock table, and equipment rated gating, when it is determined on the main candidate topology that the residual quantity meets the safety budget, an execution list containing sequence, interlock, secondary confirmation, and rollback strategies is generated, written into the control room, sent and confirmed via the power distribution automation dedicated communication network, and bound to the fingerprint value.

[0047] Simultaneously generate evidence indexes and write them to the audit repository, recording step sequence information and version information;

[0048] The entire process is executed according to idempotent, sequential, and deduplication strategies, and is managed by version locking.

[0049] On the other hand, the present invention provides a power distribution network automation control system with fault self-identification, comprising:

[0050] Time base alignment module: used to extract intrinsic anchor and communication anchor to form a dual anchor time base, set up out-of-order buffer to advance the water level and replay for correction, generate idempotent keys based on device identifier, time anchor identifier, window sequence number, template version, and feature signature, and complete alignment and deduplication;

[0051] Snapshot Acquisition Module: Used to issue temporary additional acquisition when anchoring and exceeding the limit triggering, to collect zero sequence, negative sequence, voltage transient, phasor phase-locked loop results and spectrum segments around the trigger window, generate an event snapshot matrix and bind idempotent keys;

[0052] Topology self-verification module: It is used to calculate the zero-sequence and negative-sequence equivalent incremental impedance distribution on the candidate topology based on snapshots, and to score the consistency of each candidate topology according to the directional consistency grid to obtain the topology credibility of each candidate topology. The candidate topology whose topology credibility meets the preset conditions is determined as the main candidate topology.

[0053] Homogenization feature extraction module: used to implement homogenization gating on the main candidate, extract zero-sequence current-voltage hysteresis area transition, phasor phase-locked amplitude-phase transition, slope-curvature mutation spectrum, and form fault evidence vector;

[0054] Location determination module: used to search for the most consistent sub-path on the main candidate based on arrival time sequence and amplitude attenuation law, determine the fault type and calculate the segment location confidence;

[0055] The execution verification and audit module is used to perform linear power flow residual verification on the primary candidate, outputting an execution list containing four items: sequence, interlock, secondary confirmation, and rollback, solidifying the evidence index, and recording the execution log.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] 1. By constructing a dual-anchor time base and using out-of-order buffering, water level advancement and playback correction to generate idempotent keys to complete alignment, combined with a trigger-based snapshot matrix, tomography based on incremental impedance and directional consistency lattice to determine the main candidate topology, and extracting anti-backflow characteristics of hysteresis area transition, amplitude-phase transition and slope-curvature abrupt change under homogenization gating, combined with arrival time and attenuation law to achieve segment positioning, and then outputting an execution list and evidence chain containing sequence, interlock, secondary confirmation and pullback after residual verification, so as to achieve reliable self-identification and positioning of difficult-to-detect faults under the conditions of sparse measurement points, communication jitter and grid backflow, and stably convert the identification results into auditable and directly executable distribution network control actions;

[0058] 2. By combining the tomographic criteria based on natural disturbances with temporary additional snapshots, and with the full-link traceability mechanism of version locks, idempotent keys and cross-warehouse evidence indexes, it can be implemented without relying on external signals and high-precision unified clocks and with only minor modifications. It forms a robust process with a defined order and deduplication and seamlessly connects with the existing power distribution automation system, thereby reducing the cost of modification and operation and maintenance, improving the robustness and scalability of the system, and achieving compliant and traceable engineering deployment within the security boundary. Attached Figure Description

[0059] Figure 1 This is a flowchart illustrating a power distribution network automation control method with fault self-identification according to the present invention.

[0060] Figure 2 This is a schematic diagram of the structure of a power distribution network automation control system with fault self-identification according to the present invention. Detailed Implementation

[0061] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] Example 1: Figure 1 A flowchart illustrating a distribution network automation control method with fault self-identification according to the present invention is provided. The method includes:

[0063] S1. Extract intrinsic anchors and communication anchors to form a dual-anchor time base, set an out-of-order buffer to advance the water level and replay for correction, and generate idempotent keys according to device identifier, time anchor identifier, window sequence number, template version, and feature signature to complete alignment;

[0064] S2. Under the condition of anchoring and over-limit triggering, issue temporary additional sampling, collect zero sequence, negative sequence, voltage transient, phasor phase-locked results and spectrum segments around the trigger window, form an event snapshot matrix and bind idempotent keys;

[0065] S3. Calculate the zero-sequence and negative-sequence equivalent incremental impedance distribution on the candidate topology based on the snapshot, and score the consistency of each candidate topology according to the directional consistency grid to obtain the topology confidence of each candidate topology. The candidate topology whose topology confidence meets the preset conditions is determined as the main candidate topology.

[0066] S4. Implement homogenization gating on the main candidate to extract zero-sequence current-voltage hysteresis area transition, phasor phase-locked amplitude-phase transition, and slope-curvature mutation spectrum to generate fault evidence vector.

[0067] S5. Based on the arrival time sequence and amplitude decay law, search for the maximum consistent sub-path on the main candidate to complete the fault type determination and give the segment location confidence.

[0068] S6. Perform linear power flow residual verification on the primary candidate, generate an execution list consisting of four items: sequence, interlock, secondary confirmation, and rollback, solidify the evidence index, and record the execution log.

[0069] The technical connections and implementation logic of the six steps are as follows:

[0070] S1 first establishes a dual-anchor time base using intrinsic and communication anchors, placing multi-source events into an out-of-order buffer. Events are then advanced and replayed for correction according to the waterline. Within the same window, an idempotent key is generated based on device identifier, time anchor identifier, window sequence number, template version, and signature to achieve alignment, providing a unified time and unique index for subsequent stages. S2, under "anchoring" and over-limit triggering, temporary additional sampling is issued. Zero-sequence, negative-sequence, voltage transients, phasor phase-locked loop results, and spectral segments are collected around the trigger window, forming an event snapshot matrix bound with the same idempotent key, supplementing the short-window evidence required for subsequent physical criteria. S3, based on the snapshots, calculates the equivalent incremental impedance distribution of zero-sequence and negative-sequence events on the candidate topology. Consistency scoring is performed using a directional consistency grid composed of active and reactive power increment symbols and phasor sequence relationships to obtain topology credibility and determine the primary candidate, thus constraining subsequent judgments. Based on a trusted network structure; S4 implements homogenization gating on the primary candidates, extracting zero-sequence current-voltage hysteresis area transitions, phase-locked amplitude-phase transitions, and slope-curvature mutation spectra only when the topology credibility meets the standard, generating fault evidence vectors bound to time anchor identifiers and idempotent keys, and separating fault transients from load steps and backflow disturbances; S5 searches for the maximum consistent sub-path on the primary candidates based on the arrival time sequence and amplitude attenuation law along the line, completes the fault type determination and outputs the segment location credibility, making the features consistent with the line geometry and impedance distribution, and the location verifiable; S6 performs linear power flow residual verification on the primary candidates, checking whether the proposed action falls within the safety budget within a short window. If it meets the standard, an execution list containing four items including sequence, interlock, secondary confirmation, and rollback is generated and issued, while the evidence index is solidified and the execution log is recorded. If it does not meet the standard, a rollback strategy suggestion is given. The entire process is connected by time anchors and idempotent keys. The records, snapshots, and judgments of the previous step constitute the triggering and gating conditions for the next step, ensuring that the processing order is unique, the evidence is auditable, and the results are directly executable.

