Work order cooperative treatment method based on power distribution network transient perception and topology aggregation

Abstract

This invention discloses a collaborative work order handling method based on distribution network transient sensing and topology aggregation, relating to the field of distribution network operation and maintenance technology. The method includes the following steps: First, monitoring the telemetry extreme values ​​of the D5000 system backbone nodes to identify physical transient changes and generate disaster-affected grid boundaries; then, capturing work orders according to the boundary orientation; next, anchoring the work orders to the electrical connectivity topology and folding them along the upstream path to generate physical root cause work orders; subsequently, associating them with the overload ledger and low voltage time series to form a risk state index, compressed repair time limit, and dynamic time stamp; finally, executing voice signaling preemptive scheduling and generating a power transfer operation sequence based on the aforementioned results. This method achieves work order entry convergence, merging of duplicate repair requests, dynamic adjustment of handling priorities according to the grid's pressure status, and collaborative linkage between communication scheduling and power transfer decisions. It reduces interference from irrelevant work orders and conflicts from multiple sources, improving the coherence, targeting, and scheduling preparation efficiency of distribution network fault handling.

Description

Technical Field

[0001] This invention relates to the field of distribution network operation and maintenance technology, specifically to a collaborative work order processing method based on distribution network transient sensing and topology aggregation. Background Technology

[0002] The business processes mainly involve urban and industrial park power distribution network fault handling, user power outage reporting, power supply service command, emergency repair dispatch, and restoration notification. They typically involve the 95598 system, power distribution automation system, dispatch control system, GIS system, and on-site emergency repair teams. Most existing business processes start with the collection of repair work orders or multi-source fault information, and then perform topology analysis, equipment operation information, and geographic display to form fault judgment results, and organize dispatch, emergency repair, and power outage / restoration information dissemination.

[0003] Chinese patent document CN10542639A discloses a method for processing emergency repair work orders in a distribution network based on responsibility areas. First, a hierarchical organizational structure is established for the distribution network, dividing responsibility areas by district and county. Repair work orders are received in real time, and power outage assessments are performed based on the fault information in the work orders to determine whether work order merging requirements are met. A high-, medium-, and low-voltage network topology model is established, statically establishing topology connections based on equipment information and dynamically establishing a topology tree based on real-time operating status. Then, upstream and downstream topology analysis is performed on the repair work orders. Based on planned power outage information, outage transformers, user files, and the "substation-feeder-transformer-user set-user" model, the outage equipment, outage users, and outage buildings are analyzed, and the coordinate range is determined using GIS geographic information. Finally, a similarity comparison is performed using a power outage event database to determine whether a work order is merged or a new emergency repair work order is formed, and finally, work orders are dispatched according to responsibility areas.

[0004] The aforementioned technical approach can analyze and dispatch work orders based on existing repair requests and power outage assessments. However, in scenarios involving sudden, rapid, and concurrent repair requests from multiple channels in the distribution network, the following limitations still exist: First, the starting point for handling is after the arrival of the repair work order and the power outage assessment. The generation of the fault object depends on user repair information, event database accumulation, and similarity comparison; there is a time difference between physical disturbance and work order generation. Second, the basis for responsibility area dispatch is still organizational zoning. Although topology analysis is introduced, it only determines the power outage range and merges work orders, making it difficult to intuitively reflect the urgency of handling different nodes during the fault evolution process. Third, when a physical fault has multiple complaints and its status changes rapidly, repeated reception, repeated assessment, and repeated transfer will increase the consumption of the scheduling link.

[0005] Therefore, the core technical challenges faced by this application can be summarized as follows: rapid evolution of distribution network faults, identification of fault objects in scenarios with concurrent work orders from multiple channels, and consistency of merging and scheduling. Summary of the Invention

[0006] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a collaborative work order processing method based on distribution network transient sensing and topology aggregation. The method anchors work orders to the electrical connectivity topology and folds them along the upstream path to generate physical root cause work orders. It then associates these with overload ledgers and low-voltage timing sequences to form a risk state index, compressed repair time limits, and dynamic time stamps. Finally, based on the aforementioned results, it executes voice signaling preemptive scheduling and generates a power transfer operation sequence. This method achieves work order entry convergence, merging of duplicate repair requests, dynamic adjustment of processing priorities according to grid pressure status, and coordinated linkage between communication scheduling and power transfer decision-making, thus solving the technical problems described in the background art.

[0007] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: The work order collaborative processing method based on distribution network transient sensing and topology aggregation includes: monitoring the telemetry extreme values ​​of the D5000 system backbone nodes, identifying physical transient changes and generating disaster-affected grid boundaries, and capturing work orders according to the disaster-affected grid boundaries. The work order is anchored to the electrical connectivity topology and folded along the upstream path to generate a physical root cause work order; the heavy load ledger and low voltage timing of the corresponding node of the physical root cause work order are associated to generate a risk status index, compressed repair time limit and dynamic time tag; voice signaling preemption scheduling is performed based on the risk status index, compressed repair time limit and dynamic time tag, and a transfer operation sequence is generated. In response to the work order completion request, retrieve the waveform of the corresponding node for dynamic time adjustment verification. After the verification is passed, complete the physical root cause work order and generate ledger data.

[0008] Furthermore, physical transient changes are identified based on the three-phase current, three-phase voltage, and switch position signals of the D5000 system backbone nodes, abnormal root nodes are determined, and disaster-affected grid boundaries are generated based on the downstream service range of the abnormal root nodes in the electrical connectivity topology, and corresponding acquisition time windows are formed accordingly.

[0009] Furthermore, when performing targeted crawling for the disaster-affected grid boundary, the bus scheduling server writes the disaster-affected grid boundary and the collection time window into the crawling task header, and wakes up the RPA execution node to crawl work orders located within the disaster-affected grid boundary and whose arrival time falls within the collection time window from the marketing system, supply service system, 95598, 12345 and WeChat entry point respectively.

[0010] Furthermore, address segments, phase segments, and arrival times are extracted from the captured work orders. The extraction results are then correlated with the electrical connectivity relationships of substations, feeders, branch boxes, transformer areas, and user access points in the D5000 system to obtain the electrical landing point of each work order in the electrical connectivity topology.

[0011] Furthermore, starting from the electrical landing point, we trace back along the upstream power supply path to identify candidate convergence nodes that can jointly explain the power outage phenomena of multiple work orders. Work orders that point to the same candidate convergence node and satisfy the requirements of consistent power outage description, consistent phase description, and consistent arrival order are folded into the same physical root cause work order.

[0012] Furthermore, the system reads the overload log, low voltage timing, and communication margin around the candidate convergence node corresponding to the physical root cause work order. Based on the reading results, a risk status index is generated, and a compressed repair time limit and dynamic time tag are formed in combination with the current handling status. The physical root cause work order is then written into the corresponding state machine transition node.

[0013] Furthermore, the number of idle channels, queued tasks, and occupied channel duration of the voice gateway are read and used together with the risk status index, compression repair time limit, and dynamic time stamp to calculate the ranking result of the outbound call objects, and the corresponding physical root cause work orders are allocated voice channels according to the ranking result.

[0014] Furthermore, the distribution network interconnection line ledger is retrieved around the candidate convergence node corresponding to the physical root cause work order, the connection path corresponding to the power failure load is searched, power flow verification and switch operability verification are performed on the candidate connection path, and the transfer operation sequence corresponding to the physical root cause work order is generated based on the verification results.

[0015] Furthermore, upon receiving a work order completion request, the completion transaction of the physical root cause work order is suspended, the actual waveform corresponding to the candidate convergence node is retrieved, and an expected recovery template is generated based on the power transfer operation sequence. Dynamic time warping verification is performed on the actual waveform and the expected recovery template to determine whether the corresponding power supply status has been restored.