[0071] S1. Extract intrinsic anchors and communication anchors to form a dual-anchor time base, set an out-of-order buffer to advance the water level and replay for correction, and generate idempotent keys according to device identifier, time anchor identifier, window sequence number, template version, and feature signature to complete alignment. The specific implementation is as follows:

[0072] To obtain a verifiable and unified time under conditions of sparse measurement points and message jitter in the distribution network, a dual-anchor time base was constructed in parallel using intrinsic anchors and communication anchors. The intrinsic anchor system is based on repeatable physical characteristics formed by voltage and frequency, including a sequence of inflection points composed of power frequency micro-amplitude swings and voltage gaps or step clusters. It is taken from the voltage, frequency, and zero-sequence quantity of the feeder head and key branches, with units of volts, hertz, and amperes, respectively. The acquisition rhythm can be set to several times per second and marked with millisecond-level tolerances. The communication anchor system is based on signaling characteristics formed by dedicated communication network messages, including batch verification points and round-trip delay transitions. It is taken from the communication records between the station and the master station with batch numbers and time tags. The time uses local time and can be set to simultaneously record time zone offset identifiers to eliminate cross-domain deviations. The time resolution can be set to the range of one to ten milliseconds. The recording and deviation statistics use the same resolution caliber.

[0073] Dual-anchor time base refers to a unified time base that two types of anchors simultaneously satisfy within the same observation window. The anchoring strength is a weighted score that measures the stability of anchor signs and the consistency of multiple measurement points, with a value ranging from zero to one. The number of participating measurement points can be set to no less than two. When it falls below this lower limit, the observation window is automatically shortened and a retry process is initiated. The anchoring threshold can be set as a threshold based on quantile statistics of similar scenarios. The statistical period and sample size are locked along with the template version. The statistical window can be set as a rolling interval of natural days from the most recent three to the most recent thirty days. The sample size can be set to no less than twenty anchored windows. The weights and default values ​​of the stability component and the consistency component are managed by the threshold version. Version switching is based on the anchoring start time. Preferably, the stability component can be set as a normalized combination of the recurrence rate of anchor signs and the amplitude fluctuation range within the observation window. The consistency component can be set as a normalized combination of the anchoring time difference and the amplitude difference of multiple measurement points. The weights and default values ​​of the two are managed by the threshold version and are recorded with each version.

[0074] The acquired measurements and signaling first enter the out-of-order buffer, which is a buffer area for temporarily storing asynchronously arriving entries. The water level line advances monotonically according to the time anchor identifier and window sequence number, which is used to define the entries that can enter alignment. Playback correction rearranges and corrects entries written prematurely or delayed within the same window, ensuring that the same event is only retained at a single time position. When alignment takes effect, an alignment window record with an idempotent key is formed. The idempotent key consists of the device identifier, time anchor identifier, window sequence number, template version, and feature signature. The device identifier is used to uniquely identify the station or device, the time anchor identifier uniquely identifies the current anchoring event, the window sequence number indicates the order of the current observation window in the anchoring sequence, the template version indicates the effective version of the field and caliber, and the feature signature indicates the statistical fingerprint of the observed quantity within this window.

[0075] Alignment window records are written to the time-series warehouse as structured entries. The time-series warehouse is an append-only, non-reversible, time-series persistent storage. Record fields include window identifier, time anchor identifier, start time, end time, anchoring strength, template version, threshold version, and fingerprint value. The window identifier is a unique text number composed of the station number, feeder number, time anchor identifier, and window sequence number, and follows the format rules in the template version. Format verification failure is recorded as an idempotent conflict and enters a secondary adjudication. The fingerprint value forms a traceable reference to the evidence entries of this window and prevents subsequent modification with a verification value. It also covers the version number and window identifier to ensure traceability consistency. It is used for non-reversible verification and traceability, and the specific implementation method is not limited, but it should comply with the one-way summary requirements of the local power industry audit record-keeping specifications. The alignment trigger condition is that both types of anchors are simultaneously satisfied within the same observation window and the anchoring strength reaches the threshold. The stopping condition is anchor failure or the end of the observation window. The observation window covers several power frequency cycles, and its capabilities and range cover station-end measurements and private network communication at the feeder head and key branches, without relying on new high-sampling hardware.

[0076] The dedicated communication network can be configured as an industry-specific wireless private network or a fiber optic Ethernet private network, employing reliable transmission with arrival acknowledgment and timeout retransmission mechanisms. The arrival acknowledgment timeout can be set to one to five seconds, and the timeout retransmission limit can be set to three to five times. The end-to-end latency limit can be set to one minute. Concurrency is handled using a multi-channel independent advancement and rate limiting strategy, with feeders or station ends as channel units. Each channel independently maintains its water level and advances under a shared version library. The channel concurrency limit can be set to one to ten, and the queue limit per channel can be set to ten to one hundred entries, dynamically adjusted according to the rate limiting strategy. The statistics window scrolls by natural day and is locked along with the template version. If an intrinsic anchor and a communication anchor conflict within the same observation window, the fallback rule uses the communication anchor's timestamp as the standard and records the deviation. Simultaneously, the subsequent observation window is shortened, and the system retryes within a limited number of attempts. The maximum number of retry attempts can be set to three and managed by a threshold version. If the maximum is exceeded, the system reverts to the previously verified dual-anchor state and records the reason for the fallback and the deviation value. Both anchor inconsistencies and retries are recorded in the audit repository. These records can be configured to include the alignment basis, threshold version, idempotent key, number of retries, reason for the fallback, timestamp, and operator identifier—a non-fallback record.

[0077] The idempotency strategy ensures that the same idempotent key is written only once; the sequential strategy ensures alignment before storage; the deduplication strategy uses idempotent keys and time anchors to identify freshness-based retention items, and performs secondary adjudication using window and device identifiers when collision detection fails, establishing a subordinate reference relationship for overwritten items using fingerprint values. The verification criteria include multi-source arrival deviation and alignment failure rate. Multi-source arrival deviation is the maximum difference minus the minimum difference at the arrival times of the same event at various measurement points; the deviation recording unit can be set to milliseconds and consistent with the observation window unit. The alignment failure rate is the percentage of windows within a fixed statistical window that are not anchored or whose alignment is incomplete. The security and compliance boundaries are: not exceeding protection settings, not triggering unauthorized actions, and not collecting personal information unrelated to operation; station and device identifiers are only used for traceability, and retention periods and access permissions are set to the minimum according to local regulations.

[0078] The effective boundary of the version is defined by the anchoring start time. Old version records are not rewritten; only new mappings are added. Each anchoring and alignment generates a fingerprint value and version mapping in the time sequence warehouse. Downstream, only idempotent keys and time anchor identifiers are used for retrieval and inheritance, without changing the record content. Preferably, the observation window can be set to ten cycles, the allowed clock drift can be set to two to five seconds, and the end-to-end alignment delay does not exceed one minute. More preferably, the intrinsic anchor can be set as a combination of voltage step clusters and power frequency inflection points to improve the stability of anchoring strength. More preferably, the time delay inflection criterion of the communication anchor can be set as a fixed multiple of deviation superimposed on the historical median as the baseline. To enhance substitutability, the intrinsic anchor can also be set as a feature cluster based on zero-sequence quantities or a feature cluster based on specific harmonic phase alignment, provided that the occurrence frequency is stable, consistent across measurement points, and reproducible within the observation window. If these conditions are not met, it will automatically downgrade to a voltage step cluster scheme and leave a trace.

[0079] S2. Under the condition of anchoring and exceeding the limit trigger, a temporary additional sampling is issued. Around the trigger window, zero sequence, negative sequence, voltage transient, phasor phase-locked loop results, and spectrum segments are collected to form an event snapshot matrix and bind idempotent keys. The specific implementation is as follows:

[0080] Temporary additional sampling is issued when the anchor is established. Anchoring is defined as the simultaneous fulfillment of intrinsic anchor and communication anchor, with anchor strength reaching a threshold. Intrinsic anchor refers to repeatable physical characteristics in voltage and frequency, while communication anchor refers to the stable transition between message batch verification points and round-trip delay. Anchor strength is a zero-to-one index based on characteristic stability and multi-point consistency. The applicable boundary for temporary additional sampling is the trigger window within the alignment window. The alignment window is determined by the start and end times of the dual anchors and is constrained by the water level line, which serves as a progress marker for window advancement. One replay correction is allowed within the time limit before the water level line is exceeded; after exceeding the water level line, the window freezes. The trigger window is determined by the over-limit conditions, which are any one of zero-sequence current, negative-sequence voltage, phasor amplitude-phase change rate, or spectral energy segment exceeding the threshold registered in the operation safety budget. The operation safety budget refers to the set of rated constraints, protection settings, over-limit thresholds, and interlocking rules for feeders and equipment. It is fixed with the template version and threshold version and is stored in read-only format. The operation safety budget is periodically reviewed by the operation and maintenance entity, and change records are generated and fixed with version tracking.

[0081] Temporary sampling is performed using burst frames, which are sampling frames that temporarily increase the sampling rhythm and resolution for a short period of time, preferably lasting for several tens of milliseconds, with a rhythm of several tens of frames per second. Zero-sequence current, negative-sequence current, voltage transients, amplitude and phase of phasor lock, and spectral energy segments are collected around the trigger window, with units of amperes, volts, angles, and hertz, respectively. Time stamps use millisecond-level alignment times; the tolerance of the time stamps can be set to several milliseconds. Exceeding the tolerance is recorded as an alignment anomaly and marked in the log. Preferably, the time stamp tolerance can be set to five milliseconds. All segments are first buffered out of order, and alignment, bandwidth limiting, and noise suppression are completed within the alignment window. When missing measurement points occur, continuous supplementation is performed according to the measurement trend of adjacent moments within the same window, and the confidence level of the supplemented points is marked. The confidence level is an index of zero to one, reflecting the comprehensive reliability of segment completeness, noise level, and alignment quality.

[0082] The short window determination employs a combination of threshold and consecutive count mechanism. If the exceeding amplitude reaches the threshold and recurs within a specified consecutive count, a snapshot is confirmed and retained. If the threshold is not reached or the consecutive count is not met, the current short window terminates. Upon completion, an event snapshot matrix is ​​generated. This matrix is ​​a structured collection of multiple measurement fragments gathered around the trigger window, including window identifier, measurement name, value, unit, time, confidence level, template version, and threshold version. This matrix is ​​bound by idempotent keys, which consist of device identifier, time anchor identifier, window sequence number, template version, and feature signature. The feature signature is a digest fingerprint formed by the combination of participating measurements and rules, used to uniquely identify this snapshot and implement a deduplication strategy. Only one valid write is allowed for the same idempotent key; duplicate writes are registered as idempotent conflicts, and the conflict reason and time stamp are retained.

[0083] The event snapshot matrix is ​​persisted to the snapshot repository. The snapshot repository uses an idempotent key as the primary index and generates an evidence index as a secondary index. The evidence index consists of an idempotent key, a time anchor identifier, a version number, a timestamp, and a device list, and is fixed in the audit repository with a hash sequence number. The hash sequence number is a one-way digest identifier calculated from the evidence index field set, used for tamper-proof verification, and does not restrict the specific implementation method. The audit repository is an append-only archive partition used to store the hash sequence number, version number, and timestamp of the evidence index, and is prohibited from being modified or deleted. The connection with the upstream is achieved by the time base alignment step publishing alignment window records through a dedicated communication network after the water level advances. This step is accessed and started based on the idempotent key. The connection with the downstream is achieved by the topology self-verification step retrieving data from the snapshot repository through a dedicated communication network based on the idempotent key and the time anchor identifier. In terms of time and resource constraints, the end-to-end processing latency can be set to several hundred milliseconds, the concurrent channels can be set to a fixed number based on the feeder or station, the number of retries triggered can be set to three, and no further burst retries are performed after falling back to normal sampling.

[0084] The criteria for sudden failures are limited to three categories: terminal rejection, link unreachability, and resource occupancy. Failure logs must include at least the failure time, device identifier, failure cause enumeration, idempotent key, template version, threshold version, and evidence index number. Failure events are uniformly registered as incomplete snapshots, and the minimum retention period for logs is determined by local security procedures. To improve robustness, one replay correction is allowed within the time limit before the waterline is exceeded. This is used to reassemble segments and cover non-final entries under the same idempotent key. If the requirements are still not met after replay, the snapshot is marked as a candidate snapshot. Candidate snapshots are entries whose snapshot completeness does not reach the set coverage ratio but still have reference value. They are used in subsequent stages with reduced weights and cannot replace the master snapshot under the same idempotent key. The publicly stated definition of snapshot integrity is, within the alignment window, the expected coverage formed by the required sampling channel and the window duration is used as the benchmark, and the comparison is made with the actual effective recording volume and retained to one decimal place; the publicly stated definition of frame loss rate is, within a burst period, the difference between the expected number of sampled frames and the actual number of frames stored is used as the benchmark and retained to one decimal place.

[0085] The boundaries of safety and compliance are clearly defined. Temporary additional collections must not exceed the protection settings and equipment ratings, must not change the system's operating status, and must not collect personal information unrelated to operation. All templates, thresholds, and fields are version locked. New versions are applied from the moment they take effect, while old versions are read-only and included in the evidence index. Preferably, the number of consecutive times can be set to three, with the threshold determined by a certain multiple of the historical median. The burst frames last for a certain number of milliseconds, the rhythm is a certain number of frames per second, and the end-to-end latency does not exceed a certain number of hundreds of milliseconds.

[0086] S3. Calculate the zero-sequence and negative-sequence equivalent incremental impedance distributions on the candidate topologies based on snapshots, and score the consistency of each candidate topology according to the directional consistency grid to obtain the topology reliability of each candidate topology. The candidate topologies whose topology reliability meets the preset conditions are determined as the main candidate topologies. The specific implementation is as follows:

[0087] Physical consistency self-verification of candidate topologies is performed within the alignment window. The alignment window is a time window triggered by dual-anchoring and closed when the water level reaches the tail marker; the window number is incremented from the start time. The dual-anchor time base is jointly established by intrinsic anchors and communication anchors, used for fault-tolerant alignment of multi-source records. The candidate topology set is a finite connectivity state set generated based on ledgers, operation tickets, remote signaling of switches, and connectivity relationships. Its applicable boundary is the same feeder and the reachable branch range of its active and reactive power. The snapshot matrix is ​​a structured set of zero-sequence current, zero-sequence voltage, negative-sequence current, negative-sequence voltage, phasor amplitude and phase, power increment, and time identifier archived by station and branch measurement points around the alignment window. All entries inherit the alignment and denoising results under the dual-anchor time base and are associated with window identifiers, time anchor identifiers, and version information using idempotent keys.

[0088] The observation window is a continuous time period formed by connecting the two short windows before and after the disturbance. The length of the short window and the sampling rhythm are given by the aperture template. Preferably, the observation window covers several power frequency cycles, and the baseline is taken from the stable short window before the alignment window. The stability criterion is that the amplitude fluctuation and phase drift do not exceed the limit set by the aperture template. When the stable short window is not available, it can be set to call the baseline in the aperture template and reduce the weight according to the completeness of evidence. Within the same observation window, the equivalent incremental impedance distribution of zero-sequence and negative-sequence is estimated on each candidate topology. The equivalent incremental impedance is the sequence component impedance increment relative to the baseline, used to characterize the local electrical changes caused by topology and events. Subsequently, a directional consistency grid is constructed. The directional consistency grid is a grid constraint system composed of active power increment symbols, reactive power increment symbols, and phasor sequence relationships at each measurement point. A unified symbol and sequence convention is adopted: active power positive direction is from the feeder to the branch, reactive power positive direction is determined according to the power factor direction in the ledger, phasor sequence is a fixed order of A, B, C, phase angle is in degrees, current transformer polarity is marked according to the ledger, grid points with power increments lower than the template threshold are treated as invalid and are not included in the denominator of evidence completeness, and are only recorded in the audit warehouse.