[0016] Furthermore, the completed transactions of the physical root cause work order are released only when the dynamic time regularization verification passes, and the abnormal root node, disaster grid boundary, first and last handling time, transfer operation sequence and waveform verification result of the physical root cause work order are input into the streaming derivation process to generate the corresponding ledger data.

[0017] (III) Beneficial Effects This invention provides a collaborative work order processing method based on distribution network transient sensing and topology aggregation, which has the following beneficial effects: By collecting extreme values ​​of telemetry extreme values ​​from the D5000 system backbone nodes, disaster-affected grid boundaries are generated. Work orders are then collected according to these boundaries, directly linking work order entry points to physical disturbances in the distribution network. This prevents work orders from irrelevant areas from entering the handling process, providing initial data with clear sources and convergent scope for subsequent handling. By anchoring work orders to the electrical connectivity topology and folding them along the upstream path to generate physical root cause work orders, duplicate repair requests from multiple channels and with different descriptions are placed under the same object. Scheduling, transfer, and expediting are all based on the same fault source, avoiding object splitting and judgment errors in parallel handling processes. By corresponding the heavy load ledger and low voltage timing of the nodes corresponding to the physical root cause work orders, a risk status index, compressed repair time limits, and dynamic time tags are generated. This ensures that handling is driven by the actual pressure state of the faulty node, avoiding priority distortion caused by fixed time limits for handling.

[0018] By using risk status index, compressed repair time limit, and dynamic time stamp to preemptively schedule voice signaling and generate transfer operation sequence, communication resource allocation and distribution network transfer decision-making are linked together. The front-end expediting order and back-end operation preparation are completed simultaneously, shortening the connection time from judgment to handling.

[0019] By retrieving node waveforms and performing dynamic time warping verification before the work order completion request enters the closed loop, and binding the field response and electrical reset status, it is possible to identify actual recovery and early completion. This ensures that the physical root cause work order closure conditions are consistent with the actual power restoration, improving the reliability of the closed-loop results. Furthermore, by generating ledger data from the aforementioned disaster-affected grid boundaries, physical root cause work orders, risk status indices, power transfer operation sequences, and dynamic time warping verification results, transient sensing, topology aggregation, status compression, preemptive scheduling, and closed-loop accounting are linked into a collaborative chain. This integrates the same technical object throughout the entire process of data collection, processing, and data accumulation, enhancing the overall systematicness and feasibility of the solution. Attached Figure Description

[0020] Figure 1 This is the overall architecture diagram of the power distribution network fault work order collaborative handling system of the present invention; Figure 2 This is a flowchart of the cross-domain work order collection process based on transient triggering and disaster-affected grid boundary constraints of the present invention; Figure 3 This is a schematic diagram of the generation of a work order for electrical anchoring and topology folding within the disaster-affected grid boundary according to the present invention; Figure 4 This is a risk state splicing and dynamic time-limit state machine diagram of the physical root cause work order of the present invention; Figure 5 The present invention takes into account the signaling preemption and forwarding of channel back pressure to generate a dual-track collaborative diagram; Figure 6 This is a flowchart of the work order completion verification and ledger streaming derivation process based on waveform regression confirmation in this invention. Detailed Implementation

[0021] 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.

[0022] Please see Figures 1-6 This invention provides a collaborative processing method for work orders based on distribution network transient perception and topology aggregation, including: Step 1, replacing the timed polling trigger source with the physical transient extreme value of the backbone node, and constraining the cross-domain work order collection range with the disaster-affected grid boundary, so that concurrent access changes from global average net casting to local targeted entry.

[0023] In the initial stage of a distribution network fault, the change is not in the number of work orders, but in the voltage and phase relationship of the backbone nodes. If the power supply system, marketing system, and D5000 system are still polled at a fixed cycle, invalid interface access will be repeatedly triggered when the physical state is stable. It will also miss the handling window within a few seconds when transient disturbances occur, causing the old business system to be overwhelmed by concurrency, and the work order set obtained in subsequent steps will be mixed with irrelevant regional data.

[0024] Therefore, step one is uniformly executed by the bus scheduling server. With the help of the D5000 side OT monitoring probe, GIS mapping component and RPA execution node, the disturbed area is first identified from the underlying physical transient, and then the area is used as the subsequent capture boundary, so that the software layer acquisition action follows the power grid field fluctuation trajectory, and provides cross-domain work order input for the topology folding and aggregation in step two.

[0025] In power distribution network maintenance, 95598 repair requests, 12345 referrals, WeChat messages, and power supply dispatch are not naturally synchronized. The upstream events that truly impact these service entry points simultaneously are usually short-term drops, inrush currents, or three-phase imbalance expansion at a specific electrical node. If the data collection logic only considers the work order text itself, it can only determine the scope after user complaints have already accumulated, placing undue pressure on the database connection pool and RPA session pool simultaneously.

[0026] Therefore, we first define physical transient changes as pre-triggering conditions, and then define the disaster-affected grid boundary as a hard constraint for data acquisition, so that cross-system capture directly follows the disturbance propagation path of the power grid side.

[0027] The bus scheduling server first receives telemetry data unidirectionally from the OT listening probes on the D5000 side, including data from the 35kV nodes, 110kV nodes, and their downstream feeders. The received data only includes three-phase current, three-phase voltage, switch position signals, and timestamps; no control commands are written back to the D5000. The monitored data first enters the edge buffer, and then the bus scheduling server constructs transient trigger strengths for the same backbone node. When transient trigger strength Exceeding the trigger threshold At that time, the system does not immediately capture all work orders, but continues to map abnormal nodes to the GIS power supply zones, forming the disaster-affected grid boundaries. Only the boundary of the disaster-affected grid. Once identified, the RPA execution node is awakened and proceeds according to the disaster-affected grid boundary. The system accesses the service and marketing systems by defining the specific streets, buildings, and distribution areas. Therefore, physical judgment comes first, spatial constraints follow, and concurrent data collection occurs last, avoiding the systemic waste of data collection followed by filtering.

[0028] Transient triggering is not simply a comparison of whether a single phase current exceeds its limit, but rather a simultaneous observation of three types of information: amplitude abrupt change, phase consistency, and duration. The bus scheduling server maintains a short window sequence within the edge buffer, calculates the drop slope and phase deviation between adjacent sampling points for each backbone node, and then compresses the three-phase results into a single transient trigger intensity. : Among them, transient triggering strength The overall disturbance level of a backbone node within a short window, with a value greater than 0; a larger value indicates a higher necessity for triggering; time base. The current determination time is given directly by the timestamp; the time window width... Transient observation interval, with positive values, used to limit calculations to cover only the initial fault segment; three-phase current. Parting At any moment The current measurement value reflects load-side impact and voltage drop; three-phase voltage Parting At any moment The voltage measurement value reflects the voltage drop and recovery trend; current weighting With voltage weight Both are non-negative coefficients, and their sum is 1, used to distinguish the dominance of overloaded events and underpressure events; phase coordination factor The degree of uniform convergence of the three-phase phase deviation, with a value range of [value range missing]. The value increases when all three change in the same direction of the fault.

[0029] when At this time, the bus scheduling server only releases the capture token associated with that node, and does not globally open all capture threads; when At this time, the RPA execution node is in a dormant state, with only the D5000 side listening. In this way, throttling is not passively initiated by interface load, meaning that the entry point is brought under control before the physical transient spreads into a massive number of work orders.