[0089] The system synthesizes a consistency score by combining the degree of agreement of the equivalent incremental impedance distribution with the satisfaction of the directional consistency grid. The consistency score is a scalar value ranging from zero to one. Based on this, a topological credibility score is generated by combining the completeness of evidence, the freshness of evidence, and the consistency of the caliber according to a fixed priority. The completeness of evidence is the proportion of all collected items specified in the caliber template and is normalized to zero to one. The freshness of evidence is the timeliness of the snapshot relative to the alignment window and is normalized to zero to one. The consistency of the caliber is the degree of matching between the measurement unit, rhythm, and template version and is normalized to zero to one. The topological credibility score is also normalized to zero to one. Candidates whose topological credibility score reaches the threshold are marked as primary candidates, while those that do not reach the threshold are added to the waiting list, and the scoring process, weight parameters, and evidence citations are simultaneously retained. When scores are tied, the satisfaction of the directional consistency grid, the freshness of evidence, and the consistency of the caliber are compared in turn. If they are still tied, the freshness score is used as the final decision based on the time anchor. The inclusion criteria for the candidate topology set are that the ledger and operation ticket are within the freshness range specified by the template and the connectivity relationship is reachable; the elimination criteria are that key evidence is continuously missing or the direction consistency grid is not satisfied within the continuous window; an upper limit is set for the number of candidates, and when the upper limit is exceeded, the candidates are pruned according to the topology credibility and evidence freshness, and the evidence index and version signature of the pruned records are retained.

[0090] Topology records are written to the topology repository in a structured manner and undergo version locking and evidence chain solidification. Each topology record includes at least the candidate number, topology credibility, consistency score, evidence completeness, evidence freshness, consistency of interpretation, number of evidence items, primary candidate identifier, time anchor identifier, window identifier, template version, threshold version, and fingerprint value. The topology repository is a window-segmented, write-only logical storage domain, using append-only writing and prohibiting overwriting and in-situ modification. The audit repository is an append-only evidence retention domain used to solidify fingerprint values ​​and version signatures; deletion and recycling are prohibited, and the media format can be set to log storage with sequential write characteristics. The location and feature analysis stages access topology records using the primary candidate identifier and aligned window identifier. When multiple primary candidates exist within the same window, they are re-evaluated according to the direction consistency priority rule. If they are still tied, only the record with the highest topology credibility is subject to homogenization gating; other tied records are retained in shadow form for offline review.

[0091] Topological self-verification begins after the window closes, with computation delay not exceeding the alignment window duration. Parallel granularity is set with an upper limit based on feeder division. A warehouse is considered unreachable if three read / write attempts time out or the connection fails within the alignment window. The number of retries can be set to three with fixed backoff; exceeding this limit registers the warehouse as unreachable. Cross-task conflicts are resolved using mutexes and idempotent keys. Evidence conflicts occur when different sources point to opposing key conclusions within the same observation window. When an evidence conflict occurs, a topological evidence contradiction is registered, triggering a replay correction. The replay cooldown time is no less than a short window; replays cease after exceeding the upper limit. After re-evaluation, dual records are retained, and priority is assigned to the direction consistency grid for adjudication.

[0092] The idempotency policy stipulates that the same idempotent key is only written to the main record once. Subsequent writes within the same window overwrite the main record according to the freshness of the time anchor identifier and the threshold version, while the overwritten content is retained as a shadow record. Cross-window references follow a fixed order of increasing window number, and cross-order access is prohibited. The interface is unified as follows: candidate number, time anchor identifier, window identifier, template version, threshold version, and evidence index are text types; topological credibility, consistency score, evidence completeness, evidence freshness, and consistency are decimal types from zero to one; the number of evidence items is a count type; and the main candidate identifier is a Boolean value. The phasor amplitude unit is consistent with the corresponding measurement: voltage is in volts, current is in amperes, and phase angle is in degrees. The return status is mapped to success, insufficient evidence, evidence conflict, version mismatch, and repository unreachable, and corresponding text descriptions are recorded in the audit repository.

[0093] The caliber template is uniformly configured, including measurement name, unit, sampling rhythm, short window length, observation window width, baseline stability limit, scoring weight, threshold value, parallel upper limit, playback cooldown time, candidate number upper limit, and version identifier. Template changes only affect the alignment windows generated subsequently. The weight and threshold are locked by the template version, with a value range of zero to one. Preferably, the caliber template lists the upper limit of uncertainty and calibration period for each measurement; when the upper limit is exceeded, the relevant evidence is processed with reduced weight. The idempotent key consists of device identifier, time anchor identifier, window sequence number, template version, threshold version, and feature signature, used to uniquely identify the same window record throughout the entire link. Any duplicate writes with the same key are overwritten according to the freshness of the time anchor identifier and the threshold version, and a shadow record is retained. The shadow record is a read-only copy of the overwritten historical master record, used only for auditing and backtracking, and does not participate in online adjudication or downstream gating. The freshness of the time anchor identifier is a quantitative index of zero to one, used to characterize the closest time anchor identifier within the alignment window, preferably represented by the normalized difference between the latest anchoring time within the window and the end time of the window.

[0094] Candidate set management follows template parameters. The upper limit for the number of candidates, the replay cooldown time, and the parallel limit are positive integers and configurable. When the upper limit is exceeded, candidates are pruned according to topology credibility and evidence freshness, while retaining the evidence index and version signature. When the cross-feeder communication status changes and enters a new window, the main candidate of the previous window cannot be pushed out of the window to the new topology. Batch statistics retain the correct hit rate and false positive rate in the audit warehouse. Preferably, the sample size of each batch is not less than the lower limit of the template, and the confidence description caliber is recorded. When both anchors fail simultaneously, only the candidate set and record writing are retained, without triggering scoring and adjudication, and the anchor failure is registered in the audit warehouse. Privacy and retention policies declare the minimum retention period and access permission level in the caliber template. All access follows the site security control and operation isolation. Historical windows are not retrospectively rewritten. The evidence chain is solidified in the audit warehouse at the window granularity, thus forming a closed loop from evidence acquisition, physical self-verification, credibility quantification to record retention.

[0095] S4. Implement homogenization gating on the primary candidate, extract zero-sequence current-voltage hysteresis area transition, phasor phase-locked amplitude-phase transition, and slope-curvature abrupt change spectrum to generate fault evidence vectors. The specific implementation is as follows:

[0096] Under the constraint of the primary candidate topology, homogenization gating is implemented. Homogenization gating is a gating mechanism that triggers feature extraction only within the primary candidate topology determined by the physical consistency score in the previous step and within the alignment window boundary. This is used to limit the judgment to the real connection relationship and unified time base to reduce cross-layer misjudgment. The physical consistency score is a normalized quantity synthesized by the equivalent incremental impedance distribution consistency score and the directional consistency grid score of the candidate topology according to a fixed priority and weight within the alignment window. The value range is zero to one. The consistency score is used to measure the matching degree between the measurement snapshot and the candidate topology in the zero-order and negative-order incremental impedance distribution. The directional consistency grid score is used to measure the grid constraint satisfaction of the active and reactive increments and phasor order relationship. The weights, synthesis methods and thresholds of the two scores are managed by the version lock of the template version and the threshold version and are fixed along with the scoring record. The scoring record includes the candidate number, the two scores, the synthesis result, the version number and the evidence index.

[0097] The primary candidate topology is the network connection relationship in the candidate set that reaches a confidence threshold and is identified as a preferred scheme, including path length, node order, impedance distribution, and switching status; the alignment window is a time segment formed by dual-anchor time base alignment, including window identifier, start time, end time, and anchoring strength. The anchoring strength is a normalized quantity of the degree of satisfaction between intrinsic anchors and communication anchors, with a value range of zero to one. The threshold is placed in the preferred embodiment and locked along with the template version; the snapshot is a set of multiple measurement segments formed around the trigger segment and stored in the snapshot warehouse, which has been bound by idempotent keys. The idempotent keys are identified by the device identifier, The combination of time anchor identifier, window sequence number, template version, and feature signature is used to uniquely identify the same event throughout the entire link. The feature signature is a short text fingerprint generated in a fixed order by the measurement item name, frame sample rhythm, device identifier, and current window correction parameters, and is managed by version lock. The template version refers to the version identifier of the feature item set, field name, unit, and record format, which is used to ensure consistent interpretation across windows and stations. The version lock is the version management rule for templates, thresholds, and parameters. Once a version is effective, it cannot be overwritten. Only new versions are allowed. The effective boundary is marked by a timestamp, and historical records are not retroactively recalculated.