[0030] For example, if a tree obstructs the contact wire in a 10kV feeder substation area, causing a simultaneous drop in three-phase voltage, the D5000 side OT monitoring probe will record a sudden drop in voltage before the on-duty personnel receive continuous calls. Based on this, the bus scheduling server will only release the collection token for the power supply zone where the feeder is located. The repair entry points of other streets in the power supply service system will not be scanned at the same time. Work orders from relevant areas will start to enter in a concentrated manner, while interfaces in unrelated areas will remain inactive.

[0031] From an implementation perspective, edge buffers can employ circular memory queues or shared memory pages with timestamp indexes; phase coordination factor It can be formed by the three-phase phase angle offset, or it can be derived by the zero-sequence component and the negative-sequence component. As long as the discrimination principle of multi-phase in-direction disturbance enhancement triggering and single-point isolated jitter suppression triggering remains unchanged, it falls into the parallel implementation method of this step.

[0032] Once the triggering conditions are met, the system does not directly use the administrative region name as the scope of data collection. Instead, it first traces the abnormal backbone nodes back to the power supply path, and then projects the power supply path onto the GIS power supply partition to form the boundary of the disaster-affected grid. Disaster grid boundary Once formed, it is simultaneously written into the RPA task header, interface query message, and local deduplication key, ensuring consistency in what is collected, where it is collected from, and where it stops within a single boundary object. Disaster-affected grid boundary. The generation can be described as follows: Among them, the boundary of the disaster grid The allowed spatial acquisition range for this transient event is output as a set of continuous GIS grid cells; the anomaly root node. : By transient triggering strength First time crossing the trigger threshold Electrical nodes serve as the starting point for path backtracking; electrical connectivity graph : A set of connectivity relationships between substations, feeders, branch boxes, and distribution areas, used to determine the direction of power transmission; a geographic mapping set. The mapping relationship between electrical nodes and streets, buildings, user addresses, and coordinate grids is used to map electrical paths to geographical locations; extended radius. : The scale of the boundary envelope extending outward from the anomaly root node, with a positive value, used to cover repair points of the same origin near the boundary; mapping operator First, trace back the disturbed path along the electrical connectivity graph, then perform a composite operation to generate a spatial envelope based on the geographic mapping set, and output isolated land parcels that do not have a power supply relationship with the abnormal root node.

[0033] The mapping operator Used to identify abnormal root nodes Mapped to the boundary of the disaster-affected grid Specifically, this includes: along the electrical connectivity diagram Extract the abnormal root node All downstream service nodes Read each service node in the GIS mapping database. The corresponding service polygon Perform a union operation on all service polygons to obtain the initial disaster area; then expand the distance according to the boundary. Perform geometric expansion on the initial disaster-stricken area; finally, adjust according to the grid edge length. Rasterize the expanded region and output the disaster-affected grid boundary. ; Its mathematical form can be supplemented as follows: abnormal root node The electrical node is determined by the first electrical node in step one where the transient trigger strength exceeds the trigger threshold, and is derived from the D5000 telemetry event determination result; electrical connectivity diagram. It consists of the connection relationships between substations, feeders, branch boxes, transformer substations, and user access points, used to represent the upstream and downstream power supply topology; GIS mapping library. Stores the mapping relationship between each electrical node and its serving polygon, geographic coordinates, and administrative address; a set of downstream service nodes. : Obtained by traversing the electrical connectivity graph, representing the set of nodes reachable downstream from the anomalous root node and affected by its power supply; service polygon :node The corresponding power supply service area in the GIS mapping database; boundary extension distance. : The amount of edge expansion used to cover repair points near the same source near the boundary, derived from operation and maintenance configuration or historical service radius settings; grid edge length Used to discretize continuous geographic areas into crawlable and comparable grid cells, derived from the system grid configuration; disaster-affected grid boundaries. The final output set of grid numbers or grid polygons is used for subsequent targeted crawling work orders; When the downstream service areas of multiple abnormal root nodes overlap, they are merged according to the trigger time difference and the upstream level of the execution event, retaining only one final abnormal root node; when a service node lacks a valid service polygon, its parent node's service polygon is read backwards and marked as a degraded mapping in the result; when the output area contains multiple unconnected sub-regions, the disaster-affected mesh boundary is... It allows for the composition of multiple discrete sub-regions.

[0034] RPA execution nodes obtain the boundaries of the disaster-affected grid. Subsequently, asynchronous queries were initiated to both the power supply system and the marketing system. The power supply system's query message carried the street, building, and power outage label; the marketing system's query message carried the household number's district, transformer substation identifier, and time window; and the WeChat entry point filtered the results based on both address keywords and coordinate grid conditions. All query results were first cached locally, and then processed according to the disaster-affected grid boundaries. The user address primary key and time window are used to perform deduplication. The purpose of this is not to complete the work order folding in step two ahead of time, but to isolate noise complaints that are clearly outside the power supply path from the boundary in step one. Within the same feeder, if the downstream service polygon overlap ratio of two triggering nodes exceeds a preset ratio, and the trigger time difference is less than the merging window, then the upstream node with shallower depth and higher transient trigger strength is taken as the abnormal root node. The remaining events are merged into the same event cluster.

[0035] For example, a short-term fluctuation in a busbar section of a 110kV substation is transmitted to two adjacent distribution areas via a main feeder. The bus dispatch server determines the fault based on the abnormal root node. Tracing back to the same power supply path, the GIS mapping component covers the path to two consecutive cells. The RPA execution node only captures the newly added 95598 repair requests and service dispatch orders within these two cells on the same day; non-faulty work orders generated due to unpaid fees in another commercial street in the same city will not be included in this collection set.

[0036] As an alternative, GIS power supply zoning can be carried out using regular grids, road fences, or transformer service polygons; RPA execution nodes can access old pages via browser-driven methods or access new servers via interface proxies; as long as the input is still subject to transient triggering intensity... and the boundary of the disaster grid The dual constraints, and the fact that the output is still a set of cross-domain work orders within the limited area, are consistent with the processing mechanism of this step.

[0037] When using it, first construct the transient trigger strength. Regenerate the disaster-affected grid boundary This method transcribes physical disturbances at the power grid site into start / stop conditions and spatial boundaries for data acquisition, preventing legacy systems from undergoing repeated scanning during stable periods. By breaking down data acquisition token release, GIS path projection, and asynchronous queries into sequentially linked single-chain actions, it enables the abnormal root node... Disaster grid boundary The data collection time window can be passed on to the meta-work order generation stage.

[0038] Step 2: Define the boundaries of the disaster-affected grid. Work orders from the same source, which are dispersed to different business entry points, are first anchored to the same electrical connectivity diagram, and then folded along the upstream power supply path into physical root cause work orders that can be directly transferred.

[0039] Redundancy in power outage scenarios isn't just about an increase in the number of work orders, but also about the repeated appearance of the same fault in different wording. Residents might report a power outage for the entire building, businesses might report a store outage, service representatives might record an anomaly in a transformer area, and the WeChat entry might include the building number and landmark. If aggregated only by address string, the same feeder might be split into multiple groups; if merged only by transformer area name, independent faults in boundary areas might be mistakenly merged. Therefore, this step first projects the address fragment, phase fragment, time fragment, and source fragment from the work order text onto the electrical connectivity diagram, and then, based on the anomaly root node... With the boundary of the disaster grid Perform upstream convergence judgment to transform the textual description into a power supply common source relationship.

[0040] After obtaining the cross-domain work order set from step one, the bus scheduling server first writes it to the anchor buffer instead of the persistent layer. Each work order in the anchor buffer contains the source channel, original text, fetch time, address field, and affected grid boundary. Subsequently, the server calls the work order text parser to segment the address number, building number, village / community name, transformer alias, phase description, and power outage description. Then, it calls the substation-feeder-branch box-transformer area-user access point connectivity relationships from the D5000 topology library to find candidate landing points for each work order. Only when a candidate landing point falls within the disaster-affected grid boundary... And the abnormal root node If the connection is established, the work order will proceed to topology backtracking; otherwise, it will be temporarily stored as an external record.

[0041] The role of electrical anchoring is to transform natural language repair requests into definite locations on the electrical connectivity diagram. The bus scheduling server, running in a Linux environment, calls the work order text parser. First, it segments the address into four levels: administrative region, road name, house number, and building number. Then, it uses a phase dictionary to identify expressions such as single-phase, two-phase, whole-building power outage, and yellow voltage, mapping them to phase clues. Afterward, the server processes the work order text... With candidate nodes Each one is compared individually to form an anchor score. : Among them, the anchoring score : No. Work order text Pointing to candidate nodes The degree of credibility, with a value range of . Work order text Original repair request content from a specific channel, providing address, phase, and description of the power outage; candidate nodes. : Substation nodes, feeder nodes, branch box nodes, transformer area nodes, or user access points in the electrical connectivity diagram; address weights Phase weights Geographic weight Time weight All are non-negative coefficients and satisfy the following conditions: ; Address matching operator Address fragments and candidate nodes The degree of consistency between the ledger addresses, with a value range of [value range missing]. Phase matching operator Phase segments and candidate nodes The consistency of wiring information, with a value range of [value range missing]. Geographic matching operator Work order location and candidate nodes The GIS location relationship, with a value range of 100%. Time matching operator Work order arrival time With the acquisition time window The degree of overlap, with a value range of . .

[0042] The address matching operator is defined as follows: in, For work order text The set of address fragments obtained after address normalization; Candidate nodes A collection of address fragments from the ledger; Let Jaccard similarity be the sum of the similarities between the two sets. This is a marker indicating whether the work order alias and the standard ledger name match exactly; a match is assigned 1, and a non-match is assigned 0. The work order address number; For candidate node ledger address numbers; This is the normalization coefficient for the difference in house numbers; , , The values ​​are weights, all of which are not less than 0, and their sum is 1.

[0043] The phase-matching operator is defined as follows: in, For work order text Extracted phase segments; Candidate nodes The ledger wiring phase; For work order phase category; The phase category is the candidate node; the phase category is divided into single-phase, two-phase, and three-phase.

[0044] The geographic matching operator is defined as follows: in, From work order location to candidate node The minimum distance to the boundary of the service polygon; when the work order location falls within the service polygon, ; This is the geographical distance normalization coefficient.

[0045] The time matching operator is defined as follows: in, This refers to the arrival time of the work order; For the data collection time window; For a moment Time window for data collection The time distance to the nearest boundary; This is the time distance normalization coefficient. , , The initial values ​​are given by the system configuration. The initial weights are set with equal weights and calibrated offline based on historically labeled work order samples. The server is located at the boundary of the disaster-affected grid. Internal and anchored scores The largest candidate node will be used as the electrical landing point for this work order.

[0046] For example, a resident reported through 95598 that the elevator in Building 3 was out of service and the hallway lights were not working, while another business owner wrote on WeChat that all the shops in the Xingfu Road market were dark. The two posts are different, but the address fragments fall within the same service area, the phase fragments both point to overall low-voltage pressure loss, and the capture time is within the data acquisition window. Therefore, the server anchors both work orders to the same distribution node. If another opinion record for a street light that is always on during the day has a similar address but does not meet the requirements for power outage description and time matching, that record will not be included in this round of anchoring results.

[0047] In parallel implementations, the address matching operator Alternatively, segment-by-segment comparison can be performed after word segmentation, or a standard address database can be backfilled before comparison; geographic matching operator The judgment can be made by point-to-surface detection or by overlaying the road section fence with the service polygon of the transformer area, as long as the output is still the boundary of the disaster-affected grid. The single electrical point within maintains the principle of this technology.

[0048] Once each work order has an electrical endpoint, the bus scheduling server does not determine merging based on text similarity. Instead, it traces upstream along the electrical connectivity graph to find an upstream convergence node that can simultaneously explain the power outages of multiple work orders. The server adds the anchored work orders to the backtracking queue and traverses in reverse order of user access point - transformer area - branch box - feeder, checking each candidate convergence node. Calculate the topological convergence potential : Among them, topological convergence potential Candidate convergence nodes For the disaster-affected grid boundary Internal, acquisition time window The aggregation capacity of all anchored work orders within the scope, with a value greater than 0; collection time window. From the trigger time Prefetch duration and duration The starting point was determined jointly. The endpoint is the transient triggering strength. continuously below the release threshold And continue The moment. Work order index set. Simultaneously satisfy the disaster-affected grid boundary Constraints and Acquisition Time Window Constrained work order set; channel weight : No. The strength of evidence regarding the source channel of a work order is positive; topological distance. Work order landing point to candidate convergence node The edge distance in the electrical connectivity graph takes a non-negative integer value; the convergence check factor. : No. The consistency verification results of the work order in terms of power outage description, phase clues, and timing arrival are within the range of [value range missing]. Work order placement : No. Electrical nodes of a work order; candidate convergence nodes The transformer substation, branch box, or feeder node traced back along the upstream path.

[0049] When a candidate convergence node topological convergence potential When the convergence threshold is exceeded, the server selects the candidate convergence node. Generate physical root cause work order and set all pointers to that candidate convergence node. The original work order is folded and linked to the physical root cause work order. Below. Folded physical root cause work order The original work order number, source channel, first arrival time, last arrival time, main address fragment, and phase summary are retained, but only one processing entry, one expedited object, and one subsequent state machine identifier are retained in the flow layer.

[0050] For example, if three buildings along a feeder line report consecutive power outages within a short period, the server first anchors each of the three buildings to two adjacent transformer substation nodes. Then, tracing back upstream, it finds that these two substation nodes converge to the same branch box node, and the outage descriptions and arrival order of each work order are consistent. Therefore, this branch box node is selected as a candidate convergence node. Three batches of work orders were folded into the same physical root cause work order. Anti-folding rule: Folding is allowed only when multiple work order corresponding nodes first converge at the same candidate convergence node along the upstream path, and the power outage timing difference, phase representation, and switching event direction all meet the consistency condition. If any condition is not met, multiple physical root cause work orders are retained.

[0051] As an alternative, topological distance It can be obtained by traversing the adjacency list or by querying the shortest path in the attribute graph database; convergence check factor It can be generated by a rule engine or a sequence matcher. As long as the folding is based on the upstream convergence relationship on the electrical connectivity graph, rather than simple text similarity, it belongs to the parallel expansion of this step. The address normalization preprocessing sub-process includes: first, normalizing road names, community names, and building numbers according to the standard address database; then, mapping historical old transformer area names, popular names, and property abbreviations to standard names using an alias dictionary; finally, extracting the house number, building number, and unit number in a hierarchical manner. Only the address fragments after normalization are entered into the anchor score calculation.

[0052] In practice, by first performing electrical anchoring and then upstream backtracking, the cross-domain work order set output in step one is transformed into a set revolving around the anomaly root node. and the boundary of the disaster grid A convergent, structured fault set reduces out-of-bounds noise from entering subsequent flows. This is achieved through topological convergence potential. Triggering physical root cause work order This step compresses repetitive repair requests from multiple channels into a single processing entry point, enabling step three to focus on physical root cause work orders. Directly load the overloaded ledger and time-compression rules.

[0053] Step 3: Process the physical root cause work order The approach has been changed from being driven by fixed time limits to being driven by risk conditions, so that the countdown for handling is directly influenced by the load pressure on the fault node and the low voltage propagation trajectory.