[0098] After entering the homogenization gating, bandwidth limiting and baseline correction are first performed within the measurement point range covered by the main candidate topology. Bandwidth limiting restricts the measurement content to the target frequency band to suppress irrelevant components. The target frequency band can be set to be centered on the zero-sequence principal component and the negative-sequence principal component of the fundamental frequency and cover several low-order harmonic intervals. The upper and lower limits of the frequency band and the harmonic order are uniformly listed in the preferred embodiment and version locking is performed. Baseline correction is to deduct background drift and DC bias and unify the reference reference. The a priori baseline refers to the amplitude and phase reference calculated based on the stable frame samples of the reference segment within the same alignment window. The reference segment can be set as several frame samples at the beginning of the window. The specific number of frames is set in the preferred embodiment. The a priori baseline is locked along with the template version.

[0099] Subsequently, four types of feature indicators are extracted according to the frame sample rhythm within a short-time window. The short-time window can be set as a continuous time period spanning several cycles of the power frequency. The frame sample is a set of single measurements collected within the short-time window at a fixed step size and then subjected to bandwidth limiting and baseline correction. The frame sample rhythm can be set to several tens of frames per second, with specific values ​​provided in the preferred embodiment. The four types of feature indicators include zero-sequence current-voltage hysteresis area transition, phase-locked loop amplitude-phase transition, slope abrupt change, and curvature abrupt change. Among them, the zero-sequence current-voltage hysteresis area transition is the area enclosed by the zero-sequence current and zero-sequence voltage phase diagrams within the short-time window in adjacent phases. The increment between sub-windows is used to characterize nonlinear energy release; phase-locked amplitude-phase transition is the discrete transition of amplitude and phase change of negative-sequence or zero-sequence phasors under amplitude-phase locking conditions. The amplitude-phase locking conditions can be set to ensure that the phase fluctuation is not higher than a preset angle threshold and the amplitude drift is not higher than a preset amplitude threshold within the same short-time window. The reference phase and reference amplitude are derived from the prior baseline within the alignment window, and the relevant thresholds are version-locked; slope mutation and curvature mutation are the intensity of first-order and second-order shape changes of the measured curve near the edge, used to identify the discontinuous characteristics of intermittent arcs;

[0100] Features are characterized by amperes, volts, angles, and dimensionless quantities; angles are uniformly defined as electrical angles, and the zero-phase reference is determined based on the prior baseline within the alignment window, with the aperture locked according to the template version. Triggering is based on a combination of threshold and duration criteria. The threshold is the minimum amplitude or shape intensity required for each indicator to reach the judgment, and the duration is the minimum count that reaches the threshold within a certain number of consecutive frames of samples. Feature extraction is initiated when the confidence of the primary candidate topology is not lower than the threshold and the snapshot integrity reaches the specified level. The coverage of key measurement items refers to the available ratio and synchronization ratio of four types of measurements—zero-sequence quantity, negative-sequence quantity, voltage transient, and amplitude-phase locking result—within the short-time window. Both must simultaneously meet the preset ratio to be considered complete, and the ratio threshold is placed in the preferred embodiment. If any condition is not met, it is registered as a gate failure and kept pending. The registration code for gate failure adopts a unified enumeration, which includes at least three categories: threshold failure, insufficient snapshot integrity, and resource throttling.

[0101] To enhance resistance to backflow, a backflow immunization strategy is adopted: using the equivalent source of the main candidate topology as the prior baseline, the stable backflow component is projected onto a subspace orthogonal to the fault-sensitive direction before feature determination; a stable backflow indicator refers to the situation where a continuous number of frames of samples show a reverse active power increment accompanied by a reactive power increment changing in the same direction on the main candidate path. When this indicator is met, the backflow immunization strategy is triggered, and the number of frames and the criterion are placed in the preferred embodiment; when the residual amount after orthogonal projection still exceeds the preset threshold, it is downgraded to an amplitude-phase transition indication of the channel differential, and a downgraded identifier is written into the record. The downgraded identifier participates in the determination in the downstream positioning stage according to the weight reduction rule, and the reduction ratio is placed in the preferred embodiment and associated with the evidence index in the audit warehouse;

[0102] After feature extraction, a fault evidence vector is formed. The fault evidence vector is a set of evidence items consisting of hysteresis area transitions, amplitude transitions, phase transitions, slope abrupt changes, and curvature abrupt changes under the same alignment window and the same primary candidate topology. It is accompanied by window identifier, primary candidate identifier, idempotent key, template version, and threshold version to record the source, caliber, and version locking relationship. The evidence vector is written into the evidence repository and provided to the positioning stage for invocation according to the window identifier and primary candidate identifier through the publish-subscribe channel. The publish-subscribe channel is an internal message channel with arrival confirmation and secondary confirmation mechanisms as well as retry control. Arrival confirmation is a message from the receiver confirming that the record has arrived. Secondary confirmation is a confirmation message from the master station within the same channel for the idempotent key of the record to check the channel consistency in a closed loop. Both arrival confirmation and secondary confirmation are subject to timeout control and retry limit. The timeout for arrival confirmation and secondary confirmation can be set to a default value in the hundreds of milliseconds. The default value can be adjusted through the parameter table and is managed by version lock. The parameter table is an internal read-only configuration list containing default values, upper and lower limits, units, and version numbers. It can be issued for updates, but it can only take effect after arrival confirmation and secondary confirmation are completed.

[0103] To ensure consistent order and deduplication, only one valid write is allowed for the same idempotent key. When duplicate entries are found, the latest record is retained based on the freshness of the time anchor identifier. The freshness of the time anchor identifier refers to the priority determined by the combination of the time anchor identifier version and the time identifier, prioritizing records with newer versions and later time identifiers. To ensure timeliness, the end-to-end latency of feature extraction is capped, preferably in the hundreds of milliseconds range. Concurrency can be set to multi-channel partitioned by feeder. When resource contention occurs, entries are placed in a queuing queue, which adopts a first-in, first-out (FIFO) approach. If the queue times out, a resource throttling code is registered and feature extraction for that window is terminated. Within the allowed range, one replay is allowed to fill short-term gaps. Replay is a mechanism to re-evaluate reached entries within the same window without modifying the original records, and a replay entry is associated with the evidence index to indicate that a record has been kept.

[0104] Regarding security and compliance boundaries, this step does not issue control actions, does not change the status of primary equipment, and does not exceed protection settings. Full-link access is limited to measurement reading and record storage. Each run generates an evidence index and stores it in the audit repository. The evidence index includes idempotent keys, time anchor identifiers, template versions, threshold versions, and participating device numbers. The participating device numbers adopt a structure of station number plus device number plus loop number and are kept unique across the entire network. The field format and value range are set in the preferred implementation. Templates, thresholds, and rules are all version locked. New versions are applicable to subsequent windows from the moment they take effect, and historical records are not backed up and recalculated.

[0105] Abnormal situations are registered using codes. The error code field consists of a code and a description. The code includes four starting items: unmet gating threshold, insufficient snapshot integrity, resource rate limiting, and unconfirmed arrival. The description records the version number, idempotent key, and time anchor identifier. New codes should be registered in the version lock. After all conditions are met and the evidence vector is successfully written, a consumable notification is sent through the publish-subscribe channel, and the state transition to readable is completed in the evidence warehouse. This enables the positioning process to carry out spatiotemporal consistency constraints and segment positioning under the premise of primary candidate topology, unified time base, and clear caliber.