[0054] In power distribution network emergency repair scenarios, what truly determines the order of fault handling is not the order in which work orders enter the system, but rather whether the faulty node continues to lose voltage, whether the upstream interconnection node is approaching its power transfer limit, and whether the low voltage continues to spread along the downstream user side. If all physical root cause work orders are considered... By uniformly applying fixed repair time limits, the state machine can only recognize operational actions such as dispatched orders, on-site arrivals, and restorations, but it cannot recognize electrical hazards that could spread to adjacent transformer areas if intervention is not initiated. Therefore, this step first uses the candidate convergence node output from step two as the state observation center, and then connects the D5000 heavy load ledger, low voltage event sequence, and tie-line transferable capacity to the same state machine input surface, ultimately making the software layer time limit conform to the grid-side pressure state.

[0055] The bus scheduling server receives physical root cause work orders. Subsequently, instead of immediately issuing a unified set of reminders, they first relied on physical root cause work orders. The bound candidate convergence nodes read load rate records, low voltage alarm records, and daily peak records from the corresponding feeders, branch boxes, transformer substations, and upstream interconnection nodes. The read data first enters a risk splicing buffer, where it is expanded into load sequences, low voltage sequences, and interconnection margin sequences along the same sampling time axis. Subsequently, the state machine engine retains only one processing object: the physical root cause work order. However, by attaching a rolling risk status index and a compressed time limit label to the object being addressed, step four deals not with abstract expedited tasks, but with alarm tasks already bearing physical sorting results. Risk status concatenation is not simply adding together several ledger values, but rather first focusing on the physical root cause work order. A disaster path profile is established for candidate convergence nodes, and then it is identified whether three phenomena occur simultaneously: power loss at faulty nodes, interconnection nodes approaching the upper boundary of the load, and low voltage continuing to spread along the user side. The bus scheduling server, within the risk splicing buffer, processes each physical root cause work order. Forming a risk status index : Among them, the risk status index Physical Root Cause Work Order The current overall electrical emergency level is categorized into the following ranges: Load weight Low voltage weight Contact weight All are non-negative coefficients and satisfy the following conditions: Failure load rate Physical Root Cause Work Order The ratio of the current load of the corresponding candidate convergence node to the upper limit of its rated capacity; upstream load rate. The ratio of the current load of the upstream connection node connected to the candidate convergence node to the upper limit of the rated capacity; Load squeezing operator :right and The common bearing result after piecewise saturation mapping has a value range of [value range missing]. Low voltage sequence During the data collection time window Internally, with physical root cause work orders An ordered set of low-voltage events recorded by user access points along the same path; low-voltage diffusion operator. The result of expanding the sequence after sequentially accumulating the arrival order and coverage area, with a value range of [value range missing]. ; Contact capacity : Remaining path capacity available upstream of candidate convergence nodes; margin compression operator The extent to which the contact margin reduces the handling leeway, with a value range of [value range missing]. .

[0056] The load squeezing operator Defined as: Among them, the failure load rate The ratio of the current load to the rated upper limit of the candidate convergence node corresponding to the physical root cause work order; upstream load rate. The ratio of the current load of the upstream connection node connected to the candidate convergence node to its rated upper limit; load threshold. The initial threshold for entering the load compression judgment is greater than [value missing]. and less than The and It is calculated from the current load data of the D5000 system and the upper limit of the rated load of the nodes.

[0057] The low-voltage diffusion operator Defined as: Among them, low voltage sequence For the collection time window An ordered set of low-voltage events in the downstream power supply path of the candidate convergence node along the inner edge, arranged in topological order; The number of low-voltage events; For the first Each node is within the data collection time window The flag indicating whether a low voltage event has occurred is set to 1 if it has occurred and 0 if it has not. The diffusion attenuation coefficient has a value greater than 0 and not greater than 1; the low voltage sequence It is formed by extracting low voltage alarm records from the D5000 system or low voltage events sent from the terminal.

[0058] The residual compression operator Defined as: Among them, the remaining capacity of communication The upper bound of the carrying capacity of the upstream connection path of the candidate convergence node is the difference between the current carrying capacity and the current carrying capacity; reference margin. This is the standard margin for the communication path, and its value is greater than 0; the communication margin It is obtained by combining the power flow verification results of the distribution network interconnection line ledger with the candidate connection path.

[0059] When the risk status index Written into the physical root cause work order Afterwards, the state machine engine does not immediately change the external state name, but internally first changes the physical root cause work order. The status is marked as observation, imminent, or preemptive. For example, after a feeder branch box goes out of service, multiple downstream buildings continuously report low voltage, while the upstream tie line is responsible for supplying power to nearby transformer areas. The on-duty personnel will not see a simple overdue work order on the dispatch screen, but rather a work order with a preemptive status tag indicating the physical root cause. At this time, the repair team had not yet arrived, but the state machine engine had already placed the object in the priority processing sequence.

[0060] In parallel implementations, low-voltage sequence The data can originate from the D5000 alarm log or from the low-voltage quality record sent by the user-side concentrator; load squeeze operator Piecewise linear mapping or monotonically bounded exponential mapping can be used, as long as the lifting relationship between the fault node and the upstream connecting node when they approach the bearing boundary is still reflected, the mechanism of this step remains unchanged.

[0061] In obtaining the risk status index Then, the state machine engine begins to rewrite the original fixed time limits. It doesn't directly overwrite the basic time limits in the policy ledger, but rather rewrites them on the physical root cause work orders. Internally reserved basic repair time limit Then based on the risk status index Generate compressed repair time limit With dynamic time tags .

[0062] Because of the basic repair time limit It can still serve as a reference boundary in general scenarios, while compressing the repair time limit. It is responsible for transmitting electrical hazards to subsequent communication preemption. Its generation relationship is as follows: Among them, the compression repair time limit Physical Root Cause Work Order Current allowed retention period; basic repair timeframe Standard repair time preset by the system; compression factor Risk Status Index The compression intensity of the handling time is taken as a value greater than 0; Risk Status Index To assess the overall electrical emergency level; State correction function Current status of handling Compressed repair time limit The phased correction takes a value greater than 0 and is specifically a piecewise constant function: maintaining the original compression range in the stage where the repair has been dispatched but not yet arrived, further tightening in the stage where the repair has arrived but not yet isolated, and moderately slowing down in the stage where the repair has been isolated and is waiting to be powered.

[0063] The state correction function Used to characterize the current status of the situation. Compressed repair time limit The phased correction is defined as a piecewise constant function: The current handling status is determined jointly by the dispatch terminal's workflow record and the status returned by the emergency repair terminal; when the current handling status is... When a change occurs, the state machine engine re-values ​​the state correction function. Based on this, the compression and repair time limit will be updated. .

[0064] The state machine engine is based on the compression and repair time limit. Generate dynamic time tags and put the dynamic time tag Write back to physical root cause work order The flow records, team-assigned repair terminals, and subsequent voice signaling queues. The dynamic time stamps here... It is not an independent formula variable, but rather determined by the current moment and the compressed repair time limit. and state correction function The expiration description field consists of the stage results.

[0065] For example, when the emergency repair team receives a repair assignment on a mobile terminal, the same screen simultaneously displays candidate convergence nodes, tie line pressure warnings, and dynamic time stamps. After the work team enters the site and completes the segmented isolation, the terminal sends the status change again, and the state machine engine immediately recalculates the compressed repair time limit. The remaining processing time seen by the dispatcher changes accordingly, instead of using the fixed countdown from the initial dispatch time.

[0066] As an alternative, the state correction function It can be generated by a rules engine or by a state transition table; state change signals can be entered by the dispatcher or transmitted back by a mobile terminal, as long as the input always follows the physical root cause work order. Changes in the treatment phase and continuous output are key to shortening the repair time. With dynamic time tags The linked updates are a type of synchronous implementation.