[0106] S5. Based on the arrival time sequence and amplitude attenuation law, search for the maximum consistent subpath on the primary candidate to complete the fault type determination and provide the segment location confidence. The specific implementation is as follows:

[0107] Under the alignment window, location determination is performed within the path length and equivalent impedance range of the primary candidate topology. The alignment window is a time interval formed by the time base alignment module under dual-anchor time base conditions; the time anchor identifier is used to uniquely mark this interval. The idempotent key consists of the device identifier, time anchor identifier, window sequence number, template version, and feature signature, used for order constraints and deduplication. The primary candidate topology is a reliable topology that meets the threshold based on the directional consistency grid and incremental impedance tomography scoring by the topology self-verification module. The directional consistency grid is a set of grid constraints composed of active power increment symbols, reactive power increment symbols, and phasor order relationships. Incremental impedance tomography compares the consistency between the equivalent incremental impedance distribution formed by snapshots within a short window before and after a fault and the calculated distribution of the candidate topology.

[0108] The evidence vector consists of hysteresis area transitions, amplitude transitions, phase transitions, slope abrupt changes, and curvature abrupt changes, and is solidified in the evidence bin using idempotent keys and time anchors. The arrival time sequence is the order of the evidence vector at each measurement point, ordered according to the unified time identifier within the alignment window; the adjudication order for parallel times is based on the fixed order of the equipment identifiers. The amplitude attenuation law refers to the monotonically decreasing or stepwise retreating trend of zero-sequence quantities, negative-sequence quantities, and the aforementioned characteristic amplitudes along the path length and the direction of increase in equivalent impedance. The judgment caliber uses the amplitude ratio and attenuation direction of adjacent measurement points within the same alignment window. The short window is a finite time period covering several power frequency cycles, preferably ten cycles; the freshness of the time anchor identifier is graded according to the time interval from the end of the alignment window to the present, used for the retention and overlay adjudication of parallel records, and recorded along with the version lock; the playback window is limited to the section within the alignment window that has not reached the water level and is still in the out-of-order buffer, allowing only one retrieval and solidification using the original idempotent keys;

[0109] The localization process begins with geometric consistency screening, eliminating entries that are not sequential or whose attenuation direction is opposite to that of neighboring points, and merging them according to path length steps. Then, within the observation window of adjacent short windows before and after a coverage failure, the matching of arrival order and attenuation trend with the path metric of the main candidate topology is used as a trigger. When the evidence coverage reaches the minimum coverage ratio and the number of occurrences meets the requirements, categories and segments are formed. The observation window refers to a continuous time period formed by merging adjacent short windows in chronological order within the same alignment window, with its boundaries defined by the start and end times of the alignment window. Evidence coverage is the percentage of effective measurement points covered by evidence on the main candidate path; consistency residual is the deviation between the evidence entry and the path metric under the constraints of arrival order and attenuation; and gating satisfaction is the percentage of entries whose feature thresholds are met. The weights and calculation methods of these three are uniformly managed by the template version, threshold version, and weight version. The weight version uses the same version numbering system as the aforementioned versions and is fixed with the same hash along with the evidence index.

[0110] If the coverage is below the minimum coverage ratio or a continuous path is not formed, it is marked as pending. Within the allowed range, a replay is triggered to fill the missing parts within the same window; if a continuous path is still not formed, it is registered as location non-convergence. The location record includes segment identifier, fault type, location reliability index, time anchor identifier, idempotent key, primary candidate identifier, template version, threshold version, weighted version, and evidence index. The evidence index is hashed to solidify the evidence source, version, and device list, and is written to the audit repository; the location reliability value ranges from zero to one, and is obtained based on a weighted combination of coverage, consistency residual, and gate satisfaction. Records are written to the location repository and can be retrieved and subscribed by idempotent key and time anchor identifier. The same idempotent key is only validly written once; when duplicates arrive, the freshness of the time anchor identifier and the priority of the threshold version are used to decide whether to retain or overwrite, and all changes are recorded. Communication uses a sequential channel with a sequence mark; the sequential channel is a reliable transmission channel that maintains the order of messages, carries a sequence mark, and provides one retransmission capability; it can be set to retransmit once when there is network jitter.

[0111] The end-to-end delay limit can be set as the upper bound of a short window time. Concurrency is limited at the feeder level. Insufficient resources result in queuing and recording the duration. Unprocessed entries in the queue are marked as overdue for processing. The applicable boundary is limited to the reachable section covered by the primary candidate topology. Evidence exceeding the boundary is not included in the consistency calculation and its source is marked. Switch position, path length, and equivalent impedance are jointly provided by the ledger and telemetry, with units of meters and ohms, respectively. Electrical angles are uniformly expressed in "degrees (electric angles)" and synchronized in the positioning and auditing warehouses with the same version number. On-site verification is based on section hit rate, misclassification rate, and positioning coverage rate. The sample size can be set to the work order size that meets the statistical confidence requirements. Preferably, the positioning confidence threshold can be set to 0.9, the number of consecutive satisfactions can be set to three, and the representative value is placed in the preferred embodiment. The safety and compliance boundaries do not exceed the protection settings and operation permissions. When there is a conflict with interlocking or locking rules, the positioning record can still generate and solidify the evidence index. The execution stage refuses to issue the record according to the rules and registers the conflict reason in the auditing warehouse.

[0112] In terms of robustness, individual measurement points arriving in opposite directions can be included in the consistency score with low weight; when the distance between adjacent measurement points differs significantly, the attenuation direction weighted by path length can be used to determine the caliber. An alternative approach is to use the difference constraint between adjacent switches to replace the attenuation law determination in scenarios with sparse measurement points, thus maintaining the sequential relationship and traceability mechanism of the positioning process. The entire link is constrained by version locking and the solidification of the evidence chain; any changes to the caliber, weight, and threshold are limited to the effective time and retain the entire history to ensure consistency and traceability of the review.

[0113] S6. Perform linear power flow residual verification on the primary candidate, generate an execution list consisting of four items: sequence, interlock, secondary confirmation, and rollback, solidify the evidence index, and record the execution log.

[0114] Linear power flow residual verification on the primary candidate topology involves a lightweight verification of potential limit-crossing risks within the safety budget, generating a traceable execution list. The primary candidate topology refers to the preferred network structure determined based on physical consistency scoring. The safety budget refers to the acceptable boundaries set for current upper limits, voltage upper and lower limits, allowable temperature rise, breaking capacity, closing capacity, backfeed risk, and transfer capacity. Residual capacity refers to the minimum margin of key constraints relative to the corresponding safety budget under the action of the proposed execution sequence, with units consistent with the constraint quantities. The alignment window is a decision window defined by time anchor identifiers under dual-anchor time bases, used to carry the measurement and recording of the same time segment.

[0115] To perform verification, positioning records are obtained from the self-positioning warehouse, primary candidate topologies are obtained from the topology warehouse, and current measurements are collected. The positioning records include section identifiers, fault categories, positioning reliability, time anchor identifiers, version information, and evidence indexes. The primary candidate topologies include switch status, impedance parameters, reachable path sets, and version information and evidence indexes. Current measurements cover current, voltage, frequency, phase, and temperature at the station end and feeder head, key branches, and candidate switching points, and the measurement freshness is marked. Measurement freshness refers to the lag threshold between the measurement acquisition time and the judgment time, which can be set to the second level. When the threshold is exceeded, the measurement is derated according to a conservative coefficient and participates in the residual judgment. When the upper limit is exceeded, the measurement does not participate in the judgment and is recorded as an entry. The conservative coefficient refers to the derating ratio of related measurements participating in the judgment when the measurement freshness exceeds the threshold, and is managed as a numerical parameter in the threshold version. The control repository, audit repository, location repository, and topology repository are all logical storage domains, which respectively store the execution list, evidence index and execution log, location record, and main candidate topology. All items are associated with an idempotent key, which consists of device identifier, time anchor identifier, window sequence number, template version, and feature signature.

[0116] Before entering the verification process, gating is performed: the execution sequence is compared item by item with the interlock table, interlock table, and equipment rating. The interlock table shows the sequence and mutual constraints between switches, the interlock table is a set of steps prohibited due to maintenance, defects, or operating conditions, and the equipment rating is the boundary of the circuit breaker, load switch, and disconnector's capacity. After gating is passed, residual capacity is judged within a short window that covers the alignment window and extends backward to the effective period of the step sequence. The short window duration, threshold, and duration are versioned parameters, and the capacity and range can be set to cover several to dozens of power frequency cycles, with the duration being a specified integer number of times. Secondary confirmation is triggered when it involves tie closing, crossing important busbars, or approaching the protection setting boundary.

[0117] When the residual capacity reaches the threshold and the required number of durations is met, an execution list is generated. The execution list is a set of actionable steps, including four components: sequence, interlock, secondary confirmation, and rollback strategy. The sequence refers to the order of opening and closing steps that must be completed sequentially. Secondary confirmation refers to checkpoints confirmed by the upper-level system or on-duty personnel before critical steps. The rollback strategy is a reset plan in case of verification failure or execution obstruction. The template version refers to the version of the execution list structure and step sequence template; the threshold version refers to the version of the threshold and duration; and the time anchor identifier is the time label under a dual-anchor time base. The execution list is written to the control module and sent to the station-end device via the dedicated communication network for distribution automation, requiring a confirmation receipt. The dedicated communication network for distribution automation is a dedicated communication network used for distribution automation. The confirmation receipt is a confirmation message returned by the station-end device after receiving the list. Both the list and the receipt carry fingerprint values, which are tamper-proof identifiers generated based on the message content and time, used for link verification and auditing.

[0118] If the residual capacity does not reach the threshold or the gating is restricted, a rollback strategy suggestion is generated, specifying the rejection reason, triggering items, and alternative steps. This suggestion is then written to the audit repository after being associated with the location record and the version information of the primary candidate topology. The rollback strategy suggestion is a non-executable item and is only for decision-making reference. The communication side specifies end-to-end time limits, concurrency, and retry strategies. The end-to-end time limit is the maximum allowed time from initiating residual capacity verification to receiving confirmation from the station, and can be set to an upper limit that conforms to the station's operational specifications. Concurrency is limited at the station level. Retry can be set to a specified number of times and employ a backoff strategy. The backoff strategy can be set to fixed backoff or incremental backoff to control the retry interval and avoid link congestion. The number of retry attempts is managed as a persistence count field in the threshold version. If no confirmation is received after cumulative retries, a rollback strategy suggestion is triggered and the missing confirmation number is registered.

[0119] The entire process employs idempotency, sequential, and deduplication strategies: the same idempotent key is only allowed to take effect once in the control warehouse; the sequence is fixed as gating first, then verification, then distribution, and confirmation before execution; the control warehouse determines duplicates based on idempotent keys, while the station determines duplicates based on a combination of time anchor identifiers and device identifiers. The audit side generates an evidence index and stores it in the audit warehouse. The evidence index includes idempotent keys, time anchor identifiers, device lists, version numbers, and fingerprint values; the execution log synchronously records the step sequence time, station number, device identifier, step sequence status, threshold version, evidence index, and error number. Error numbers use a unified enumeration table, including verification exceeding budget, step sequence being locked, missing receipts, fingerprint value conflicts, and inconsistent time bases, for subsequent review and statistics. Version locking covers the interlock table, locking table, security budget, threshold, template, and communication protocol. Any update is delimited by the effective time and historical versions are read-only. All records are linked together by the evidence index to form a traceable evidence chain.

[0120] The safety and compliance boundaries are clearly defined: protection settings must not be exceeded, reverse power transmission is prohibited, and unauthorized unlocking of equipment is forbidden; steps involving personnel confirmation are only executed after secondary confirmation; records involving station and personnel identification are managed according to permissions and retention periods. In terms of capability and scope, this verification can provide a rapid and traceable execution list for isolation, transfer, and interconnection looping scenarios within the primary candidate topology without adding or only adding a small number of sensors, maintaining stability under conditions of communication jitter, minor measurement loss, and clock drift; parameters and thresholds can be configured according to station capabilities. Preferably, the end-to-end time limit can be set to a time period within the upper limit conforming to the station's operating specifications, the number of retries can be set to a limited number, concurrent current limiting can be configured according to station capabilities, and the measurement freshness threshold can be set to the second level; when the credibility of the primary candidate topology decreases, the confirmation level can be temporarily increased and the executable steps restricted; when the residual judgment is insufficient, manual secondary confirmation can be introduced before issuing the execution list to ensure that an executable and auditable distribution network automation control closed loop is formed within the established safety budget.

[0121] Example 2: Figure 2 A schematic diagram of a distribution network automation control system with fault self-identification is provided according to the present invention. The distribution network automation control system with fault self-identification is used to implement the aforementioned distribution network automation control method with fault self-identification, comprising:

[0122] Time base alignment module: used to extract intrinsic anchor and communication anchor to form a dual anchor time base, set up out-of-order buffer to advance the water level and replay for correction, generate idempotent keys based on device identifier, time anchor identifier, window sequence number, template version, and feature signature, and complete alignment and deduplication;

[0123] Snapshot Acquisition Module: Used to issue temporary additional acquisition when anchoring and exceeding the limit triggering, to collect zero sequence, negative sequence, voltage transient, phasor phase-locked loop results and spectrum segments around the trigger window, generate an event snapshot matrix and bind idempotent keys;

[0124] Topology self-verification module: It is used to calculate the zero-sequence and negative-sequence equivalent incremental impedance distribution on the candidate topology based on snapshots, and to score the consistency of each candidate topology according to the directional consistency grid to obtain the topology credibility of each candidate topology. The candidate topology whose topology credibility meets the preset conditions is determined as the main candidate topology.

[0125] Homogenization feature extraction module: used to implement homogenization gating on the main candidate, extract zero-sequence current-voltage hysteresis area transition, phasor phase-locked amplitude-phase transition, slope-curvature mutation spectrum, and form fault evidence vector;

[0126] Location determination module: used to search for the most consistent sub-path on the main candidate based on arrival time sequence and amplitude attenuation law, determine the fault type and calculate the segment location confidence;

[0127] The execution verification and audit module is used to perform linear power flow residual verification on the primary candidate, outputting an execution list containing four items: sequence, interlock, secondary confirmation, and rollback, solidifying the evidence index, and recording the execution log.

[0128] The technical connections and implementation logic of the six modules are as follows:

[0129] The entire process follows a closed-loop linkage of "alignment → authentication → self-certification topology → feature separation → location → execution": The time base alignment module first extracts intrinsic anchors and communication anchors to form a dual-anchor time base, advances and replays corrections within the out-of-order buffer using a waterline approach, and simultaneously generates idempotent keys based on device identifiers, time anchor identifiers, window sequence numbers, template versions, and feature signatures to form reusable alignment window records; When anchoring is achieved and triggering exceeds limits, the snapshot acquisition module performs temporary additional acquisition of zero-sequence, negative-sequence, voltage transients, phasor phase-locked loop results, and spectrum segments around the trigger window, generating an event snapshot matrix and binding it with the same idempotent key; The topology self-certification module uses snapshots to calculate the equivalent incremental impedance of zero-sequence and negative-sequence on the candidate topology set. The distribution is analyzed, and combined with directional consistency grid scoring, to obtain topological credibility and determine the main candidates. The homogenization feature extraction module only enables homogenization gating on the main candidates, extracting zero-sequence current-voltage hysteresis area transitions, phase-locked amplitude-phase transitions, and slope-curvature mutation spectra, which are combined into fault evidence vectors. The location determination module uses the arrival time sequence and amplitude attenuation law of the evidence vectors on the main candidates as constraints to search for the maximum consistent sub-path, output the fault type, and calculate the segment location credibility. The execution verification and auditing module performs linear power flow residual verification with the main candidates based on the location results. After passing the verification, an execution list containing four items—sequence, interlocking, secondary confirmation, and rollback—is generated, while the evidence index is solidified and the execution log is recorded. The entire link is connected by time anchor identifiers and idempotent keys. The record of each step upstream serves as the unique reference object for the next step, ensuring unidirectional judgment order, traceable evidence, and direct results to the field control side.