[0067] When using it, the risk status index By applying the fault load rate, upstream interconnection node stress state, and low voltage propagation trajectory to the same state input surface, this step enables the physical root cause work order to be implemented. The urgency level no longer depends on the time of a single repair request, but on the continuous evolution of electrical hazards. This is achieved by compressing repair timeframes. With dynamic time tags The linkage generation process directly transmits the grid-side pressure status to the dispatching and subsequent communication processes, avoiding the use of the same expediting rhythm for high-risk faults and general faults.

[0068] Step 4: Based on the physical root cause work order Risk Status Index Compressed repair time limit With dynamic time tags First, the physical sorting of the limited voice channels is completed, and then a transfer operation sequence matching the candidate convergence node is generated before the shift arrives.

[0069] When multiple physical root cause work orders Simultaneously, when entering the follow-up phase, the real scarcity is not the alarm information itself, but the dialing channels that the voice gateway can simultaneously occupy, and the number of transfer schemes that the dispatcher can verify in a short time. If all work orders use the same alarm template, the voice channels will be exhausted by general follow-ups, while high-risk faults will be relegated to the back of the queue; if all transfer actions are left to be verified on-site before being performed, the pressure on the tie line, the spread of low voltage downstream, and the heavy load on the backbone node will continue to accumulate. Therefore, the physical dangers given in step three should first be transcribed into a channel contention order, and then this order should be linked to the distribution network interconnection line ledger.

[0070] The bus scheduling server receives physical root cause work orders. Next, first set the risk status index Compressed repair time limit Dynamic time tags Candidate convergence nodes and attached work orders are written to the signaling preparation area. The voice gateway scheduler reads the outgoing call objects from the signaling preparation area and simultaneously reads the number of available channels, queuing depth, and occupied channel duration to form a channel backpressure snapshot. Subsequently, the power flow solver uses the same physical root cause work order... The corresponding candidate convergence nodes are generated in advance by extracting adjacent feeders, tie switches, normally open points and current load distribution from the distribution network interconnection line ledger database, and generating transfer candidate paths in advance.

[0071] Voice signaling sequencing is not simply about comparing which timeout occurs first, but rather comparing which physical root cause work order is being processed. Delayed processing will more quickly lead to supply transfer difficulties, low voltage propagation, or increased upstream overload. Within the signaling preparation area, the voice gateway scheduler generates a signaling preemption value for each pending outbound call. : Among them, signaling preemption value Physical Root Cause Work Order The contention intensity of the current voice channel ranges from [value range missing]. Risk weights Time Limit Weight Back pressure weight State weights All are non-negative coefficients and satisfy the following conditions: ; Risk Status Index That is, the overall electrical hazard level output in step three; time-limit compression mapping. Compressed repair time limit With dynamic time tags The degree of urgency between them; channel back pressure Congestion level is calculated by combining voice gateway queuing depth, occupied channel duration, and the number of currently idle channels; backpressure mapping. Channel back pressure Amplified signaling contention results; processing status Physical Root Cause Work Order Stage at which it is located; state mapping The necessity of expediting the process during the handling phase.

[0072] The time-limited compression mapping Defined as: Among them, the compression repair time limit This is the current allowed repair time output in step three; the base repair time limit. Pre-set standard repair time for the institutional side; dynamic time tags The emergency level label output in step three; For the dynamic time tag The corresponding level value; The preset maximum level value; weight. , All are not less than And sum to The , , It is read directly from the current output of the state machine engine.

[0073] The channel back pressure Defined as: Among them, the number of queued tasks The number of outbound calls currently waiting for the voice gateway; maximum queuing capacity. The maximum number of queued tasks allowed by the voice gateway; the total duration of the occupied channel. This represents the sum of the usage times of all currently occupied voice channels; reference usage time. Reference duration for channel occupancy configured for the system; number of idle channels Number of currently unused voice channels; Total number of channels The total number of voice channels configured for the voice gateway; weights , , All are not less than 0, and their sum is 1; , , and All are obtained directly from the voice gateway status information.

[0074] The back pressure mapping Defined as: Among them, channel back pressure The congestion level obtained from the aforementioned synthesis; when When the value is greater than 1, it is truncated to 1. If the value is not greater than 1, the original value is retained.

[0075] The state mapping Defined as a piecewise constant function: Among them, the status of handling Physical Root Cause Work Order The current transition phase in the state machine; the processing state The status is determined jointly by the dispatch terminal's workflow record and the status returned by the emergency repair terminal; when the status is... When changes occur, the voice gateway scheduler re-retrieves the state mapping. .

[0076] Voice gateway scheduler preempts based on signaling value Outbound call recipients are sorted in descending order. When there are sufficient idle channels, the top recipients directly enter the voice call queue; when there are insufficient idle channels, the top recipients occupy the dialing channel first, and those in the lower queues are subject to signaling preemption. Objects below the current channel threshold will be immediately switched to WeChat article push and dispatch terminal text reminders.

[0077] For example, a work order for the physical root cause of a fault in a feeder branch box. The repairman is not present; risk level is [not specified]. High, and dynamic time stamps The remaining processing time has reached its limit; another work order with similar feedback needs to have its repair time reduced during the same period. It wasn't very urgent, so the voice gateway scheduler first occupied the dialing channel and directly switched to WeChat text and image push.

[0078] In the parallel implementation, the channel back pressure amount It can be synthesized from queue length and average call duration, or from queue length, idle channel ratio, and number of failed redials; state mapping A rule table or a piecewise function can be used, as long as the sorting criterion is always subject to... , , This, along with the channel back pressure, aligns with the principle of this step.

[0079] After the voice gateway scheduler completes the sorting, the bus scheduling server sorts the same batch of physical root cause work orders. The data is fed into the power flow solver. The solver does not wait for the field team to report the isolation results before starting calculations. Instead, it first uses candidate convergence nodes as fault interruption points and searches the distribution network interconnection line ledger for available tie feeders, switchable normally open points, and recoverable downstream load blocks. For each candidate transfer path... The power flow solver generates and supplies adaptation values. : Among them, the transfer of adaptation value Candidate transfer path Physical Root Cause Work Order The degree of executability, with a value range of . Residual weight Step sequence weights Association weight Load weight All are non-negative coefficients and satisfy the following conditions: Path margin Candidate transfer path The remaining carrying capacity after the transfer is executed; surplus mapping Path margin Characterization of safe acceptance capability; step length Complete the candidate transfer path Number of switching actions to be performed; step sequence mapping Step length The result of reducing implementation complexity; association mapping Candidate transfer path With physical root cause work order The degree of topological fit between the candidate convergent nodes; Path load Candidate transfer path Estimated load level after load transfer; load mapping Path load Characterization of subsequent power supply stability.

[0080] Power flow solver from supplying adapter values The switching actions are extracted from the highest candidate power transfer path to form a power transfer operation sequence that includes switch number, execution order, expected power receiving range, and conditions prohibiting parallel operation. This sequence is then linked to the physical root cause work order on the dispatch terminal. Displayed side-by-side. The preemption target is limited to unconnected tasks, including those in queue and ringing states; for low-priority calls that have been connected but not completed, only the eligibility for subsequent redialing is cancelled, and no forced hang-up is performed.

[0081] The surplus mapping Defined as: Among them, path margin For candidate transfer paths For the remaining storage space after the transfer is executed, select the candidate transfer path. The upper limit of the load capacity minus the predicted load after transfer; reference margin The standard margin set for the system is greater than 0; the path margin It is obtained by combining the distribution network interconnection line ledger with the power flow verification results.

[0082] The step sequence mapping Defined as: Wherein, step length To complete the candidate transfer path Number of switching actions required; reference step length The standard number of actions set for the system, with a value greater than 0; the step sequence length It is obtained from statistics generated during the transfer operation sequence generation process.

[0083] The association mapping Defined as: in, For candidate transfer paths A set of recoverable loads; For physical root cause work orders; For the physical root cause work order The corresponding set of power-loss loads; For candidate transfer paths The actual number of target power loss loads that can be recovered; The total number of the power-loss load set; and It is determined by the electrical connectivity topology, the boundary of the disaster-affected grid, and the mapping results of power outage users.

[0084] The load mapping Defined as: Among them, path load For candidate transfer paths Predicted load after load transfer; safe load threshold For candidate transfer paths Allowable upper limit of safe load; normalization coefficient The load overload penalty coefficient, with a value greater than 0; the path load The power flow solver determines the candidate transfer paths. The state after the transfer is calculated.

[0085] For example, a physical root cause work order The associated candidate convergence node is located in the path of the 10kV mid-section. The power flow solver finds two connection paths from the interconnection line ledger, one of which is closer but has a path margin. Small, and then the length of the next step sequence The path is long, but it can withstand the load of downstream buildings. The dispatch terminal will prioritize the switching order of the next path.

[0086] As an alternative approach, the power flow solver can employ either a Newton-style iterative power flow solver or a forward-backward iterative solver; the search for candidate transfer paths can be based on graph traversal or directly generated from the ledger index. This is all contingent on the output being a physical root cause work order. The specific transfer operation sequence, rather than abstract textual prompts, is used to maintain the implementation boundaries of this step. Candidate transfer paths must simultaneously meet four conditions: the network maintains a radial structure after the switch; the predicted load of each line segment within the path does not exceed the allowable value; the voltage of downstream key nodes falls within the allowable range; and all switches involved in the switch are in an operable state and not blocked. Paths that do not meet any of the conditions are directly eliminated.

[0087] When in use, high-risk individuals are given priority access to the scarce voice channel, while general individuals are redirected to the WeChat text and image channel. This is achieved through the transfer of adaptation values. Candidate convergence nodes, path margins, step lengths, and path loads are placed in the same power flow screening chain, presenting an executable switching sequence to the dispatch terminal before the shift arrives. This is achieved by having the voice gateway scheduler and power flow solver work around the same physical root cause work order. Synchronous operation: This step treats communication preemption and distribution network transfer as the same dual-track action, pre-determining the power supply status to be restored and the target load range for the completion verification based on electrical waveforms in step five.

[0088] Step 5: Process the physical root cause work order The completion action was changed from manual click confirmation to waveform restoration confirmation, and the processing results after verification were directly converted into continuously appendable ledger records.

[0089] The most distorted aspect of power distribution network emergency repairs isn't before a work order is dispatched, but rather the instant the work order is about to be closed. After the team completes isolation, power transfer, or power restoration, the "completed" button on the mobile terminal is merely a work action; it doesn't automatically equate to electrical restoration. If the system only closes the physical root cause work order based on this button... The compression repair time limit formed in step three The power transfer operation sequence generated in step four will lose its physical landing point in the last hop. Meanwhile, if all work orders for the day are scanned back in a full table manner each night to generate the "10kV Line Overload Ledger" and "Power Supply Monthly Report," the structured results obtained in the previous steps will be re-disassembled. Therefore, this step first freezes the completion requests, and then retrieves the high-frequency waveforms corresponding to the candidate convergence nodes from the D5000 side, replacing manual declarations with electrical restoration evidence.

[0090] After receiving the completion request from the mobile terminal, the bus scheduling server first checks the physical root cause work order. Bound candidate convergence nodes, transfer operation sequences, and dynamic time stamps A waveform tracing task is generated. This task only reads voltage, current, and switch position signals before and after the claimed completion time from the D5000 waveform acquisition interface, without writing control commands back to the primary equipment. After obtaining the sequence, the waveform verification engine first checks for a closing inrush current segment consistent with the transfer operation sequence, and then checks whether the steady-state load after the inrush current has fallen back to the expected recovery range. Only when the transient and steady-state characteristics close simultaneously will the completion request continue to be passed down; otherwise, the request remains suspended and is written to the exception log.

[0091] The waveform verification engine does not interpret every current rise as a recovery. Instead, it first uses the power transfer operation sequence output in step four as a reference to determine the expected closing section, expected power receiving range, and expected load direction. Then, it matches these expectations with the actual waveform segment by segment. To avoid second-level offsets between the on-site execution time and the scheduling record, the waveform verification engine performs dynamic time warping on the voltage sequence, current sequence, and switch position signal before judgment, forming electrical reset values. : Among them, electrical reset value Physical Root Cause Work Order The actual recovery level of the corresponding fault point near the claimed completion time is within the range of [value range missing]. Distance weight Inrush weight steady-state weights All are non-negative coefficients and satisfy the following conditions: ; Dynamic time warping distance Actual waveform segment With expected recovery template The distance after time axis elastic alignment is greater than or equal to 0; actual waveform segment The sequence of three-phase voltage, current, and switch position signals collected before and after the claimed completion time; the expected recovery template. : Standard recovery trajectory generated based on the transfer operation sequence and candidate convergence nodes; surge mapping The degree of similarity between the inrush current and the load current after closing the circuit breaker, with a value range of [value range missing]. Steady-state mapping After the inrush current subsides, whether the steady-state voltage and steady-state current enter the expected recovery range, with the value range being: ; steady state segment : The recovered steady-state sequence extracted after dynamic time warping and alignment.

[0092] The surge mapping Defined as: Among them, the set of surge characteristics This is the set of features extracted from the actual waveform within a preset inrush window after the closing time. This represents the actual peak inrush current. This refers to the actual moment when the peak flow occurs; This refers to the actual inrush flow decay time. , , These are the reference peak current, reference peak time, and reference decay duration corresponding to the expected recovery template, respectively. , , These are normalization coefficients, all of which take values ​​greater than 0; The range of values ​​is .

[0093] The steady-state mapping Defined as: Among them, steady-state segment The actual waveform segment within the preset steady-state window after the surge subsides; For the steady-state segment The average voltage; For the steady-state segment The average current; , These are the reference average voltage and reference average current within the steady-state window corresponding to the expected recovery template, respectively. , These are normalization coefficients, all of which take values ​​greater than 0; The range of values ​​is .

[0094] When electrical reset value Only when the regression threshold is exceeded will the bus scheduling server release the suspended completed transaction and release the physical root cause work order. The status is rewritten to complete after waveform verification passes; if the electrical reset value If the regression threshold is not exceeded, then the physical root cause work order is issued. The process remains incomplete, and the abnormal moment, waveform summary, and execution team identifier are written to the verification pool. The waveform verification engine prioritizes extracting millisecond-level waveforms from feeder terminal units (FTUs), transformer terminal units (TTUs), station-end fault recording devices, or distribution automation event buffers; D5000 only provides the index, device identifier, and time alignment reference. When the above high-frequency sources are missing, the system degenerates into a conservative verification mode of 'switching timing + post-recovery steady-state range', and inrush current characteristic determination is not performed.

[0095] If the mobile terminal clicks and the dispatcher does not immediately detect the work order's closed status, it first retrieves the high-frequency waveform of the candidate convergence node through the waveform verification engine. If only a short switch appears in the waveform and there is no subsequent stable load recovery, it will not pass through the work order closing interface.