[0130] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0131] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0132] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission; wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission includes infrared, microwave, etc. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center containing one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0133] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0134] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0135] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0136] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0137] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0138] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0139] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A power distribution network automation control method with fault self-identification, characterized by, Comprise: S1, extract the intrinsic anchor and the communication anchor to form a double-anchor time base, set the reorder buffer advance water line and playback correction, generate an idempotent key according to device identification, time anchor identification, window number, template version and feature signature to complete alignment; S2, under the condition of anchor and limit trigger, issue temporary increase sampling, collect zero sequence, negative sequence, voltage transient, phasor phase-locked result and spectrum segment around the trigger window, form event snapshot matrix and bind idempotent key; S3, based on the snapshot, calculate the zero sequence and negative sequence equivalent incremental impedance distribution on the candidate topology, and score the consistency of each candidate topology according to the direction consistency character, get the topology credibility of each candidate topology, and determine the main candidate topology when the topology credibility meets the preset condition; S4, implement the same layer gating on the main candidate, extract the zero sequence current voltage hysteresis area transition, phasor phase-locked amplitude phase transition, slope curvature mutation spectrum, and generate fault evidence vector; S5, search for the maximum consistent sub-path on the main candidate according to the arrival time sequence and amplitude attenuation law, complete the fault type judgment and give the section positioning confidence; S6, execute linear flow residual amount check on the main candidate, generate an execution list composed of four items of sequence, interlocking, secondary confirmation and withdrawal, solidify the evidence index and record the execution log; Wherein S1 comprises: Build a double-anchor time base in parallel with the intrinsic anchor and the communication anchor, and determine the anchor when both anchors meet the threshold in the observation window; Write measurements and signals into the reorder buffer, advance the water line according to the time anchor identification and window number, and playback correction in the same observation window; Generate an idempotent key according to device identification, time anchor identification, window number, template version and feature signature, form an alignment window record, write it into the time sequence warehouse and generate an evidence index; When the two anchors are inconsistent, decide according to the communication anchor time mark and register the deviation; Wherein S3 comprises: In the alignment window, based on the event snapshot matrix bound by the double-anchor time base, estimate the zero sequence equivalent incremental impedance distribution and the negative sequence equivalent incremental impedance distribution on each candidate topology; Refer to the baseline of the alignment window to construct the direction consistency character composed of active incremental sign, reactive incremental sign and phasor sequence relationship; Generate a consistency score according to the preset weight, and store the consistency score, window identification, time anchor identification, template version and threshold version in the topology warehouse in association with the idempotent key; Based on the generated consistency score, generate the topology credibility according to the preset priority combining evidence integrity, evidence freshness and caliber consistency; Determine the main candidate topology when the topology credibility meets the threshold; When the scores are equal, decide in turn according to the direction consistency character satisfaction, evidence freshness and caliber consistency; Generate a topology record containing candidate number, topology credibility, consistency score, main candidate identification, time anchor identification, window identification, template version, threshold version and fingerprint value, and store it in the topology warehouse in association with the idempotent key, and solidify the shadow record and audit warehouse fingerprint value; Wherein S4 comprises: In the main candidate topology and within the alignment window, when the main candidate topology credibility and snapshot integrity reach the threshold, call the snapshot from the snapshot warehouse according to the idempotent key, perform bandwidth limiting and baseline correction in turn according to the template version and the threshold version, and extract the zero sequence current voltage hysteresis area transition, phase-locked amplitude phase transition, slope mutation, and curvature mutation to generate the fault evidence vector; Detect stable backflow signs and execute the moisture immune strategy and generate the degradation identifier when the criterion is met; Solidify the fault evidence vector to the evidence warehouse together with the window identifier, the main candidate identifier, the time anchor identifier, the idempotent key, the template version, and the threshold version, and complete the arrival confirmation and secondary confirmation through the publication and subscription channel; The same idempotent key is only valid once, and the later record is covered according to the time anchor identifier freshness and is solidified to leave a trace in the audit warehouse; Wherein S5 comprises: Within the alignment window, based on the path length and equivalent impedance of the main candidate topology, call the evidence vector solidified with the idempotent key and the time anchor identifier, and implement geometric consistency screening according to the arrival time sequence and amplitude decay law; When the evidence coverage, consistency residual error, and gate satisfaction meet the criterion, search for the maximum consistent sub-path and form a positioning record, which includes the section identifier, the fault type, the positioning confidence, the time anchor identifier, the idempotent key, the main candidate identifier, the template version, the threshold version, the weight version, and the evidence index; The evidence beyond the reachable section of the main candidate topology is not included in the consistency calculation.

2. The power distribution network automation control method with fault self-identification according to claim 1, characterized in that S2 Comprises: When the anchor is formed and the out-of-limit condition is met, issue the temporary increased burst frame collection instruction within the alignment window; Place the zero sequence current, negative sequence current, voltage transient, and phasor phase-locked result and the frequency spectrum energy segment into the out-of-order buffer area, advance according to the water line and replay the correction to complete the alignment, perform bandwidth limiting processing and noise suppression, continuously fill in the missing points and mark the confidence; According to the threshold and the number of continuous times, confirm the retained snapshot, generate the event snapshot matrix, and bind it with the idempotent key, which is composed of the device identifier, the time anchor identifier, the window serial number, the template version, and the feature signature, the same idempotent key is only valid once; Write the snapshot matrix to the snapshot warehouse and generate the evidence index solidified in the audit warehouse; Allow one-time playback correction when the water line has not been crossed, and mark it as a candidate snapshot and leave a trace in the audit warehouse if the retention condition is still not met; Register the burst frame failure as an incomplete snapshot and leave a trace in the audit warehouse.

3. The power distribution network automation control method with fault self-identification according to claim 1, characterized in that S6 Comprises: Within the alignment window, based on the time anchor identifier, associate with the idempotent key, obtain the positioning record from the positioning warehouse, obtain the main candidate topology from the topology warehouse, and collect the current measurement, and participate in the linear power flow residual amount check according to the measurement freshness and apply a conservative coefficient; Under the interlocking table, the locking table, and the device rated gate, when the residual amount meets the safety budget on the main candidate topology, generate an execution list containing the sequence, interlocking, secondary confirmation, and rollback strategy, write it to the control warehouse, issue it through the power distribution automation special communication network, and return the confirmation, and bind the fingerprint value; Synchronously generate the evidence index and write it to the audit warehouse, record the step sequence information and version information; The full link is executed according to the idempotent, sequence, and deduplication strategy, and is managed by the version lock.

4. A power distribution network automation control system with fault self-identification, used to implement the power distribution network automation control method with fault self-identification according to any one of claims 1-3, characterized in that, Comprises: Time base alignment module: used for extracting the double-anchor time base composed of intrinsic anchor and communication anchor, setting up the reorder buffer to promote the waterline and playback correction, generating idempotent key according to device identification, time anchor identification, window serial number, template version, feature signature, and completing alignment and deduplication; Snapshot collection module: used for issuing temporary increase collection when the anchor is formed and the limit is triggered, collecting zero sequence, negative sequence, voltage transient, phasor phase-locked result, and spectrum segment around the trigger window, generating event snapshot matrix and binding idempotent key; Topology self-certification module: used for calculating the equivalent incremental impedance distribution of zero sequence and negative sequence based on snapshots on the candidate topology, and scoring the consistency of each candidate topology according to the direction consistency, obtaining the topology credibility of each candidate topology, and determining the candidate topology whose topology credibility meets the preset condition as the main candidate topology; Same layer feature extraction module: used for implementing same layer gating on the main candidate, extracting zero sequence current and voltage hysteresis area transition, phasor phase-locked amplitude and phase transition, and slope curvature mutation spectrum, and forming fault evidence vector; Positioning determination module: used for searching the maximum consistent sub-path on the main candidate according to the arrival time sequence and amplitude attenuation law, determining the fault type and calculating the section positioning confidence; Execution review and audit module: used for linear power flow residual review on the main candidate, outputting the execution list containing sequence, interlocking, secondary confirmation, and withdrawal, solidifying evidence index and recording execution log.

Images