[0096] In the parallel implementation, the template is expected to be restored. It can be generated from historical waveforms of similar types, or from regularized templates; dynamic time warping distance. The three-phase sequence can be calculated as a whole or calculated separately and then synthesized. As long as the principle of determining the regression after transient alignment remains unchanged, it falls within the scope of this step. A two-level template generation mechanism is added: the first priority is to select the central template from the set of historical recovery waveforms that have been verified and are from the same feeder and the same type of switch action; the second priority is to generate a rule template according to the three-stage rule of switch closure - short-term inrush current - steady-state regression when historical samples are insufficient.

[0097] After a completed transaction is released, the bus scheduling server does not first write the data to the regular work order database and then have it fully scanned by the nighttime tasks. Instead, it writes the physical root cause work orders. Electrical reset value Candidate convergence nodes, transfer operation sequences, and first and last processing times are directly fed into the streaming derivation engine.

[0098] The streaming derivation engine uses route number and date as keys to continuously append verified records along the same path, generating ledger derivation values. : Among them, ledger derivative value Physical Root Cause Work Order The validity of data entering the route ledger and monthly report statistics is categorized into value ranges. Regression weights Time-series weights Topological weights All are non-negative coefficients and satisfy the following conditions: Electrical reset value The recovery determination result output by processing method A; Timing Mapping Physical Root Cause Work Order First and last processing time, shortening repair time With dynamic time tags The degree of consistency, with a range of values. Topology mapping : Abnormal root node With the boundary of the disaster grid The accuracy of the attribution in the ledger is within a certain range. .

[0099] The timing mapping Defined as: Among them, physical root cause work orders The start time of processing The moment the first order enters the processing chain; the moment of completion. This refers to the completion time after waveform verification passes; compression and repair time limit. This is the current allowed repair time output in step three; The normalization coefficient for the duration difference is greater than 0; The range of values ​​is .

[0100] The topology mapping Defined as Among them, abnormal root node The trigger node determined in step one; The abnormal root node The set of downstream service nodes in an electrical connectivity topology; For service nodes Service polygon; disaster-affected grid boundary The disaster-stricken area generated in step one; symbol Indicates the area of ​​the region; the The range of values ​​is .

[0101] Streaming Derivative Engine Derivative Values ​​by Ledger Whether the accounting threshold has been reached determines the status of the physical root cause work order. Should it be added to the "10kV Line Overload Ledger" append area, or only added to the anomaly review area? The streaming engine output fields should at least include: line number, feeder number, candidate convergence node number, and anomaly root node. Disaster grid boundary The identifier, first order arrival time, completion time, and compression / repair time limit. Whether the power supply is transferred, the transfer path number, waveform verification result, and electrical reset value. , Entry mark.

[0102] If a 10kV line is handled twice on the same day, the first physical root cause work order will be issued. After waveform verification and supply recovery, the streaming engine will add it to the day's entries; the second physical root cause work order. Therefore, it cannot be used in the monthly report summary.

[0103] As an alternative approach, the streaming derivation engine can be deployed as an in-memory message stream processor or as an event handler with window aggregation; the ledger attribution key can be a combination of line number and date, or a combination of feeder number, transformer area number, and shift date, as long as the entry prerequisite always depends on the electrical reset value. Instead of a manual completion statement, this maintains the technical boundaries of this step.

[0104] When in use, an electrical reset value is introduced before the work order is closed following the completion request. Determine whether this step involves physical root cause work orders. The final closure is based on waveform evidence, enabling the risk prioritization and transfer actions formed in the preceding steps to achieve a physical loop at the end. This is achieved by allowing the ledger to derive value. Directly undertake electrical reset value Abnormal root node With the boundary of the disaster grid This step ensures that the line ledger and monthly statistics only absorb the processing results confirmed by the waveform.

[0105] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0106] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0107] 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 units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units 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 units may be electrical, mechanical, or other forms.

[0108] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0109] 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 technical scope 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.

Claims

1. A work order collaborative processing method based on distribution network transient sensing and topology aggregation, characterized in that: include, Monitor the telemetry extreme values ​​of the D5000 system backbone nodes, identify physical transient changes and generate disaster-affected grid boundaries, and capture work orders according to the disaster-affected grid boundaries. The work order is anchored to the electrical connectivity topology and folded along the upstream path to generate a physical root cause work order; the heavy load ledger and low voltage time sequence of the corresponding node of the physical root cause work order are associated to generate a risk status index, compressed repair time limit and dynamic time tag. Based on the risk status index, compressed repair time limit, and dynamic time stamp, voice signaling preemption scheduling is performed, and a transfer operation sequence is generated; In response to the work order completion request, retrieve the waveform of the corresponding node for dynamic time adjustment verification. After the verification is passed, complete the physical root cause work order and generate ledger data.

2. The work order collaborative processing method according to claim 1, characterized in that: Based on the three-phase current, three-phase voltage, and switch position signals of the D5000 system backbone nodes, physical transient changes are identified, abnormal root nodes are determined, and disaster-affected grid boundaries are generated according to the downstream service range of the abnormal root nodes in the electrical connectivity topology, and corresponding acquisition time windows are formed accordingly.

3. The work order collaborative processing method according to claim 2, characterized in that: When performing targeted crawling for the disaster-affected grid boundary, the bus scheduling server writes the disaster-affected grid boundary and the collection time window into the crawling task header, and wakes up the RPA execution node to crawl work orders located within the disaster-affected grid boundary and whose arrival time falls within the collection time window from the marketing system, supply service system, 95598, 12345 and WeChat entry point.

4. The work order collaborative processing method according to claim 3, characterized in that: The address segment, phase segment, and arrival time of the captured work order are extracted. The extraction results are then matched with the electrical connectivity relationships of substations, feeders, branch boxes, transformer areas, and user access points in the D5000 system to obtain the electrical landing point of each work order in the electrical connectivity topology.

5. The work order collaborative processing method according to claim 4, characterized in that: Starting from the electrical landing point, trace back along the upstream power supply path to identify candidate convergence nodes that can jointly explain the power outage phenomena of multiple work orders. Work orders that point to the same candidate convergence node and satisfy the same power outage description, phase description, and arrival order are folded into the same physical root cause work order.

6. The work order collaborative processing method according to claim 5, characterized in that: The system reads the overload log, low voltage timing, and contact margin around the candidate convergence node corresponding to the physical root cause work order. Based on the reading results, it generates a risk state index and combines it with the current handling status to form a compressed repair time limit and dynamic time tag. The physical root cause work order is then written into the corresponding state machine transition node.

7. The work order collaborative processing method according to claim 6, characterized in that: The system reads the number of idle channels, queued tasks, and occupied channel duration from the voice gateway. It then uses these data, along with the risk status index, compression repair time limit, and dynamic time stamp, to calculate the ranking of outbound call targets. Based on this ranking, the system prioritizes allocating voice channels to the corresponding physical root cause work orders.

8. The work order collaborative processing method according to claim 7, characterized in that: The distribution network interconnection line ledger is retrieved around the candidate convergence node corresponding to the physical root cause work order. The connection path corresponding to the power failure load is searched, and power flow verification and switch operability verification are performed on the candidate connection path. Based on the verification results, the transfer operation sequence corresponding to the physical root cause work order is generated.

9. The work order collaborative processing method according to claim 8, characterized in that: Upon receiving a work order completion request, the completion transaction of the physical root cause work order is suspended, the actual waveform corresponding to the candidate convergence node is retrieved, and an expected recovery template is generated based on the power transfer operation sequence. Dynamic time warping verification is performed on the actual waveform and the expected recovery template to determine whether the corresponding power supply status has been restored.

10. The work order collaborative processing method according to claim 9, characterized in that: The physical root cause work order is released only when the dynamic time regularization check passes. The abnormal root node, disaster grid boundary, first and last handling time, transfer operation sequence and waveform check result of the physical root cause work order are input into the streaming derivation process to generate the corresponding ledger data.

Images